Skip to main content

Full text of "DTIC ADA311740: Symposium Proceedings Polymeric/Inorganic Interfaces II Held in San Francisco, California on 18-20 April 1995. Volume 385."

See other formats


rorm Appro vea 

OMB NO. 0704-0188 

Pufi* raoenng Ouraan nr «m eoMcoon ot rtomnaon a framo to ivtngi 1 hour oar raaoonaa, wem** mo tana tor mMiig RKnocm saararunQ nmg dua wkm 

couMDon et niomMOon. rauang auggaaoona tar rtauan* oua WaanmQton Hoaoouartara SatvwM. Oiracaorata lor aaormaaon Ooaraaono ana Aaootta. 1213 Jottoraon 

Davaragrmay. Su*a 1204. Artngton, VA22202-4302. anatoa>* Who* ot Uanagamont anaSudgat Pww^fl«JuCU»Pn5^(0T0WlM). Waawgtcm. 0C 20503. 

II H R 1: 1 iHii 


Materials Research Society Symposium Proceedings, Vol 385. 
Polymeric/Inorganic Interfaces II 




John B. Ballance (principal investigator) 


Materials Research Society 

Pittsburgh, PA 15237 




U.S. Armv Research Office 

P.O. Box ’122I1 

Research Triangle Park, NC 27709-2211 



ARO 34543.2-MS-CF 


The views, opinions and/or findings contained in this report are those of the authors) and should not be construed as 
an official Department of the Army position, policy or decision, unless so designated by other documentation. 


Approved for public release; distribution unlimited 


13. ABSTRACT (Maximum 200 wonts) 

This volume addresses various aspects of poiymer/inorganic interfaces, such 
as surface preparation and treatment, characterization, and performance of inter¬ 
faces. In addition, it also discusses applications where the interface and its 
properties play a significant role, such as biointerfaces, microelectronics, polymer 
composites, and interpenetrating polymer networks. " 









NSN 7540-01 -280-5500 Standard Form 296 (Rav. 2-8s> 

PrMcntMd by ANSI Std. 239-1S 

Polymer/Inorganic Interfaces II 

Materials Research Society 
Symposium Proceedings Volume 385 

Interfaces II 

Symposium held April 18-20, 1995, San Francisco, California, U.S.A. 


Lawrence T. Drzal 

Michigan State University 
East Lansing r, Michigan, U.S.A. 

Robert L. Opila 

AT&T Bell Laboratories 
Murray HilL New Jersey; U.S.A. 

Nicholas A. Peppas 

Purdue University 
West Lafayette, Indiana, U.S.A. 

Carol Schutte 

National Institute of Standards and Technology 
Gaithersburg, Maryland, U.S.A. 

19960812 023 





Pittsburgh, Pennsylvania 

This work was supported in part by the Office of naval Research under Grant 
Humber N00014-95-1-0529. The United States Government has a royalty-free 
license throughout the world in all copyrightable material contained herein. 

This work was supported in part by the Army Research Office under Grant number 
DAAH04-95-1-0150. The views, opinions, and/or findings contained in this report 
are those of the author(s) and should not be construed as an official Department of 
the Army position, policy, or decision, unless so designated by other 

This work was supported in part by the Automotive Composites Consortium, 
Physical Electronics Incorporated, and Surface/Interface, Inc. 

Single article reprints from this publication are available through 
University Microfilms Inc., 300 north Zeeb Road, Ann Arbor, Michigan 48106 


Copyright 1995 by Materials Research Society. 

AH rights reserved. 

This book has been registered with Copyright Clearance Center, Inc. For further 
information, please contact the Copyright Clearance Center, Salem, Massachusetts. 

Published by: 

Materials Research Society 
9800 McKnight Road 
Pittsburgh, Pennsylvania 15237 
Telephone (412) 367-3003 
Fax (412) 367-4373 

Library of Congress Cataloging in Publication Data 

Polymer/inorganic interfaces II: symposium held April 18-20, 1995, San Francisco, 
California, U.S.A. / editors, Lawrence T. Drzal, Robert L. Opila, Nicholas A. 
Peppas, Carol Schutte 

p. cm.—(Materials Research Society symposium proceedings, 

ISSN 0272-9172 ; v. 385) 

Papers presented at the Symposium on Polymer/Inorganic Interfaces at the 
Spring Meeting of the Materials Research Society. 

Includes bibliographical references and index. 

ISBN 1-55899-288-X (hardcover : alk. paper) 

1. Polymers—Surfaces—Congresses. 2. Polymers—Mechanical properties— 
Congresses. 3. Micromechanics—Congresses. 4. Microstructure—Congresses. 

I. Materials Research Society Meeting (1995 : San Francisco, Calif.) 

II. Symposium on Polymer/Inorganic Interfaces (1995 : San Francisco, Calif.) 

III. Series: Materials Research Society symposium proceedings ; v. 385. 

TA455.P58P683 1995 95-32996 

620.1’920429—dc20 CIP 

Manufactured in the United States of America 


Preface. lx 

Materials Research Society Symposium Proceedings.x 


Photopolymerization Dynamics Studies of Acrolein on Al, Ni, 

and Au Substrates Using Surface Second Harmonic Generation.3 

Fang Chen, Suchitra Subrahmanyan, and Hilary S. Lackritz 

•Bonding and Penetration at Metal/Self-Assembled Organic 

Monolayer Interfaces. 11 

D.R. Jung; A. W. Czanderna, and G.C. tlerdt 

In Situ Study of the Exposure of Polycarbonate to an Argon 


5. Vailon, B. Drevillon, F. Poncin-Epailiard, and J.C. Rostaing 

Nondestructive Characterization of Fiber-Matrix Adhesion in 

Composites by Vibration Damping.33 

Weiqun Gu, Guo-Quan Lu, li. Felix Wu, Stephen L. Kampe, 

P. Ross Lichtenstein, and David W. Dwight 


•Integrating Cell Transplantation and Controlled Drug Delivery 
Technologies to Engineer Liver Tissue .43 

D.J. Mooney, K. Sano, P.M. Raufmann, li. McHamara, 

J.P. Vacanti, and R. Langer 

Natural Poly(hydroxybutyrate-hydroxyvalerate) Polymers as 

Degradable Biomaterials.49 

Cyril Chaput, L'tlocine Yahia, Amine Selmani, and Charles-liilaire Rivard 


Interfacial Water and Adhesion Loss of Polymer Coatings on a 

Siliceous Substrate.57 

T. Hguyen, E. Byrd, D. Alsheh, W. McDonough, and J. Seiler 

On the Microstructure of the Epoxy/Adherend Interphase . 65 

M. Libera, W. Zukas, S. Wentworth, and A. Patel 

•Invited Paper 


Adhesion and Thermal Deformation of Ceramic/Polymer 
Heterostructures .71 

Cynthia Madras, Peter Y. Wong, and loannis N. Miaoulis 


*XPS Study of Buried Metal/Polymer and Polymer/Metal 

P.K. Wu 

Adhesion Between Polyimide Films and Al 2 0 3 Substrate.91 

Y. Nakamura, Y. Suzuki, Y. Watanabe, and S. tlirayama 

Sticking Probability and Step Coverage Studies of Si0 2 and 

Polymerized Siloxane Thin Films Deposited by Plasma Enhanced 

Chemical Vapor Deposition.97 

Jeremy A. Theil 

•Fourier Transform Infrared Spectroscopy of Polymer-Metal 
Interface Reactions . 103 

B.ti. Cumpston, J.P. Lu, B.G. Willis, and K.F. Jensen 

In-Situ XPS Study of the Aluminum Poly(p-phenylenevinylene) 


K. Konstadinidis, P. Papadimitrakopoulos, M. Galvin, and R. Opiia 


Plasma Polymerized Primers for Rubber to Metal Bonding: 

Characterization of the Interphase . 125 

Y.M. Tsai, P.J. Boerio, and Dong K. Kim 

The Role of Coupling Agents in Composite Durability . 131 

K.S. Macturk, C.L. Schutte, C.R. Schultheisz, D.L. tiunston, 
and M.J. Tarlov 

Adhesive Systems Based on Functionalized Block Copolymers .137 

Randall S. Saunders and Michael S. Kent 

Composition-Structure-Properties Relationship and Durability 

of Modified Organosilicate Polymeric Composite .147 

Svetlana V. Tchouppina and Larisa N. Krasil'nikova 


•Environmental Effects on Interface Behavior in Graphite/Epoxy 

Single Fiber Composites . 155 

Linda S. Schadler, Michael J. Koczak, and Maher S. Amer 

•Invited Paper 


Interactions of Phenol Resin Precursor and Calcium Aluminates.167 

Masaki Hasegawa, Dinilprem Pushpalal, Tomonori Takata, 

Naomi Maeda, and Tadashi Kobayashi 

Rheological Properties of Silica Filled Poly (methyl methacrylate).173 

Derek P. Rucker and Stacy G. Bike 

Synthesis and Characterization of Organic/Inorganic 

Interpenetrating Polymer Networks.179 

Barry J. Bauer, Catheryn L. Jackson, and Da-Wei Liu 

Effects of Surface Modification on the Structure of Adsorbed 

Block Copolymer Monolayers. 185 

Rahool S. Pai-Panandiker and John R. Dorgan 


Reactions of Defined Oxidized Carbon Fiber Surfaces with Model 
Compounds and Polyurethane Elastomers... 195 

Charles U. Pittman, Jr., Steven D. Gardner, Guoren He, Lichang Wang, 

Zhihong Wu, C. Singamsetty, Biahua Wu, and Glyn Booth 

'Tailoring the Structure of Polymer Brushes Through Copolymer 
Architecture . 201 

Dilip Gersappe, Michael Fasolka, Rafel Israels, and Anna C. Balazs 

Polymer Brush-Lined Membranes for Flow and Filtration Control . 213 

Edith M. Sevick, Frank A. Bruce, and David R.M. Williams 

Scanning Conduction Microscopy: A Method of Probing Abrasion 
of Insulating Thin Films on Conducting Substrates. 221 

J.T. Dickinson and K.W. Hipps 

The Morphology of Polytetrafluoroethylene (PTFE) Thin Films 

Formed by Pulsed-Laser Deposition .227 

M. Grant Norton, Wenbiao Jiang, and J. Thomas Dickinson 

Electroplating of Fluoropolymers Using ECR Plasma Deposited TiN 
as Interlayer. 233 

A. Weber, A. Dietz, R. Pockelmann, and C.-P. Klages 

Metal Deposition on Laser Modified Teflon Surfaces. 239 

Stefan Latsch, Hiroyuki Hiraoka, and Joachim Bargon 

Removal of Poly(dimethylsiloxane) Contamination from Silicon 

Surfaces with UV/Ozone Treatment. 245 

F.D. Egitto, L.J. Matienzo, J. Spalik, and S.J. Fuerniss 

Author Index. 251 

Subject Index. 253 

'Invited Paper 



This volume contains contributions that were presented in the symposium 
on polymer/inorganic interfaces during the 1995 MRS Spring Meeting in San 

This volume addresses various aspects of polymer/inorganic interfaces, such 
as surface preparation and treatment, characterization, and performance of inter¬ 
faces. In addition, it also discusses applications where the interface and its 
properties play a significant role, such as biointerfaces, microelectronics, polymer 
composites, and interpenetrating polymer networks. Tailoring surface properties 
of substrates, and hence interfaces, involves both the use of self-assembled 
monolayers and tethered copolymers (brushes) for controlling surface energy, 
lubricity, biocompatibility, adhesion, and topography. Polymeric brushes have the 
ability to modify their morphology as the solvent strength or pH changes. Primers 
influence adhesion of polymers to metals either by altering the chemical inter¬ 
actions which form reactive species that hasten chemical bonding or altering acid 
base interactions, or by altering the mechanical performance, such as tensile 
strength and viscoelastic properties of that primer coating. 

Characterization techniques allow for the further understanding of properties 
at interfaces. For example, ellipsometric characterization of ultrathin polymeric 
films yields data on the swelling of these films on exposure to solvent compared 
to that in thicker films. The direct measurement of polymeric and interfacial 
energies allows for further understanding of adhesion and surface science of 
polymeric coatings. The use of transmission electron microscopy and differential 
scanning calorimetry enables the study of the fiber/matrix interfacial morphology 
and may contribute towards understanding of properties of composite interfaces. 
X-ray photoelectron spectroscopy measures the bond formation between metal- 
carbon as well as metal-oxygen-carbon interactions that yield information regard¬ 
ing the chemical characteristics of metal/polymer interfaces and may assist in 
optimization of their adhesion. Furthermore, near-edge x-ray absorption fine 
structure elucidates the orientation of bonds at buried polymer/metal interfaces. 
This nondestructive analysis allows for the measurement of interfacial chemical 
interactions between the polymer and metal, and may be useful in optimizing 
adhesion in these types of systems. 

The contributions of this volume examine various inorganic material/polymer 
interfaces with emphasis on composites, epoxy resins, polyurethanes, poly (methyl 
methacrylate), polysiloxanes and various hydrophilic polymers. They have been 
arranged in seven units that identify major research efforts in specific fields. We 
hope that this volume will be a small, albeit important, addition to the already 
growing literature in this field. 

Finally, the success of the symposium was due in large part to the financial 
support of the Office of naval Research, the Army Research Office, the 
Automotive Composites Consortium, Physical Electronics Incorporated, and 
Surface/Interface, Inc. 

Lawrence Drzal 
Robert L. Opila 
Nicholas A. Peppas 
Carol Schutte 

May 1995 


Materials Research Society Symposium Proceedings 

Volume 352—Materials Issues in Art and Archaeology IV, P.B. Vandiver, 

J.R. Druzik, J.L. Galvan Madrid, I.C. Freestone, G.S. Wheeler, 1995, 
ISBN: 1-55899-252-9 

Volume 353—Scientific Basis for Nuclear Waste Management XVIII, T. Murakami, 

R. C. Ewing, 1995, ISBN: 1-55899-253-7 

Volume 354—Beam-Solid Interactions for Materials Synthesis and 

Characterization, D.E. Luzzi, T.F. Heinz, M. Iwaki, D.C. Jacobson, 
1995, ISBN: 1-55899-255-3 

Volume 355—Evolution of Thin-Film and Surface Structure and Morphology, 

B. G. Demczyk, E.D. Williams, E. Garfunkel, B.M. Clemens, 

J.E. Cuomo, 1995, ISBN: 1-55899-256-1 

Volume 356—Thin Films: Stresses and Mechanical Properties V, S.P. Baker, 

P. Borgesen, P.H. Townsend, C.A. Ross, C.A. Volkert, 1995, 

ISBN: 1-55899-257-X 

Volume 357—Structure and Properties of Interfaces in Ceramics, D.A. Bonnell, 

U. Chowdhiy, M. Riihle, 1995, ISBN: 1-55899-258-8 
Volume 358 —Microcrystalline and Nanocrystalline Semiconductors, R.W. Collins, 

C. C. Tsai, M. Hirose, F. Koch, L. Brus, 1995, ISBN: 1-55899-259-6 
Volume 359—Science and Technology of Fullerene Materials, P. Bernier, 

D. S. Bethune, L.Y. Chiang, T.W. Ebbesen, R.M. Metzger, 

J. W. Mintmire, 1995, ISBN: 1-55899-260-X 

Volume 360—Materials for Smart Systems, E.P. George, S. Takahashi, 

S. Trolier-McKinstry, K. Uchino, M. Wun-Fogle, 1995, 

ISBN: 1-55899-261-8 

Volume 361—Ferroelectric Thin Films IV, S.B. Desu, B.A. Tuttle, R. Ramesh, 

T. Shiosaki, 1995, ISBN: 1-55899-262-6 

Volume 362—Grain-Size and Mechanical Properties—Fundamentals and 

Applications, N.J. Grant, R.W. Armstrong, M.A. Otooni, T.N. Baker, 

K. Ishizaki, 1995, ISBN: 1-55899-263-4 

Volume 363—Chemical Vapor Deposition of Refractory Metals and Ceramics III, 
W.Y. Lee, B.M. Gallois, M.A. Pickering, 1995, ISBN: 1-55899-264-2 
Volume 364—High-Temperature Ordered Intermetallic Alloys VI, J. Horton, 

I. Baker, S. Hanada, R.D. Noebe, D. Schwartz, 1995, 

ISBN: 1-55899-265-0 

Volume 365—Ceramic Matrix Composites—Advanced High-Temperature 

Structural Materials, R.A. Lowden, J.R. Hellmann, M.K. Ferber, 

S.G. DiPietro, K.K. Chawla, 1995, ISBN: 1-55899-266-9 
Volume 366—Dynamics in Small Confining Systems II, J.M. Drake, S.M. Troian, 

J. Klafter, R. Kopelman, 1995, ISBN: 1-55899-267-7 
Volume 367—Fractal Aspects of Materials, F. Family, B. Sapoval, P. Meakin, 

R. Wool, 1995, ISBN: 1-55899-268-5 

Volume 368—Synthesis and Properties of Advanced Catalytic Materials, E. Iglesia, 
P. Lednor, D. Nagaki, L. Thompson, 1995, ISBN: 1-55899-270-7 
Volume 369—Solid State Ionics IV, G-A. Nazri, J-M. Tarascon, M. Schreiber, 1995, 
ISBN: 1-55899-271-5 

Volume 370—Microstructure of Cement Based Systems/Bonding and Interfaces in 
Cementitious Materials, S. Diamond, S. Mindess, F.P. Glasser, 

L. W. Roberts, J.P. Skalny, L.D. Wakeley, 1995, ISBN: 1-55899-272-3 
Volume 371—Advances in Porous Materials, S. Komarneni, D.M. Smith, J.S. Beck, 

1995, ISBN: 1-55899-273-1 

Volume 372—Hollow and Solid Spheres and Microspheres—Science and 

Technology Associated with their Fabrication and Application, 

M. Berg, T. Bernat, D.L. Wilcox, Sr., J.K. Cochran, Jr., D. Kellerman, 
1995, ISBN: 1-55899-274-X 

Volume 373—Microstructure of Irradiated Materials, I.M. Robertson, L.E. Rehn, 

S. J. Zinkle, W.J. Phythian, 1995, ISBN: 1-55899-275-8 

Materials Research Society Symposium Proceedings 

Volume 374—Materials for Optical Limiting, R. Crane, K. Lewis, E.V. Stryland, 

M. Khoshnevisan, 1995, ISBN: 1-55899-276-6 
Volume 375—Applications of Synchrotron Radiation Techniques to Materials 
Science II, L.J. Terminello, N.D. Shinn, Q.E. Ice, K.L. DAmico, 

D. L. Ferry, 1995, ISBN: 1-55899-277-4 

Volume 376—Neutron Scattering in Materials Science II, D.A. Neumann, 

T.P. Russell, B.J. Wuensch, 1995, ISBN: 1-55899-278-2 
Volume 377—Amorphous Silicon Technology—1995, M. Hack, E.A. Schiff, 

M. Powell, A. Matsuda, A. Madan, 1995, ISBN: 1-55899-280-4 
Volume 378—Defect- and Impurity-Engineered Semiconductors and Devices, 

S. Ashok, J. Chevallier, I. Akasaki, N.M. Johnson, B.L. Sopori, 

1995, ISBN: 1-55899-281-2 

Volume 379—Strained Layer Epitaxy—Materials, Processing, and Device 

Applications, J. Bean, E. Fitzgerald, J.Yloyt, K-Y. Cheng, 1995, 

ISBN: 1-55899-282-0 

Volume 380—Materials—Fabrication and Patterning at the Nanoscale, 

C.R.K. Marrian, K. Kash, F. Cerrina, M. Lagally, 1995, 

ISBN: 1-55899-283-9 

Volume 381—Low-Dielectric Constant Materials—Synthesis and Applications in 

Microelectronics, T-M. Lu, S.P. Murarka, T.S. Kuan, C.H. Ting, 1995, 
ISBN: 1-55899-284-7 

Volume 382—Structure and Properties of Multilayered Thin Films, T.D. Nguyen, 

B. M. Lairson, B.M. Clemens, K. Sato, S-C. Shin, 1995, 

ISBN: 1-55899-285-5 

Volume 383—Mechanical Behavior of Diamond and Other Forms of Carbon, 

M.D. Drory, M.S. Donley, D. Bogy, J.E. Field, 1995, 

ISBN: 1-55899-286-3 

Volume 384—Magnetic Ultrathin Films, Multilayers and Surfaces, A. Fert, 

H. Fujimori, G. Guntherodt, B. Heinrich, W.F. Egelhoff, Jr., 

E. E. Marinero, R.L. White, 1995, ISBN: 1-55899-287-1 

Volume 385—Polymer/Inorganic Interfaces II, L. Drzal, N.A. Peppas, R.L. Opila, 

C. Schutte, 1995, ISBN: 1-55899-288-X 

Volume 386—Ultraclean Semiconductor Processing Technology and Surface 

Chemical Cleaning and Passivation, M. Liehr, M. Hirose, M. Heyns, 
H. Parks, 1995, ISBN: 1-55899-289-8 
Volume 387—Rapid Thermal and Integrated Processing IV, J.C. Sturm, 

J.C. Gelpey, S.R.J. Brueck, A. Kermani, J.L. Regolini, 1995, 

ISBN: 1-55899-290-1 

Volume 388—Film Synthesis and Growth Using Energetic Beams, H.A. Atwater, 

J. T. Dickinson, D.H. Lowndes, A. Polman, 1995, 

ISBN: 1-55899-291-X 

Volume 389—Modeling and Simulation of Thin-Film Processing, C.A. Volkert, 

R.J. Kee, D.J. Srolovitz, M.J. Fluss, 1995, ISBN: 1-55899-292-8 
Volume 390—Electronic Packaging Materials Science VIII, R.C. Sundahl, 

K. A. Jackson, K-N. Tu, P. B0rgesen, 1995, ISBN: 1-55899-293-6 
Volume 391—Materials Reliability in Microelectronics V, A.S. Oates, K. Gadepally, 

R. Rosenberg, W.F. Filter, L. Greer, 1995, ISBN: 1-55899-294-4 
Volume 392—Thin Films for Integrated Optics Applications, B.W. Wessels, 

D. M. Walba, 1995, ISBN: 1-55899-295-2 

Volume 393—Materials for Electrochemical Energy Storage and Conversion— 
Batteries, Capacitors and Fuel Cells, D.H. Doughty, B. Vyas, 

J.R. Huff, T. Takamura, 1995, ISBN: 1-55899-296-0 
Volume 394—Polymers in Medicine and Pharmacy, A.G. Mikos, K.W. Leong, 

M.L. Radomsky, J.A. Tamada, M.J. Yaszemski, 1995, 

ISBN: 1-55899-297-9 

Prior Materials Research Society Symposium Proceedings available by contacting Materials Research Society 

Part I 

Experimental Probes of Interfaces 


School of Chemical Engineering, Purdue University, West Lafayette, IN 47907-1283 


The dynamics of the gas phase photopolymerization of acrolein on aluminum (Al), 
nickel (Ni), and gold (Au) substrates were studied in-situ and in real time using surface 
second harmonic generation (SSHG) and monitoring vapor pressure decay. The Al and Ni 
substrates had a significant effect on the apparent rate of polymerization. A dark reaction 
after irradiation occurred in the absence of metal (Al or Ni) substrates, but no dark reaction 
was observed when the metal substrates were present. The significant differences in the 
metal/monomer interactions during p hotop olymerization are indicated by SEMs of the 
polyacrolein, as well as evidence from FTIR-ATR spectra. The SSHG intensities from the 
Au, AI, and polyacrolein surfaces were obtained. 


Solvent-free gas phase photopolymerization on metal substrates shows potential for 
creating defect-free thin films with excellent electrical and optical properties for commercial 
applications such as environmental protective coatings and electrical insulators. The 
photopolymerization-formed thin films are of particular interest in applications such as 
microlithography, 1 photoresists, 2 and semiconductor devices.3 Understanding the 
photopolymerization kinetics is imperative for enhanced control of thin film growth and the 
determination of film mechanical and electrical properties. A second order nonlinear optical 
process, SSHG, has been developed to determine the surface dynamics in-situ . 

Vinyl monomers such as acrolein 4 and methyl methacrylate 5 are reported as 
photoinitiable monomers near the ultraviolet (UV) region. Researchers 6 - 7 have studied the 
effects of physical parameters including monomer pressure, temperature, and light intensity 
on the physical properties of polymer films. However, little is known about the 
polymerization kinetics and their ability to be modified by processing methods. 

The photopolymerization of vinyl monomers is a free radical reaction. 8 Radical 
production and polymerization in the gas phase can be spatially directed by laser-based 
processing techniques and light exposure procedures. Upon exposure to a suitable light 
source, photopolymerization of a monomer vapor in contact with a metal substrate produces a 
thin film with good uniformity and mechanical properties. 9 Free radical photopolymerization 
involves three main steps: initiation, propagation, and termination. In the presence of a metal 
substrate, photoinitiation can occur in the gas phase, on the metal substrate, or by a 
combination of both. The conventional method for studying gas phase reaction kinetics is to 
determine the apparent rate of polymerization by monitoring the time dependence of 
monomer vapor pressure as a function of light intensity. The surface dynamics during the 
photopolymerization are currently unclear because of the lack of surface probing tools with 
high sensitivity and good time resolution. In recent years, SSHG has been proven 10 * 11 in its 
ability to study surface dynamics with submonolayer sensitivity and high surface specificity. 

We employed SSHG to study the dynamics of acrolein polymerization at Al and Au 
substrates with the polymer film growth characterized in-situ and in real time. SSHG is a 
laser-excited coherent optical process with advantages of high spectral and spatial resolution. 
Its good sensitivity and selectivity in the interfacial region facilitates the potential for 
examining the surface reaction dynamics by in-situ mapping of molecular composition of the 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

surface monolayer. By monitoring the surface second order nonlinear susceptibility, surface 
reaction rates can be determined in-situ. 

A medium with inversion symmetry forbids a second order nonlinear optical process. 
SSHG is based on the principle 12 that such a symmetry will be broken at an interface, where 
the nonlinear processes are allowed. This indicates that the SSHG will be highly surface 
specific and without any contributions from the bulk materials. The nonlinear polarization 

induced by the incident laser field is given by: 13 Pp } (2co) = X$ff :E(co)E(©), where E(co) is 

the electric field at frequency co, and x^ff is the effective surface second order nonlinear 
susceptibility. The surface second harmonic intensity, few, in the reflected medium with 
nonlinear polarization P( 2 )(2co) as a source term, is proportional to the square of the 
fundamental light intensity and surface second order nonlinear susceptibility. 

The surface photopolymerization kinetics is determined by monitoring the SHG 
intensity, I, from the surface as the reaction proceeds, i.e., I, is measured simultaneously 
along with the pressure change during the photopolymerization. The unreacted surface 

monomer concentration can be deduced by : 14 1^ Xs?eff = xi? + + 0 “ e )Xp 2) > where 

are the equilibrium second order nonlinear susceptibilities of metal, monomer, 

and polymer surfaces, respectively. 0 is the unreacted monomer coverage on the surface. 

By simultaneously studying the gas phase kinetics and surface dynamics during 
photopolymerization, the photoinitiation sites and reaction rate law can be determined. 
Furthermore, the relation between the polymer film properties and photopolymenzation 
processing has been studied using variou s spec troscopy such as the Fourier transform 
infrared-attenuated total internal reflection (FTIR-ATR), X-ray photoelectron spectroscopy 
(XPS). Scanning electron microscopy (SEM) was used to characterize the polymer 
topography. Polymer glass transition was also determined. 


Prior to use, acrolein (ACR, Aldrich, structure CH 2 =CH-CHO) was passed through an 
inhibitor removal column, then vacuum distilled and degassed via freeze-pump-thaw cycles 5 
times. Au, Al, and Ni substrates (Johnson-Matthey) were polished with micropolish alumina 
(0.05CR) and degreased with acetone. 

Purified monomer was introduced into the reaction cell by thermoequilibnum between 
a Schlenk tube and the reaction cell. Photopolymerization was carried out in a specially 
designed reaction cell (made of aniodized aluminum) which enabled the in-situ measurement 
of SSHG and gas phase pressure in real time. The photopolymerization was then initiated 
using the Hg-Xe arc lamp. A UG-11 filter which transmits 95% light between 250 nm and 
400 nm was placed before the reaction tube to minimize thermal polymerization. The 
transmitted light intensity was measured with a coherent power meter. Pressure m the 
reaction cell was measured with a transducer which had a linear response from 0 to 1.034x10 

N/cm 2 . 


SSHG was used to measure the surface decay of Xs.eff during the reaction. A 
nanosecond pulsed laser (1064 nm) is the source of fundamental light. The filtered laser 
beam passes through a polarizer and a half wave plate. The incident light is focused onto the 
sample at 70° to the surface normal. The SSHG signal generated at the surface was detected 

by a photomultiplier. . . __ ArrD 

XPS was used to determine the metal substrate elemental composition, FIIK-aik 
was empolyed to identify the polymer structure. Differential Scanning Calorimetry 
(5°C/min.) was used to study the thermal stability of polymer films. SEM was utilized to 
characterize the film topography. 



Surface Characterization of Al, Ni, and Au 

To understand the effects of Al, Au, and Ni substrates on the photopolymerization 
dynamics, their surface compositions have been studied using XPS. Fig. 1 shows the XPS 
spectrum of an Al surface. The atomic composition analysis indicates that : On the Al 
surface, the percentage of carbon (C), oxygen (0), and Al are 64.8%, 22.9%, and 12.3% 
respectively; On the Ni surface, the percentage of C, O, and Ni are 50.6%, 33.9%, and 15.6%, 
respectively; On the Au surface, the percentage of C, O, and Au are 54.3%, 11.9%, and 
33.8%, respectively. 

Binding Energy (eV) Wavelength (nm) 

Fig.l. XPS spectrum of an Al substrate Fig.2 UV absorbance spectrum of acrolein 

Pressure Decay During Photopolvmerization 

The rate of apparent polymerization in the gas phase was followed by measuring the 
monomer vapor pressure drop in the reaction cell. Fig. 3 shows the time dependence of 
monomer pressure in the gas phase during the photopolymerization of acrolein. An induction 
period, indicated by a delay in the onset of pressure decay, was observed in all samples. 
When an Al substrate was present, there was a 30 min induction period followed by an 
exponential decay of the monomer pressure, demonstrating first-order photopolymerization 
kinetics. After approximately 1.5 hours the pressure appears to decay linearly with time, 
which might be caused by the deposited layers of polymer on the Al substrate as the reaction 
proceeded. Similar induction period, first order kinetics, and subsequent linear pressure decay 
were observed in the case of Ni substrate. There was no possibility of a dark reaction being 
initiated by the irradiation as shown in Fig. 3 by the corresponding flat portions of the curves. 
A shorter induction period was found without a metal substrate. The reaction rate was found 
to be first-order in monomer concentration after a 15 min induction period. It is also 
interesting to note the existence of the possibility of a dark reaction being initiated by the 
irradiation which is in contrast to the case of Al and Ni substrates reactions. This is partially 
due to the fast quenching effect of the metal substrate on free radicals. 


Time (min.) 

Fig. 3. Time dependence of pressure. The irradiation was turned on and off as indicated. 

Steady State Surface Second Harmonic Generation Intensit 

The surface reaction rate law can be determined by monitoring the time dependence of 
surface monomer concentration decay during photopolymerization. SSHG from a metal 
surface arises from its free-electron-like polarizability. The adsorbates on metals reduce the 
free electron response due to the partial localization of the electrons upon adsorption. This 
means SSHG intensity can differentiate between the metal, metal/polymer, and 
metal/monomer surfaces. The SSHG intensities from Al, Au, and polyacrolein surfaces were 
obtained. Fig. 4 shows that the SSHG intensities from the Al and Au surfaces are larger than 
that from polyacrolein. The difference enables the surface unreacted monomer coverage to be 
monitored by following the SHG intensity decay during photopolymerization. 

The SSHG measurement had no effect on the photopolymerization kinetics indicated 
by the UV-VIS absorbance spectrum of acrolein at the reaction temperature as shown in Fig. 
2. Acrolein has zero absorbance at the fundamental light (1064 nm) and the SSHG (532 nm). 

Fig. 4. SHG intensity from Au, Al and PA/A1, Au Fig.5. Film thickness vs. light intensity 


Light Intensity Dependence of Final Film Thickness 

Photopolymerization of acrolein onto an A1 substrate was carried out for 30 min at 
room temperature (22°C) as a function of light intensity and the dependence of the final film 
thickness was obtained. It was reported 15 that thermopolymerization can be minimized by 
keeping the power/area less than 100 mW/cm 2 . A UG-11 filter kept the light intensity below 
100 mW/cm 2 . The final film thickness was calculated from the film weight gain and surface 
area by assuming a constant density of the polymer film (polyacrolein density is 1320 
kg/m 3 ) 16 . Fig. 5 shows that the polymer film thickness increases with the increase of light 

Structure of Polvacrolein Formed on Al, Ni, and Au 

The IR spectra of polyacrolein formed on Al, Ni, and Au are shown in Fig. 6. In all 
the spectra, polyacrolein chains contained bonds corresponding to ether, acetal, carbonyl, and 
unsaturated hydrocarbonyl groups in various amounts. The hydrocarbon, carbonyl, and -CO- 
H peaks from the spectrum of polyacrolein on Au were observed at different wavenumbers 
than those from polyacrolein formed on Al and Ni. The spectra of polyacrolein formed on Al 
and on Ni exhibited peaks at the same wavenumber with different intensities. It is not clear at 
this stage whether Al, Ni, and Au substrates have catalytic effects on the photo¬ 
polymerization. However, a metal surface does affect the polymer structure. 

Polyacrolein has been reported as a complex, crosslinked polymer, often referred to as 
"disacryl', with cyclic acetal and hemiacetal (tetrahydropyran) structural units as well as free 
aldehyde. 17 The polymer formed on an Al substrate was insoluble in all common solvents 
and became swollen after extensive soaking in a base bath and acidic solution. The glass 
transition temperature of polyacrolein formed on an Al substrate was 118.7“C. It was 
observed that at temperature above 250°C the polymer charred instead of melting also 
indicating a crosslinked polymer structure. 

Fig. 6. IR specta of polyacrolein formed on Al, Ni, and Au substrates 

Microspherical Structure of Polvacrolein Formed on Al. Ni. and Au 

The SEMs of polyacrolein formed on Al, Ni, and Au substrates are shown in Fig. 7. 
An irregular array of ellipsoidal particles with size ranging from 1-5 pm were found from the 
polyacrolein formed on Al and Ni. However, the micrograph of polyacrolein formed on Au 
has uniformly distributed leaf-like particles. Evidence of particle aggregation was observed at 


2000x magnification. In the gas phase photopolymerization, particle size and distribution 
were limited by the collision of radicals with other radicals, with a metal substrate, and with 
the wall of the reaction cell. The deposition of particles on all the surfaces in the reaction cell 
indicated that initiation occurred in the gas phase. However, the insoluble, continuous 
polyacrolein film demonstrated that the polymer growth occurred in the monomer adlayer on 
Al, Ni, and Au substrates as opposed to the film being formed from the deposition of polymer 
particle aggregates in the gas phase. The different types of micrographs observed in 
polyacrolein formed on Au and Al also confirmed the effect of substrate structure on the 
formation of polymer. 

Fig. 7. SEMs of polyacrolein formed on (a) Al (x 2,000), (b) Ni (x 2000), and (c) Au (x 2,000) 


The determination of photopolymerization dynamics is imperative to the enhanced 
control of thin film growth and film properties. A unique method was established here to 
study the photoinitiation and dynamics of photopolymerization on Al, Ni and Au substrates 
by monitoring the gas phase kinetics and surface dynamics in-situ and in real time. An 
induction period was observed in the gas phase of the acrolein photopolymerization. The Al 
and Ni substrate had a significant effect on the apparent rate of the reaction. No dark reaction 
was observed when Al and Ni substrates were present, but a dark reaction occurred in the 
absence of a metal surface. The effects of Al, Ni, and Au substrates on polymer film 
properties were studied using SEM and FTIR-ATR. The significant differences in the 
metal/monomer interaction during photopolymerization were illustrated by the SEM 
micrographs of polyacrolein formed Al and Au substrates, as well as that from FTIR-ATR 
spectra. These preliminary results demonstrated the importance in studying the surface 
photopolymerization dynamics. SSHG was developed here and provides an approach for 
studying the surface dynamics. The SHG intensities from Au, Al, polyacrolein on Al, and 
polyacrolein on Au were obtained. Future work will emphasize characterizing the surface 
dynamics of photopolymerization using SSHG. 


1. A. Barraud, C. Rosilio, A.R. Texier, Thin Solid Films 68, 91 (1980) 

2. H. Ringsdorf, Agrew Chem.Int. Ed. 15, 764 (1976) 

3. B. Maun and H. Kuhn, J. Appl. Phys. 42,4398 (1971) 

4. F.E. Blacet, G.H. Fielding, J.G. Roof. J. Am. Soc. 59,2375 (1935) 

5. H.W. Melville, Proc. Royal. Soc. A. 163,511 (1957) 


6. F.D. Lewis, MJ. Nepras, H.L. Hampsch, Tetrahedron 43 (7), 1635 (1987) 

7. A.N. White, in Polymer surfaces , edited by D.T. Clark and W.J. Feast (John Wiley & 
Sons, New York, 1978), p.155 

8. G. Odian, Principle of Polymerization Chemistry, ed. (John Wiley and Sons, New York, 
1981) p.179 

9. J.Y. Tsao and D.J. Enrlich, Appl. Phys. Lett. 42 (12), 997 (1983) 

10. R.M. Com and D.A. Higgins, Chem.. Rev. 94, 107 (1994) 

11. F.R. Aussenegg, A. Leitner, H. Gold, Appl. Phys. A 60, 97 (1995) 

12. Y.R. Shen, Annu. Rev. Phys. Chem. 40, 327 (1989) 

13. T.F. Heinz, Nonlinear Surface Elactromagnetic Phenomena, edited by H.F. Ponatch and 
G.I. Stegence (Elserier Science Publishers, 1991) 

14. G. Berkovics, Th. Rasing, Y.R. Shen, J. Chem. Phys. 85 (12), 7374 (1986) 

15. X.L. Zhuo, X.Y. Zhu, J.M. White, Surface Science Report 13,73 (1991) 

16. R.F. Fishier, in Acrolein, edited by C.W. Smith (John Wiley & Sons, New York, 1962) 


17. A. Usanmaz and R.D. Dogan, Polymer J. 22 (3), 233 (1990) 




National Renewable Energy Laboratory, 1617 Cole Blvd., Golden, CO 80401 


The purpose of research on metals (M) deposited onto self-assembled monolayers 
(SAMs) is to understand the interactions between the metal and eventually metal oxide 
overlayers on well-ordered organic substrates. Applications of M/SAM and inorganic/SAM 
research results to the understanding of real inorganic/organic interfaces in vacuum and under 
environmental conditions can potentially play a key role in the development of advanced 
devices with stable interfacial properties. The results of selected M/SAM studies to date are 
reviewed, and M/SAM combinations ranked according to reactivity and penetration. Specific 
examples of reactive interfaces (Cu/COOH, Cr/several groups) and nonreactive interfaces 
with penetration (Ag/CH 3 , Ag/COOH) are used to illustrate the extremes. 


This review summarizes part of recent research 1 ' 11 to identify metal/organic functional 
group interactions at metal/self-assembled monolayers (SAM) interfaces, primarily by 
means of X-ray photoelectron spectroscopy (XPS ), 1-4 * 8 ' 11 or to elucidate penetration rates of 
metal overlayers through the SAM using ion scattering spectroscopy (ISS ). 5-7 As illustrated 
in Figure la, SAMs have fulfilled a need for model systems with highly ordered organic 
surfaces of uniform chemical composition for studying interactions at metal or metal 
oxide/organic interfaces . 1-11 Chemical bonding at metal/organic interfaces plays an important 
role in the reliability and durability of many technological devices . 12 Until recently, bonding at 
metal/organic interfaces was typically studied either by surface analysis of organic species 
adsorbed onto single crystal metal surfaces 13 * 14 or by surface analysis of metalized 
polymers . 15 * 16 The chemical environment studied in the first type of experiment bears little 
resemblance to that present at polymer/metal (oxide) interfaces. In contrast, the lack of 
structural and chemical order of polymer surfaces complicates the location and type of 
interactions with metal or oxide overlayers. 

Reactive transition metals (e.g., Cr and Ti) are frequently used to promote adhesion at 
metal/polymer interfaces, and have been the subject of many studies . 17-20 Less reactive 
metals (e.g., Ag and Cu) are frequently used as metalization coatings of polymers for mirror 
applications or as grid lines in photovoltaic cells, which may contain polymer/metal interfaces. 
For studying the stability or reactivity of interfaces, SAMs are ideal because they form highly 
ordered, thermally stable arrays of close-packed aliphatic chains terminated by a surface of 
uniform chemical functionality as originally cited in the work of Allara and Nuzzo 21 " 22 and 
recently reviewed by others (Table l ). 23 * 24 SAMs with a variety of terminating functional 
groups are now available, making studies of the interaction between an organic surface of a 
known functionality and a known metal species possible. Progress in studying alkane thiol 
SAMs terminated with CH 3 , OH, COOH, COOCH 3 , and CN with Ti, Cr, Cu, Ni, Al, K, Na, 
Ag overlayers has been summarized and critically reviewed . 10 


Mat. Res. Soc. Symp. Proc. Vol. 385 0 1995 Materials Research Society 

Objectives of Recent Research 

In support of the main goal of M/SAM 
research (understanding the interactions 
between M and MO overlayers on SAMs), 
the objectives of recent work have been to: 

(a) identify new chemical compounds or 
complexes, and intermediate stages in 
compound formation, if any; (b) measure as a 
function of the ratio of deposited metal atoms 
to SAM organic functional groups (OFGs), 
the three main configurations of metal that 
may be present, as illustrated in Figure lb: S 
(smooth metal that is chemically bonded or 
complexed with the SAM and probably 
spread out on top of the SAM, C (clustered) 
nonbonded metal that is on top of the SAM 
and probably in the form of large clusters, 
and P (penetrating) nonbonded metal that 
significantly penetrates the SAM and whose morphology could take many forms; (c) compare 
quantities S, C, and P for deposition at different rates on surfaces at different temperatures, 
and thereby search for evidence of activated processes, metastable/nonequilibrium states, 
and results that depend on the degree of SAM ordering; (d) obtain images of clusters on the 
SAM (by scanning probe or electron microscopy), and use the distribution of cluster sizes, 
the rate of deposition, and the surface temperature as parameters to model the kinetics of 
cluster nucleation and growth; (e) characterize the degree of ordering and close packing of the 
SAM before and after metallization (are pinholes playing a role?); and (f) measure rates of 
penetration of metal through the SAM. 


Some Properties of Thiol/Au SAMs as 

Summarized by Dubois and Nuzzo 24 

Strength of Au thiolate bond -44 kcal/mol 

Superlattice of S head groups (V3xV3)R30° 

S-S nearest-neighbor spacing 4.99 A 

Tilt angle of methylene chain axis -25-30° 
from surface normal 

Twist angle of the C-C-C plane -52° 

from the plane containing the 
chain axis and the surface normal 

T at which significant chain -380 K 

disordering occurs 

T at which a C22 thiolate monolayer -500 K 

Critical Questions in M/SAM Research 

The lowest energy state of a M/SAM system is that for which all the deposited metal 
forms between the substrate and the SAM, because the surface free energies of metals are 
much higher than those of organics. Thus, two critical questions about M/SAM research are: 
are the configurations S and C of Fig. lb stable at room temperature, and what factors 
determine that stability? 

Chemical interactions (objective (a)) surely play a major role in the formation of a 
M/SAM interface. The scheme depicted in Fig. lb, and our corresponding identification of 
species S with chemical bonding and species C and P with nonbonding, are intended as 
idealizations (e.g., we neglect the role of metal carbide formation which could occur along the 
entire chain). Studies that address objective (a) in combination with objectives (b), (d), or 
(f) can yield information on the correlation of the chemistry of the M/SAM system with the 
way that the metal is distributed in or on the SAM. 

Objectives (c) and (e) are concerned with how a M/SAM system attains the observed 
chemical state and spatial distribution. The process of adlayer nucleation and growth, a 
competition that is kinetically dependent on surface temperature and deposition rate, may be 
considerably more complex than that found for two-dimensional systems if there is diffusion 
into (and not just over) the SAM. The mechanism of penetration may involve different 
possible rate-limiting steps, for example, lateral diffusion leading to penetration at defect 
sites vs. direct vertical diffusion that might be enabled by thermal motion (because of ambient 


Deposited metal 


X—X—X— x- • • *x —X 

Metal of 

OFG of 



A. »A’ •• «A> •• >A Attachment 

. . . v v v , ^ v\ ~ t0 substrate 



He diffraction 

HP=7| M 

/ I 


A IR, Raman, X-ray 
diffraction, NEXAFS 

. Y LEED, TEM electron 
(SJ diffraction, STM 


X = CH 3 CO2CH3 CH2OH 

co 2 h 

Fig. 1 (a) Idealized scheme for using an organized 
molecular assembly to form an interface between a 
metal overiayer and an organic functional group. The 
thickness of the SAM is d, and is the length of the 
molecule at perpendicular attachment. For an MUA 
SAM on gold, the attachment group is a thiolate and 
the methylene chain is 28° from the normal; (b) 
cross-sections of the principally observed results to 
date in which there is (S) strong metal-OFG 
interaction, (C) weak metal-OFG interaction, and 
(P) very weak metal-OFG interaction. The 
designations S, C, and P stand for smooth, clustered, 
and penetrating, respectively. 

Fig. 2. (a) Schematic side-view of the orientation of 
a single, long-chain thiol molecule adsorbed on gold. 
The tilt angle (a) is with respect to the surface 
normal, whereas the twist angle (p) is with respect 
to a plane established by the chain axis and the 
surface normal vectors. The techniques listed to the 
right are those used to determine the structure of 
that portion of the molecule, (b) The surface 
projections of several representative group. (From 
Dubois, L. H. and Nuzzo, R. G. (1992), Ann. Rev. 
Phvs. Chem.. 43, 437; With permission.) 


thermal energy or transient thermal energy imparted by atom impact), chemical reaction with 

and saturation of the OFG, or disordering due to strong chemical effects.. 

The sorting of the various possibilities of chemical states and spatial distributions will 
require, in most cases, experimentation with both structural and spectroscopic probes under 
varying conditions of deposition rate and surface temperature. After this information is 
available, some general conclusions about processes at M/SAM interfaces may emerge that 
can be extended to predict processes at M/polymer interfaces. Detailed process information 
may result only from quantitative comparison with theoretical calculations and simulations of 
these systems. 

Experimental Evidence for the Order and Structure of Alkan e-Thiols on Gold 

The properties of thiol monolayers on gold (Fig. 2) have been the subject of a recent 
review article 24 and are briefly summarized in Table 1. The characteristics of the ordered 
SAM include: (1) the presence of the desired OFG at the vacuum/SAM interface; (2) a low 
density of pinholes in the film; (3) uniform film thickness; (4) uniform average orientation of 
constituent molecules; (5) a low population of gauche bonds in the methylene chains; (6) 
lateral ordering (real-space Bravais lattice or translational symmetry vectors) and interna 
structure of the unit cell of the SAM monolayer; and (7) a domain size of coherent lateral 

order. , , , , 

Of characteristics (1) through (7), (1) has been conveniently probed by measuring the 

contact angles of sessile drops of water and other liquids on the SAM surface; (2) has been 
probed by several techniques including STM, ISS, and cyclic voltammetry and show no 
(significant) evidence of pinholes; (3) has been shown by single-wavelength ellipsometry; 
(4) and (5) have been indicated from Fourier transform reflection absorption infrared 
spectroscopy (FT-RAIRS), surface Raman spectroscopy, and near-edge X-ray absorption 
fine structure (NEXAFS) spectroscopy; (6) has been deduced from diffraction patterns from 
the scattering of electrons, X-rays (XRD, or X-ray diffraction) and low-energy helium atoms 
(HAS, helium atom scattering); and (7) has been determined for the alkane thiolate SAM on 
Au to* be approximately 60 A by transmission electron diffraction (TEM), but results from 
XRD and HAS show 89 and 44 A, respectively. Both STM and AFM studies have yielded 
images exhibiting V3 x V3 lattice patterns consistent with the diffraction results for these 
SAMs. The robust mechanical properties of SAMs, which have been deduced from interfacial 
force microscopy (IFM) studies, are particularly relevant for the potential uses of SAMs for 
surface modification as well as the question of damage or disordering that may occur in 
M/SAM studies. These and other film characteristics (e.g., their wettability by water and 
other liquids) are described further by Dubois and Nuzzo.^4 

Definitions Used In This Article 

AFM atomic force microscopy; BE, binding energy; FT-RAIRS, Fourier transform 
reflection absorption infrared spectroscopy; HAS, helium atom scattering; IFM, interfacial 
force microscopy; ISS, ion scattering spectroscopy; M or MO, metal or metal oxide; M/SAM, 
metal/self-assembled monolayer; NEXAFS, near-edge X-ray absorption find structure; 
NREL, National Renewable Energy Laboratory; ODT, octadecanethiol, HS(CH 2 )nCH 3 ; 
OFG, organic functional group; RAIRS, reflection absorption infrared spectroscopy; SAM, 
self-assembled monolayer; SE, spectroscopic ellipsometry; (S)TEM, (scanning) 
transmission electron microscopy; STM, scanning tunneling microscopy; UV, ultraviolet; XPS, 
X-ray photoelectron spectroscopy; XRD, X-ray diffraction. 



A detailed list and critique of the experimental measurements used to characterize 
SAMs is available, 10 including all those identified in the preceding section. For the typical 
results presented in this article, the use of XPS and ISS will be described. The sample 
preparation methods described below are similar to those reported by other groups. 

SAM Preparation 

The SAMs were formed by immersing clean gold thin film samples in ethanolic solutions 
of HS(CH 2 )nCN, HS(CH 2 )i 7 CH 3 , HS(CH 2 )i 0 COOH, or HS(CH 2 )i 5 COOH, using the 
well-known self-assembly process for alkyl thiols. 21 * 24 ’ 25 Some of the gold films were 
prepared by evaporating 80 nm Au directly onto Ar + ion bombardment-cleaned Si wafer 
coupons, 26 while others were prepared by evaporating a 9 nm Cr adhesion layer followed by a 
200 nm Au layer onto solvent-cleaned Si wafers. The hillock sizes of these two types of gold 
films have been measured by scanning tunneling microscopy (STM) and atomic force 
microscopy (AFM) to be 30-50 nm wide and 10 nm high for the Ar + ion sputtered substrates 
and about 100 nm wide and 10 nm high for the Cr-adhered Au films. In most cases, the Au 
surfaces were placed in a UV-ozone cleaner for 5 min to remove trace organic contaminants 27 
prior to immersion in the thiol. Water contact angles of less than 10° were observed on the 
Au surface only if such a cleaning procedure was used. Following incubation times of 2-4 
days at 295 K, a sample was removed, rinsed in ethanol, dried, and checked by water contact 
angle for the presence of the SAM. Contact angles of 10°-20 o ±2°, 64°±2°, and 112°±2° 
observed in this study for the COOH-, CN- and CH 3 -terminated SAMs, respectively, are in 
good agreement with published results. 25 

Evaporative Deposition of the Metal 

Evaporative deposition of the metal onto the SAMs and subsequent X-ray 
photoelectron spectroscopy (XPS) and ISS analysis were carried out using a Leybold- 
Heraeus LHS-10 surface analysis system. The LHS-10 is equipped with an analysis 
chamber and a sample preparation chamber through which a sample introduction rod can be 
translated without breaking vacuum. The sample preparation chamber is equipped with two 
vacuum feedthroughs that provide power to crucibles used for evaporative deposition. 
Deposition rates are measured with a quartz crystal oscillator using an IC 6000 deposition 
monitor. A negligible amount of sample heating occurs during deposition because the sample 
to source distance is approximately 38 cm. An X-ray source for XPS analysis, an ion gun for 
sputtering and for ISS, and a concentric hemispherical energy analyzer (CHA) are attached to 
the analysis chamber. The CHA is used to determine the energies of scattered ions in ISS. 
A complete description of this apparatus and its use for XPS and ISS have been presented 
previously. 28 

With the sample mounted onto a hot/cold transfer rod (initially at room temperature) 
and inserted into the LHS-10 chamber that is maintained at a base pressure of 2-5x10’ 8 Pa, 
deposition of the metal took place in an attached preparation chamber that permits in situ 
transfer of the sample to and from the XPS (or ISS) analysis chamber at constant 
temperature. The sample was allowed to outgas at room temperature for 1 h prior to heating 
or exposure to X rays. In our standard procedure, the sample was: characterized by XPS at 
295 K, heated or cooled as desired, recharacterized by XPS, exposed to the first incremental 
deposition of metal, and again characterized by XPS. 1 ’ 4 * 6 Repeated 


deposition/characterization cycles were performed in some cases, resulting in total coverages 
of up to 20 A. Typically, a time period of up to 2 h was required for each cycle of XPS 
characterization involving a survey spectrum and narrow scan spectra. In this paper, we refer 
to the amount of metal deposited as a uniform coverage, e.g., 1.0 nm Ag, as though it is 
uniformly covering the surface of the SAM. We are aware that clustering may occur, but 
must provide a frame of reference. In all experiments less than 3 min elapsed between the 
end of metal deposition and the beginning of the XPS or ISS analysis. 

The photoelectron spectra were recorded using a non-monochromatic Mg Ka X-ray 
source operated at 240 W. Narrow scans were taken using constant pass energies of 50 eV 
for C Is, O Is, Au 4/, Cu 2 p, Ag 3 d, and Cr 2 p (for coverages of 2 A or more) and 200 eV for 
Cr 2 p (for coverages below 2 A) and N Is. Survey scans were taken in the constant 
transmission mode with a retardation ratio of 3. For determining XPS peak areas and binding 
energies, the C Is narrow-scan line shapes were fit with a linear background and 85% 
Gaussian-15% Lorentzian (G/L) product functions of appropriate peak widths for each of the 
high resolution scans. 

ISS compositional depth profiles were measured using 1 keV 3 He + ions at a current 
density of 0.2 pA/cm 2 and temperatures from 113 to 293 K. The ion beam was rastered over 
a 2 mm x 2 mm area with the detected signal gated at a 70% aperture to minimize the signal 
from the etch-pit walls. During ISS analysis. He from the ion gun caused the chamber 
pressure to increase from a base pressure below 2x10'^ Torr to a carefully maintained 
pressure of 4x1 O ’ 7 Torr. Each ISS spectrum was acquired over a period of approximately 40 
s. Detection of C with 3 He + is limited by the number of H atoms bonded to the C. Carbon in 
CH 3 or CH 2 is difficult to detect with ISS, but C or CH exposed to the ion beam are 
detectable. Consequently, we did not detect C in our compositional depth profiles. Sulfur 
from the thiol has been observed by ISS in previous work. Accordingly, data were taken over 
an energy range E/Eo of 0.850 to 1.0, to obtain depth profiles with good time resolution and 
good signal-to-noise. 


We present and discuss in this section a summary of the M/OFG combinations studied 
to date and rank them according to their relative reactivity and degree of penetration into the 
SAM . 10 We then present and discuss typical results for the reactivity of Cr/CN and 
Cu/COOH and the rate of penetration of Ag/CH 3 and Ag/COOH. 

Ranking of M/OFG Combinations According to Reactivity and Penetration 

In Table 2, based on the XPS and FT-RAIRS results, we have ranked the chemical 
reactivity of the various M/SAM systems according to the OFG of the SAM. Likewise, in 
Table 3, we have ranked the penetration of the M/SAM systems according to the XPS results 
reported for the different OFGs and different chain lengths of the SAMs. A dependence of the 
degree of penetration on the temperature of the sample, that is, lower penetration for lower 
temperatures, has been found for the systems of Cr/CN, Ni/CN, and Cu/CN (Reference 4; T = 
173 K), Ag/CH 3 (References 5-7; T > 90 K), and Ag/COOH (Reference 32). A dependence 
on the length of the thiolate molecule may also affect the XPS results shown in Table 3, that 
is, shorter molecules show evidence for permitting greater penetration. 

We conclude from these rankings that the degree of penetration increases with 
decreasing reactivity. In order for the adsorbed metal atom or cluster to penetrate, it first 
must encounter the OFG. If there is a reaction there, then penetration is unlikely. If there is 


TABLE 2. Ranking of the Reactivity of M/SAM Systems by Metal and OFG 



Organic Functional Group 

Type of Bonding 


Very high 



Oxide and carbide 


Very high 



Oxide and carbide 



Ti, Cr 

ch 3 


4, 8,9 




Nitride and carbide 




COOCH 3 (preheat-treatment) 





COOCH 3 (postheat-treatment) 






Unidentate metal oxide 



K, Cu 





Ni, Cu 




No reaction 

Ag, Al, Na 

CH 3 , COOH 

- 5,7,30,31,32 

Notes: References are given to FT-RAIRS or XPS results, except for Na/CH 3 . The degree of reactivity is 
based on the authors’ interpretation of the results. 

TABLE 3. XPS Evidence for Penetration in M/SAM Systems (All Au\S-(CH2)n-OFG) at Room 
Temperature (RT) and at Lower Temperatures (LT). 

Metal Shift 

C Is attenuation 



















Very low 










Very low 





ch 2 oh 





Very low 










Very low 




Ti, Cr 



















































CH 2 OH 














Slightly low 



















Quite low 







CH 3 




























Notes: The metal shift refers to whether a core level has been measured at low coverages to exhibit a high 
binding energy or low binding energy shift, HBE or LBE, respectively, with respect to the bulk metal core 
level. “C Is attenuation” refers to the attenuation of the C Is level compared to that expected for a uniform 
metal overlayer. References to the sources of these XPS results are given. Question marks (?) indicate that 
the information has not been reported. The degree of penetration is based on the authors’ interpretation of the 
available XPS results, in which “very high” means the overlayer essentially resides in the same plane of the 
OFG and “high” means the overlayer penetrates to the Au/SAM interface. 


no reaction, then penetration can proceed in some cases. Rates of penetration have only been 
measured in the cases of Ag 5 ~ 7 and Na 30 * 31 on CH 3 -terminated SAMs and for Ag on COOH- 
SAMs . 32 Table 2 shows that the reactivity is highest for the oxygen-containing OFGs, 
(except for Ag/COOH) intermediate for the CN OFG, and lowest for the CH 3 OFG. For the 
metals, Ti is the most reactive, followed by Cr, Al, K, Ni, Cu, Na, and Ag, although the data 
for Al, K, Ni, and Na are scant. The decreasing reactivity within the transition metal series, 
Ti, Ni* and Cu, and the low reactivity of the noble metals, Cu and Ag, is in the same order as 
the filling of the d orbitals. The relative reactivity of trivalent Al and the monovalent alkali 
metals (Na, K) is not so clear at this point because so few of these M/SAM systems have 
been explored. Evaporating oxygen-free Al films has not been successful, as yet . 33 The 
M/SAM studies involving Al evaporation may have been affected by rapid oxidation of the 
deposit 33 which would have had a severe effect on the chemical interactions at the interface. 

There is some evidence that the reactivity of the SAM with the incoming M is also 
dependent on the SAM temperature . 4 ’ 11 Activation of M/SAM reactions may be because of 
electronic, steric, or structural factors and is not unexpected. Further temperature-dependent 
studies are warranted in order to learn more about activation energies in M/SAM reactions. 

Reactive System: C I s XPS Line Shapes for Cr/CN 

The C Is line shapes are shown in Figs. 3(a) and 3(b) for 0.6 and 6 A Cr coverages, 
respectively, for deposition onto a series of different Au/S(CH 2 )nCN samples held at 173, 
295, 323, and 373 K, along with the C Is line shape for a bare SAM for comparison. The most 
striking difference in Fig. 3 is the steady increase from 295 to 373 K in the -283 eV low 
binding energy (LBE) shoulder [Fig. 3(b)] that we assign to a Cr carbide species. A shift in 
the high binding energy (HBE) peak also is observed. 

The HBE C Is peak of the CN-terminated SAMs is found at 286.85±0.1 eV for zero Cr 
coverage and is attributed to the end-most, nitrile carbon and its nearest (a) neighbor along 
the chain. At 6 A Cr coverage the HBE peak shifts to 286.45±0.1 eV for lower temperatures 
(173-295 K) and to 286.25±0.15 eV for higher temperatures (323-373 K). The larger shift of 
this HBE C Is component (about 0.2 eV) shows there is a slight increase in Cr(CN) 
reactivity in the higher temperature range. Similar results are found at lower Cr coverages, 
as shown in Fig. 3(a). The shift with Cr deposition probably results from a net charge 
transfer from Cr to CN molecular orbitals that increases the charge in the valence levels of 
the two endmost carbons. More detail about curve fitting of the C Is line shapes, estimates of 
the chromium carbide stoichiometry (Cr 3 C 2 ), and control experiments for assessing the 
influence of X-ray damage is available . 11 

Reactive System: Binding Energies of the X PS Cr 2p Levels 

In Ref. 4, we discussed the correlation of metal penetration into or possibly through the 
SAM with the position of the Cr 2 p BEs. The Cr 2 p positions are shown in Fig. 4. They are 
shifted to HBE for Cr deposited at 295 and 173 K onto the CN SAM [Fig. 4(b)], consistent 
with little or no penetration. They are shifted to LBE for Cr deposited at 295 K onto Au [Fig. 
4(a)], consistent with thin Cr deposits on a conducting substrate. They are not shifted, but 
are approximately at the bulk position, for Cr deposited at 323 and 373 K [Fig. 4(b)]. These 
results show smaller, negligible HBE shifts for 323 and 373 K compared to 295 K. In terms of 
the core hole lifetime/final state effect supported by Wertheim for the shifts of metal clusters 
on insulating substrates , 34 our result is consistent with some degree of penetration at 323 
and 373 K. 


288 285 282 

Binding Energy (eV) 

Fig. 3. C Is XPS line shapes for the bare 
Au/S(CH 2 )l lCN SAM at 295 K (top of a and b), 
and for (a) 0.6 A Cr and (b) 6 A Cr deposited on a 
series of Au/S(CH 2 )l lCN SAMs at temperatures of 
173, 295, 323, and 373 K, as labeled. The plots at 6 
A coverage have been increased in intensity by a 
factor of 1.5 for clarity. The position of the solid 
vertical line is 284.8 eV. 






| foM 










590 580 570 

Binding Energy (eV) 

Fig. 4. Cr 2 p XPS line shapes (a) for 0.5 and 50 A 
Cr on Au at 295 K, and (b) for 0.6 A Cr deposited 
onto a series of Au/S(CH 2 )nCN SAMs at 173, 295, 
323, and 373 K. as labeled. The solid vertical lines 
at 583.45 and 574.25 eV mark the peak positions for 
bulk Cr 2 p. 


Reactive System: O Is XPS Line Shapes f or Cu/COOH 

Czanderna et al. deposited Cu on the HS(CH 2 )ioCOOH SAM and performed XPS as a 
function of coverage. 1 Changes were reported in the O Is line shape (Fig. 5) that showed a 
constant component corresponding to the C=0 oxygen, as well as a component due to the C- 
oxygen that shifted to lower BE with increasing Cu coverage. In addition, they observed a 
strong HBE shoulder on the low-coverage Cu 2p 3/2 peak (Fig. 6), no Cu 2p shake-up 
satellites due to Cu(II), and a Cu Auger peak consistent with a Cu(I) oxidation state. All of 
these data together were cited in support of the formation of a unidentate Cu(I)-0 complex at 
the single-bonded O atom of the SAM. The leveling off of growth of the HBE Cu 2p 
component was interpreted as evidence that the Cu-COOH interaction was complete at a 
coverage of 0.2 to 0.6 nm Cu. This was the first M/SAM work published by the NREL group, 
although work with Cu on CH 3 -, OH-, COOH-, and CN-terminated SAMs was begun in 
1988 and yielded preliminary results that were reported in 1989. Further detailed discussion 
of these results is available. 1 

Penetrating Systems: TSS CDP of Ag/Ofr and Ag/COOH from 113 K to 293JC 

ISS compositional depth profiles were obtained from 113 to 293 K to study the 
penetration of Ag into ODT, MUA, and MDA, as a function of temperature. For all of these 
experiments, 1.0 nm Ag was deposited at a rate of approximately 0.01 nm/s onto the SAM 
prior to depth profiling. The elapsed time after starting metal deposition and completion of the 
first ISS spectrum was approximately 300 s (100 s, deposition; 160 s, sample transfer; and 40 
s, first ISS spectrum). Thus, 5 min elapsed for time-dependent processes to proceed before 
our “initial” spectrum was taken during depth profiling. An ion beam current density of 0.2 
pA/cm 2 was used for all ISS depth profiles. XPS and ISS measurements were carried out on 
several samples at 193 K and verified that no detectable oxygen signal could be obtained from 
the ODT samples. These data indicate that no ice formed on the samples during low 

temperature experiments. .... ... 

The plots in Fig. 7 show the fraction of Ag (Ag intensity divided by Ag intensity plus 
Au intensity) in the ISS depth profiles as a function of erosion time at temperatures from 113 
to 293 K. The data are consistent with an increasing rate of Ag penetration to the ODT/Au 
interface at progressively higher temperatures, which results in greater Au intensities that 
are detected concomitantly with Ag when the Ag is at the ODT/Au interface. The time to 
erode through a “bare” ODT SAM to the ODT/Au interface is about 1000 s at 0.2 pA/cm 2 . 
The plots in Fig. 7 are consistent with a model of Ag residing on the methyl end group of ODT 
at liquid nitrogen temperatures and at the ODT/Au interface at 293 K. 2 The data also 
elucidate the temperature regime over which the rate of penetration increases rapidly. The 
residual Ag signal for erosion times greater than about 2000 s, which corresponds to the 
location of the ODT/Au interface, indicates the presence of Ag clusters at the ODT/Au 

interface. . 

The integrated initial ISS Ag peak intensity is shown as a function of temperature in 
Fig. 8 and provides a measure of the maximum ISS Ag signal obtainable with our apparatus 
for 1.0 nm Ag before extensive penetration into the SAM has occurred. In contrast to Fig. 7, 
the initial ISS Ag signal decreases above and below 153 K. The reproducibility of these data 
was carefully checked for coverages of 1.0 nm Ag. Two additional ISS compositional depth 
profiles were taken for a coverage of 5.0 nm Ag at 113 K to clarify the unexpected results 
below 153 K. The initial ISS signal intensity at 113 K and 5.0 nm Ag coverage is 
approximately the same magnitude as the data taken at 153 K with 1.0 nm Ag coverage. We 
have not determined if the Ag peak intensity is a maximum at 153 K (1.0 nm Ag) and 113 K 
(5.0 nm Ag). 


Binding energy (eV) 

Fig. 5. Curve-resolved O Is photoemission peaks for 
the HS(CH2 )io COOH SAM on gold before (a) and 
after (b) deposition of 0.05 nm of copper. 

0 2000 4000 6000 8000 

Erosion time (s) 

Fig. 7. Fraction of Ag signal (see the text) as a 
function of erosion time at 0.2 jiA/cm^ for 1.0 nm 
Ag on ODT/Au at temperatures of 113, 153, 193, 
233, and 293 K. 

Binding energy (eV) Temperature (K) 

Fig. 6. Copper photoemission peaks for 0.05 nm of 
copper on (a) 80 nm of gold and (b) the 
HS(CH 2 )iqCOOH SAM on gold. 

Fig. 8. Initial ISS peak intensities for 1.0 nm Ag on 
MHA and ODT after deposition at temperatures 
from 113 K. 


The data for Ag/MUA and Ag/MHA in Fig. 8 also show a larger initial ISS signal is 
obtained. Data taken that is similar to that in Fig. 7 show the rate of Ag penetration is 
slower for MU A than for ODT, and even slower for MHA. For example, the ISS peak area 
was measured at four different positions on the same sample for 1.0 nm of Ag on MUA and 
MHA as a function of time after deposition at 295 K after waiting for about 5, 15, 35, and 65 
min after deposition; a fifth ISS Ag peak intensity was measured after 900 min for both MUA 
and MHA. The data show that after deposition Ag remains on the surface of MUA for at 
least 5 min. and some Ag is on the MUA surface for more than 1 h. For MHA, Ag remains on 
the surface for more than 1 h (i.e., the Ag ISS intensity is unchanged), and the ISS intensity 
is only reduced to 68 % of its initial value after 15 h, which shows that complete penetration to 
the SAM/Au interface is greatly retarded by the longer chain (Cig vs Cn) MHA.. By 
comparison, 1.0 nm of Ag penetrates ODT in less than 5 min at this temperature (Fig. 7). Ag 
on MUA is an intermediate case in which Ag remains on the surface for 5 to 15 min but then 
penetrates into the MUA, but 15% of the deposit is still on the surface after 1 h. 

The time dependence of the slower Ag penetration through MUA and MHA compared 
with that for ODT at 295 K provides some indication of the mechanism of metal penetration 
through SAMs. First, weak interactions between Ag and the COOH group may retard the 
onset of Ag transport through MUA and MHA. We did not detect evidence for formation of a 
Ag unidentate complex with O, even though Ag-0 interactions of 18 kcal/mol to 69 kcal/mol 
are known to exist . 39 Secondly, we speculate that hydrogen bonding between surface COOH 
groups results in a more tightly packed SAM and makes penetration of Ag through MUA and 
MHA more difficult. The CH 3 group on ODT, by comparison, is in free rotation at 295 K and 
does not retard transport of Ag to the ODT/Au interface. The increased rate of penetration of 
Ag through MUA relative to that for MHA is attributed to an enhanced defect density in the 
less tightly packed shorter chain alkanethiol. A third possibility is that the presence of the 
COOH group on MUA and MHA reduces the number of pinhole defects in the SAM, so that 
fewer transport pathways are available in these films, again with the MHA more defect free 
than for MUA. 

In Fig. 8 , the initial ISS Ag peak intensities from CDPs of 1.0 nm Ag on MHA and ODT 
are compared for temperatures from 113 K to 295 K. The decrease in intensity below 150 K 
for ODT was discussed in our previous paper . 6 As is seen in Fig. 8 , a similar decrease 
occurs for MHA, though below 200 K, and we think the reasons for the decrease are the same 
as for the Ag on ODT . 6 The initial ISS Ag peak at 295 K for 1.0 nm Ag on MHA is more 
intense than the most intense initial ISS peak observed for 1.0 nm Ag on ODT. This result 
indicates that the Ag initially remains longer on the COOH surface of MHA, the ISS may be a 
maximum for any given temperature. Differences in Ag coverage on ODT and MHA probably 
result from differences in wetting behavior (and thus clustering) as well as penetration of Ag 
through the two SAMs. Even though SAMs are model systems for studying interactions at 
metal/organic interfaces, we emphasize that the systems being studied are complex when 
metal penetration occurs. Factors dictating the specific structure of metal overlayers on 
SAMs might include metal nucleation behavior, SAM defect density and domain size, the 
chemical functionality of the SAM, and the available free volume between alkanethiol chains. 

Important Concerns about M/SAM Studie_s 

Several important and crucial questions exist for M/SAM studies. These include: (a) 
are contaminants present in or on the film, (b) what kind of defects are in the SAM and at 
what density, (c) does the degree of penetration (Table 3) increase with increasing SAM 
defect density, (d) does the SAM disorder significantly even for low metal overlayer 
coverages, and (e) what problems are caused by X-ray and e-beam exposures? None of the 
questions (a) - (e) seem unmanageable based on a recent review by Jung and Czanderna . 10 



In the research to date, the interactions of several M/SAM systems have been 
identified , along with qualitative relationships among the effects of chemical reactivity, 
temperature, and penetration of a metal/organic interface. With further research, the task 
remains to understand quantitatively how M/SAM interfaces are formed. This will require 
more complete characterization of the reactants, the structures, and the kinetics of the 
reactants and the structures of M/SAM systems. 

The initial research on M/SAM systems has served as a test of their use as model 
systems, and of whether M/SAM research will have benefits in applied areas, for example, 
real M/polymer interfaces. The results described above are clearly more readily interpreted 
and more detailed than most of their counterparts in the metal/polymer interface literature. 
Based on the presence of similar reactants at a M/polymer interface, such as OFGs and 
metal atoms, this research approach succeeds in identifying factors that affect the initial bond 
between metal and polymer as the interface is formed in vacuum. 

From the initial results, a continued broad effort is warranted for studying a matrix of 10 
to 15 metals and metal oxides with about 10 different OFGs to evaluate the full potential of 
the approach and of this subfield of surface science. This should lead to the identification of 
submatrices targeted at specific classes of materials and applications. 

Future characterization needs include obtaining (1) detailed, molecular orbital 
descriptions of new metal-organic bonds in a unique environment (because of the difference 
between bringing individual molecules to a metal surface compared with bringing individual 
metal atoms to an organic surface), (2) diffusion processes and rates for metals on organic 
surfaces, and (3) penetration processes and rates for metals into organic surfaces. These 
advances will be based on further work with valence band probes, lateral imaging probes, and 
probes that can measure rates of penetration, as discussed above. Control of defects and a 
capability for the characterization of defects will be necessary to understand the relative role 
of intrinsic vs. extrinsic effects in chemical and physical processes in M/SAM systems. 

Finally, the results of M/SAM studies may be applied to the understanding of 
M/polymer interfaces, and for this reason, M/SAM interfaces are often described as “model” 
M/polymer interfaces. Studies of metal oxides, semiconductors, and other inorganics 
deposited onto SAM substrates should provide results than can be applied to the 
understanding of other types of inorganic/organic interfaces. Further relevance to 
inorganic/organic interfaces actually used in practice and to interfacial reaction mechanisms 
under ambient conditions can be gained by exposure of inorganic/SAM interfaces to different 
atmospheres, temperature extremes, radiation, and chemical reagents. Inorganic/organic 
interfaces with well-defined and stable properties are needed in advanced devices. 36 ' 38 
Research to identify and modify reactions at inorganic/SAM interfaces will provide the 
fundamental understanding to fill that need. 


The authors are pleased to acknowledge support by the U.S. Department of Energy 
under Contract DE-AC36-83CH10093. We are especially grateful for the helpful discussions 
with D. L. Allara, P. Zhang, R.W. Collins, and D.E. King and to D.L. Allara and G. 
Whitesides for providing most of the alkane thiols we used in our experimental work. We 
thank D.L. Allara, P. Zhang, R.L. Opila, K. Konstadinidis, M.D. Porter, M. Grunze, H.G. 
Rubahn, L.H. Dubois, R.G. Nuzzo, M.J. Tarlov, and P. Zhang for permission to use figures 
from their work in ref. 10, and R.W. Collins, G. Scoles, A.J. Bard, J.E. Houston, K.H. Gray, 
R.W. Linton, S.M. Lindsay, G.N. Robinson, A.N. Parikh, K. Edinger, and F. Eisert for 
providing copies of their publications. 



1. A.W. Czanderna, D.E. King, and D. Spaulding, J. Vac Sci. Technol. A 9, 2607 (1991). 

2. D.R. Jung, D.E. King, and A.W. Czanderna, Appl. Surf. Sci. 70/71, 127 (1993). 

3. D.R. Jung, D.E. King,, and A.W. Czanderna, J. Vac. Sci. Technol. A 11, 2382 (1993). 

4. D.R. Jung and A.W. Czanderna, Mater. Res. Soc. Symp. Proc. 304, 131 (1993). 

5. G. Herdt and A.W. Czanderna, Surf. Sci. Lett. 297, L109 (1993). 

6 . G. Herdt and A.W. Czanderna, J. Vac. Sci. Technol. A 12, 2410 (1994). 

7. M.J. Tarlov, Langmuir 8, 80 (1992). 

8 . P. Zhang, Ph.D. thesis, Department of Materials Science and Engineering, Pennsylvania 
State University, 1993. 

9. R.L. Opila, K. Konstadinidis, D.L. Allara, and P. Zhang (private communication). 

10. D.R. Jung and A.W. Czanderna, Crit. Rev. Solid State Mater. Sci. 19, 1 (1994). 

11. D.R. Jung and A.W. Czanderna, J. Vac. Sci. Technol. A 12, 2402 (1994). 

12. E. Sacher, J.J. Pireaux, and S.P. Kowalczyk, (1990), in Metallization of Polymers, ACS 
Symposium Series, series editor, M. J. Comstock, (American Chemical Society, 
Washington, DC) and references. 

13. F.P. Netzer and M.G. Ransey, Crit. Rev. Solid State Mater. Sci. 17, 397 (1992). 

14. M.R. Albert and J.T. Yates, The Surface Scientist’s Guide to Organometallic Chemistry 
(American Chemical Society, Washington, DC, 1987). 

15. J. M. Burkstrand, J. Vac. Sci. Technol. 20, 440 (1982). 

16. J.M. Burkstrand, J. Appl. Phys. 50, 1152 (1978); 52, 4795 (1982). 

17. K. Kostanididis, R.L. Opila, J.A. Taylor, and A.C. Miller, in Ref. 4., 83-90. 

18. S.G. Anderson, J. Leu, B.D. Silverman, and P.S. Ho, J. Vac. Sci. Technol. A 11, 368 

19. M.J. Goldberg, J.G. Clabes, and C.A. Kovac, J. Vac. Sci. Technol. A 6, 991 (1988). 

20. J.L. Jordan, C.A. Kovac, J.F. Morar, and R.A. Pollack, Phys. Rev. B 36, 1369 (1987). 

21. R.G. Nuzzo and D.L. Allara, J. Am. Chem. Soc. 105, 4481 (1983). 

22. D.L. Allara and R.G. Nuzzo, Langmuir, 1, 45 (1985); 52 (1985). 

23. A. Ulman, An Introduction to Ultrathin Organic Films (Academic, New York, 1991). 


24. L.H. Dubois and R.G. Nuzzo, Annu. Rev. Phys. Chem. 43, 437 (1992). 

25. C.D. Bain, E.B. Troughton, Y.-T. Tao, J. Evall, G.M. Whitesides, and R.G. Nuzzo, J. 
Am. Chem. Soc. Ill, 321 (1989). 

26. D.E. King and A.W. Czanderna, Surf. Sci. Lett. 235, L329 (1990). 

27. J.R. Vig, J. Vac. Sci. Technol. A 3,1027 (1985). 

28. J.R. Pitts, Ph.D. dissertation, Department of Physics, University of Denver, CO, 1985. 

29. D.L. Allara, D.R. Jung, and P. Zhang, Metal atom reactions with self-assembled 
monolayers, paper presented at 39th Natl. Symp. American Vacuum Society, Chicago, 
November 9 to 13, 1992. (D.L. Allara, private communication.) 

30. K. Bammel, J. Ellis, and H.-G. Rubahn, Chem. Phys. Lett. 201, 101 (1993). 

31. F. Balzer, K. Bammel, and H.-G. Rubahn, J. Chem. Phys. 98, 7625 (1993). 

32. G. Herdt and A.W. Czanderna, J. Vac. Sci. Technol. A 13, (1995) In Press. 

33. Independent observations by D.L. Allara, et al. at Penn State U., R.L. Opila, et. al. at 
ATT-Bell Labs, and A.W. Czanderna, et. al. at NREL. 

34. G.K. Wertheim, Z. Phys. B 66, 53 (1987). 

35. D. Spaulding, M.S. thesis. Materials Science Dept., University of Denver, Denver, CO, 

36. A.W. Czanderna and R.J. Gottschall, Eds., Mat. Sci. Engr. 53, 1-168 (1982). 

37. J.D. Swalen, D.L. Allara, J.D. Andrade, E.A. Chandross, S. Garoff, J. Israelachvili, T.J. 
McCarthy, R. Murray, R.F. Pease, J.F. Rabolt, K.J. Wynne, and H. Yu, Langmuir 3, 932 

38. A.W. Czanderna and A.R. Landgrebe, Eds., Current Status, Research Needs and 
Opportunities in Applications of Surface Processing to Transportation and Utilities 
Technologies, Crit. Rev. Surf. Chem. 2 (Nos. 1-4) and 3 (No. 1), 1993. 

39. A.W. Czanderna, J. Vac. Sci. Technol. 14, 408 (1977). 



*Laboratoire de Physique des Interfaces et des Couches Minces (CNRS UPR 258), 

Ecole Polytechnique, 91128 Palaiseau, France 

**Laboratoire de Chimie et Physicochimie Macromoldculaire (CNRS URA 509), 

University du Maine, Avenue Olivier Messiaen, 72017 Le Mans, France 
***L'Air Liquide CRCD, BP 126,78350 Les-Loges-en-Josas, France 


The exposure of polycarbonate to an argon plasma is studied using in situ ellipsometry from 
the UV to the IR, nuclear magnetic resonance and light scattering measurements. An increase in 
the refractive index and the existence of two populations of different molecular weights show that 
structural changes occur in the polymer. They are correlated with modifications at the polymer 
unit scale, such as formation of new polar groups and decrease in dimethyl groups. Two 
simultaneous reaction mechanisms must be considered to account for these changes. The adhesion 
of a silica layer on treated polycarbonate is then discussed. 


Polycarbonate (PC) is a transparent polymer with good mechanical properties, but its use is 
limited by poor resistance to abrasion and to UV irradiation. This drawback could be overcome by 
using a hard protective coating acting as an anti-UV filter, such as a stack of plasma deposited 
silica and silicon nitride layers. However the adhesion of coatings is usually poor, because of the 
weak surface energy of polymers. For this reason a surface treatment is required before coating 
the polymer. 

The effect of inert gas plasmas on polymers has been known for a long time 1 ’ 2 . Vacuum 
ultraviolet photons (^<175 nm) can break C-H or C-C bonds, creating free radicals. Further 
reaction with other radicals or with other chains leads to recombination, unsaturation, branching 
and crosslinking. The last behavior may improve the bond strength of the surface by forming a 
very cohesive skin. 

In the present paper we study the interaction of an argon plasma with PC, in order to improve 
the adhesion of a silica layer. Structural changes in the polymer are investigated by solubility, 
light scattering and in situ UV-visible ellipsometry measurements.The latter provide the refractive 
index of the material. The correlation is made with modifications at the polymer unit scale, which 
are studied by nuclear magnetic resonance (NMR) and in situ IR ellipsometry. 


The 3 mm thick PC samples are argon plasma treated in a dual mode microwave- 
radiofrequency plasma reactor of the surface wave type described elsewhere 3 . The microwave 
excitation generates a high concentration of active species in the gas phase, and the radiofrequency 
allows us to control the ion bombardment on the substrate. In this study the plasma conditions are 
as follows: pressure = 0.080 mbar; microwave power = 200 W; radiofrequency power = 0; 
treatment time = 10 s to 1 hr. 

Proton NMR spectroscopy is run in CDCI 3 solution with TMS as reference on a 400 MHz 
Bruker apparatus. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

Spectroscopic ellipsometry 

Let us recall that the ellipsometric angles H* and A are defined by p = r p /r s = tan T where 
r D and r s are the complex reflexion coefficients for the parallel and perpendicular polarizations 
respectively. In the case of a homogeneous sample, and A are related to the complex refractive 
index (n - i k) by 

e = (n - i kf = Eo sin 2 4 



tan <j>o 

( 1 ) 

where £ is the dielectric function of the material, n the real refractive index, k the extinction 
coefficient, £o the dielectric function of the ambient medium (£o = 1) and <j> 0 the angle of incidence 
(<j) 0 = 70°). In the case of an inhomogeneous sample (one or more layer on a substrate), Y and A 
are still related to the complex refractive indices of the various materials. 

In the infrared, the dielectric function is dominated by the vibrational contributions Ae ot the 
chemical bonds, estimated from the lorentzian expression: Ae(a) = F/(c 0 2 - a 2 + ira), where a = 
1/X is the wavenumber, and a 0 , T and F are the frequency, the width and the strength ot the 
oscillator respectively. In the case of a thin film (thickness d f << X/4n) on a substrate, we present 
IR ellipsometry measurements in the form of the ellipsometric density D - In (p s /p), where p 
refers to the sample and p s to the bare substrate 4 . D is directly related to the complex dielectric 
function e f of the film by 

D = i4^Vio tan»Sinope, df([ . £f) (2) 

(e s -eo)(e s -eotan 2 <|>o) 

where £ s is the substrate dielectric function. Considering that D is roughly proportional to i df (Es - 
Ef), and taking into account eqn. (2), a chemical bond in the film can be identified by a maximum 
of Re D together with a inflexion point (negative slope) in Im D, while a vibrational mode of the 

substrate results in the opposite behavior. . tnrTP _ T ,. v 

UV-visible ellipsometry measurements are performed m real time using a Jobin-Yvon 

phase-modulated ellipsometer 5 in the 1.5-5 eV range. The FTIR ellipsometer is based on a double 
modulation system 6 and covers the range 900-4000 cm 1 . 


Fig. 1 shows the effective refractive index n of PC measured by UV-visible ellipsometry. A 
progressive increase in n during the argon plasma treatment is observed. N is calculated for a 
homogeneous material (see eqn (1)), but actually the treated sample consists of a modified 
overlayer on an unchanged substrate, and n has to be considered as an average on the overlayer 
and the substrate. The progressive increase is probably related to an increase in the overlayer 
thickness. To estimate the thickness, an inhomogeneous layer model had to be used with ex situ 
multiple angle of incidence measurements (not shown). A first estimation gives a typical depth ot 

variation of the refractive index of a few microns. .. 

A possible interpretation for the increase in the refractive index is a densification: according to 
the effective medium theory 7 , a given material has a smaller index if it contains inclusions of void; 
inversely a densification will result in a higher index. Moreover, densification would be in 
agreement with crosslinking, a well-known effect of inert gas plasmas on polymers 2 . 

Besides, the A spectra show a growing structure in the UV range (Fig. 2). Such a structure is 
related to the absorption of certain chemical groups (chromophores) 8 , and the progressive 
increase in A indicates the emergence of new chromophores. Their origin will be discussed later. 

In order to check whether the increase in the refractive index is related to crosslinking, the 
solubility of untreated and treated PC in THF has been measured (Table I). The increase in the 
insoluble fraction shows the appearance of an interchain bridging, already after a 10 s treatment 









. ' 


120 240 

Treatment time (s) 


Fig. 1. Effective refractive index at 3.5 eV measured by ellipsometry in real time during the Ar 
plasma treatment. 

Photon energy (eV) 

Fig. 2. Ellipsometric angle A in the UV range for untreated PC (solid line) and Ar plasma 
treated PC for 6 min (dotted line) and 1 hr (dashed line). 

Table I. Insoluble fraction in THF, and integration of the two dimethyl singlets (NMR) 
referred to the untreated sample, for Ar treated PC. 

Treatment time 

Insoluble fraction(%) 

(CH 3 ) 2 C (%) 


56.8 1 


10 s 



6 min 



1 hr 




The insoluble fraction increases between 10 s and 6 min, and saturates after 6 min. Thus some 
crosslinking occurs in PC during the first 6 min of the Ar plasma treatment. 

Light scattering measurements in dioxane have been performed to estimate the size of the coils: 
their size should increase in case of crosslinking. However it turns out that the treated samples 
contain two different populations: one of bigger size than the untreated sample, corresponding to 
the crosslinked fraction of the polymer; and the other one, in greater quantity, of smaller size than 
the untreated sample. The latter results from chain breakings in the polymer. This means that both 

the degradation and crosslinking reactions occur during the plasma treatment. 

NMR measurements have been performed in order to elucidate the degradation and 
crosslinking mechanisms. The NMR spectra of the proton show for the untreated sample a 
multiplet at 7.2 ppm (aromatic protons) and two singlets at 1.7 and 1.6 ppm (dimethyl groups in 
the middle and at the end of a chain resp.) (Fig. 3). OH groups at the end of the chains have not 
been detected. The plasma treatment results in a progressive decrease in the integration of the 
dimethyl singlets (Table I), particularly at the chain ends. Thus methyl groups seem to be 

involved in the degradation and/or in the crosslinking reaction. 

Further information about the chemical mechanisms is given by in situ IR ellipsometry. As 
explained in the experimental section, ¥ and A spectra of the treated samples are referred to the 
untreated sample spectra to form the ellipsometric density D. The major changes (i. e. D ^ 0) are 
observed in the C=0 stretching mode region (Fig. 4). The carbonate C=0 mode (1771 cm 1 ) 
appears as a negative peak in Re D and as an inflexion point with positive slope in lm D. Thus 
this mode, already present in PC, is in smaller quantity in the modified material (see eqn. (2)). On 
the contrary, new modes appear at 1690 and 1750 cm -1 as positive peaks in Re D and as inflexion 
points with negative slope in Im D. They correspond to new bonds in the modified material and 
can be attributed to less oxidised C=0 groups than the carbonate group (carbonyl and ester 
groups). The three modes, hardly visible after 10 s treatment, become more and more intense after 

6 min and 1 hr. t . , . A „ „ 

Moreover, XPS measurements show a decrease in the carbonate signal after a 10 s treatment, 
without any further evolution 9 . As XPS is only sensitive to the surface, while IR ellipsometry is 
also sensitive to the bulk, we can conclude that the reaction occurs first at the surface, and then 
propagates towards the bulk. 



f <r' °--f-° H 

ch 3 o ch 3 

Fig. 3. PC chain. 





Fig. 4. Ellipsometric density D for 6 min (dashed lines) and 1 hr (solid lines) Ar treated PC. 
The arrows indicate vibrational modes appeared during the treatment. 



Two reactions, degradation and crosslinking, seem to be in competition during the Ar plasma 
treatment. As we do not use any radiofrequency plasma in this study, and as the effect is not 
limited to the surface, we can infer that ion bombardment is not responsible for these reactions, 
but that their cause is UV light emitted by the plasma. 

The literature concerning the photodegradation of PC mentions two photodegradation 
mechanisms: one through direct photochemistry, and the other through induced photochemistry 8 . 
The first one (Fig. 5) is due to the excitation of the polymer chromophores by short-wavelength 
photons, the absorption band of PC being centered at 265 nm. After breaking of the carbonate 
bond, two successive photo-Fries rearrangements occur. The phenylcarbonate units rearrange into 
phenylsalicylate units (Li), which then rearrange into dihydroxybenzophenone groups (L 2 ). 
Concurrently the radicals coming from the carbonate breaking can recombine after decarbonyl- 
ation or a decarboxylation, generating various photoproducts (L 3 ). Li, L 2 and L 3 are not stable in 
the combined presence of oxygen and UV light, but during the plasma treatment there is no 
oxygen. Hence the direct mechanism could be responsible for the decrease in the coil size, as 
chain breaking is one of the consequences of these reactions; for the new C=0 modes (Fig. 4), 
from Lj and L 2 ; and for the increased absorption in the UV range (Fig. 2), due particularly to L 2 . 

However crosslinking and the decrease in dimethyl groups are not accounted for by this 
mechanism. The second mechanism mentioned in the literature is induced by defects or impurities 
of the material. By absorption of long-wavelength photons, they produce R* radicals (Fig. 6 ). 
These radicals react with PC by removing a hydrogen atom from the methyl group, and this new 
radical isomerises into a tertiary macroradical R 3 to increase its stability. Then in the presence of 
oxygen, hydroperoxydes are formed and lead to degradation. In our case the macroradical could 
react with an other chain, leading to crosslinking. 

The adhesion of a silica layer on PC has been measured using a microscratch test 10 . The 1 |im 
thick silica layer was plasma deposited, after an Ar treatment or directly on PC. Although an 
improvement is already obtained after 30 s, the best result is given by the 6 min treatment. Two 
reasons may be considered to account for the enhanced adhesion: crosslinking and creation of 
new polar groups (Li and L 2 ). As most changes in surface polar groups have occured after 10 s, 
the second mechanism is not prevailing. In spite of concurring degradation, crosslinking thus 
improves the adhesion of a silica layer on PC, probably through the formation of a cohesive skin. 





+Q- oh 


0 \ 

hv I photo-Fries 

+©-°' + ’S- < ^F 

R1 O R 2 

+Q- oh 


L 2 

chain breaking 





+@- 0H 

+©- H 

Fig. 5. Direct photodegradation mechanism of PC. 


Absorbing impurities 

f + — -©-\-©-°- c u -°- + RH 

ch 3 o CH 3 O 

I Isomerisation 


R, ° 

Fig. 6. Production of macroradicals (R 3 ) through the induced photodegradation mechanism of 


In this paper we have studied the interaction of an argon plasma with PC S ^ ct ^ a i^f ^ 
the polymer have been investigated by light scattering measurements and by UV-visible 
ellipsometry, which provides the real-time refractive index of thematenal. The s ^cmrd chafes 
have been correlated with modifications at the polymer unit scale, which were studied by NMR 
and in situ IR ellipsometry. It appears that the exposure of PC to an argon plasma produces two 
simultaneous reactions: a degradation reaction, that results in chain breakings, formation of new 
polar groups (carbonyl and ester) and increased absorption in the UV range; and a crosslinkmg 
reaction that probably involves methyl groups. The later mechanism seems to be responsible o 
the enhanced adhesion of a silica layer on PC. 


The authors are grateful to A. Gheorghiu and C. Sdndmaud for XPS measurements and to 
L. Martinu and J. E. Klemberg-Sapieha for adhesion tests. 


1 R H Hansen and H. Schonhom, J. Polym. Sci., Polym. Lett. Ed. B4, 203 ( 19 66)- 

2. J.R. Hollahan and A.T. Bell, Techniques an d Applica tions of Plasma Chemist ry (Wiley, New 

3. LC^Rostaing, F. Coeuret, B. Drevillon, R. Etemadi, C. Godet, J. Hue, J.Y. Parey and V. 

4 . ^^^Drl^llmi^ Appl. Phys- Le«. 59, 950 (1991); J. Non-Cryst. Solids 137- 

5. R Drlvillon, Prog, in Cryst. Growth and Charact of Mat 27,1 (1993) 

6 A Canillas, E. Pascual et B. Drevillon, Rev. Sci. Instr. 64, 2153 (1993). 

1. DJE. Aspnes, Thin Solid Films 89, 249 (1982) 

8 . A. Rivaton, D. Sallet and J. Lemaire, Polym. Photochem. 3,463 (1983). 

9 A Gheorghiu and C. Sdndmaud (private communication). 

10. L. Martinu and J.E. Klemberg-Sapieha (private communication). 





Department of Materials Science and Engineering, Virginia Polytechnic Institute and 
State University, Blacksburg, VA 24061 
Composites Innovation, Owens-Coming Science and Technology Center, 

Granville, OH 43023 


Adhesion at fiber-matrix interface in fiber-reinforced composites plays an 
important role in controlling the mechanical properties and overall performance of 
composites. Among the many available tests applicable to the composite interfaces, 
vibration damping technique has the advantages of being nondestructive as well as highly 
sensitive. We set up an optical system to measure the damping tangent delta of a 
cantilever beam, and correlated the damping data in glass-fiber reinforced epoxy-resin 
composites with transverse tensile strength which is also a qualitative measurement of 
adhesion at fiber-matrix interface. Four different composite systems containing three 
different glass-fiber surface treatments were tested and compared. Our experimental 
results showed an inverse relationship between damping contributed by the interface and 
composite transverse tensile strength. 


It is well-known that the fiber-matrix interfacial adhesion has a major effect in 
achieving superior mechanical properties of a composite. The tensile strength of the 
composite is dependent on the ability of the composite to transfer the tensile load from 
the broken fibers to the surviving ones through shear in the matrix and at the interface. 
Thus, a method that is capable of determining the interfacial adhesion strength is needed 
to evaluate the mechanical performance of composite materials. 

Numerous experimental techniques have been developed for measuring interfacial 
adhesion strength in fiber-reinforced composites. These methods include the single fiber 
pull-out test [1, 2], microbond test [3-5], the single fiber fragmentation test [6-9], the 
microindentation test [10, 11], and some nondestructive evaluation techniques, such as 
vibration damping [12]. Vibration damping is a promising nondestructive technique 
because it is simple and quite sensitive to the interfacial region. The method has a 
potential to be used by materials industry for in situ monitoring of the mechanical 
performance of composites. 

According to the theory of energy dissipation [13], the quality of the interfacial 
adhesion in composites can be evaluated by measuring the part of energy dissipation 
contributed by the interfaces; the interface part can be obtained by separating the fiber 
and matrix from the total composites. Zorowski and Murayama [12] were the first to 


Mat. Res. Soc. Symp. Proc. Vol. 385 ° 1995 Materials Research Society 

develop a method for the quality of the interfacial adhesion in the reinforced rubber 
through energy dissipation measurements based upon the following relationship: 

tanS in = tan5 comp -tanS, 

tan 8, = 

tan8 f E f V f + tan 5 m E m V m 
E m V m +EjV f 



where tan5 is the internal energy dissipation due to poor adhesion from the interface, 
which can be used for evaluating the interfacial adhesion; tan8, is the effective loss 
tangent for a composite with perfect interfacial adhesion, tan8 comp is the measured 
internal energy dissipation of the composite system. E is the Young s modulus and V 
represents volume fraction. Subscripts/and m refer to the fiber and matrix, respectively. 
By measuring the total system energy dissipation in terms of tan 8 and knowing tan 8 
and the dynamic moduli of the components as well as the volume fraction of fibers, the 
dissipation due to the poor interfacial adhesion can be determined. 

In this paper, we describe an optical setup for measuring vibration damping of 
cantilevers specimen. The setup was used to measure the damping of glass-fiber 
reinforced polymer composites, and the damping data were correlated with the transverse 
tensile strength of composites. 


2.1 Composite Sample Preparation 

Composite laminate specimens were fabricated at the Owens-Coming Science and 
Technology Center. D.E.R. 331 epoxy resin from Dow Chemicals Company and 
Lindride 66 curing agent from Lindau Chemicals Inc. were selected as the matrix material 
commonly used in filament winding process. The reinforcements were E-glass fibers 
with a diameter of 10 pm. Four fiber systems that contained different surface treatments 
were investigated in this study as listed in Table 1. To make a composite laminate, 
D.E.R. 331 epoxy (100 parts by weight) was mixed with Lindride 66, curing agent, (85 
parts by weight). Composite laminates were fabricated through a filament winding 
machine and samples were cured for two hours at 120°C and two hours at 180 C under a 
hot press machine with a 1.43 MPa constant pressure. The composite laminates were 
then cut into 30 mmx 4 mmx 0.5 mm specimens. The actual length of the cantilever 
beam was 25 mm. 

2.2 Optical Setup 

A schematic of the optical setup designed to measure the deflection and vibration 
dynamics of a cantilever beam is shown in Figure 1. The construction consists of a 1 
mW solid state laser (670 nm), a mirror, a beam splitter and a position sensitive 
photodetector. A sample is mounted by clamping it vertically between two plates such 
that the protruding part forms a cantilever beam. An electronically triggered pin is used 


to generate an initial deflection on the sample; vibration of the sample is initiated by 
retracting the pin. Vibration curves are obtained by bouncing a laser beam off the sample 
to the photodetector. 


Figure 1. Schematic diagram of the optical system 

The damping factor, tan8 , is calculated from decaying-oscillatory damping curve 
by the following [14] 

tang = ln(VA) 


where n is the number of cycles of the vibration, A 0 is the amplitude of the first 

vibration, and A n is amplitude of the n th vibration. The term ^ , also known as 


the logarithmic decrement A, can be obtained by fitting the experimental data to the 
following formula [14] 

A(t) = B 0 exp(-^co r t) cos(co d t - cj>) + B x (3) 

where co, is the resonant frequency of vibration, £ = A / v(2ti) 2 + A 2 ~ M 2% when 
damping is small, co rf = (l-^ 2 ) l/2 co r , B 0 , B i and <j> are constants. 

Table 1. A List of the Samples Used in this Study 

Specimen Type 

Fiber Volume Fraction 

Description of Surface Treatment 



No.l sizing with silane 



Untreated fibers 



No.l sizing without silane 



No. 2 sizing with silane 



Figure 2 is a typical example of the results that obtained from the optical system 
described above, in which a composite specimen without any fiber surface treatment 
(Type B specimen) was tested. The value of tan 8 was found to be 3.81 x 10 . This 
value is typical for this material system. Other experimental results are summarized in 
Table 2. The values presented here are the averages that obtained from at least five 

U.OU Vi. UU 

Time, t, (Sec) 

Figure 2. An example of the vibration damping curves from the optical system 

Table 2. Measured Resonant Frequencies, Damping Factors, 
and Transverse Tensile Strength_ 



(s' 1 ) 

tan5 c ,„„ p (xlO 3 ) 

tan8 ifJ (xlO 3 ) 

a,, (MPa) 


4061 (1.2%)* 

2.46 (4.9%) 


56 (8.8%) 


3733 (2.0%) 


2.24 n 

28 (14.7%) 


3980 (2.7%) 

2.75 (3.6%) 


45 (8.5%) I 


4038 (1.2%) 

2.73 (0.7%) 


46 (9.3%) 

♦Numbers in parenthesis represent coefficient of variation. 


From the Bemoulli-Euler beam equation [14-16], it can be shown that the 
Young’s modulus, E, of the material is related to its frequency of vibration. The equation 
used for calculating E for a beam specimen is described as follows [14]: 

e 12 pco 2 r L 4 


where go r is the resonant frequency of the first mode of vibration, L and t are the length 
and the thickness of the beam, and p is the density. The density of the epoxy resin is 
1.115 g/cm 3 [17]. 

It has been reported that the damping factor also varies with frequency [18, 19]. 
By changing the beam length, a resonant frequency of about 3900 s’ 1 was obtained, which 
is comparable to the frequencies of other composites that were tested. The measurements 
from our optical system show that E m =2.4 GPa and tan6 m = 25x 10“ 3 . It is also 
known that Ej = 72.3 GPa and tandy = 1 x 10~ 3 [20]. These data were used to calculate 
tanS in which are also given in Table 2. 

Equation (1) shows that with a higher value of tan 5 m the poor adhesion exhibits 
in the system. Results show that Type B specimen appears to have the weakest fiber- 
matrix interfacial adhesion among the four composite systems. However, Type A 
specimen which contains No. 1 sizing and silane show the best interfacial adhesion. It is 
also interesting to note that the Type C specimen (having No. 1 sizing and no silane) and 
Type D specimen (having No. 2 sizing and different silane) seem to have an equal 
magnitude of interfacial adhesion. 

It is known that the poor interfacial adhesion exhibits a low transverse tensile 
strength, o lr . The measured a tr from the same composite systems are also listed in 
Table 2. It can easily be seen that the observations from tanS ;>1 are consistent with the 
results from c lr , and both show the same magnitude of the interfacial bonding. In other 
words, tan5 in and c lr are highly inversely correlated as shown in Fig. 3. 

Figure 3. The relationship between tan 5 in and ct tr . Error bars represent ±1 
standard deviation. 


It is suggestive that a common process among the defect-related mechanisms 
involving a) transfer of kinetic energy of structural motion to potential energy of a defect 
and b) dissipation of the potential energy in the form of heat to its surroundings. 
Therefore, a sample contains more defects should have a higher damping factor; and also 
the more loosely configured a defect is, the higher its contribution to damping. At a 
strongly bonded interface, there are fewer loosely bonded defects or centers which can 
easily absorb kinetic energy; therefore, it should have smaller damping. At a weak 
interface, there are likely more loosely bonded defects or centers to absorb kinetic energy; 
thus, it should have higher damping. 


An optical setup was constructed to measure the damping factor of a cantilever 
beam and used to characterize adhesion at fiber-matrix interfaces in glass-fiber reinforced 
polymer-resin composites. Tested samples had three different fiber-surface treatments 
and a controlling sample. The results show that the composite system having No. 1 
sizing and silane exhibits the best fiber-matrix interfacial adhesion. The system having 
No. 1 sizing without silane turns out to be an equal magnitude of interfacial adhesion as 
the system having No. 2 sizing and different silane. Samples with untreated fibers have 
the weakest interfacial adhesion. The experimental results showed that a strong inverse 
relationship between damping characteristics of the fiber-matrix interface and transverse 
tensile strength of composites. 


1. L. J. Broutman, Interfaces in Composites. STP 452 . edited, by M. J. Salkind 
(American Society of Testing and Materials, Philadelphia, 1969), p. 27. 

2. P. S. Chua and M. R. Piggott, Composites Science and Technology, 22, 33 (1985). 

3. B. Miller, P. Muri, and L. Rebenfeld, Composites Science, and Technology, 28,17 

4. H. F. Wu and C. M. Claypool, Journal of Materials Science Letters, 10, 260 (1991). 

5. H. F. Wu and C. M. Claypool, Journal of Materials Science Letters, 10, 1072 (1991). 

6 . H. F. Wu, G. Biresaw, and J. T. Laemmle, Polymer Composites, 12 (4), 281 (1991). 

7. A. N. Netravali, Z. -F. Li, W. Sachse, and H. F. Wu, Journal of Materials Science, 26, 
6631 (1991). 

8 . J. P. Favre and J. Perrin, Journal of Materials Science, 7, 1113 (1972). 

9. L. T. Drzal, M. J. Rich, M. F. Koenig, and P. F. Lloyd, Journal of Adhesion, 16, 1 

10. J. F. Mandell, E. J. H. Chen and F. J. McGarry, Intemat. J. Adhesion Adhesives, 1, 40 

11. H. F. Wu and M. K. Ferber, Journal of Adhesion, 45, 89 (1994). 

12. C. F. Zorowski and T. Murayama, Proc. 1st Int. Conf. on Mech. Behav. of Mater., 5, 
Soc. of Mater. Sci., Kyoto, Japan, 28 (1972). 

13. L. E. Goodman. Structural Damping . ASME. 36 (19591. 


14. L. Meirovitch, Elements of Vibration Analysis , edited by B. J. Clark and M. E. 
Margolies, McGraw-Hill, 1975. 

15. E. Volterra and E. C. Zachmanoglou, Dynamics of Vibrations . Charles E. Merrill, 
Columbus, OH, 321 (1965). 

16. K. Clark, Dynamics of Continuous Elements . Prentice-Hall, Englewood Cliffs, 
NJ, 75 (1972). 

17. M. Weller and Hassel Ledbetter, J. Mater. Res., 5 (5), 913 (1990). 

18. L. B. Crema, A. Caastellani and A. Serra, Journal of Composite Materials, 23, 

19. R. F. Gibson and R. Plunkett, Journal of Composite Materials, 10, 325 (1976). 


Part II 

Biointeractions and Biointerfaces 




*Dept. Chemical Engineering, Massachusetts Institute of Technology, Cambridge MA 02139 
"Dept. Surgery, Harvard Medical School and Children’s Hospital, Boston MA 02115 


Engineering liver tissue using hepatocyte transplantation may provide a new approach for treating a 
variety of liver diseases. However, techniques to transplant hepatocytes and promote their survival 
must be developed. We have developed systems to transplant hepatocytes on highly porous 
(95%), biodegradable sponges, and to regulate the survival of cultured hepatocytes by releasing 
specific growth factors in the cellular environment. Sponges were fabricated from poly (L, lactic 
acid) (PLLA) and polyvinyl alcohol using a particulate leaching technique. Epidermal growth 
factor and insulin, critical factors for hepatocyte growth and survival in culture, were incorporated 
into microspheres fabricated from poly (lactic-co-glycolic acid) (PLGA) utilizing a double emulsion 
technique. The incorporated factors were released in a controlled manner over one month in vitro, 
and the released factors maintained their biological activity, as measured by their ability to promote 
hepatocyte growth and survival in culture. The growth factor-containing microspheres could be 
transplanted with hepatocytes using the porous sponges, and the localized, sustained release of 
these factors improved hepatocyte engraftment 2-fold. These studies suggest that hepatocyte- 
containing tissues can be engineered using cell transplantation, and that regulating the 
microenvironment of transplanted cells can control their engraftment. 


Whole liver transplantation is currently the established therapy for end-stage liver disease 
but it suffers from a great limitation in donor organs. Approximately 30,000 people still die each 
year in the United States of liver disease (1), and a large number of these people could be treated if 
there was an available supply of liver tissue. Transplantation of hepatocytes, the major liver-cell 
type, may replace liver function in patients suffering from liver failure, and provide an alternative 
to whole liver transplantation (2). The challenges in engineering liver tissue with hepatocyte 
transplantation include the necessity of delivering a large number of cells to a specific anatomic 
location, and providing a suitable environment for long-term hepatocyte engraftment. 

Highly porous sponges fabricated from biodegradable polymers of die lactic and glycolic 
acid family may be ideal cell delivery devices (3). Large numbers of hepatocytes can be seeded in 
a highly porous device, and transplanted to the desired location. The pores provide a route for 
fibrovascular tissue ingrowth from the surrounding host tissue (4), and the ingrown blood vessels 
may provide for the metabolic needs of the implanted cells as they develop into a new tissue. 
Highly porous sponges have been previously fabricated from poly-L-lactic acid utilizing a 
particulate leaching technique, and coated with polyvinyl alcohol to enable even and reproducible 
seeding with hepatocytes (3). These devices have been utilized to tranpslant hepatocytes into 
experimental animals, where a number of the transplanted hepatocytes engrafted (histologically 
appeared healthy) and survived for the limited time of the experiment (1 week). However, it will 
be critical to provide the proper environment if these cells are to engraft over long time frames. 

A variety of growth factors have been identified that regulate the survival and proliferation 
of hepatocytes. These include epidermal growth factor (EGF) and insulin (5). A critical role for 
insulin is suggested by the finding that trophic factors from islet cells improve the survival of 
transplanted hepatocytes (6), as does co-transplantation of islets with hepatocytes (7). However, it 
is difficult to attribute these effects to the presence of specific factors such as insulin due to the 
number of factors secreted by islets, and the potential role of islet-hepatocyte communication in this 
process. If techniques to reproducibly deliver specific factors were developed it would allow one 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

to investigate the role of various factors, alone and in combination, in the survival and growth of 
transplanted hepatocytes. This may allow one to define an optimal environment for the survival ot 
transplanted cells, and move hepatocyte transplantation closer to a clinically relevant therapy. 

In this report we describe a technique to deliver biologically active factors such as EGF and 
insulin over extended periods to hepatocytes transplanted into heterotopic sites using biodegradable 
polymer sponges. This approach modulates the microenvironment of transplanted hepatocytes to 
improve engraftment. 


Microspheres containing insulin and EGF were prepared by a modification of a previously 
described double-emulsion technique (8). In brief, a 75/25 copolymer of poly-(D,L-lactic-co- 
glycolic) acid (Resomer RG 75R, intrinsic viscosity 0.2 dl/g; Henley Chem. Inc, Montvale, NJ) 
was dissolved in ethyl acetate (Fisher Scientific) to yield a 5% solution (w:v). Mouse EGF 
(Collaborative Research; Bedford, MA) and recombinant human insulin (Sigma) was dissolved m 
water to yield a solution of 2 mg/ml (EGF) and 1 mg/ml (insulin), and 50 pi of the solution was 
added to 1 ml of the polymer solution. The resulting solution was sonicated continuously at 10 
watts (Vibracell; Sonics and Materials, Danbury, CT) for 15 sec to yield a single emulsion. An 
equal volume of an aqueous solution containing 1% polyvinyl alcohol (MW 25,000, 88% 
hydrolyzed; Polysciences Inc., Warrington, PA) and 7% ethyl acetate was added to the single 
emulsion, and the resulting solution was vortexed (Vortex Mixer; VWR) for 15 sec at the high 
setting to yield the double emulsion. This double emulsion was transferred to a rapidly stirring 
250 ml beaker containing 150 ml of an aqueous solution of 0.3% polyvinyl alcohol/7% ethyl 
acetate. The ethyl acetate was allowed to evaporate over the ensuing 3 hr to yield polymer 
microspheres with entrapped EGF. The microspheres were then filtered, washed with water, and 
microspheres with a size between 32 and 0.4 mm were collected. The microspheres were 
lyophilized (Labconco Freeze Dryer, Kansas City, MO), and stored at -20°C until use. Control 
beads were prepared with the same procedure, but the aqueous solution used to form the first, 

single emulsion (water in organic) contained no insulin or EGF. 

To determine the efficiency of factor incorporation, and the kinetics of release from the 
microspheres, * 25 I-labeled mouse EGF (260 mCi/mg; Biomedical Tech. Inc., Stoughton, MA) and 
125 I-labeled insulin (New England Nuclear, Boston MA; 110 mCi/mg) were utilized as tracers. 
Approximately 1 pCi of labeled drug was added to the aqueous drug solution before formation of 
the single emulsion, and the beads were prepared as described above. After bead fabrication, a 
known mass of beads was counted in a LKB CliniGamma 1272 (Wallac, Gaithesburg, MD), and 
the incorporated cpm was compared to that of the initial aqueous solution to calculate the drug 
incorporation. Drug release was quantitated by placing a known mass of beads in a known volume 
of phosphate buffered saline (PBS) solution containing 0.1% Tween 20 (Sigma Chem. Co.), and 
incubating at 37°C. At set times, the solution was centrifuged to concentrate the beads at the 
bottom of the vial, and samples were removed and counted in a gamma counter. Sink conditions 
were maintained during the release study. 

For scanning electron microscopic examination, samples were first gold coated using a 
Sputter Coater (Desk n, Denton Vacuum, Cherry Hill, NJ), and imaged using an environmental 
scanning electron microscope (ElectroScan; Wilmington, MA). Photomicrographs were taken with 

Polaroid 55 film. ,. . 

Cultured hepatocytes were utilized to determine whether the drugs incorporated into and 
released from microspheres had retained their biological function. Hepatocytes were isolated from 
Lewis rats using a two-step collagenase perfusion, and purified using a Percoll gradient as 
previously described (9). Hepatocytes were plated at a density of 10, 000 cells/cm 2 on 24 well 
tissue culture dishes coated with 1 mg/cm 2 of type I collagen (Collagen Corp., Palo Alto, CA) 
using a carbonate buffer coating technique (9). Serum-free William's E medium (Gibco, Grand 
Island, NY) containing insulin (20 mU/ml; Sigma), dexamethasone (5 nM; Sigma), sodium 
pyruvate (20 mM; Gibco), a mixture of penicillin and streptomycin (100 U/ml; Irvine Scientific, 


Santa Ana, CA), and ascorbic acid (50 mg/ml, fresh daily; Gibco) was used for all experiments. 
For conditions in which drug released from microspheres was utilized, medium with no drug was 
incubated with drug containing microspheres for 24-96 hr to allow release of known amounts of 
drug (calculated using the known release kinetics), the solution was centrifuged, and the medium 
containing the released drug was removed and used in subsequent experiments. To analyze cell 
entry into S phase of the cell cycle, tritiated thymidine autoradiography was utilized. Cultured 
hepatocytes were refed 48 hr after plating with medium containing 1 mCi/rnl 3 H-thymidine (NEN; 
Boston, MA). At 72 hr cells were twice washed with PBS to wash out any non-incorporated 3 H- 
thymidine, fixed with glutaraldehyde, and dehydrated with 100% methanol. Culture wells were 
overlaid with NTB-2 emulsion (Kodak; Rochester NY), and the dishes were allowed to expose for 
7 days in complete darkness. Dishes were developed with D-19 developer (Kodak). 

Isolated and purified hepatocytes were mixed with EGF-containing or control 
microspheres, and seeded onto 95% porous cylindrical sponges (diameter=2.15 cm, thickness=l 
mm) in a petri dish. The sponges were fabricated from poly-(L, lactic) acid (Medisorb; Cincinnati, 
OH), and coated with polyvinyl alcohol as previously described (3). Each sponge received 0.4 ml 
x 50X10 6 hepatocytes/ml + 10 mg of microspheres. Cell-polymer devices were implanted into the 
mesentery of laboratory rats as previously described (10). Implants were removed after 14 days, 
fixed in formalin, and processed for sectioning. Sections of implants were stained with 
hematoxylin and eosin, and engrafted hepatocytes were identified by their large size, large and 
spherical nuclei, and distinct cytoplasmic staining. All animals were housed in the Animal 
Research Facility of Children's Hospital, and NIH guidelines for the care and use of laboratory 
animals (NIH Publication #85-23 Rev. 1985) have been observed. 


Microspheres fabricated with the double-emulsion technique had diameters ranging from 10 
to 30 pm (the approximate size of suspended hepatocytes) (Fig. 1). The average diameter was 

19±12 pm. The efficiency of insulin and EGF incorporation into microspheres was 42*10% and 
53*11%, respectively. Insulin was released from the microspheres in a large initial burst, followed 
by a steady release over the ensuing 10 days. EGF release also had an initial burst, although not as 
large as insulin, and a steady release was again noted after this time (Fig. 2). 

Figure 1. Photomicrograph of polymer microspheres. A scanning electron microscope was 
utilized to image microspheres. A size bar is shown on the photomicrograph. 


Time (days) Tlme < d0 y s > 

Figure 2. Release kinetics of 125 I-labelled insulin (A) and EGF (B) from polymer microspheres. 
The amount of protein released was normalized to the total radiolabelled protein incorporated into 
the mass of microspheres utilized in the release study. Results from a representative experiment 
are shown. 

Studies with cultured hepatocytes were subsequently performed to determine whether the 
released factors retained their biological activity. EGF and insulin released from microspheres 
stimulated hepatocyte entry into the synthetic (S) phase of the cell cycle, indicating that they 
retained their biological activity. Both insulin and EGF normally stimulate hepatocyte growth in a 
dose-dependent manner (Fig. 3). Quantitation of the number of hepatocytes entering S phase 
revealed that insulin released from the microspheres still had retained significant activity, although 
less than control insulin (Fig. 3A). In contrast, the same dose of EGF either released from 
microspheres or control EGF showed a similar stimulation (Fig. 3B). EGF released from 
microspheres was also shown to promote the long-term survival of cultured hepatocytes (11). 

To determine whether EGF released from microspheres could positively influence the 
engraftment of hepatocytes transplanted to a heterotopic site, hepatocytes (2xl0 7 cells/sponge) and 
microspheres (10 mg/sponge) were mixed together and seeded onto porous, biodegradable 
sponges fabricated from poly-(L-lactic) acid (Fig. 4). The ratio of polymer/salt used in the sponge 
fabrication process was set at 0.03 to achieve a device porosity of 95*2%. Cell/microsphere- 
seeded devices were implanted into the mesentery of laboratory rats. Retrieval of implants after 
two weeks, followed by histological preparation and observation, revealed that animals which 
received EGF containing microspheres had a large number of engrafted hepatocytes (not shown). 
Quantitation of the number of engrafted hepatocytes revealed that animals which received EGF- 
containing microspheres contained a statistically significant greater number of engrafted 
hepatocytes than animals which received control microspheres (11). 


A system has been developed to both deliver large numbers of hepatocytes to specific 
anatomic locations, and to release growth factors at the site of hepatocyte transplantation from 
polymer microspheres. Specific hepatotrophic factors (insulin and EGF) can be incorporated into 
polymer microspheres. These factors are released over extended time periods when the 


Figure 3. Stimulation of hepatocyte growth by EGF (A) and insulin (B). Hepatocytes were 
cultured in medium containing varying concentrations of EGF or insulin that was not incorporated 
into microspheres (Control), or a known amount of EGF or insulin released from polymer 
microspheres (Released). Released EGF stimulated hepatocyte growth to the same extent as 
control EGF, while released insulin had a lower stimulatory effect than control insulin. The 
growth of cells was analyzed using 3 H-thymidine autoradiography. Cells in S phase of the cell 
cycle (Growing cells) incorporate the radiolabelled thymidine and can be identified by their black 
nuclei following autoradiography. 

Figure 4. Photomicrograph of a biodegradable sponge seeded with hepatocytes prior to 
implantation. The sponge was visualized using scanning electron microscopy, and a size bar is 
shown on the photomicrograph. 


microspheres are placed in an aqueous environment, and the released factors maintain their 
biological activity. 

Hepatocytes can be transplanted in a variety of manners (reviewed in 12). The PLA 
sponges utilized in this study allow large numbers of hepatocytes to be efficiently transplanted into 
experimental rats at a desired anatomic location. Significant numbers of transplanted hepatocytes 
survive and form a new tissue with the ingrowing fibrovascular tissue and the polymer device (3). 
Fibrovascular tissue invasion leads to the formation of a vascular network which supplies the 
metabolic needs of the developing tissue. 

Delivery of hepatotrophic factors via sustained release from microspheres may prove a 
valuable and flexible technique to control the local environment of transplanted cells. A known 
dose of a factor can be delivered with this approach, and the time over which a drug is released 
from a polymer matrix can typically be regulated by the drug loading, the type of polymer utilized, 
and the exact processing conditions (13). A critical feature is that the released protein(s) must 
retain its biological activity for this approach to be useful. The biological activity of proteins such 
as insulin can be compromised by the processing conditions, due to aggregation and denaturation 
of the drug (14). This may account for the decreased activity of insulin released from the 
microspheres in this study. The biological activity of the EGF incorporated into and released from 
microspheres in this study did not appear to be adversely affected, but additional studies will be 
necessary to confirm that this does not occur at later stages of the release. The large burst release 
of protein at the beginning of the release studies is likely related to surface-associated protein. It is 
unclear, however, why the initial burst was much greater for insulin than EGF. 

In summary, we have developed a system to transplant hepatocytes and control the 
presence of specific growth factors in the local environment of the transplanted cells. This system 
may greatly improve the engraftment of transplanged hepatocytes, and other cell types, and move 
cell transplantation closer to a clinically relevant therapy. 


The authors gratefully acknowledge the assistance of Dr. B. Schloo, who prepared the 
histological sections. This work was supported by grants from the National Science Foundation 
(BCS9202311), and Advanced Tissue Sciences. 


1. American Liver Foundation, Vital Statistics of the United States, 1988:Vol.2 (A). 

2. R. Langer and J.P. Vacanti, Tissue engineering, Science, 260, 920 (1993). 

3. D.J. Mooney, S. Park, P.M. Kaufmann et al, J. Biomed. Mat. Res., in press. 

4. A.G. Mikos, G. Sarakinos, M.D. Lyman, D.E. Inger, J.P. Vacanti, and R. Langer, Biotech. 
Bioeng., 42, 716 (1993). 

5. N. Fausto, Prog. Growth Factor Res. 3, 219 (1991). 

6. V. Jaffe, H. Darby, A. Bishop, H.J. Hodgson, Int. J. Exp. Path. 72, 289 (1991). 

7. P.M. Kaufmann, K. Sano, S. Uyama, and J.P. Vacanti, Transplantation Proc. 26 (1994). 

8. S. Cohen, T. Yoshioka, M. Lucarelli, L.H. Hwang, and R. Langer, Pharm. Res. 8, 713 

9. D. Mooney, L. Hansen, J.P. Vacanti, R. Langer, S. Farmer, and D. Ingber, J. Cell Phys. 
151, 497 (1991). 

10. S. Uyama, P.M. Kaufmann, T. Takeda, and J.P. Vacanti, Transplantation 55, 932 (1993). 

11. D.J. Mooney P.M. Kaufmann, K. Sano, et at., submitted. 

12. L.K. Hansen, and J.P. Vacanti in Current controversies in biliary atresia , edited by M.A. 
Hoffman (R. G. Landes, Austin, TX 1993), p. 96. 

13. R. Langer, Science 249, 1527 (1990). 

14. V. Sluzky, J.A. Tamada, A.M. Klibanov, and R. Langer, Proc. Natl. Acad. Sci. 88, 9377 

(1991). Thjs artic | e a | SO appears in Mat. Res. Soc. Symp. Proc. Vol 394 




*Biomedical Engineermg Institute, Ecole Polytechnique, Montreal, Qc, CANADA; 

**Chemical Engineering Department, Ecole Polytechnique, Montreal, Qc, CANADA; 
***Department of Surgery, Ste-Justine Hospital, Montreal, Qc, CANADA. 


Poly(fi-hydroxybutyrate-co-fl-hydroxyvalerate) have been recently proposed as 
degradable biomaterials for drug delivery systems, sutures, bone plates and short-term implants. 
Three PHBYHV (7, 14 & 22 % HV) films were analyzed for in vitro cytotoxicity and aqueous 
accelerated degradation, in vivo degradation and tissue reactions. The PHB/HV materials and 
extracts elicit few or mild toxic responses, do not lead in vivo to tissue necrosis or abscess 
formation, but provoke acute inflammatory reactions slightly decreasing with the time. The 
degradation of PHB/HV polymers present low rates in vitro as well as in vivo. The weight loss 
rate generally increases with the copolymer composition (HV content) and ranges from 0.15- 
0.30 (in vitro ) to 0.25 %/day (in vivo). Compositional and physico-chemical changes in 
PHB/HV materials were rapidly detected during the accelerated hydrolysis, but were much 
slower to appear in vivo. The structural and mechanical integrity of PHB/HV materials tend to 
disappear early in vitro as well as in vivo. After 90 wks in dorsal muscular tissues of adult 
sheep, there was no significant dissolution of the PHB/HV polymer, 50-60% of the initial 
weight still remaining. PHB/HV polymers are biodegradable materials, either by hydrolysis or 
implantation, but with extremely low dissolution or degradation rates. 


Poly(fi-hydroxybutyrate-fi-hydroxyvalerate) materials [PHB/HV] are members of the 
Polyhydroxyalkanoate [PHA] family, a group of new polymeric materials based on bacterial 
polyesters. PHA materials are intra-membranarly produced by prokaryotic cells as carbon and/or 
energy reserves. PHB and polypropylene have been often compared in terms of physico-chemical 
(melting point, crystallinity level) and tensile properties, however PHB/HV polymers are reported as 
environmentally-and/or-bio-degradable materials [1]. These biotechnological thermoplastics are of 
extreme interest as 100% compostable and ecologic plastics for packaging materials. 

Bacterial PHB and PHB/HV polyesters have been introduced and evaluated as biodegrada¬ 
ble, biocompatible and natural thermoplastics for medical or surgical devices. Drug delivery systems, 
sutures, bone plates and restorative absorbable devices have been proposed with PHB homopolymer 
[2,3], PHB membranes have been equally applied for guided gingival tissue regeneration and studied 
in vivo for thoraco-abdominal tissue reconstruction in animals [2,3,4], The PHB/HV copolymers 
have been reported as presenting improved processibility, tactility and accelerated biodegradation 
properties over the PHB homopolymer. They have been recently preferred for evaluations of 
biomedical implantable systems such as orthopaedic resorbable composites or tissue engineering 
substrates. In orthopaedic surgeiy, PHB/HV biopolymers have been considered for their piezo- 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

electrical properties, low and progressive rate of hydrolysis in comparison to other synthetic 
resorbable polymers and potential bioactivity in regard to the depolymerization into (3- 
hydroxybutyric acid (HB), a natural constituent of the human blood. 


Three commercial PHB/HV (HV 7, 14 & 22 %) copolymers were processed by dissolution 
in chloroform solutions (10 % w/v) and casting on the glass surface of petri dishes. The solvent was 
allowed to evaporate for at least 96 hrs, and all PHB/HV films were stored under anhydrous 
conditions at room temperature. The cytotoxic responses of the PHB/HV films were controlled 
according to the direct contact and agar diffusion cell culture tests [ASTM F813-83(88) and F895- 

In vitro degradation studies 

The accelerated degradation studies were performed in Phosphate Buffered Solutions (PBS), 
pH 7.4, at 70°C. The PHB/HV films were cut in rectangular-shaped specimens (70°C), and the 
sample size was measured using a micrometric caliper. The film specimens were then sealed in glass 
bottles with 50 ml of PBS solution. The accelerated hydrolysis of the PHB/HV samples was carried 
out in an oven (70°C), and the PBS solutions were checked daily to avoid excessive medium 
evaporation. To analy ze the hydrolytic degradation effects, samples of the PHB/HV specimens were 
taken at 0, 5, 10, 20, 40, 60, 80, 110, 140, 160 and 220 days (70°C). Once removed from the 
aqueous medium, the PBS excess was eliminated on absorbent paper, the specimens were washed in 
distilled water and 70% ethanol, air-dried at 40°C, and then stored under anhydrous conditions at 
room temperature. 

In vivo Implantations. 

The PHB/HV materials for implantations were sampled in rectangular film specimens. The 
specimen sizes (width x length x thickness) were measured using a digimatic micrometnc caliper. 
The PHB/HV specimens were sterilized by Ethylene Oxide (EtO). The PHB/HV films were 
implanted in dorsal muscular tissues of adult sheep for 1, 5, 8, 11, 15, 20, 25, 30, 52, 66, 70 and 90 
weeks. The animals were anesthezied using Acepromazine R and Thiopenthal R 5.0%, and maintained 
during surgery by Halothane R 1.5%. The dorsal region was incised on each side of the spine, three 
independant pouches were created in the muscular bundles along and beside the spine. The PHB/HV 
film was disposed in the pouch and attached to the muscles using non-resorbable Prolene R sutures. 
The muscular pounches and dorsal incisions were closed with absorbable Dexon R sutures. The sheep 
were admisnistrated analgesics and antibiotics (Demerol R /Permlong R ) for a two-day post-op period. 
All animals were sacrificed by Euthanyl R overdose and the muscular tissues surrounding the 
PHB/HV specimens were excised and freshly maintained in saline solutions. The muscular tissues 
were immediately dissected to extract the PHB/HV samples. Samples of the muscular tissues and 
fibrous capsules around the polymeric films were excised and fixed in 10% formalin solutions. The 
membranes were histologically analyzed for the 1, 5, 11, 20, 66, 70 and 90 week periods by paraffin 
embedding, thin slides (2-4 mm) and Hematoxylin & Eosin (HE) staining. The muscular tissues 
remaining around the PHB/HV specimens were enzymatically digested using collagenase (37°C, 1 
week), then separated from the PHB/HV materials in a density gradient column with distilled water. 


All PHB/HV specimens were washed using distilled water and ethanol, air-dried at 40°C, then stored 
at room temperature. 

Biodegradation of PHB/HV polymers 

After being washed and dried, the degraded PHB/HV film samples were weighed precisely 
using an analytical balance (weight variation < 0.01%) and the weight loss was calculated as being 
(m-mynio (m<>: initial weight, nr weight of the film at the period t). The linear correlation of the 
weight loss data with the degradation time was analyzed using a Pearson Product-Moment 
Correlation analysis with a 95% confidence level The degradation effects on the degraded PHB/HV 
specimens were monitored using Gel Permeation Chromatography (GPC, Refractive Index), 
Nuclear Magnetic Resonance ( 13 C-NMR), and Differential Scanning Calorimetry (DSC). The 
PHB/HV surfaces were also studied by Scanning Electron Microscopy (SEM). 


The cytotoxic responses to PHB/HV solid materials were quoted as mild reactions with few 
cell lysis and no lysis propagation outside the surface area beneath the solid samples. The preliminary 
toxicity tests suggest that the solid PHB/HV materials as processed in the laboratory elicit mild to 
moderate cellular reactions in vitro, however the potential cytotoxicity of PHB/HV extracts vary 
with the medium, weight or surface/volume ratio, time or temperature. 

Table I; The weight loss rates of PHB/HV polymers dining the in vitro and in vivo degradations. 




(Initial dry weight) 

(100 g) 


_ (i<x>g) 

K*i (in vitro, 70°C) 
(milligrams per day) 




K„i (in vivo) 
(milligrams per day) 




Table H: The number-average molecular weight (IVE) and Mv/M n ratio verus the hydrolysis time. 


t (days) 

MJM n 




in vitro (70°C) 












Hydrolysis time t (days) 

Figure 1: The mole HV content (%) of the PHB/HV polymers versus the hydrolysis time. 

Table III: The thermal parameters of the PHB/22%HV polymer versus the hydrolysis time. 


t (days) 

T m (°C) 

AH m (J/g) 




in vitro (70°C) 













In vitro: 

The mean weights of the PHB/HV materials are 103.5 (HV7%), 88.7 (HV14%) and 101.9 
mg (HV22%). These materials were sampled in 40mm x 15mm rectangularly-shaped specimens. 

Whatever the copolymer composition, all PHB/HV films clearly show pronounced signs of 
degradation by aqueous hydrolysis at high temperature. The weight loss rates [K^ (in vitro, 70°C)] 
were calculated from the correlation analysis (Pearson’s correlation coefficients > 0.92), and vary 
from 170.2 (HV7%) to 283.1 mg/day (HV22%) [Table I]. Undoubtedly, they tend to increase 
during the in vitro accelerated hydrolysis with the [HVJ content in the copolymers. 

The number-average molecular weights, M n , (R.I.), and M„/M w ratio rapidly decrease with 
the hydrolysis time [Table Df|. The kd hydrolysis rate was calculated, as previously defined by Doi et 
al (1990) for PHB/HV polymers, from the hydrolysis time, the number-average degree of 
polymerization at time zero [DP n (0)] and the ratio of the number-average molecular weights at time 
zero and time t [Mn(0)/M„(t)] [5]. The molecular weight-based degradation rate kd was equal to 
3.58xl0' 5 day" 1 for aqueous hydrolysis of the PHB/22%HV polymer at 70°C (Pearson’s Moment 
correlation coefficient: 0.90). The mole HV contents, defined as being the ratio of the NMR band 
intensities of the Methyl groups of the HB and HV units, clearly decrease at 70°C with the 
hydrolysis time. These drops in HV content seem not to be linearly correlated with the time, but 
were dependant upon the initial HV content (nominal) [Figure 1]. The crystallinity level of 
PHB/22%HV materials clearly increases during the hydrolytic degradation ranging from 28.3% 
(t=0) to 68.4% (220 days), when calculated with using a heat of melting for the crystalline region 


about 146 J/g (pure PHB) [Table HI]. Those results demonstrate that PHB/HV polymers 
chemically degrade in vitro by accelerated aqueous hydrolysis. The attacks take place in all polymer 
molecules, however the M w molecular weights decreased slightly fester than the molecular 
weights. The molecular weight distribution was still unimodal and tend to progressively narrow with 
the hydrolysis time (Mu/M,,). The quite linear relationship between the average number of bond 
cleavages per original polymer molecule and the hydrolysis time supports the feet that the decrease 
in M n can be associated with randon chain scissions [5], 

In vivo 

The mean weight of the PHB/HV samples was about 284+/-1405 g and the mean surface 
area about 2715+/-591 mm 2 (N=30). The specimens rapidly broke up in vivo at the suture 
attachments and started to fold up on themselves and to be infiltrated by tissues. 

No abscess formation and tissue necrosis was observed in the vicinity of the PHB/HV 
specimens. Acute inflammatory reactions with numerous macrophages, neutrophils, lymphocytes 
and fibrocytes were observed at 1 week in the capsule at the interface between the muscular tissues 
and implants. These reactions were quite less intense at 11 weeks for all implants and the density of 
inflammatory cells was much lower, but inflammatory cells (lymphocytes) still remain in the capsule 
and muscular tissues. The fibrous capsules were dense and well-vascularized, organized early with 
crimped oriented fibers and fibroblastic cells. Those connective tissue cells were aligned in parallel 
with the implant surface, especially at the interfeces with the PHB/HV films . Some muscular bundles 
were detached from the adjacent muscular tissues, and surrounded by the fibrous tissues. The fibrous 
tissues, which represented the interface between the dorsal muscles and the PHB/HV film, were 
subdivided into secondary fibrous membranes going deeply through the dense muscular bundles. A 
great number of isolated or grouped fetty cells were observed in the capsule, at the interface 
between the capsule and muscular tissues, and in the adjacent muscles for long-term implantation 
periods (70 and 90 wks). There were few differences in terms of capsules and tissue characteristics 
or cellular activity between the three PHB/HV polymers. 

During the implantations, the PHB/HV specimens have presented irregular and weak values 
in term of dry weight loss, and most of the PHB/HV materials were still present at 70 and 90 wks, 
suggesting very low degradation and dissolution rates in vivo. The data analysis showed that the 
weight loss rates [K^ {in vivo)] vaiy from 38.7 to 44.1 mg/day, increasing slightly with the HV 
content. Some compositional and physico-chemical changes were detected, more particularly on the 
PHB/22%HV materials which appear to be the most sensitive to the in vivo degradation: the 
number-average molecular weight (decreasing Mn), polydispersity (increasing Mv/Mn), mole HV 
content (decreasing [HV]) and heat of fusion (increasing AH™) were clearly affected in vivo , and 
showed either increasing or decreasing tendencies with the implantation time. However, the 
degradation rates in vivo were much slower than those associated with the in vitro accelerated 
hydrolysis. The degradation rate ka was determined as previously described for the hydrolytic 
degradation and found to be about 8.98xl0' 7 day' 1 for the PHB/22%HV polymer (Pearson’s 
Moment correlation coefficient: 0.99). There were not clear signs of surface degradation activity on 
the implanted polymeric films, and systematic examinations demonstrated that there was no clear 
difference in terms of texture or microporosity between the virgin and degraded surfaces. 

Bacterial PHB/HV polyesters chemically degrade when submitted to hydrolytic attacks in 
physiological media, however they provide few significant signs of degradation at the physiological 
temperature. On the other hand, the use of high medium temperatures (50, 60 70°C and up) 
improves the ability to screen the hydrolytic degradation effects of such materials. At 70°C, the 
PHB/HV polymers undergo clear physico-chemical or compositional changes in terms of number- 


and weight-average molecular weights, mole [HV] fraction, melting point and heat of fusion. The 
nominal copolymer composition and initial average molecular weights have a strong effect on the 
degradation kinetics. In vivo, those physico-chemical and compositional changes are not detected 
(HV 7 & 14%) or are very weak (HV 22%). At 70 and 90 wks implantation, no clear changes in 
terms of[HV] content, melting temperature or heat of fusion are observed. However, if the apparent 
mechanical performances rapidly change during the hydrolysis, the weight loss or dissolution rate 
appears to be extremely limited in vitro as well as in vivo. The hydrolysis rate kd of a PHB/22%HV 
polymer is about 3.6xl0' 5 day' 1 in vitro in PBS solutions at 70°C, and about 9.0xl0' 7 day 1 in vivo m 
muscular tissues of sheep. The hydrolytic degradation of the PHB/HV films may be associated with 
random chain scissions throughout the whole polymer matrix. 


The PHB/HV copolymers are chemically sensitive to aqueous hydrolysis, but provide low 
degradation rates in vitro, even at high temperatures (accelerated conditions). At 70°C, the 
degradation of PHB/HV films in PBS solutions, pH 7.4, were quantified by weight loss rates of 
about 0.15-0.30 mg/day. The hydrolyzed PHB/HV materials rapidly show significant changes in 
terms of n umb er-average molecular weights and M^/M* ratio, mole HV content and thermal 
parameters (melting points, heat of fusion and crystallinity levels). Our observations suggest that 
PHB/HV materials also degrade in vivo, but present very low dissolution rates, approximately seven 
times slower than those observed for the accelerated hydrolysis. The in vivo degradation effects on 
PHB/HV polymers probably require very long implantation periods (years). The cytotoxic responses 
toward PHB/HV solid materials or extracts in MEM (Minimum Essential Medium), saline or 
DMSO media generally were quoted as mild to moderate. In vivo, all short-term (1-11 wks) tissue 
reactions were characterized by the presence of acute inflammation with the presence of numerous 
inflammatory cells. For implantations of 11 weeks and up, the fibrous capsules were variable in 
thickness, but were dense well-vascularized tissues with oriented crimped collagen fibers and 


1. H. Brandi, R. A. Gross, R W. Lenz, R. C. Fuller. In: Advances in Biochemical 
Engineering. Biotechnology . Vol. 41, edited by A. Fiechter (Springer-Verlag Berlin Heidelberg, 
1990) pp77-93. 

2. T. Malm, S. Bowald, S. Karacagil, A. Bylock, C. Busch. Scand. J. Thor. Cardiovasc. 
Surg. 26, 9 (1992). 

3. T. Malm, S. Bowald, A. Bylock, T. Saldeen, C. Busch. Scand. J. Thor. Cardiovasc. 
Surg. 26, 15 (1992). 

4. N.R Boeree, J. Dove, J.J. Cooper, J. Knowles, G.W. Hastings. Biomaterials 14, 793 

5. Y. Doi, Y. Kanesawa, M. Kunioka, T. Saito. Macromolecules 23, 26 (1990). 

6. S. Gogolewski, M. Jovanovic, S.M. Perren, J.G. Dillon, M.K. Hughes. J. of Biomedical 
Materials Research, 27, 1135 (1993). 

This article also appears in Mat. Res. Soc. Symp. Proc. Vol 394 


Part III 

Adhesion and Interface Durability 



National Institute of Standards and Technology, Gaithersburg, MD 20899 


Water is often the main cause of adhesion loss of a polymer coating/substrate system. 
The buildup of the interfacial water layer and the loss of adhesion of polymer-coated siliceous 
substrates exposed to liquid water has been investigated. The thickness of the interfacial water 
layer was measured on epoxy-coated Si0 2 -Si prisms using FTIR-multiple internal reflection 
(FTIR-MIR) spectroscopy. Adhesion loss on flat siliceous substrates was determined by a wet 
peel test on epoxy-coated Si0 2 -Si wafers and adhesion loss of composites was obtained by 
measuring the interlaminar shear strengths of epoxy/E-glass fiber composites. Both untreated and 
0.1% silane-treated substrates were used. Little water was observed at the interface of the silane- 
treated samples but about 10 monolayers of water have accumulated at the interface of the 
untreated samples after 100 h of exposure to 24 °C water. Untreated, flat substrates lost most of 
their bonding strengths within 75 h of exposure but silane-treated specimens retained 80% of their 
adhesion after 600 h of exposure to 24 °C water. Adhesion loss of untreated composites immersed 
in 60 °C water was greater than that of treated samples; however, the rate of loss of both silane- 
treated and untreated composites was much lower than that of flat substrates. Adhesion loss was 
found to follow the same trend as interfacial water buildup. 


The detrimental effects of water and water vapor on the adhesion of polymer-coated 
metals 1 , adhesive bondings 2 , polymer/fiber composites 3 and asphalt pavements 4 are well 
documented. However, there has not been a study that examined the relationship between the 
water layer at a polymer coating/substrate interface and the adhesion loss. The main reason for 
this is the lack of quantitative information on the interfacial water layer. A recent development 
of a technique to quantify the water layer at the polymer coating/substrate interface 5 has made 
it possible to study the linkage between these two phenomena. In the present study, the thickness 
of the interfacial water layer and the loss of adhesion of polymer coating/siliceous substrate 
systems as a function of time of exposure to water has been investigated. The interfacial water 
layer was measured on specimens of coatings applied on flat substrates, while adhesion loss was 
determined on both flat substrates and unidirectional polymer/E-glass fiber composites. 

Materials and Specimen Preparations 

For measuring water at the polymer coating/siliceous substrate interface, specimens of 
an epoxy coating applied on untreated and silane-treated, 50x10x3 mm spectroscopic grade, Si 
parallelogram prisms were prepared. For adhesion loss measurements on flat siliceous substrates, 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

specimens of coatings on untreated and silane-treated, 100-mm diameter Si wafers were used. 
Adhesion loss of the composites was determined from unsized, untreated and silane-treated 16+2 
^m diameter E-glass fibers impregnated with an epoxy resin. 

The polymer coating was a stoichiometric mixture of a low molecular weight, diglycidyl 
ether of bisphenol A (epoxide equivalent weight = 189 g) and a polyethertriamine (amine 
equivalent = 83 g) curing agent. This epoxy coating was used for both flat substrates and 
composites. Si prisms and wafers were cleaned with acetone followed by methanol and dried with 
hot air before use. The surfaces of these substrates had a native Si0 2 layer of about 2.5 nm thick, 
as determined by an ellipsometer. Under ambient conditions (24 °C and 45% relative humidity), 
they should be covered with silanol groups (SiOH) and adsorb water 6 . The hydroxylated, Si0 2 - 
covered Si (Si0 2 -Si) prisms and wafers, and E-glass fibers are designated as the siliceous 
substrates in this study. Silane-treated surfaces were prepared by immersing cleaned Si prisms, 
Si wafers, and E-glass fibers for 30 minutes in an acidified (pH=4) water solution containing 
0.1% aminoethylaminopropyltrimethoxysilane. The treated substrates were dried for 10 minutes 
at 110°C before use. After mixing thoroughly the two components and letting them stand for 30 
minutes at ambient conditions, the coating was applied on the prisms and wafers using the 
drawdown technique, similar to that described in Reference 5. All coated samples were cured 
at 80 °C for four hours followed by two weeks at ambient conditions. The thicknesses of the 
coatings on the prisms and wafers (determined at the conclusion of the experiment) were in the 
130-150 fim range. The quality of all coated specimens was good and no visible pinholes or air 
bubbles were observed (with a naked eye) on the sample surfaces. Tg of the cured epoxy films 
was 83 °C (by differential scanning calorimetry, DSC). 

Unidirectional composites were prepared by impregnating E-glass fibers with the same 
epoxy in between two transparent polyethylene (PE) sheets. Unsized, untreated and silane-treated 
fibers approximately 250 mm long were carefully laid unidirectionally on one PE sheet. After 
fixing one end of the fibers onto the PE sheet (by an adhesive tape), the epoxy was poured 
liberally on the fixed end of the fibers. After placing the other PE sheet over the resin-soaked 
fibers, the resin was spread by a glass plate placed on the PE sheet. The resin was spread along 
the fiber direction repeatedly until the fibers and the resin formed a transparent mat. The 
impregnated fibers were cut to size and placed in open-ended, aluminum molds having an inside 
dimension of 200x8x6.5 mm. The molds were placed in an autoclave for three hours at 80 °C 
and under a pressure of 1.4 MPa above atmospheric pressure, followed by post curing at 100 °C 
for 12 h. After removing the composites from the molds and conditioning them at ambient 
conditions for two weeks, they were polished and cut to a specimen size of 38x7.9x6.35 mm for 
exposures and testings. All specimens were transparent and there was no evidence of visible air 
voids, as observed by the naked eye. The fiber fraction of the composites was 40 % (determined 
by thermogravimetric analysis) and Tg of the epoxy in the composites was 84 °C (by DSC). 

Measurement of Water at the Polymer Coating/Siliceous Substrate Int erface 

Specimen configuration and Fourier transform infrared-multiple internal reflection (FTIR- 
MIR) procedure used for the measurement of water at the polymer coating/siliceous substrate 
interface are similar to those described in Reference 5. Briefly, a water chamber was attached 
to each cured, coated prism. The specimen with the water chamber attached was placed vertically 
in an attenuated total reflection (ATR) accessory holder and measurement of water at the 
interface was carried out using an FTIR spectrometer with a variable ATR accessory. After 
filling the chamber with distilled water at 24 °C, FTIR-MIR spectra were taken automatically 
every 15 minutes without disturbance to the specimens or the spectrometer. For the specimen 


configuration employed in this study, the only pathway for water migration from the environment 
to the interface is through the coating film thickness. All spectra were the co-additions of 32 
scans and taken at a resolution of 4 cm 1 . Unpolarized light at a 45° incident angle and purged 
dry air were used. 

The thickness of the water layer, /, at the polymer coating/siliceous substrate interface 
was determined using the following expression 5 : 

I = 


( 1 ) 

where A is the FTIR-MIR absorbance; A* is the maximum infrared absorbance of water 
(provided by FTIR-MIR analysis of water on a coating-free siliceous substrate); c w is the fraction 
of water sorbed in the coating within the FTIR-MIR probing depth (extrapolated from water 
uptake in coating); d pw and d pQ are the penetration depths of the evanescent wave (calculated from 
internal reflection theory) in water and coating, respectively. For a Si substrate, d pw is 0.22 pm 
and d pc is 0.24 /*m. Equation 1 is still valid for the case where the water layer at the 
coating/substrate interface is not continuous, e.g., discrete droplets. 

Assuming water is uniformly distributed over the entire surface area of the specimen, the 
amount of water at the coating/substrate interface is given by 

Qi = 1 a P 

( 2 ) 

where a is the area in contact with water and p is the density of water at the interface. 

Measurement of Adhesion Loss 

The adhesion loss of epoxy-coated Si0 2 -Si wafers as a function of exposure to water at 
24 °C was measured using an improved version of a wet, 90° peel adhesion tester described in 
Reference 7. The apparatus consists of a linear bearing slider fixed to a computer controlled 
universal testing machine fitted with a 2.0 kg load cell. The improvement arose from lighter 
materials used and a low-friction moving slider. This resulted in a friction reduction from 200 
g to 120 g. At each pre-specified exposure time, epoxy-coated wafers were removed from the 
test chamber and each wafer was immediately incised with a razor into five 12x65-mm 
specimens. Each specimen was carefully peeled from the substrate, leaving a 35-mm length of 
the specimen unpeeled. The wafers were then positioned in the test apparatus and peeled at a rate 
of 20 mm/min. All peel tests were conducted at room temperature and took approximately 30 
minutes for each wafer from the moment it was taken out from the test solution. Ten specimens 
on two wafers were tested and the results averaged. 

Adhesion loss of the composites was determined from the interlaminar shear strengths 
(ILSS), which is a measure of the adhesion between the resin and the fiber. ILSS values were 
calculated from the peak load of the short-beam, three-point bending test (ASTM D2344): 

ILSS = 3F/4bd (3) 


where F is the peak load in kN, and b and d are the width and thickness of the specimen in m. 
Specimens of 38x7.9x6.35 mm immersed in distilled water at 60 °C were removed from the 
container and tested at pre-specified time intervals. The specimen was supported by a 25-mm 
span and tested at a displacement rate of 0.5 mm/min. To ensure accuracy, the calibration of 
testing machine was verified at each testing period. Except for two cases where only two 
specimens were used, the results were the average of three or four specimens. 


In order to provide quantitative information on water at the coating/substrate interface, 
FTIR-MIR difference spectra were acquired. This was done by subtracting spectra taken before 
exposure from those collected at different exposure times. Figures la and lb display difference 
spectra in the 2800-3800 cm 1 region (water OH stretching) of epoxy-coated untreated and 0.1 
% silane-treated Si0 2 ~Si prims exposed to water for several representative times. The high signal- 
to-noise ratio of these spectra is due to the low concentration of detected water in these samples. 
The bands in the 3000-3650 cm 1 region are due to OH stretching mode of water, as verified 
previously 8 . The band peaking near 3400 cm' 1 was chosen for the analysis, and its intensity 
changes as a function of exposure time are presented in Figure 2 for untreated and silane-treated 
specimens. The intensity of the water bands of both untreated and silane-treated samples 
increased with exposure time up to approximately 100 h then leveled off. These changes are the 
result of water entering the coating/substrate interfacial region and interacting with the evanescent 
wave 8 . Further, it can also be seen that the intensity of water detected for the untreated specimen 
is higher than that for the silane-treated one. 

Wavenumber, cnr 1 Wavenumber, cm’ 1 

Figure 1. FTIR-MIR difference spectra of epoxy-coated untreated (a) and 0.1% silane- 
treated (b) Si0 2 -Si prisms exposed to 24°C water for several time intervals. 

Substituting FTIR-MIR absorbance data (A) given in Figure 2, and c w , A*,, d pc and d pw 
values, which are taken from Reference 5, into Equations 1 and 2, thickness and amount of the 
water layer at the interface for epoxy coated untreated and 0.1 % silane-treated Si0 2 -Si substrate 
are determined. The results are given in Figure 3, which shows that essentially no or little water 
had entered the interface of the silane-treated specimens, but about 10 monolayers (one 
monolayer of water is about 0.3 nm) had built up at the interface of the untreated samples after 
100 h exposure. The results on water at the interface for the silane-treated specimens suggest that 
the bond strengths between the treated surface and epoxy resin were stronger than those between 
the treated surface and water. Therefore, water molecules could not displace these bonds and thus 


no or little water entered the interface. This result, which has been confirmed by a duplicate 
specimen, is supported by the adhesion loss data given in Figure 4. It is also consistent with the 
chemical bonding theory, which has been proposed as an explanation for the interfacial 
reinforcement mechanism and enhanced hydrolytic stability of composites made with silane- 
treated fibers 9 . It is noted that water is a weakly-sorbed species: the magnitude of water-oxide 
bonds are in the 40-65 kJ/mol range 10 and thus incapable of replacing stronger chemical bonds 
between the silane and the siliceous surface or between the silane and the epoxy resin. 

It is important to point out here that the results of Figure 3 provide the first evidence that 
Equation 1 is valid for calculating water at the polymer/substrate interface. This equation, which 
was derived from the internal reflection theory, was based on a two-layer model consisting of 
the water layer at the interface and water sorbed in the coating film within the FTIR-MIR 
probing depth. If no water was at the interface, the total water detected (e.g. FTIR-MIR 
absorbance given in Figure 2) is due to only water in the polymer layer near the interface. This 
is the case for the coated, silane-treated specimens. These specimens had a hydrolytically stable 
interface, as evidenced by the small reduction in the adhesion after prolonged exposure (Figure 
4) and corroborated by the well-established mechanism of silane coupling agents on glass fibers. 
Such an interface should allow little or no water entering this region. This is consistent with the 
results calculated from Equation 1 (Figure 3). 

Figure 2. Intensity change of water band with time exposed to 24°C water for epoxy coated, 
untreated and 0.1% silane-treated Si0 2 -Si prisms (each dot represents a data point). 

Figure 3. Thickness and amount of the interfacial water layer of epoxy-coated untreated and 
0.1% silane-treated Si0 2 -Si prisms exposed to 24°C water (each dot represents a data point). 


0.0 -i- T-1 - 1 - 1 - 1- 1 

0 100 200 300 400 500 600 

Immersion Time, hours 

Figure 4. Peel adhesion of epoxy-coated untreated and 0.1% silane-treated Si0 2 -Si wafers 
exposed to 24 °C water (bars on curves indicate one standard deviation). 

Figure 4 presents adhesion changes as a function of time exposed to 24 °C water for 
epoxy-coated untreated and 0.1% silane-treated Si0 2 -Si wafers. Untreated specimens lost most 
of their adhesion within 75 h of exposure, while silane-treated samples retained 80% of their 
adhesion even after 600 h immersion. These results, which are in good agreement with interfacial 
water data given in Figure 3, provide the first experimental evidence to show that the gradual 
buildup of a water layer at the interface is the main cause of adhesion loss of polymer coatings 
on an untreated siliceous substrate. The results of Figures 3 and 4 again illustrate the great 
affinity of untreated siliceous surfaces for water and that these surfaces must be modified to 
improve the durability of a polymer/siliceous substrate system. 

Figure 5 displays ILSS data of composites prepared with untreated and silane-treated E- 
glass fibers immersed in distilled water at 60 °C for different times. Although at a lesser extent, 
the results follow the same trend with the adhesion loss of flat Si0 2 -Si substrates; that is, 
composites of untreated fibers lost more of their shear strengths and at a faster rate than the 
composites of treated fibers did. Untreated and treated composites lost 55 and 23%, respectively, 
of their original shear strengths after three-month immersion. This compares with 83% adhesion 
loss after 75 h and 25 % loss after 600 h exposed to water at 24 °C for untreated and treated flat 
substrates, respectively. The results also show that the larger amounts of water at the 
polymer/substrate interface corresponded to the greater ILSS losses of the composites, suggesting 
that water at the resin/fiber interface may also play a role in the adhesion loss of composites. 

Figure 5. ILSS as a function of time in 60 °C water of untreated and 0.1 % silane-treated 
epoxy/E-glass fiber composites (bars on curves indicate one standard deviation). 


The results of this study and other data for solvent-free organic coatings on siliceous 
substrates 11 indicate that larger amounts of water at the coating/substrate interface generally 
correspond with greater loss of adhesion. Additional data are needed to determine more precisely 
the range of the interfacial water layer corresponding with the range of the adhesion loss. If a 
relationship between adhesion loss and interfacial water layer exists, FTIR-MIR technique could 
be useful for estimating the adhesion loss at the molecular level and for predicting the durability 
of polymer/siliceous substrate systems. 


Water is often the main cause of adhesion loss of a polymer coating/substrate system. 
This study investigated the accumulation of water at the interface, and adhesion loss of epoxy 
coatings on flat siliceous substrates and of epoxy/E-glass fiber composites. Both untreated and 
0.1 % silane-treated flat substrates and fibers were included. Silane treatment effectively 
prevented water from entering the interface but a water layer of about 3 nm thick has built up 
at the interface of the untreated Si0 2 -Si specimens after 100 h exposure to 24 °C water. 
Untreated, flat substrates lost most of their bonding strengths within 75 h exposure but silane- 
treated specimens retained 80% of their adhesion after 600 h exposure. Adhesion loss of 
untreated composites immersed in 60 °C water was greater than that of treated samples; however, 
the rate of loss of both silane-treated and untreated composites was much lower than that of flat 
substrates. Adhesion loss of both flat siliceous substrates and composites was found to follow the 
same trend with interfacial water buildup. If further analysis confirms that a relationship between 
the adhesion loss and the interfacial water laydr exists, the FTIR-MIR technique could be useful 
for estimating the adhesion loss at the molecular level and for predicting the durability of 
polymer coating/siliceous substrate systems. 


We thank Dr. David Dwight of Owens-Corning for providing the glass fibers, and Drs. 
Carol Schutte and Wolfgand Haller for their valuable comments on experiments with composites, 
Ned Embree for testing the composites, and Bill Roberts for DSC analysis. 


I. H. Leidheiser, Jr. and W. Funke, J. Oil & Colour Chemists' Assoc. 70, 121 (1987). 

2 A.J. Kinloch (ed.) Durability of Structural Adhesives. (Applied Sci., N.Y., 1983). 

3. C.L. Schutte, Materials Sci. and Eng. R13, 265 (1994). 

4. M.A. Taylor and N.P. Khosla, Transportation Research Record 911, 150 (1983). 

5. T. Nguyen, D. Bentz, and W.E. Byrd, J. Coatings Technol. 66, no. 834, 39 (1994). 

6. T. Nguyen, E. Byrd, and D. Bentz, J. Adhesion 48, 169 (1995). 

7. D, Alsheh, T. Nguyen, and J.W. Martin, Proc. Adhesion Society Conference, February, 
1994, p. 209. 

8 T. Nguyen, W.E. Byrd, and C. Lin, J. Adhesion Sci. Technol. 5, 697 (1991). 

9. For brief reviews on this subject, see Reference 3 and J.L. Koenig and H. Emadipour, 

Polymer Composites 6, 142 (1985). 

10. P.A. Thiel and T.E. Madey, Surface Sci. Rep. 7, 211 (1987). 

II. T. Nguyen, D. Alsheh, D. Bentz, and W.E. Byrd, Proc. Adhesion Society Conference, 
February, 1995, p. 252. 



M. Libera,* W. Zukas,** S. Wentworth,** and A. Patel* 

* Stevens Institute of Technology, Hoboken, NJ 07030 

** U.S. Army Materials Tech. Laboratory, Watertown, MA 02172 


It is now recognized that there is a region at the epoxy/adherend interface known as the 
interphase whose chemistry and structure are different from those of bulk epoxy. There is, 
however, no adequate understanding of its microstructure. This paper describes differential 
scanning calorimetry (DSC) and transmission electron microscopy (TEM) studies of the 
interphase between an aromatic amine cured epoxy and alumina/oxidized-aluminum surfaces. 

DSC results show dramatic differences in resin-cure behavior in the presence of particulate 
alumina. TEM results on microtomed cross-sectional specimens of anodized aluminum wire 
embedded in epoxy show regions of incomplete epoxy infiltration and interfacial stress. Ru0 4 
staining combined with high-angle annular-dark-field STEM imaging indicates that there are 
structural variations within the bulk epoxy at lengths of ~10-30nm. The magnitude of these 
fluctuations decreases in the near the adherend interface, and there is a simultaneous variation in 
the average epoxy structure. A plausible interpretation of these observations is that the interphase 
region suffers a variation in curing-agent concentration of unknown magnitude and there is a 
higher concentration of homopolymerized epoxy there relative to the bulk. 


Epoxies are particularly important as the matrix in fiber-reinforced composites, as durable 
coatings for inorganic materials, and as adhesives. These applications involve one or more 
interfaces between the epoxy and an adherend. Several researchers (1-4) have shown that there is 
a region at the epoxy/adherend interface known as the interphase whose chemistry and structure 
are different from that of bulk epoxy. This interphase can be critically significant. The ability of 
an epoxy matrix to transfer stress to a strong adherend in, for example, a fiber-reinforced 
composite is ultimately controlled by the interphase. The interphase has been studied by various 
optical spectroscopies (5-9), TEM (10), and XPS(11). Previous calorimetric studies by Zukas, 
Wentworth, et al. (12) using particulate alumina as a model adherend surface has shown that 
reactions involving only the epoxy become significant in the interphase for stoichiometric 
formulations of resins cured with aromatic amines. Further work (13) indicated that the alumina 
clearly leads to an interphase structure that is different from the bulk and also dependent on the 
cure history and water content of the system. Despite this progress, however, there is not yet an 
understanding of the interphase microstructure. This work presents first results using 
transmission electron-optical techniques coupled with DSC to address what specific 
microstructural features might be found in the interphase. 


This research used epoxy resins based on diglycidyl ether of bisphenol A (DGEBA) including 
DER (Dow) and Epon (Shell) resins. Several primary aromatic amine curing agents were 
investigated including: diaminodiphenyl methane (DDM); 3,3’-diaminodiphenyl sulfone (33DDS); 
4,4'-diaminodiphenyl sulfone (44DDS); and meta-phenylene diamine (mPDA). 


Mat. Res. Soc. Symp. Proc.Vol. 385 ® 1995 Materials Research Society 

DSC analyses were performed with a Perkin-Elmer DSC-2C calorimeter using 5-50mg samples 
encapsulated in stainless steel pans. Cure exotherms were obtained using dynamic scans at 
constant heating rates between 0.5 to 40°C/min. Neat resin formulations were prepared by mixing 
the epoxy resin with curing agent at various stoichiometries at room temperature. Similar samples 
which were filled with alumina particulate used Brockmann I activity neutral aluminum oxide 
(Aldrich, surface area =155m 2 /g, pore size = 5.8nm at 150mesh). Filled samples were prepared 
by first dissolving the curing agent in the neat resin at elevated temperature, cooling to room 
temperature, then adding the alumina. Pre-reaction of the formulation before the calorimetry 
experiments was minimized. 

Transmission electron microscopy was done using a Philips CM20 FEG TEM/STEM in both 
traditional bright-field TEM mode and high-angle annular-dark-field (HAADF) STEM mode (14). 
Thin specimens were prepared by ultramicrotomy, supported on copper TEM grids with or 
without holey-carbon films, and stained. Two different staining procedures were used: (i) a 
several-step uranyl-acetate protocol developed by Di Filippo et al (15); and (ii) exposure to Ru0 4 
vapor (room temperature, 15min). Specimens were cut from tensile bars prepared by Prof. L.T. 
Drzal. These consisted of a single 125pm diameter anodized Al wire embedded in epoxy. The 
wire was anodized using phosphoric acid in general accordance with SAE Aerospace 
Recommended Practice 1524 REV A. The treated wire was dried and then encapsulated in Epon 
828 cured with mPDA. This procedure produced a single wire axially aligned in a dogbone 
tensile coupon having a gauge section of 3mm x 1.5mm x 25mm (fig. 1). 

Figure 1 - Schematic illustration of the tensile specimen with a single axial anodized Al wire from 
which thin sections were cut for TEM characterization of the Al/epoxy interphase region. 


The general cure exotherm observed for a dynamic DSC scan of a neat resin mixture is a single 
exotherm (12, 13) proportional to the amount of amine present. This indicates that epoxy/amine 
reactions take place during dynamic cures of these resins. Measured experimental features such 
as the fact that there is only a single isotherm, the exotherm peak temperature, and the enthalpy of 
reaction are consistent with other observations reported in the literature on similar epoxy systems 
(16-19). Figure 2a shows a typical result. 

Filling the epoxy samples with activated alumina particulate has a significant effect on the DSC 
exotherms (figure 2b). Whereas the unfilled (neat) resin shows only a single exotherm, the filled 
resin shows three: two low-temperature exotherms and an exotherm somewhat lower in 
temperature than the unfilled exotherm. The two additional low-temperature peaks are 


attributed to the action of the activated 
alumina. It appears to be acting as a catalyst 
for amine-epoxy reactions and as an initiator 
for epoxy-epoxy homo-polymerization 
reactions. The highest temperature 
exotherm in the unfilled neat resin is shifted 
to lower temperatures in the filled samples. 
This exotherm has been attributed to 
reactions involving excess epoxy (12,13). 


Bright-field TEM images of uranyl acetate 
stained cross sections of the dogbone 
specimens are shown in figure 3. In addition 
to the fine-scale roughness induced by the 
anodized layer, there are burrs and other Al 
protuberances. There is also evidence of 
stress-induced wrinkling of the specimen 
near the interface (A), thickness striations 
from the microtomy (B), and incomplete 
infiltration of the epoxy (C). Staining using 
the Di Filippo uranyl-acetate protocol failed 
to show significant contrast within the epoxy 
both in bright-field TEM and HAADF 
STEM modes. 


Figure 2 - DSC scans for neat (upper) and 
alumina-filled (lower) epoxies cure exotherms. 

Figure 3 - Bright-field TEM images of the interface region between epoxy and an anodized 
aluminum wire. Uranyl acetate staining provides no electron-optical contrast within the epoxy. 


Ru-stained specimens display substantial contrast within the epoxy when imaged via HAADF 
STEM. This imaging method is very effective at resolving variations in average atomic number. 
Regions rich in Ru (atomic number 44) appear bright in HAADF images. Representative results 
characterizing the bulk epoxy well away from the A1 wire are presented in fig. 4. The contrast 
and brightness of the image are optimized to show the structure of the epoxy lying above one of 
the holes in the carbon support film. This image shows regions of light and dark contrast with 
fluctuations on a ~10-30nm length scale. There do not appear to be any obvious variations in the 
average HAADF signal scattered from the bulk epoxy. A similar microstructure was observed in 
several different areas of the bulk epoxy film. 

HAADF STEM images characterizing the Al/epoxy interface are different than those from the 
bulk. Fig. 5a is a low-magnification image where the contrast and brightness are optimized to 
show the general structure. The darker circles correspond to areas where the electron beam 
passes only through epoxy. Fig. 5b shows a high-magnification image collected from one such 
region in the interphase. Here the contrast is optimized to show the structure of the epoxy. 

These conditions necessarily wash out the image where there is aluminum or carbon film. As in 
fig. 4, the epoxy shows regions of modulated contrast on a length scale of ~10-30nm. There is 
also a general variation in the average contrast perpendicular to the oxide/epoxy interface. The 
HAADF signal was recorded using a storage oscilloscope interfaced to the annular-dark-field 
detector as described in reference (20). Figure 5c plots this signal as a function of distance from 
the oxide/epoxy interface. The interface is not sharp. The epoxy region begins in the vicinity of 
x=60-80nm. Beyond that the signal rises from a minimum to a maximum at approximately 
x=170nm. The average signal is then roughly constant. The magnitude of the fluctuations about 
the average signal also appear smaller in the near-interface region than in die region beyond 
x~170nm. Similar behavior was observed at other locations in this same film. 


The calorimetric data show clear evidence that activated alumina particulate has a significant 
effect on the epoxy cure behavior. At least two additional low-temperature exotherms appear in 
particulate-filled specimens which are absent in identical but unfilled specimens. The activated 
alumina appears to catalyze both the epoxy homopolymerization and the amine-epoxy 


Fig 5 - HAADF STEM images of the Ru0 4 -stained wire/epoxy interface region: (a) contrast and 
brightness optimized to show general structure; (b) optimized to show epoxy structure; (c) 
HAADF detector amplified voltage as function of distance from the wire/epoxy interface. 

reactions. Intuitively, one would expect any variations in the epoxy microstructure due to 
homopolymerization and other catalytic effects to occur in the vicinity of the adherend surface. 
The HAADF STEM images of Ru-stained epoxies, in fact, indicate that there is a Ru depletion 
close to the interface with anodize A1 wire and that the average Ru concentration increases to 
some relatively constant value approximately lOOnm away from the interface. These 
measurements also suggest that the rather uniform 10-30nm modulation of Ru concentration 
observed in bulk epoxy well away from the A1 wire have a smaller amplitude near the interface. 
The important issue yet to resolve concerns what chemical or structural feature characteristic of 
the epoxy is manifested by the measured Ru distribution. 

Ru staining involves covalent bonding where Ru0 4 oxidizes particular chemical moieties in the 
polymer (21). It is particularly reactive with aromatic groups. These are contained by both the 
DGEBA resin and the mPDA curing agent used in the Al-wire dogbone specimens. Whether or 
not Ru0 4 is more reactive with the epoxy or the curing agent due to the influence of other species 
in proximity to the aromatic rings in each case is not known. These variations, if present, are 


likely to be small and may be compensated for if sufficient staining time is provided to saturate all 
reactive sites. Concluding that the Ru concentration and, hence, the HAADF STEM intensity is 
proportional to the concentration of aromatic rings thus appears reasonable. The number of 
aromatic rings per unit volume is different for the epoxy resin and the aromatic amine curing 
agent. A spatial variation in the concentration of aromatic rings could be caused by a variation in 
the concentration of curing agent. If so, the observation of a Ru/aromatic-ring depletion within 
the interphase would be consistent with the conclusion from the DSC work that the adherend 
surface catalyzes the homopolymerization reaction for which a curing agent is not needed. The 
curing agent would naturally segregate away from the interface to some distance where the 
interface no longer exerts a catalytic influence on the cure. This argument that spatial variations 
in curing agent concentration occur in the interphase region is weakened by the fact the Ru0 4 
staining procedure does not distinguish effectively between the epoxy rein or the curing agent. 
Future work will concentrate on similar electron optical analysis using epoxy systems were 
greater staining selectivity can be had or by using holographic imaging which may enable direct 
imaging of the unstained microstructure. 


The authors are grateful to Professor L.T. Drzal (Michigan State University) for generously 
providing the Al-wire dogbone specimens, to Sutinder Behai (Exxon Research) for his help with 
the ultramicrotomy, and to Ginam Kim and Krisda Siangchaew (Stevens) for help with the 
staining and data manipulation. This research is supported by the Army Research Office and uses 
microscopy facilities provided by the NJ Commission on Science & Technology and the NSF. 


1. L. H. Sharpe, J. Adhesion 4, 51-64 (1972). 

2. L.H. Sharpe, The Science and Technology of Adhesive Bonding, Proc. 35th Sagamore Army 
Materials Conf., ed. L. Sharpe and S. Wentworth, Gordon and Breach, NY (1990). 

3. L.T. Drzal, in Epoxy Resins and Composites II, Springer-Verlag, Berlin (1986). 

4. J. Miller and H. Ishida, in Fundamentals of Adhesion , ed. L. Lee, Plenum, NY (1991). 

5. A. Garton et al., J. Poly. Sci. Tech. A26, 1377 (1988). 

6. A. Garton, J. Poly. Sci: Poly. Chem. Ed. 22, 1495 (1984). 

7. J. Nigro and H. Ishida, J. Appl. Poly. Sci. 38, 2191 (1989). 

8. D. Ondrus, FJ. Boerio, and K. Grannen, in ref 2. 

9. S.L. Tidrick and J.L. Koenig, ibid. 

10. J.S. Crompton, J. Materials Science 24, 1575-1581 (1989). 

11. R. Dillingham and F.J. Boerio, J. Adhesion 24, 315 (1987). 

12. W. Zukas, K. Craven, and S. Wentworth, J. Adhesion 33, 89 (1990). 

13. W. Zukas, M. Sennett, and S. Wentworth, Proc. 16th Ann. Mtg. Adhesion Soc. 187 (1993). 

14. S.J. Pennycook, Ultramicroscopy 30, 58-69 (1989). 

15. G. Di Filippo, J. Vander Sande, and D. Uhlmann, J. Appl. Poly. Sci. 35,485-505 (1988). 

16. C.C. Ricciardi and R.J.J. Williams, J. Appl. Poly. Sci. 32, 3445 (1986). 

17. J. Galy, A. Sabra, and J. Pascault, Poly. Engr. Sci. 26,1514 (1986). 

18. A.C. Grillet, J. Galy, J. Pascault, and J. Bardin, Polymer 30, 2094 (1989). 

19. J.M. Barton, J. Macromolecular Sci. Chem. A8, 25 (1974). 

20. M. Libera, J. Ott, and K. Siangchaew, submitted to Ultramicroscopy. 

21. J.S. Treat, J.L Scheinbeim, and P.R. Couchman, Macromolecules 16, 589-598 (1983). 




Thermal Analysis of Materials Processing Laboratory, Mechanical Engineering Department, Tufts 
University, Medford, MA 


The adhesion and thermal properties of optical ceramic structures bonded by a polymer 
interlayer are explored. The mechanical, adhesion and optical properties of the heterostructure are 
dependent on the thicloiess of the bond and on the residual thermal stresses developed during the 
bonding process. The thermomechanical properties of a bonded structure over a range of 
temperatures are investigated, focusing on the thermal expansion and operating temperature limits 
of the polymer bond. Hie temperature and stress histories during manufacturing are determined 
through both numerical modeling and experimental analysis. The effect of stress relaxation and 
initial stresses on the behavior of the bonding material are examined for different processing 
conditions. The bond material relaxation time constant, activation energy, viscosity, and shear 
modulus are approximated from observation of the temperature-dependent behavior of the 


When fabricating complex layered devices for the optical or microelectronics industry, 
predicting existing thermal strains in the structure at every point during production and use is 
crucial. Thermal strains in layered devices bonded together by an organic layer, polymer or wax, 
are manifested as curvature changes in the structure similar to the curvature changes thin films 
cause when deposited on a substrate. If thermal stresses due to the actual bonding cause 
significant curvature changes, the optics, performance, and adhesion of the device may be 
compromised. These stresses are significant even at low bonding temperatures if the bonded 
materials have large differences in thermal expansion coefficients. Predicting thermal strains must 
involve complete characterization of the bonded materials and the bonding agent at its operation 
temperatures and through any thermal cycles the structure may undergo during use. With organic 
bonding layers, it is likely that inelastic deformation will occur at elevated temperatures if the bond 
layer is under thermal strain. 

In order to fully understand and predict the strain and curvature of a structure, the elastic and 
inelastic temperature- and time-dependent behavior of the structure must be characterized. This 
study combines an elastic deformation model with a viscoelastic behavior model to fit observed 
behavior of an optical structure bonded with an organic material. Heat treatments involving 
prolonged holds at elevated operation temperatures were performed in order to obtain a wide 
characterization of the bond material and the behavior of the structure. All of the heat treatments 
were done at temperatures between room temperature and the bonding temperature of the structure. 
The thermal strains resulting from the temperature changes are considerable enough to cause 
observable curvature changes and inelastic effects caused by the bonding layer. The structure is a 
polished diamond disk bonded to a zinc sulfide (ZnS) disk by an optically-transparent high 
temperature wax. The melt temperature of the wax is 100°C and the materials were bonded at this 
temperature. These materials have applications in the infrared optics industry as durable high 
temperature coatings. However, diamond and ZnS have very different thermal expansion 
coefficients, and significant stresses result due to slight temperature changes. Because of the low 
melting temperature of the wax and the relatively high thermal stresses involved, the elastic and 
inelastic response to temperature changes are readily characterizable in this structure. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

When a bonded structure is heated to a service temperature that is a significant fraction of the 
melting point of the adhesive material, the curvature of the entire structure will change over a 
period of time due to stress relaxation in the adhesive layer. Often the stresses will relax at room 
temperature. If the initial thermal stress in the bond is known when the sample is heated to its 
service temperature and the curvature change is observed, then significant properties of the system 
may be characterized. These properties are the relaxation time constant for the adhesive material, 
the activation energy for inelastic deformation, and the viscosity. There are several theoretical 
models that were used to characterize the behavior of this structure and estimate the properties as a 
function of time and temperature. The stress at a given temperature was estimated using an elastic 
strain model for layered structures developed by Townsend, [1]. This model uses the 
thermal expansion coefficients and elastic moduli of each of the layers to calculate the curvature, or 
thermal strain, and the thermal stresses throughout any layered structure. Thermal elastic stresses 
are predicted between the bonding temperature of the material, (in which the stresses are zero), and 
the service temperature at which die curvature change, or stress relaxation, is measured. The actual 
stress in the adhesive layer changes with time as it relaxes; however the initial stress is 
approximated to be elastic prior to the relaxation if the temperature is changed rapidly enough. If 
no inelastic deformation occurs in the bonded layer throughout all heat treatments, then the stress 
and curvature of the structure should be completely elastic and predictable by this model. 

Since the wax adhesive material is a relatively small and simple molecule compared to long- 
chained or thermoset bonds, the stress relaxation may be modeled according to Maxwell’s model 
of viscoelastic behavior. This model applies to a range of materials from glycerin to thermoplastics 
that are single relaxation mode polymers [2]. Maxwell's model describes the system as a spring 
and dashpot in series, where the displacement is the strain, the dashpot constant is the viscosity of 
the material, and the spring constant is the modulus of elasticity [3]: 


+ i^ = 0 

G dt 

( 1 ) 

where a is the instantaneous stress, r\ is the viscosity of the bond layer, and G is the shear 

modulus of the wax bond. r|, a, and G are all temperature-dependent, making the shape of the 
solution vary with temperature. The stress at any time as it relaxes at a constant temperature may 
be determined from the solution to Equation 1 and the initial stress [3]: 

c = <J^e /A (2) 

where a is the time-dependent stress, cs 0 is the initial stress, t is the time, and X is the time 
constant. <J Q and A, are also temperature dependent, so Equation (2) must be fitted to the material 
behavior over a range of temperatures. The time constant may be measured by observing the 
curvature relax from an initial value to a value close to zero as a function of time since stress and 
curvature are directly related. If o Q and X are known, then the stress in the bond may be known at 
any time for a given temperature and the change in curvature of the structure may be predicted as a 
function of time. 

Another property inherent to each material is the relaxation activation energy, Ajj., describing the 
free energy activation barrier for molecular motion [2]. Molecular motion is the relaxation 
mechanism. The relaxation time constant is related to the activation energy, temperature, initial 
viscosity, and shear modulus [2]: 



HT) = 



where X is the time constant, rjo is the initial viscosity, G is the shear modulus, Ap* is the Vogel- 
Fulcher activation energy for stress relaxation, k is Boltzmann’s constant, and T is the temperature. 
The activation energy is ideally constant for a given material. However, Ap in reality changes with 

temperature as the material expands and the free volume increases, but a constant A|i is a 
reasonable approximation [2]. 

If the initial stress is known and the curvature change is observed as a function of time and 
temperature, then the other properties may be determined from this model and applied to predicting 
further curvature changes and stress relaxation in other layered structures. 


Samples of ZnS and diamond 2.54 mm (1 inch) in diameter were bonded together using a wax 
polymer bond with melting temperature of 100°C. Both disks had optically polished surfaces on 
both sides. The ZnS thickness was 880 pm and the diamond layer thickness was 210 pm. A 
high melting temperature wax of polyethylene structure was used as the bonding agent between the 
two optical disks. Several beads of wax were placed on the ZnS surface and melted on a hot plate. 
The temperature was measured by a thermocouple positioned at the edge of the ZnS piece. Once 
the wax was fully melted and evenly spread across the surface, the diamond disk was pressed on 
top of the melted wax surface and die structure was removed and placed on a cool (room 
temperature) surface in order to rapidly solidify the wax. The thickness of the adhesive layer was 
~3 pm. 

After the samples were bonded and cooled to room temperature, they were heated rapidly to 
final temperatures of 45°, 55°, and 65° and held at those temperatures while the curvature of the 
structure was measured as a function of time. 

The structure curvature was measured by the laser-beam deflection method. Since both optical 
disks had initial non-zero curvatures, the absolute curvature values measured reflect the sum of 
both the initial and the thermal-stress induced curvatures of both bonded materials. Because of 
this, all of the results are stated in terms of change of curvature as a function of time and 
temperature rather than absolute curvature values. 


Tim e Pc p cnd gn L S.trgSS. Re laxa tion 

Figure la), b), and c) show the measured curvature change of the bonded sample at 45°C, 
55°C, and 65°C respectively as a function of time. The curvature change is fitted to an exponential 
function reflecting the viscoelastic stress relaxation that is occurring at these temperatures. There is 
scatter in the measured data due to sample and temperature control variations. The time constant 

for relaxation at each temperature, X(T), is apparent from the exponential decay of the relaxation 
curves. Physically, this represents the time at which 64% of the stress has been relaxed. The 
curvature is shown to decrease as the samples flatten out to a less stressed state. 


The relaxation time constants decrease with increasing temperature as expected from Equation 
3. As the temperature approaches the melting temperature of the wax, stress relaxation is more 
rapid. Figure 2 shows the measured relaxation time constants, in seconds, as functions of 
temperature. Although difficult to detect, because the temperatures are so close and the noise in the 
curvature measurements, the time constants are decreasing along a slight exponential curve as 
would be expected. 

Figure 2. Time constants determined from curvature 
change measurements as a function of temperature. 


The initial actual thermal stresses, Go’s, in the wax bond at each temperature were not known 
explicitly, but were approximated at each temperature using the elastic strain model and an 
approximate modulus of elasticity. These approximate Go's, for a bond temperature of 75°, (where 
the curvature was measured to begin changing), were 7.4 MPa, 4 MPa, and 2.5 MPa at 45°, 55°, 
and 65° respectively. Figure 3 shows the time dependent stress relaxation at the three temperatures 
using the approximate values for the initial stresses. 

Figure 3. Approximate viscoelastic stress 
relaxation at 45°C, 55°C, and 65°C based on 
assumed initial stress and measured time 

Figure 4. Range of bond material activation energy 
(Ea) determined from the time constants and the 
approximate viscosity/shear modulus ratio. 


Activation Energy Prediction 

The Vogel-Fulcher activation energy for inelastic deformation, a value not given for this 
material in the literature, was predicted by using Equation 3, solving for the activation energy as a 
function of rio/G ratio. The xio/G ratio of the bond material was approximated using appropriate 

ranges for this ratio. Figure 4 shows the activation energy as a function of tb/G ratio ^g Ae 
experimentally measured time constants at each temperature. The region highlighted between the 
solid lines shows the appropriate range of activation energies. The range of initiall viscosity 
expected for this material within the temperature regions investigated: 85% to 91% ot the melting 
point of the bond material, is 10* to 10^ kg/m*s [4J. The shear modulus is on the order of 10 4 Pa 
[5] giving a viscosity/modulus ratio of 0.01 to 0.1 as a best approximation. The material 
activation energy is in the range of 0.25 and 0.3 eV, or 5.8 to 6.9 kcal/mol. 


A combination of elastic and inelastic stress theories were used to characterize time and 
temperature dependent behavior of an individual optical structure. The goal of thiswork was to use 
a predictive model to determine the curvature at any given temperature and tmie. The net curvature 
of the structure is a combination of elastic thermal stresses and stress relaxation m the organic bond 
layer. The wax bond behaves according to viscoelastic stress relaxation theory in which the 
structure curvature is dependent on the relaxation rate and temperature and properties of the wax 
bond From observation of the curvature change as a function of time and temperature, key 
properties of the wax may be extracted, such as the Voger-Fulcher activation energy and the 
temperature dependent relaxation time constant Knowing these properties enables prediction of 
the curvature of the structure at temperatures times other than what was observed in this study, in 
addition, the models employed here may be adapted for determining change in shape over time ol a 
range of systems in which single relaxation mode thermoplastic polymers are used to bond 
different materials. 


1. P.H. Townsend, D.M. Barnett, and T.A Brunner, J. Appl. Phys. 62 (11), 44384444 (1987). 

2. Shiro Matsuoka, Relaxation Phenomena in Polymers, (Oxford University Press, New York, 1992), p.9, pp. 45- 

3. N.G. McCrum, C.P. Buckley, and C.B. BucknaU, Principles of Polymer Engineering, (Oxford University Press, 
Oxford, 1988), pp. 101-166. 

4. R. Byron Bird, Warren E. Stewart, and Edwin N. Lightfoot, Transport Phenomena, (John Wiley & Sons, New 
York, 1960), pp. 14-18. 

5. M.F. Ashby and D.R.H. Jones, Engineering Materials I, (Pergamon Press, Oxford, 1980), p. 31. 


Part IV 

Interface Characterization 



Department of Physics, Southern Oregon State College, Ashland, OR 97520. 


Metal/Polymer systems have potential applications as interconnect materials in integrated 
circuits. Polymers with low dielectric constants, if used as interlayer dielectric, can reduce the 
RC time constant. Device speed can be doubled if a polymer can replace the present dielectric 
material, Si0 2 . However, the problem of weak metal/polymer adhesion must be understood 
and resolved. In situ deposition and analysis are the most controlled means to study an 
interface formation process. However, for practical reasons, i.e., application, time, cost, and 
flexibility, it is critical to study metal/polymer interface ex situ. X-ray Photoelectron 
Spectroscopy (XPS), can be used to determine the composition and bonding structures of 
buried interfaces. This is achieved by examining peeled surfaces, thin overlayer, and low- 
energy-ion sputtered surfaces. The possible adhesion mechanism or failure mode is 
determined by correlating XPS results with adhesion strength. Based on these results, 
adhesion enhancement methods, such as substrate surface treatments, can be formulated. The 
product of these treatments can be evaluated using the same analysis. These techniques for 
studying buried interfaces using XPS are reviewed and results of their applications to the 
metal/Teflon AF 1600 interface is presented. 


Dielectric materials in interconnect technology are important to consider when 
electronics device operation frequency enters the GHz regime. The RC time constant of the 
interconnect ultimately limits the device speed. Just as important is the cross-talk between 
interconnect lines which will reach unacceptable levels as device dimensions decrease to 
submicron levels. These problems can be eliminated by using materials with low dielectric 
constants. Polymers are the most practical candidate for the proposed application. The 
present dielectric material, Si0 2 has a dielectric constant of 3.9. Teflon, for example, has a 
dielectric constant of 2.1. Polyimides and fluorocarbon polymer, with dielectric constants 
below 2.5, are being examined as potential insulators.[l-4] 

For a material to be suitable as a new dielectric, it should have a low dielectric constant 
and be able to withstand 450°C, the highest currently used processing temperature, while in 
contact with the metal. For devices to function reliably, polymers used in interconnects must 
have good adhesion with metal. However, polymers such as fluorocarbon polymers, because 
of their low surface energy, show only weak polymer/metal adhesion. It is thus crucial to 
understand the adhesion mechanism as well as to develop means of improving metal/polymer 
adhesion. [5-14] 

The ultimate adhesion strength between a polymer and metal depends on several 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

1. Mechanical Anchoring - Adhesion improves with a rough interface which provides a 
large surface area for bonding as well as sufficient mechanical interlocking. [16,17] 

2. Chemical Bonding - Chemical bond formation at an interface improves the adhesion 
between the two materials. [18] 

3. Electric Double Layer - When a polymer comes in contact with a metal, charge 
transfer between the two surfaces can result in the formation of a double electric layer, 
i.e., a capacitor, at the interface. [19] Because of electrostatic forces, energy is required to 
separate such an interface. [20] 

4. Weak Boundary Layer - Any contamination, reaction or diffusion between two 
materials in contact may result in the formation of multiple interfaces with different 
compositions as a function of distance from the interface. Adhesion failure at such an 
region will occur at the layer with the weakest adhesion strength. [21,22] 

One or more of these mechanisms can be operative at any one time. All adhesion mechanisms 
operate at or near the interfacial region between the two materials. Establishing the relative 
importance of the adhesion mechanism can shed light on the choice of adhesion enhancement 
method. Adhesion can be enhanced by altering the interfacial region, for example, by 
increasing surface roughness and introducing bonding sites [16, 17], by enhancing 
interdiffusion or forced mixing [5], or by eliminating weak boundary layers. Surface 
modification of fluoropolymers is of great importance to improve adhesion and a review can 
be found in reference [23]. 

To understand the contribution from various mechanisms to the ultimate adhesion 
strength, one needs to determine the composition as a function of distance from the interface 
and the chemical state of the elements which contain information on the bonding structure and 
bonding sites. The latter is especially important in the case of polymers. Metal can 
preferentially bond to certain functional groups. If these groups can be identified, special 
copolymers can be synthesized to promote adhesion without compromising other beneficial 
properties of the polymer. Copolymers of fluoropolymers have shown to have good adhesion 
with some metals. [24] 

X-ray Photoelectron Spectroscopy (XPS) is best suited for such measurements because of 
its ability to determine relative elemental concentrations as well as chemical bonding 
information. XPS application to polymer is well understood and chemical shifts are well 
documented. [25,26] The escape depth of photoelectron used in XPS is 1 to 4 nm and 
composition can be determined with good depth resolution. Unfortunately, some polymers 
will decompose on prolonged exposure to X-rays and care must be taken to ensure correct 
interpretation of XPS results. [27] 

The most controlled means of studying interfacial formation is by in situ deposition and 
examination of the process under ideal conditions, i.e., in Ultra-High Vacuum. The chemistry 
and morphology of the interface as a function of film thickness can be determined as the film 
develops. [28] Such studies are important in revealing the basic mechanism of adsorption, 
nucleation and growth of overlayers. The results of these studies, however, are not directly 
applicable to device technology. Fabrication usually occurs under less than ideal conditions, 
i.e., lower vacuum or higher contamination, higher temperature because of higher deposition 
rate, and stress can result from subsequent processing conditions. In addition, some adhesion 
enhancement techniques are performed after the interface is formed, i.e., ion beam "stitching" 
of the buried interface [5]. Furthermore, weak boundary layers can form during subsequent 
processing. It is thus important to be able to study buried interfaces. 



0 0 





The common technique for exposing a buried interface is by ion sputtering using noble 
gas. Sputtering polymers however can result in many problems. In addition to differential 
sputtering, ion bombardment can induce mixing, i.e., as a result of momentum transfer, and in 
the case of polymers, can cause decomposition. Very useful information, however, can be 
determined from sputtered surfaces by careful interpretation of the results. 

In this article, some of the techniques in applying XPS and ion sputtering to determine 
bonding structures at buried interfaces are 
illustrated using the sample system, Teflon 
AF 1600 and metals, i.e., Ag, Al, Au, and 
Cu. Teflon AF 1600 is an amorphous 
fluoropolymer synthesized by Resnick at 
Du Pont Corporation.[29] It is a 
copolymer of tetrafluoroethylene or Teflon 
and 2,2-bis(trifluoromethyl)-4,5-difluoro- 
1,3-dioxiole, see Figure 1. The bulk 
dielectric constant of AF 1600 is 
determined to be 1.9 at room temperature, 
the lowest of all polymers, including 
Teflon. [30] Deposition of amorphous AF 

1600 films was recently reported by Nason et. al.[31] and R. Chow et. al [32] using vacuum 
pyrolysis and by Blanchet using a laser ablation method. [33] The amorphous nature of these 
films is potentially superior to Teflon in mechanical properties, i.e. reduced creep, as well as 
optical properties, i.e. reduced scattering.[34,35] AF 1600 is thus potentially useful as a 
dielectric layer in integrated circuit packaging and as a thin film waveguiding material. To 
realize such applications, some practical problems have to be addressed. Chief among these is 
the metallization of AF 1600. 

We deposited metal on AF 1600 films and measured the peel strength. We found that the 
AF 1600 peel strength is less than 5 gm/mm with metals such as Ag, Au, and Cu. But the 
peel strength between Al and AF 1600 is much higher, ~ 15 gm/mm. Our goal is to determine 
the adhesion mechanism in Al/AF 1600 and the Al bonding structure with AF 1600.[37] 

cf 3 

Figure 1. The structure of AF 1600, x:y = 1:2 


AF 1600 films are deposited using either the spin-on or vapor deposition techniques. 
Vapor depositions were done in a chamber with a base pressure of 4.0 xlO' 4 Pa. Details of the 
vapor deposition procedure are described in Reference 31. Spin-on films are made from a 
solution consisting of AF 1600 in FC-75, a perfluorated solvent from 3M. Spin-on films are 
annealed at 110°C for 15 min to remove any solvent. In the present work, the thickness of the 
AF 1600 film is 50-100 A in the case of AF 1600/metal/Si samples and 3000 A in the case of 
metal/AF 1600/Si samples. 

Metals are e-beam deposited in a system with a base pressure of 7.0 xlO* 6 Pa. The 
deposition rate and film thickness used are 5 A/sec and 3000 A respectively for deposition on 
Si and 1 A/sec and 30 - 100 A respectively for deposition on AF 1600. The as-deposited metal 
films are found to be contamination free within the detection limit of XPS. All high 
resolution characteristic XPS spectra are identical to that from the pure metals. An oxide 


layer is found on A1 and Cu films as a result of exposure to atmosphere. The oxide layer can 
usually be removed by less than 30 sec sputtering with 5 keV Ar depending on the amount of 

XPS is done using a Perkin-Elmer 5500 multitechnique system. Mg-ka X-ray is used in 
all spectra shown here. Sputtering is done using either 5 keV or 300 eV Ar ions at a pressure 
of 7.0 xlO' 3 Pa. 

Atomic Concentration and Chemical Shifts 

The relative composition of the AF 1600 can be determined from an XPS broad scan 

Figure 2 XPS broadscan of a spin-on AF 1600 film. C, O, and F are identified. 

spectrum using peak area analysis. [25, 36] 
Figure 2 shows a typical broad scan spectrum 
from a spin-on AF 1600 film and Table 1 
compares the measured relative atomic 
concentration of C, O, and F, in at. %, in the 
AF 1600 thin films determined from Figure 2 
with the theoretical value calculated from the 
structure shown in Figure 1. The experimental 
result agrees closely with the theoretical 
values. Although the oxygen concentrations 
are slightly higher, they can be the result of 
contamination due to exposure to atmosphere. 

Figure 3 shows the C(ls) spectrum taken 

AF 1600 












Table 1: Comparison of the theoretical 
atomic concentration, at. %, of different 
elements in AF 1600 to that determined by 
XPS, see Figure 2. 


Figure 3. C(ls) spectrum from a AF 1600 film. Solid lines are measured intensity and 
broken lines are fitted spectra. Four peaks are identified, see text for detail. 

from a spin-on AF 1600 film. The 
spectrum is composed of 4 peaks. The two 
most prominent peaks at 297.7 and 295 eV 
can be assigned to the -CF 2 and -CF 3 
functional group respectively, see Figure 
1. The peak at 288 eV is composed of two 
peaks from the F-C-0 and the O-C-O 
functional group, see Figure 1. Figure 3 
demonstrates the ability of XPS to identify 
bonding structure by determining the 
binding energy shift of the photoelectrons. 
Because the relative concentration of each 
functional group is proportional to the 
integrated area under their respective 
peaks, their relative concentration can be 
determined. The relative concentration of 
C associated with different functional 
group as determined from Figure 3 agrees 
well with the calculated value using the 
structure shown in Figure 1. 

Effects of Sputtering 

The effects of sputtering on an 
evaporated AF 1600 film are shown in 
Figure 4. Figure 4 shows C(Is) spectra 

Figure 4. C(ls) spectra of AF 1600 after 
sputtering for 15, 240, and 900 sec. 


after sputtering with 300 eV Ar ion for 15, 240, and 900 sec. Comparing these spectra with 
that from the as deposited spectrum, Figure 3, the integrated intensity of the two peaks at the 
lower binding energy is reduced after just 15 sec of sputtering and continues to decrease as a 
function of sputtering time. The bonding structure of the carbon associated with the F-C-0 
and the O-C-O functional group, see Figure 1, represented by these two peaks is thus changed 
as a result of sputtering. The integrated intensity of the highest energy peak corresponding to 
the -CF 3 group is reduced. At the same time, a broad peak developed at around 293 eV. 

These results indicate that the sputtering process altered the structure of AF 1600 and 
specifically the 2,2-bis(trifluoromethyl)-4,5-difluoro-l,3-dioxiole in the copolymer. A new 
bonding structure is formed with some of the carbon which have a binding energy shift to 293 
eV. Figure 4 demonstrates the problem associated with a sputtered surface in which bonding 
structure can be destroyed and created. 

Depth Profiling 

Depth profiling is performed on spin- 
on AF 1600/metal/Si samples. Figure 5 
shows the F atomic concentration as a 
function of sputtering time for spin-on AF 
1600/metal/Si. Four metals, Ag, Al, Au, 
and Cu were used. Differences in F 
concentrations before sputtering are results 
of variations in film morphology, i.e. 
continuous versus patchy. Patchy films 
indicate the AF 1600 does not "wet" the 
substrate or a high interfacial energy. XPS 
measurements of the as-deposited spin-on 
AF 1600/Al/Si(100) show results similar to 
that from pure AF 1600. No Al can be 
identified from the as-deposited film 
indicating a continuous film without bare patches. 

In all cases, most of the AF 1600 is removed after 10 sec of sputtering as indicated by the 
concentration of C, < 5 at. %, and F, < 15 at. %. The AF 1600 is completely removed after 30 
sec of sputtering as indicated by the absence of C on the sputtered surface. 

In the case of Ag, Au, and Cu, F is completely removed after 30 sec of sputtering, Figure 
5. Thus any F and metal reaction product or sputtering induced F mixing occurring between 
AF 1600 and metals are limited within this sputtered depth. 

In the case of AF 1600/A1, F, < 3 at. %, is present up to 150 sec of sputtering, see Figure 
5. The extent of F mixing in the A1/A1 2 0 3 matrix is much larger than in the previous cases 
and the depth is estimated to be > 150 A. 

Because the deposition and sputtering conditions are identical in all cases, we believe the 
F in the A1/A1 2 0 3 matrix is a result of F diffusion and/or chemical reactions with Al rather 
than sputtering induced mixing. F is detached from the AF 1600 and diffuses into the 
A1/A1 2 0 3 matrix. This result is confirmed by measurements on peel off surfaces, see below. 

Figure 5 Depth-profiling of F on samples with 
structures AF 1600/metal/Si. 


Figure 6. Al(2p) spectra from an AF 1600 
film on A1 sputtered for 10, 30, 120 sec. 

A1 and AF 1600 Interaction 

Binding Energy ( eV ) 

Figure 7. F(ls) spectra from an AF 1600 
film, top; AF 1600/A1 sputtered for 10 sec. 
middle; and a thin layer of A1 on AF 1600, 

Figure 6 shows the A1 (2p) spectra from a spin-on AF 1600/AI/Si( 100) sample at 10, 30, 
and 120 sec of sputtering. The corresponding spectrum before sputtering is just the 
background noise spectrum, i.e. no A1 present, and is not shown. The bottom spectrum is 
measured after 10 sec sputtering with contribution mainly from the interface. This spectrum 
requires 3 peaks to fit, indicating three chemical states for the Al. Peak (a) is associated with 
metallic Al, (b) with AI 2 0 3 , and (c) has the correct energy shift for A1F 3 .[36] The presence of 
A1 2 0 3 is expected because the Al film is exposed to air before AF 1600 is deposited. The 
integrated intensity of peak (c) decreases while that of peak (a) increases as a function of 
sputtering time indicating that the A1F 3 concentration decreases as a function of depth from 
the surface. After 120 sec of sputtering, the Al (2p) spectrum is similar to that from a metallic 
Al sample, top spectrum in Figure 6. The presence of A1F 3 in the A1/A1 2 0 3 matrix can be a 
result of decomposition of the AF 1600 by ion bombardment and subsequent reaction with Al. 
Direct chemical reactions between Al and F must be confirmed in samples without sputtering. 
Peak (c) in Figure 6 shows the correct energy shifts for AIF 3 , this does not mean that only 


A1F 3 is present. Other species such as AlOF can also contribute to peak (c) in Figure 7. But 
the peak position of the peak indicates a substantial amount of A1F 3 is present. 

F (1 s) spectra from three samples are shown in Figure 7. The top spectrum is measured 
from a pure AF 1600. This spectrum has a single broad peak which can be fitted by two 
peaks. The lower binding energy peak is associated with the -CF 3 functional group and the 
other peak is associated with the -COF- and -CF 2 - functional groups in the AF 1600. 

The middle spectrum in Figure 7 is measured from a sample of AF 1600/A1 after 10 sec 
sputtering. This spectrum shows a broad peak with a binding energy 3 eV lower than that 
from pure AF 1600, top spectrum. Because most of the AF1600 is removed after 10 sec of 
sputtering no charging effect is observed as indicated by the A1 peak position, not shown, 
from the same surface. This peak has the correct energy shift as A1F 3 . This spectrum is in 
agreement with the previous result. F in the A1/A1 2 0 3 matrix is mainly in the form of A1F 3 . 

The bottom curve in Figure 7 is measured from a sample with a 30 A thick A1 film on AF 
1600. The thickness of this film is less than the escape depth of the F(ls) electron. 
Photoelectrons from the interface contribute to this spectrum. The curve can be fitted with 
three peaks. Peaks (a) and (b) are associated with the -CF 2 - and -CF 3 respectively and peak 
(c) is associated with A1F 3 . Because no sputtering is performed on this sample, the reaction 
between AF1600 and the A1/A1 2 0 3 matrix is spontaneous and not a result of ion 
bombardment. This reaction occurs as low as 110°C, the spin-on processing temperature and 
one of the product of this reaction is A1F 3 . 

Reaction Site on the Polymer 

To determine the origin of the F in the A1F 3 , one can compare the F(ls) spectra taken 
from a pure AF 1600 film and that from the interface of Al/AF 1600. In Figure 7, the relative 
integrated intensity of peak (b) with respect to peak (a) of the bottom spectrum, from the 
Al/AF 1600, is smaller than that from pure AF 1600, top spectrum, indicating that the relative 
concentration of the -CF 3 group is reduced in the'AF1600 in the Al/AF 1600 sample. Thus F 
in the A1F 3 should originate from the -CF 3 functional group. 

We have calculated the Gibbs free energy reduction in forming A1F 3 using F atoms from 
the CF, CF 2 and the CF 3 groups to be -15.9, -58.1 and -135.4 kcal/mole respectively. These 
calculations are based on thermodynamic data for gas phase compounds and the results should 
be used for rough comparison only. According to this calculation, it is energetically most 
favorable to form A1F 3 using F from a -CF 3 functional group. This calculation agrees with our 
assertion that the F in A1F 3 originates from the -CF 3 functional group. Our free energy 
calculation also shows that A1 2 0 3 is more stable than A1F 3 . This might be the reason for the 
deep penetration of F in the A1 matrix. If the surface of the substrate is mainly A1 2 0 3 , F has to 
be dispersed in the matrix to find the suitable Al. The copolymer, 2,2-bis(trifluoromethyl)- 
4,5-difluoro-l,3-dioxiole, provides the necessary side chain with the -CF 3 functional groups 
which are vital to the reaction between AF 1600 and Al. 

Peel off Film 

The previous results show that Al will react with AF 1600 forming A1F 3 and other 


compounds. We have not however determined the role of this reaction with adhesion 
strength. One way of looking at this problem is to peel the overlayer off the substrate and 
examine the composition and chemical structure of the separated surfaces. 

Binding Energy ( eV ) 

Figure 8. AI(2p) spectra from a Al/AF 
1600 sample with the A1 peeled off. 

We deposited A1 on AF 1600. The 
A1 film is then mechanically peeled off. 
The surface of the remaining AF 1600 is 
then examined using XPS. The A1 (2p) 
spectrum from such a surface is shown in 
Figure 8. Two peaks are required to fit 
the spectrum. These peaks correspond to 
peaks (b) and (c) in Figure 6. These two 
peaks are associated with AI 2 0 3 and A1F 3 
respectively. Thus A1F 3 formed at the 
interface adhered to AF 1600 as the A1 is 
peeled off. The oxide, as represented by 

Figure 9. 0(1 s) spectra from a sample of 
30 A Al/AF 1600, top curve, and from a 
sample of Al/AF 1600 with the A1 peeled 
off, bottom curve. 

peak (b), can either be formed during 
deposition from reaction with AF 1600 or 

oxidation of residual AI after the peel test with ambient oxygen. A1 2 0 3 and A1F 3 are also 
found on the peeled off film. Fracture thus occurs in the A1/A1 2 0 3 /A1F 3 layer. Throughout the 
rest of this paper, we will use A1/A1 2 0 3 /A1F 3 to describe this layer even though other Al, O, F 
compound may be present. This result seems to indicate that the A1/A1 2 0 3 /A1F 3 region is a 
weak boundary layer. 

The previous result indicates that fracture occurs at the A1/A1 2 0 3 /A1F 3 interfacial region. 
Therefore some mechanisms, with larger adhesion strength, must be responsible for the 
adhesion between the AF 1600 and this interfacial region. A possible bonding structure 
which resulted in the observed adhesion strength can be identified from the 0(1 s) spectra 
shown in Figure 9. The top spectrum in Figure 9 is taken from a sample of 30 A Al/AF 1600. 
The thickness of the Al is less than the escape depth of the photoelectron from the interface. 
Three peaks are identified. Peak (a) is associated with the A1 2 0 3 and peak (b) is associated 
with the AF 1600. Peak (c) is a state associated with neither pure A1 2 0 3 nor pure AF 1600. 


The bottom curve in Figure 9 is taken from a sample of Al/AF 1600 with the A1 peeled 
off. Again peak (c) is identified. The origin of this bond is not certain but it is possible that 
this bond is responsible for the adhesion between AF 1600 and the A1/A1 2 0 3 /A1F 3 region. 
More work however are required to determine the origin and the effect of this bond on 


When 100 A thick films of AF 1600 is deposited, spin-on or evaporated, on metal 
surfaces, continuous films were formed on Au and A1 surfaces, while patchy films exposing 
the substrate metal are formed on Cu and Ag surfaces indicating large surface energy or weak 
adhesion. In the case of AF 1600/A1, chemical reactions between A1 and AF 1600 is observed 
at the interface. The principle mechanism is the reaction between F, from the -CF 3 functional 
group in the AF 1600, and A1 forming A1F 3 . F detached from the -CF 3 functional group of the 
AF 1600, diffused into the A1/A1 2 0 3 matrix and formed A1F 3 . The diffusion depth of F is 
estimated to be > 150 A. An A1 film was evaporated onto an AF 1600 film. The A1 film was 
then mechanically peeled off and the remaining polymer surface examined. A1F 3 was again 
found. This result seems to indicate that the A1/A1 2 0 3 /A1F 3 region is a weak boundary layer 
where fracture develops. An O bonding state associated with neither A1 2 0 3 nor AF 1600 is 
identified at the interfacial region. 


Composition and bonding information can be determined using XPS which are crucial in 
understanding the adhesion mechanism of buried polymer/metal and metal/polymer interface. 
When ion sputtering is required to exposed the interfacial region, it is found to alter the 
chemical structure of AF 1600. Usefiil information, however, can be obtained from sputtered 
surfaces by comparative analysis. Overlayer film of thickness less than the electron escape 
depth can be use to evaluate the extent of sputtered induced effects and identified chemical 
structure in close proximity to the interface. The fracture site and its local bonding structure is 
examined by peeling off the overlayer and exposing the fracture surface. Such information 
are critical to the understanding and identifying relative importance of the various adhesion 
mechanism. Our result shows that AF1600 alone without other treatment is not suitable as an 
interlayer dielectric for A1 metallization both in terms of adhesion and F contamination. 


This work is funded by the IBM corporation and is done at the Renesselaer Polytechnic 
Institute in conjunction with Prof. T.-M. Lu. Special thanks to Dr. G.-R. Yang and Dr. X.-F. 
Ma for their expertise in sample preparation and discussion. 



1. P. Singer, Semiconductor International, 11, 50 (October 1993). 

2. J.G. McDonald, J.T. Lin, H.J. Greub, R.A. Philhower and S. Dabral, IEEE Trans. Comp. 
Hybrids. Manuf. Technol. 12,195 (1989). 

3. D.S. Soane and Z. Martynenko, Polymers in Microelectronics- Fundamentals and 
A pplications. (Elsevier, New York, 1989). 

4. L. You, G.-R. Yang, C.-I. Lang, J. A. Moore, J.F.P. McDonald, and T.-M. Lu, J. Vac. 

Sci. Technol. A, 11 (6), 3047 (1993). 

5. J.E.E. Baglin, IBM J. Res. Develop. 38 (4), 413 (July 1994) 

6. Y.-K. Kim, C.-A. Chang, and A.G. Schrott, J. Appl. Phys. 67 (1), 251 (January 1990) 

7. F.D. Egitto and L.J. Matienzo, IBM J. Res. Develop. 38 (4), 423 (July 1994) 

8. B.J. Love and P.F. Packman, J. Adhesion, 40,139 (1993) 

9. A. Buchman, H. Dodiuk, M. Rotel and J. Zahavi, J. Adhesion Sci. Technol. 8(10), 1211 

10. M. Collaud, P. Groening, S. Nowak, and L. Schlapbach, J. Adhesion Sci. Technol. 8 (10), 

11. K.-W. Lee and A. Viehbeck, IBM J. Res. Develop. 38 (4), 457 (July 1994) 

12. L.M. Siperko and R.R. Thomas, J. Adhesion Sci. Technol. 3 (3), 157 (1989) 

13. H.M. Clearfield, D.K. McNamara, and Guy D. Davis, in Adhesive Bonding, edited by L.- 
H. Lee, (Plenum Press, New York, 1991), pp. 203. 

14. M-K. Shi, B. Lamontagne, 1. Martinu, and A. Selmani, J. Appl. Phys. 74 (3), 1744 
(August 1993) 

15. W.J. van Ooij, in Phvsiochemical Aspects of Polymer Surface, edited by K.L. Mittal, 
(Plenum Press, New York, 1983) pp. 1035. 

16. C.-A. Chang, J.E.E. Baglin, A.G. Schrott, and K.C. Lin, Appl. Phys. Lett. 51 (2), 103 
(July 1987). 

17. C.-A. Chang, Appl. Phys. Lett. 51 (16), 1236 (October 1987). 

18. A.J. Kinloch, J. Mater. Sci., 15,211 (1980). 

19. A.R. Akande and J. Lowell, J. Phys. D: Appl. Phys. 20, 565 (1978). 

20. B.V. Deijaguin and Y.P. Toporov, in Phvsiochemical Aspects of Polymer Surfaces. 
(Plenum Press, New York, 1983 ) pp. 605. 

21. W.J. van Ooij, Rubber Chem. Technol. 52, 605 (1979). 

22. W.J. van Ooij, W.E. Weening, and P.F. Murray, Rubber Chem. Technol. 54, 227 (1981). 

23. L.M. Siperko and R.R. Thomas, J. Adhesion Sci. Technol. 1,157 (1989). 

24. J.M. Park, L.J. Matienzo and D.F. Spencer, J. Adhesion Sci. Technol. 5, no. 2, 153 

25. Practical Surface Analysis, edited by D. Briggs and M.P. Seah, (John Wiley & Sons Ltd, 
New York, 1990). 

26. D. Briggs and G. Beamson, Analytical Chemistry, 84 (15), 1729 (August 1992). 

27. F.-M. Pan, Y.-L. Lin, and S.R. Homg, Appl. Sur. Sci. 47, 9 (1991). 

28. L. Atanasoska and S.G. Anderson, J.M. Meyer III, and J.H. Weaver, Vacuum, 40 (1/2), 

85 (1990). 

29. P.R. Resnick, Polym. Prepr. 31, 312 (1990). 

30. A brochure detailing the technical properties of the Teflon AF 1600 product can be 
obtained from the Du Pont corporation, Wilmington, DE; for safety information 


specify "Teflon AF: Safety in Handling and Use," No. HO7805. 

31. T.C. Nason, J.A. Moore, and T.-M. Lu, Appl. Phys. Lett. 60 (15), 1866 (1992). 

32. R.Chow, M.K. Spragge, G.E. Loomis, F. Rainer, R. Ward, I.M. Thomas, and 
M.R. Kozlwski, preprint, submitted for Materials Research Society Symposium 
Proceeding, Fall 1993. 

33. G. B. Blanchet, Appl. Phys. Lett. 62 (5), 479, (1993). 

34. W.E. Hanford and R.M. Joyce, J. Amer. Chem. Soc. 68, 2081 (1946). 

35. L. Holland, J. Vac. Sci. Technol. 14,15 (1977). 

36. see for example, C.D. Wagner, W.M. Riggs, L.E. Davis, and J.F. Moulder, "Handbook 
of X-ray Photoelectron Spectroscopy". Perkin-Elmer Corp., Eden Prarie, MN. 

37. P.K. Wu, G.-R. Yang, X.-F. Ma, and T.-M. Lu, Appl. Phys. Lett. 65 (4), 508 (1994). 



Y. Nakamura, Y. Suzuki, Y. Watanabe and S. Hirayama 

Department of Materials Science and Engineering, National Defense Academy 

1-10-20 Hashirimizu, Yokosuka, Kanagawa 239, Japan 


The adhesive strength of polyimide thin films on an alumina substrate has been 
studied by the pull test. Polyimide thin films were cured at various temperatures from 
300 to 400°C, followed by etching in an oxygen plasma for 3 min. After metallization 
with chromium and copper, the pull test was carried out. It is found that the adhesive 
strength increases with an increase of the cure temperature. However the strength 
deteriorates noticeably at the cure temperature of 400°C. A possible reason for this is the 
variation of chemical states in the polyimide thin films. 


Multi-Chip Modules(MCMs) have been classified as either C, D, or L or a 
combination of these three types. However, there are only two types of MCM from this 
point of view: one case is where the circuit is deposited onto a substrate, the other uses 
lamination of individual layers without a substrate.[l] Advanced MCMs are a merge of 
chip, wire and substrate. MCM-D/C is a promising combination technology among the 
advanced MCMs technology, therefore the adhesion issue between different materials is 
a very important technical topic. 

The reliability of any complex structure such as an MCM-D/C depends on its ability 
to resist the detachment of component materials, thus a knowledge of adhesion is 
essential for the fabrication of MCM structures. Polyimide has outstanding thermal 
stability and is relatively easy to fabricate onto silicon, but it is difficult to deposit onto 
ceramics such as alumina. [2-4] The wok presented in this paper investigates and 
characterizes the adhesion between polyimide thin films and an alumina substrate, 
focusing on the effect of the cure temperature of the polyimide thin films on the adhesive 
strength at ambient temperature. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 


The polyimide precursor(RN812, Nissan Chemical Industries) for polyimide thin films 
was used as the raw material. Polyimide precursor solutions were spun coated onto an 
alumina substrate. The substrate was spun at 2000 rpm for 60 sec and cured. This 
resulted in a film thickness of about 2 pm . This process was repeated to make the 
desired thickness of films. The films were dried at 80 °C for 30 min on a hot plate in air 
and then cured at 3 00^400°C for 30 min in air. 

The surface of the polyimide thin films was etched in an oxygen plasma for 3 min, 
followed by sputtering with chromium and copper. The final thickness of 4 pm of copper 
was deposited by electroless plating. The 0.8 mm tin coated wire was soldered on the 2 
mm square patterned copper surface, then the pull tests were carried out at a speed of 
10 mm/min. 

X ray photoelectron spectroscopy(XPS), Infrared absorption spectroscopy(IR) and 
scanning electron microscopy(SEM) were used for evaluation of the materials. 


The relation between the heating and cooling rates and the adhesive strength of the 
polyimide samples of 4 pm in thickness cured at 300 °C for 30 min is shown in Fig. 1. 
The adhesive strength was not affected by the heating and cooling rates and remained at 
the level of 1.3 kg/mm^. It is found by the SEM examination that all fractures occurred 
at the interface between the polyimide thin films and the ceramic substrate. 

Heating and cooling rates (‘C/min) 

Fig. 1 Relation between heating and cooling rates and adhesive strength of samples cured 
at 300°C for 30 min. Polyimide films thickness is about 4 pm. 

The relation between the thickness of polyimide thin films and the adhesive strength 


of samples cured at 300 °C for 30 min is shown in Fig.2. The adhesion was drastically 
decreased at the film thickness of 7 and 15 p,m with the samples treated at the rates of 
4.6 and 2.3 °C/min, respectively. This abrupt decrease in strength may be explained by 
the existing cracks that were created by the thermal treatment due to the different 
coefficients of thermal expansion of polyimide and alumina. However constant strength 
was obtained by the samples at slower treated rates. Therefore, the rates of 1.2 °C/min 
were adopted for the following experiments. 

Thickness of PI films (|xm) 

Fig. 2 Relation between polyimide films thickness and adhesive strength of samples 
cured at 300°C for 30 min. 

Fig. 3 Relation between cure temperature and adhesive strength with the 
thickness of about 4 p,m. 


The effect of cure temperature on the adhesion is shown in Fig. 3. The adhesion is 
increased with increasing the cure temperature except for the treatment at 400 C. The 
maximum strength of 2.8 kg/mm 2 was obtained with the sample cured at 375°C. 

Chemical and physical interactions between the overlayer and a substrate govern the 
fracture of the resulting interface, and this, in turn, controls the adhesion of the 
overlayer/substrate system.[5] Therefore, it is important to learn the chemical states of 
surfaces of polyimide thin films. The measured concentration of chemical states in 
polyimide thin films after the cure treatment is shown in Table I. 

Table I Measured concentration of various chemical states 
in polyimide films after cure treatment. 

C 1 s 



Cure temperature 

Cure temperature 
of 3751C 



C = 0 

1 3.6 







3 0.4 

2 6.0 


5 6.0 

6 3.4 

The concentration of C=0 bonds is lower for the sample with the higher treatment 
than that of the sample treated at the lower temperature. The treatment temperature may 
affect some chemical bonds in polyimide, resulting in easy breaks by the oxygen plasma 
ablation. In turn, the broken bonds may easily react with some elements on the alumina 
substrate such as oxygen or aluminum. Therefore, the higher concentration of C-OH 
bonds in the films was obtained for the sample treated at higher temperatures. However, 
the treatment at too high a temperature will give a definitive damage to the chemical 
bond in polyimide. IR absorption results are shown in Fig. 4. It is obvious that the 
decomposition of polyimide has occurred in the samples. Therefore the weakest adhesion 
is obtained. 










2 : C=0 

3 : C-N 


Fig. 4 ER absorption spectra of polyimide films cured at (a)375°C and (b) 400°C. 


Heating and cooling rates do not affect adhesion between polyimide thin films and an 
alumina substrate. The appropriate rates do not affect the adhesion to thicknesses of 20 
p.m in polyimide thin films. The adhesion increases with increasing the cure temperature 
of polyimide films to 375 °C because of the increase in C-OH bond that is created by the 
oxygen plasma ablation. 



[1] Robert R. McBride, David F. Zarnow, and Raymond L. Brown, Advancing 
Microelectronics, 21, 16(1994). 

[2] Ming-Shyong You, A. Sanjoh, and T. Ikeda, Proc. of 8th Int'l. Microelectronics 
Conf., Ohmiya, (1994)pp.443-448. 

[3] Masahito Ishi and Hiroshi Hata, Proc. of 8th Int'l. Microelectronics Conf., Ohmiya, 
(1994)pp. 225-230. 

[4] Hyo-Soo Jeong, Y. Z. Chu, C. J. Duming, and R. C. White, Surface and Interface 
Analysis, 18, 289(1992). 

[5] G. D. Davis, B. J. Rees, and P. L. Whisnant, J. Vac. Sci. Technol. A, 12, 2378(194). 




Central Research Laboratory, Johnson Controls, Inc., Milwaukee, WI, 53209 


This paper describes a method for estimating the effective sticking probability for plasma 
enhanced chemical vapor deposition (PECVD) of hexamethyldisiloxane (HMDSO) using 
Si0 2 and polymerized siloxanes deposited on specially prepared trench structures. 
Comparison of the data with direct Monte-Carlo simulation curves provides information 
about the incorporation probability relative to film growth. It is shown that besides 
variation in gas chemistry, the choice of trench and film dimensions influences the step 
coverage. The sticking probability is shown to increase with oxygen flow rate by about 
30%, from 0:1 to 10:1 02:HMDS0 flow ratio. This flow rate dependence is found to be 
consistent with work performed on tetraethoxysilane. 


The sticking probability is one of the most important parameters for describing film step 
coverage and developing deposition models. However, due to the lack of knowledge of the flux 
and surface kinetics of the deposition species for many chemical systems, the sticking probability 
is difficult to directly quantify. Instead much work has gone into developing models to relate the 
contour of a deposited film to the sticking probability [1-8]. In particular, several simulations have 
been created to estimate the sticking probability of CVD processes. For example, Kawahara et al. 
developed a Monte Carlo model with a single depositing species of constant sticking probability 
[1]. Watanabe et al., developed a one dimensional analytical model along with a test structure to 
perform actual measurements [2], Ikegawa et al., used direct simulation Monte Carlo with finite 
element analysis techniques to determine film conformality based upon a full reaction model for 
thermal CVD processes for phosphosilicate glasses in molecular and transitional gas flow regimes 
[3]. More recently, work has been done to develop PECVD-based models, such as the string 
model developed by Yuuki, et al. for a-Si:H film growth by SiH* decomposition [4], Predictive 
models have been used for PECVD processes to determine the relative concentration of multiple 
deposition components with different sticking probabilities for SiH* deposition [4,5]. However, 
these models have only been used to determine the overall sticking probability for PECVD 
processes whose chemistries is already well understood. One objective of this work is to apply the 
results of LPCVD models to PECVD processes, and demonstrate how they may be used as a tool 
for characterizing CVD systems of less well understood chemistries and processes. The other 
objective is to show how the sticking probability and conformality change as the ratio of the plasma 
feedstock gases varies for the HMDSO/O 2 deposition system. 


The films were deposited on (110) Si wafers with trenches of various aspect ratios. The 
deposition conditions were 250 mTorr, 30 seem HMDSO, between 0 and 300 seem of O 2 , 300W, 
250 kHz rf power, autotransformer-matched impedance at 2450O, and no substrate heating. The 
samples were placed on the grounded bottom electrode of a Plasmatherm PK-1241 parallel plate 
system with a 2.5 cm electrode gap. (The plasma was confined to the region between the two 
plates.) The deposition time was varied between 10 and 12 minutes to achieve approximately 3 pm 
films. Samples were then cleaved and imaged in full cross section in SEM backscatter mode 
operating at 20 keV. The samples were not coated with a conductive layer. 

The test structure used for this experiment was made up of repeating sets of fully rectangular 
cross section trenches, produced by orientation dependent etching (ODE) of (110) Si [9-11], The 
die measured 1.014 cm by 0.512 cm, and consisted of 74 sets of trenches, each set having 5 
different widths (nominally 5, 10, 15, 20, and 30 pm) separated by nominally 10 pm wide walls, 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

for a total trench set width of 130 pm. The trench layout used repeating sets of trenches rather than 
grouping all trenches of a single width together, in order to minimize the dimension of detectable 
local chemical variations. The maximum trench width chosen was much smaller than the molecular 
mean free path length of 200 pm. The minimum Knudsen number (X/D) for a trench was 7, well 
above the value of unity that defines the molecular flow limit. Two sets of wafers were fabricated 
to produce trenches that were 5 and 15 pm deep. These two sets of trenches gave aspect ratios of 
width:depth of 1:2.5 to 6:1. It should be noted that the 1:1 and 2:1 aspect ratios exist for both 
trench depths. During deposition, one die of each etched depth was coated, so that 10 different 
trenches were simultaneously exposed for each data point. 

The trenches were formed in (110) Si using ODE with KOH [9-11]. The trench die pattern was 
transferred to an oxide mask using standard photolithographic and wet etch techniques, such that 
the trench walls were aligned with the (111). A two-step KOH etching procedure was used to 
produce a fully rectangular cross section trench with perpendicular sidewalls and flat bases, as 
shown in Figure 1 [12]. 

The estimation of the sticking probability for each data point was determined using curves 
generated by Kawahara et al., and Yuuki et al. [1,4]. The modeled sticking probability data were 
presented as a plot of isoprobability curves of film coverage as a function of trench aspect ratio. A 
limiting case equation of sticking probability equal to unity, and minimal topwall film thickness 
was also stated (for [3 = 1, and d max /D = 0, d m in/d max = [l-((W/D)/[l+(W/D) 2 ]^ 2 )]/2 (1) [4]. In 
order to systematically determine the sticking probability for the measured data, a sticking 
probability (p) parameterized equation was found by inductive reasoning and regression analysis: 

dminAW = (j^p) ^ 1 - (2) 

where, x = W/D, W is the trench width, D is the trench depth, d max is the topwall film thickness, 
and dmin is the film thickness at the bottom of the sidewall (dmin/dmax is the fractional step 
coverage). Equation (2) reduces to equation (1) at unity sticking probability. This equation 
provided a fit within 5% to the curves for sticking probabilities equal to or greater than 0.1. 


Figure 2 is a backscattered electron SEM image of a typical deposited film on a trenched 
surface. The deposition rate for the films is in the range of 3000A/min. The aspect ratio of the 
trench is 1:2.5 (width:depth) with portions of a 1:1.5 trench on the right and a 2:1 trench on the 
left. As can be seen from examining the thickness of the sidewall layers in each trench, the highest 
aspect ratio trench has the thinnest sidewall material. The texture of the film is featureless in cross 
section, except for the sidewall material. It appears to have a columnar-like structure of low 
temperature sputtered films, oriented at 45° relative to the base. Energy dispersive x-ray mapping, 
and XPS did not detect any compositional differences of the sidewall and topwall material. 

Figure 3 is a representative plot of experimental sticking probability data superimposed on a 
plot by Kawahara et al., and Yuuki et al. to show how the experimental data compare with 
modeled data [1,4]. The data come from two wafers processed simultaneously; one has 15 pm 
deep trenches and the other has 5 pm deep trenches. While the data follows the same trend as the 
modeled material, there are important differences. First, the experimental values for d m i n /d ma x 
increase more slowly with increasing aspect ratio compared to the model, instead of following a 
single sticking probability curve. In addition the data points from the 5 pm deep trenches tend to 
have a lower step coverage than the 15 pm deep trench of about the same aspect ratio. This 
difference in step coverage is within the standard error for the data collected from a single 
deposition, but is evident for every sample. The sticking probability estimation procedure uses 
Newton's iterative method to solve for (3 using equation 2, with the values of W/D and d m jn/d max . 

Figure 4 is a plot of the sticking probability as a function of the O 2 gas flow rate. The data 
points are collected from trenches with a 2:1 aspect ratio. The sticking probability ranges from 
0.39 to 0.78 for this aspect ratio. Overall, the sticking probability ranges from about 0.2 to almost 
unity. The error bars reflect the uncertainty in estimation of the sticking probability at each point. 


Figure 1: Secondary electron SEM image of edge of fully processed uncleaved 
(110) Si trench die used in experiments. Trench walls and bases consist of Si. 
Trench depth is 15 pm. 

Figure 2: Backscattered electron SEM image of cleaved 6 pm wide x 15 pm deep 
trench with 3 pm oxide. Deposition conditions include 30 seem HMDSO and 150 
seem O 2 . Edges of the 30 pm trench to the left, and the 10 micron trench to the 
right can be seen. Note the differences in sidewall film thickness between all three 


Trench Width/Depth, W/D 

Figure 3: Ratio of film thicknesses as a function of trench aspect ratio. Data comes 
from the sample shown in Figure 2. The data are super-imposed on a plot of 
modeled curves by Kawahara et al. for comparison [1]. 

Figure 4: Sticking probability as a function of O 2 flow rate. Trench aspect ratio 
2:1, 300W, 30 seem HMDSO, 3 pm top layer film thickness. The lower set of bars 
represents the maximum deviation of measured data from plotted values. 


The lower bars are data points showing the maximum difference between the 2:1 trench sticking 
probability and the estimated sticking probabilities for the other trenches. The figure shows that 
there is a drop in the sticking probability for very low O 2 flow rates, but that it is relatively constant 
above 30 seem (1:1 HMDSO:C>2). 


The flatter shape of the measured data relative to the modeled curves is caused by the presence 
of multiple deposition species. This can be shown by how the model used for deriving the curves 
is developed [4], (The measured data in Figure 3 are superimposed upon curves generated by the 
model of Yuuki et al., and Kawahara et al. [4,1].) The model used is based upon several 
assumptions, including only one sticking probability that describes all depositing species, and a 
constant film thickness/trench depth ratio. The measured data in Figure 3, however, more closely 
match data from experiments by Tsai et al. [4,13] .They examined step coverage of PECVD a-Si:H 
films deposited from SifLi, a chemical system known to have multiple depositing species. Because 
of the deviation of the single species model from experimental evidence, Yuuki et al. changed the 
model by allowing calculations using two sticking probabilities [4], The calculations are performed 
on the assumption that the different species do not interact with each other on the surface. The best 
fit is then made by adjusting the sticking probability and the relative flux of each component. The 
shape of the multicomponent curve relative to the single component model shows significant 
deviation, but matches the experimental data [4]. The data from this experiment vary in a similar 
fashion, which provides experimental evidence for the assumption that the HMDSO/O 2 PECVD 
process has multiple deposition species for film growth. 

The other important feature to note in Figure 3 is that the d m in/d max values from the deeper 
aspect ratio trenches are slightly higher than the values from the wider aspect ratio trenches, (see, 
for example, the —1:1 aspect ratio trenches, for both sets of points in Figure 3). The effect is 
evident for each sample measured. This effect is due to the occlusion of the trench opening during 
film growth as shown by Blech, and Ross and Vossen [6,7]. As the film becomes thicker, the 
trench opening narrows effectively increasing the trench aspect ratio. The cumulative effect of 
trench occlusion and aspect ratio change is characterized by the ratio of the topwall film thickness 
over the trench depth (d max /D, occlusion ratio). As d max /D increases, trench occlusion becomes 
more pronounced. For shallower trenches the effect is more pronounced than for a deeper trench, 
since a given thickness of deposited material alters the trench cross section more. The increased 
aspect ratio of the unfilled cross section increases selective filtering of the higher sticking 
probability precursor. The material on the bottom of the trench will receive a disproportionately 
higher flux of the lower sticking probability material. This explains the behavior of d m i n /d ma x 
when comparing two trenches with similar aspect ratios but different trench depths. According to 
the models, d max /D shifts the curve by less than 2% for ratios wider than 2:1 [1,7]. For the 
presented data, in which the top wall film thickness is the same, for the 15 pm deep trenches 
dmax/D ranges from 0.17 to 0.28, and from 0.42 to 0.92 for the 5 pm deep trenches. It shows that 
the occlusion effects are relatively greater for shallower trenches. The greater trench occlusion 
lowers the relative flux reaching the trench walls, thus lowering step coverage within the trench. 

When analyzing the measured data from a PECVD system, it is necessary to take into account 
the effect of multiple deposition species and the difference in trench dimensions. By choosing data 
from a single aspect ratio, the generalized behavior of the sticking probability will be self- 
consistent with respect to independent parameters, such as gas flow. In general, it is preferable to 
select trenches with dimensions close to the size of interest. Since the objective of this work is to 
understand deposition on a planar substrate, wider trenches are preferable because they receive 
fluxes that are more representative of the multiple species that a planar surface receives. Deeper 
aspect ratio trenches receive disproportionately more low sticking probability material on the trench 
sidewalls and base. Furthermore, wide trenches are relatively less susceptible to occlusion effects 
of the trench opening. However, the sensitivity of the measurement decreases by more than a 
factor of 2 for aspect ratios between 1:2 and 5:1, and a sticking probability of ~0.3. The 2:1 aspect 
ratio trench was selected for data analysis in this experiment because it is the best compromise 
between estimation sensitivity and minimization of deep trench effects. 


Examining the details of the results provides several clues about the nature of the HMDSO/O 2 
deposition process. The shape of the step coverage curve in Figure 3 implies that there is more 
than one depositing species influencing film growth, and that they have widely varying sticking 
probabilities. Earlier work shows that at low O 2 concentrations, deposited films are essentially 
polymerized methylated siloxanes, but above a dilution of about 1:5, HMDS0:02, films are 
predominantly oxide [14]. The change in slope for the sticking probability shown in Figure 4 
occurred around 1:1 ratio. The direct relationship between sticking probability and oxygen flow 
rate is consistent with the work of Raupp et al., who examined SiC>2 film growth from TEOS and 
O 2 using microwave remote PECVD [15]. They attribute the increase in sticking probability to an 
increase in the surface reaction rate caused by higher atomic O production rates. For this 
capacitively-coupled system, a similar mechanism tied to the production rate of an excited oxygen 
species is likely to hold. Therefore, it is probable that surface oxidation reactions of the deposition 
precursor play a key role in film evolution for this process. Since a change in sticking probability 
is most often related to changes in surface composition or gas composition, the results imply that 
the incorporation processes may have relatively little influence on the final film composition. 


It has been shown that it is possible to estimate the sticking probability for PECVD processes 
in which the specific chemical species are not known, by quantifying step coverage of trenches. 
An equation has been found that fits the single component model curves in the range of 0.1 to 10 
aspect ratio to within 5 %. The fit of the experimental data with the published isoprobability plots 
has been attributed to the presence of multiple deposition species and the occlusion ratio, d max /D. 
The primary criterion of selecting trench dimensions is to find trench features that approximate 
actual conditions of interest. In order to make measurements on nominally flat substrates, it is 
necessary to select a wide trench to minimize occlusion effects. Models that include multiple 
deposition species will improve the estimation process. For the HMDSO/O 2 PECVD system, it has 
been demonstrated that the sticking probability decreases as the ratio of O 2 to HMDSO decreases 
and that this does not coincide with the incorporation of methylated siloxane groups within the 
film. These findings help demonstrate that 1) there is more than one deposition species, and 2) 
surface oxidation may play a key role in oxide film growth for this chemistry. 


[1] T. Kawahara, A. Yuuki, and Y. Matsui, Jpn. J. Appl. Phys., 30(3), 431 (1991). 

[2] K. Watanabe, and H. Komiyama, J. Electrochem. Soc., 137(4), 1222 (1990). 

[3] M. Ikegawa, and J. Kobayashi, J. Electrochem. Soc., 136(10), 2982 (1989). 

[4] A. Yuuki, Y. Matsui, and K. Tachibana, Jpn. J. Appl. Phys., 28(2), 212 (1989). 

[5] C. Y. Chang, J. P. McVittie, and K. C. Saraswat, IEDM 93 Proceedings, 853 (1993). 

[6] I. A. Blech, Thin Solid Films, 6, 113 (1970). 

[7] R. C. Ross, and J. L. Vossen, Appl. Phys. Lett., 45(3), 239 (1984). 

[8] L.-Y. Cheng, J. P. McVittie, and K. C. Saraswat, 2nd Inti. Symp. on ULSI Science and 

Technology, 586 (1989). 

[9] D. L. Kendall, Appl. Phys. Lett., 26(4), 195 (1975). 

[10] K. E. Bean, IEEE Trans, on Electron Devices, ED-25(10), 1185 (1978). 

[11] J. B. Price, 2nd Int’l. Symp. on Silicon Mater. Sci. and Technol., 120(3), 339 (1973). 

[12] J. A. Theil, J. Vac. Sci. and Technol. A, submitted (1995). 

[13] C.C. Tsai, J. C. Knights, G. Chang, and B. Wacker, J. Appl. Phys., 59(8), 2998 (1986). 

[14] J. A. Theil, J. G. Brace, and R. W. Knoll, J. Vac. Sci. and Technol. A, 12(4), 1365 (1994). 

[15] G. B. Raupp, D. A. Levedakis, and T. S. Cale, J. Vac. Sci. and Technol. A, 13(4), to be 

published (1995). 


Fourier Transform Infrared Spectroscopy of 
Polymer-Metal Interface Reactions 

B.H. Cumpston, J.P. Lu, B.G. Willis, and K.F. Jensen 
Department of Chemical Engineering 
Massachusetts Institute of Technology 
Cambridge, MA 02139 


Applications of Fourier transform infrared (FTIR) spectroscopy for probing 
polymer-metal interfaces are described with examples of polyimide (Pl)-metal systems 
relevant to electronics packaging and poly(phenylene vinylene) (PPV) derivatives used in 
electroluminescent devices. Emphasis is placed on the detection and interpretation of 
interfacial reactions that influence performance characteristics ( e.g ., adhesion, stability, 
and light emission) of polymer-metal structures. In situ infrared reflection absorption 
spectroscopy (IRRAS) is used to explore the formation of Pl-on-metal interfaces, the 
hydrolytic stability of such interfaces, and the reactivity of PPV systems when exposed to 
ultraviolet light and oxygen. Differences between polymer-on-metal and metal-on- 
polymer interfaces are discussed. Models of the infrared optical processes in the thin film 
composites are used to distinguish between chemical and optical effects. The FTIR 
observations are supported by additional spectroscopic characterizations, specifically ex 
situ X-ray photoelectron spectroscopy (XPS). 


Metal polymer composites are found in a large number of technological 
applications including electronic devices, solar cells, floppy discs, mirrors, compact discs, 
and optical waveguides [1]. In the fabrication of microelectronic devices, polymers are 
gaining use as low dielectric layers in multilevel metallization schemes, in addition to 
established packaging applications. Conjugated polymers are also receiving increased 
attention for electroluminescent devices [2] including light-emitting diodes, large area flat 
panel displays, and backlights for active matrix liquid crystal displays. Precise control of 
properties at the polymer-metal interface is crucial to product performance and reliability 
in all of these applications. For example, chemical changes at the interface between the 
polymer and the metal electrode have been identified as a possible explanation for the 
relatively short operating lifetimes of electroluminescent devices, on the order of one 
hundred hours, far too short for commercial applications. 

Building a microelectronic or optical device involves either the deposition of a 
metal onto a polymer or the formation of a polymer layer on top of the metal. In the 
metal-on-polymer (MOP) case, metal is deposited by sputtering or evaporation onto a 
fully cured polymer. The polymer interacts with a metallic surface formed by atoms or 
clusters of atoms impinging on the surface. In the polymer-on-metal (POM) case, a 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

solution of the polymer precursor is typically spin coated onto the metal and the precursor 
is then thermally converted to its final form. Alternatively, the polymer can be 
codeposited from separate vaporized monomer sources in a ultra high vacuum (UHV) 
chamber [3]. 

There are three major differences between the POM and MOP cases: (1) 
precursor versus polymer interaction with the metal, (2) presence or absence of solvent, 
and (3) thermal versus no thermal treatment after formation of the interface [4]. 
Additionally, in the POM case the polymer at the interface interacts with a native metal 
oxide, whereas in the MOP case the metal is truly metallic in nature. Kowalczyk et al 
found that the force of adhesion between the metal and polyimides is greater in the POM 
case [4] and attributed this effect to chemical reactions between the curing polymer and 
the underlying metal oxide. 

X-ray photoelectron spectroscopy (XPS) has been a particularly effective tool for 
analyzing chemistry underlying the formation of MOP interfaces [5-14], Being a UHV 
technique, XPS is readily compatible with investigations of the initial stages of interface 
formation by vacuum evaporation. By measuring atomic core levels of many 
metal/polymer systems using XPS, Burkstrand [15] was able to correlate force of 
adhesion, a macroscopic quantity, to the amount of oxygen present in the polymer, a 
microscopic property. He also found that metals with a high electronegativity are more 
likely to form oxygen complexes with the polymer, thus increasing adhesive strength. 
Chou and Tang [16] developed a simple model based on free energy calculations to 
predict whether or not a metal will have strong adhesion to a polymer; their model agrees 
with Burkstrand's experimental data. 

Detailed ab initio calculations have also been performed to identify the reactions 
taking place at the polymer-metal interface and the effect they may have on interfacial 
adhesion [17]. These self-consistent field molecular orbital calculations show excellent 
agreement with XPS data for core atomic layers, and also for valence levels, as 
determined experimentally by ultraviolet photoelectron spectroscopy (UPS) [18]. Many 
interface studies have focused on polyimides because of their importance in 
microelectronics applications [19]. Studies of other polymeric systems, e.g ., 
polyethylene (PE), polystyrene (PS), polypropylene (PP), polydimethyl-siloxane 
(PDMS), silicone, polyethylene terephthalate) (PET), poly(ether ether ketone) (PEEK), 
and polyvinylalcohol (PVA), have led to analogous conclusions about interfacial 
interactions [22]. 

A small electron escape depth (< 100 A) restricts interfacial analysis using XPS to 
very thin films, and the need for a UHV environment complicates studies of POM 
systems. FTIR, being an optical technique, has the advantage of not requiring a vacuum 
and, thus, it is capable of monitoring curing, oxidation, or degradation processes of 
polymers in situ in any environment [20]. The technique is particularly useful for 
probing the polymer-metal interface because it can detect changes in the polymer 
functional groups [21]. The utility of the technique has been demonstrated in a number 
of studies [6-8, 22-24], in particular, applications of polyimides to electronics packaging. 


In this paper we focus on FTIR analysis of POM interfaces. Since no single 
spectroscopy technique suffices to explain the complex chemical interactions taking place 
at the polymer-metal interface, other techniques, specifically XPS and Auger electron 
spectroscopy (AES), are used to complement the interpretation of interfacial behavior. 
With the aid of examples from our research on polyimides and conjugated 
electroluminescent polymers, we illustrate the kind of information about the interface 
chemistry that can be derived from the application of FTIR and complementary 


All polymer films were prepared by spin coating onto metallized silicon wafers. 
Polymer film thicknesses were then determined by ellipsometry or profilometry. 
Polymer film thicknesses were maintained below 500A to obtain a significant signal from 
the interface that is not washed out by bulk absorption. The metal films, nominally 
3000A thick, were prepared by e-beam evaporation (A1 and Cr) or sputtering (Cu) onto 
(100) Si wafers. 

FTIR was performed using a Nicolet Model 800 spectrometer capable of scanning 
from 400 cm' 1 to 4000 cm' 1 . For each FTIR scan, 8192 data points were collected, 
corresponding to a spectral resolution of 4 cm" 1 . 250 scans were taken and averaged for 
each spectrum to increase the signal-to-noise ratio. In situ polyimide curing studies were 
performed using the infrared reflection absorption (IRRAS) technique [25]. For this 
characterization, the sample was clipped onto the front of an A1 plate, which was heated 
from behind by a molybdenum wire resistive heater. A K-type thermocouple monitored 
the temperature on the sample side of the plate. The sample was placed in a stainless 
steel cell which could be purged with flowing gas or placed under vacuum. Calcium 
fluoride windows on the cell allowed the transmission of the IR beam. The IR beam from 
the interferometer passed through a wire-grid polarizer to obtain p-polarized light, 
parallel to the plane of incidence. The beam was incident on the sample at a grazing 
angle of approximately 20°, specularly reflected by the metal substrate, and detected by a 
room temperature deuterated triglycine sulfate (DTGS) detector. 

Curing studies of polyimide thin films were performed by heating the films in 
vacuum to 400 °C in the following stages after softbaking in nitrogen for 30 minutes at 
90 C: heat at 160 °C for one hour, 250 °C for 30 minutes, and 400 °C for 30 minutes. 
All curing was done at pressures less than 10 6 torr to prevent the thin film from burning 
off due to the presence of trace amounts of oxygen in inert gases. FTIR spectra were 
taken at each step in the curing process after cooling to room temperature in vacuum. 
This eliminates the need to correct the spectrum for vibrational transitions from excited 
energy levels. The polyimides studied are (i) biphenyltetracarboxylic dianhydride 
(BPDA) - p-phenylene diamine (PDA or PPD) and (ii) 4,4’-(hexafluroisopropylidene)- 
bis(phthalic anhydride) (HFDA) - 4,4’-bis(4-aminophenoxy) biphenyl (APBP). The 
structures of these two polymers are shown in Figure 1. 



Figure 1. Structures of BPDA-PDA and HFDA-APBP. 

Hydrolysis studies of the cured polyimide films were performed by placing the 
fully cured polymer-on-metal samples into an oven controlled at 85 °C and 85% relative 
humidity for 100 hours. Ex situ FTIR characterization of these films was performed 
using attenuated total reflection (ATR). In this method, the IR beam was multiply 
internally reflected in a KRS-5 crystal sandwiched between two pieces of the same 
sample. ATR is the method used for hydrolysis studies since the harsh treatment also 
effects the reflectivity of the metal substrate. These changes in reflectivity would 
produce artifacts in the IRRAS technique. 

Computer simulations of infrared spectra were performed using the 2x2 matrix 
method couple with the Kramers-Kronig relationship. Details of these calculations can 
be found elsewhere [26-28]. In brief, optical constants are calculated based on the 
transmission spectrum of an isotropic polymer sample. The matrix method can then be 
used to determine IRRAS spectra for any film thickness based on the optical constants. 
This technique is used to differentiate between optical effects and true chemical or 
orientation changes in the polymer film. 

Photo-oxidation studies of electroluminescent polymers are performed by 
exposing the polymer to dry air from an FTIR purge gas generator. Films are UV 
irradiated using an unfiltered low pressure ozone-free mercury lamp. The 254 nm line is 
the strongest and the incident intensity of this line is approximately 144 pW/cm 2 . The 
irradiance of the second most intense line at 436 nm is about 15 pW/cm 2 . The idealized 
general structure of the electroluminescent polymers studied is shown in Figure 2. 

Figure 2. Structure of poly(phenylene vinylene) (PPV) derivatives used in this study. 


Polymer films were examined ex situ by XPS using a Surface Science Instruments 
SSX100 spectrometer with monochromatic A1 k« (1486.6 eV) radiation. The resolution 
of the spectrometer was about 0.8 eV and a spot size of 600 gm was used. The thin 
polymer films were slightly conducting but 5 eV charge compensation was necessary. 
The samples were referenced to the gold 4f7/2 peak at 84.0 eV and to the carbon Is peak 
at 285.0 eV. To change the signal probing depth, the take-off angle of the photoelectrons 
was varied by tilting the sample. 


Effect of Film Thickness - Interface vs. Bulk Signal 

It is appropriate to first illustrate the ability of the IRRAS technique to observe 
chemistry at the polymer-metal interface. This is done by comparing the spectra of 
polymer films of decreasing thicknesses. As the film gets thinner, the ratio of interfacial 
signal to bulk signal increases, so that interfacial effects are more clearly observed. 
When BPDA-PDA is coated onto Au, the IR spectrum is independent of the thickness of 
the film. The chemical inertness of Au prevents interactions from occurring with the 
polyimide film. However, on more reactive metals, interactions at the interface are 
observed. This is illustrated in Figure 3 for the case of fully cured BPDA-PDA on 
Cr/Cr x O. 

1925 1825 1725 1625 1525 1425 132E 1925 1825 1725 1625 1525 1425 1325 

Wavenumber (cm -1 ) Wavenumber (cm' 1 ) 

Figure 3. IRRAS spectra of BPDA-PDA on Au and Cr/Cr x O 
as a function of polymer film thickness. 

The ratio of the imide C-N stretch (1359 cm" 1 ) to the semicircle stretch of the 
aromatic rings (1510 cm -1 ) decreases for thinner films, while the absorption band around 


1620 cm ’ 1 (quadrant stretching of aromatic rings) becomes broader, indicating that there 
is some interaction between the polymer and the Cr/Cr x O substrate. A modified 
polyimide layer is formed near the interface, which is significantly different from the bulk 
film. Angle-resolved XPS experiments show a nitrogen deficiency at the interface as the 
result of a carboxylate intermediate [29]. Fully cured BPDA-PDA films do not display 
the same behavior when coated onto AI/AI 2 O 3 , suggesting that there is less interaction 
between the polymer and this substrate. 

Simulation of IR Spectra: 

Different IR sampling techniques can give significantly different results. 
Reflection spectra often show changes in peak positions and shapes relative to 
transmission spectra, particularly when studying very thin films [27, 28]. It is important, 
therefore, to have a detailed understanding of optical effects in the film so that these 
effects are not misinterpreted as chemical changes. The curing of HFDA-APBP, an 
isotropic polyimide, will be used to illustrate this point. The left side of Figure 4 shows 
experimental IRRAS spectra for fully cured HFDA-APBP films on Cr/Cr x O. 

Wavenumber (cm' 1 ) Wavenumber {cm' 1 ) 

Figure 4. Experimental (left) and simulated (right) IRRAS spectra of fully cured HFDA- 
APBP films on Cr/Cr x O as a function of film thickness. Numbers on the left-hand side 
refer to dilution ratios yielding approximately the thickness shown on the right-hand side. 

Qualitatively, the presence of all peaks is seen in each of the spectra; however, 
there are differences in the thickest film. In particular, the relative intensities of the two 
aromatic ring semicircle stretching peaks at 1510 cm'l and 1495 cnr* have reversed in 
the 2 jim film. It is also apparent that the peaks in the thickest film are at slightly lower 
frequencies than in the thinner films. The right side of Figure 4 shows simulation results 


for HFDA-APBP on Cr/Cr x O as a function of film thickness. The simulation results 
show excellent agreement with experiments, indicating that the observed spectral changes 
are the result of optical effects only, and are not due to chemical or orientation changes in 
the polyimide film. 

In Situ Investigations of Curing Chemistry 

The curing of polyamic acid precursors to form polyimides may also be studied 
using the IRRAS technique. Figure 5 shows the temperature evolution of selected 
absorption bands during the curing of BPDA-PDA on an AI/AI 2 O 3 substrate. On this 
substrate, BPDA-PDA follows a very characteristic cure. After soft-baking, the spectrum 
is dominated by bands from the polyamic acid precursor. After heating to 160 °C, imide 
C=0 and C-N peaks increase at the expense of polyamic acid features, such as the amide 
C=0. An anhydride species is also observed, which is formed as an intermediate and 
then eventually converted to the polyimide. The anhydride intermediate forms due to the 
back-reaction of the precursor to its diamine and dianhydride components. The imide 
C-N stretching mode at 1350 cm -1 is typically used as the reference for quantifying the 
imidization reaction, since there is no interference from other species in this region of the 
spectrum. The film appears to be fully imidized after heating at 400 °C, when all peaks 
associated with the precursor polymer have disappeared. 

Temperature (°C) 

Figure 5. Curing process of BPDA-PDA on AI/AI 2 O 3 . 

Thin BPDA-PDA films (~400A) cured on Cr/Cr x O behave very similarly to the 
film cured on Al. Again, imide features begin to form after heating to 160 °C and are 


fully formed after heating at 400 °C. An anhydride intermediate is also detected during 
curing, but converts completely to the imide. However, it has been shown that when 
extremely thin BPDA-PDA films (~100A) are cured on Cr/Cr x O in nitrogen, a 
carboxylate intermediate is detected in addition to the anhydride [30]. This carboxylate 
intermediate is also completely converted to the imide by 250 °C. Since the carboxylate 
species were not detected in the thicker films, we may conclude that the carboxylate form 
in small amounts very near the interface. In the case of thicker films, the stronger bulk 
signal washes out information from the interface. 

BPDA-PDA cured on Cu/Cu x O exhibits markedly different behavior. Imide 
features begin forming at 160 °C, but above this temperature we see large amounts of a 
carboxylate species forming. As the cure progresses to higher temperatures, the 
carboxylate (as well as the anhydride) intermediates are converted to the imide. 
However, even after curing at 400 °C, a small amount of a polyamic acid-like carbonyl 
persists at 1723 cm' 1 . Again, this suggests that there is a modified polyimide layer near 
the polymer-metal interface [29]. 

It is evident, since the carboxylate species is observed in these relatively thick 
films on Cu/Cu x O, that more of this intermediate is formed than when the polymer is on 
Cr/Cr x O. Based on these results, a ranking of the level of interaction can be written as 
follows: Cu/Cu x O > Cr/Cr x O > A1/A1 2 0 3 . Figure 6 shows spectra of BPDA-PDA cured 
on Cu, Al, and Cr substrates at the intermediate temperature of 160°C. The carboxylate 
intermediate is not detected when HFDA-APBP is cured on any of the substrates. This 
polyimide probably does not interact as strongly with the metal/metal oxide because of 

Figure 6. Intermediate curing step for BPDA-PDA on three different metal substrates. 


Relevance to Adhesion and Environmental Stability. 

Strong adhesion of the polymer to the metal/metal oxide interface is important in 
virtually every application of these systems. The level of chemical interaction between 
the polymer and the substrate is directly related to the force of adhesion at the interface. 
Island blister adhesion measurements [31] have been performed for BPDA-PDA on both 
AI/AI 2 O 3 , where the interaction is weak, and Cr/Cr x O, where the interaction is stronger 
[30]. The force of adhesion data range from 28 to 51 J/m 2 on AI/AI 2 O 3 and from 384 to 
721 J/m 2 on Cr/Cr x O. Although the results demonstrate a fair amount of scatter, which is 
typical for adhesion measurements, it is clear that there is greater strength at the Cr/Cr x O 
interface, thus demonstrating a higher degree of chemical interaction. 

Wavenumber (cm-1) 

Figure 7. BPDA-PDA on three different metal substrates after 100 hours at 85 °C and 

85% relative humidity. 

Environmental stability is also of great concern to the stability of the interface 
since applications may involve high temperatures and/or high humidity, which accelerate 
existing degradation processes or initiate entirely new chemical pathways. To study these 
effects, the polyimide-on-metal structures were fully cured, then removed from vacuum 
and transferred to an oven held at 85 °C and 85% relative humidity. After treatment for 


100 hours, the samples were removed and examined by ATR. ATR spectra of BPDA- 
PDA films on AI/AI 2 O 3 , Cr/Cr x O, and Cu/Cu x O after treatment, are shown in Figure 7. 

On AI/AI 2 O 3 , the intensity of the peaks associated with the imide moieties has 
decreased slightly after treatment. In addition, bands due to aluminum oxide (960 cm' 1 ) 
and hydroxide (1100 cm' 1 ) grow. This indicates that air and/or moisture is diffusing 
through the polymer film and further oxidizing the substrate. On Cr/Cr x O, the polymer 
shows a slightly greater attenuation of the imide features than on AI. There is either a 
carboxylate or carbonate forming during the hydrolysis treatment, evidenced by a broad 
feature centered around 1400 cm' 1 . When BPDA-PDA is treated on Cu/Cu x O, the film is 
virtually destroyed with virtually no imide-related bands remaining. There is a large 
formation of CU 2 O (600 cm 4 ), which results from oxidation of the substrate. These 
results show that although a moderate amount of chemical interaction at the polymer- 
metal interface is useful in promoting adhesion, too much interaction can lead to 
catastrophic failure of the interface under environmentally rigorous situations. By 
comparison, the less reactive HFDA-APBP show much greater hydrolytic stability on all 
three substrates. 

Bulk and Interfacial Reactions Relevant to Electroluminescent Polymers, 

With the recent discovery that polymers can be used in electroluminescent (EL) 
devices [ 2 ] there has come a concentrated effort to understand how these materials 
behave and how to improve device performance. These light emitting devices provide a 
technologically exciting new field in which to study polymer/inorganic interfaces. The 
basic structure of a simple polymeric light-emitting device is shown in Figure 8 . In such 
a device, there is a polymer-metal interface and a polymer/ITO interface, where ITO is 
indium tin oxide. The rigorous environmental condition that these devices are exposed to 
is a large electric field of about 1 MV/cm, which produces local heating. In addition, the 
polymers themselves are prone to photo-degradation in air. 

| | glass 
0 ITO 

H|] polymer {-1000 A) 

IB metal 

Figure 8 . Typical polymeric light-emitting diode device structure. 

The metallic electron injecting contact has proven to be particularly problematical 
in these devices. Calcium and silver/magnesium alloys have been shown to be the most 
efficient electrodes, but the reactivity of these metals dictates that special encapsulation 
steps must be taken in order to operate the device under ambient conditions. Aluminum 
electrodes have been used, but one must sacrifice light-emitting efficiency for 
improvements in lifetime. Short operating lifetimes continue to plague polymer-based 
EL devices. It is likely that some air permeates through encapsulation materials or is 


present as an impurity in the polymer; it is unclear whether the polymer and/or the metal 
become oxidized during device operation. 

We have studied both photo- and electric field-induced degradation of a family of 
polymers having a conjugated backbone and alkoxy side groups to improve the 
processability of the polymer (c/. Figure 2). IRRAS has been coupled with XPS and 
transmission IR spectroscopy to explain the bulk degradation of the polymer film under 
these conditions. When films were exposed to dry air in the absence of light, no change 
was observed by IRRAS. Also, when films were exposed to UV irradiation in a vacuum 
of less than 10~ 6 torr, no change was observed after up to 48 hours of exposure. 
However, when films were simultaneously exposed to UV and air, the polymer oxidized. 
The evolution of the spectrum for a film of poly(2-methoxy,5-(2'-ethylhexoxy)-l,4- 
phenylene vinylene) (MEH-PPV) cast onto A1 is shown in Figure 9. The most obvious 
change is the growth of the carbonyl peak around 1723 cm' 1 . At the end of 36 hours of 
exposure, the carbonyl peak has shifted from 1723 cm* 1 to 1745 cm -1 . Also, there is a 
marked attenuation of the peaks associated with the phenyl rings at 1500, 1415, and 1035 
cm -1 . Ex situ ATR spectra confirmed that the ring structure was destroyed and not 
simply reoriented in the film with reduced sensitivity of the IRRAS technique. 

Wavenumber Icm' 1 ! 

Figure 9. IRRAS spectra of 400 A films of MEH-PPV on Al. Spectra are shown for the 
as-coated film (a), and after exposure to dry air and UV irradiation for (b) 2h, (c) 5h, (d) 

12 h, (e) 24 h, and (f) 36 h. 


The band at 975 cm -1 , associated with the vinyl double bond in the polymer 
backbone, is attenuated, but not fully destroyed during the oxidation. The decrease in 
conjugation length is also observed by the growth of the phenyl band at 1600 cm’ 1 which 
becomes IR active as the symmetry of the polymer is destroyed. The position of the 
carbonyl peak along with the persistence of the C-O peak at 1200 cm' 1 is indicative of an 
ester or acid and this assignment is future supported by XPS data. 

The observations of FTIR and XPS may be explained in terms of the mechanism 
illustrated Figure 10. This mechanism involves singlet oxygen formed by energy transfer 
from the polymer during photoluminescent operation, and since photoluminescence and 
electroluminescence pathways are identical for PPV derivatives [32], electroluminescent 
operation could likewise produce excited-state singlet oxygen. This singlet oxygen then 
undergoes a 1,2-cycloaddition across the vinyl double bond in the polymer backbone, 
shortening the conjugation length and blue-shifting the light emission. This oxidized 
segment subsequently initiates a free radical process ultimately leading to the formation 
of carbonyl containing groups such as esters [33]. The carbonyl group is known to 
quench luminescence in the polymer by preventing the recombination of electron-hole 
pairs [34, 35]. 

Figure 10. Proposed mechanism for the photo-oxidation of PPV derivatives. 


The utility of Fourier transform infrared (FTIR) spectroscopy in probing and 
understanding polymer-metal interfaces has been illustrated with examples of polyimide 
(Pl)-metal systems relevant to electronics packaging and poly(phenylene)vinylene (PPV) 
derivatives used in electroluminescent devices. In situ infrared reflection absorption 
spectroscopy (IRRAS) provides insight into the formation of Pi-on-metal interfaces, the 
hydrolytic stability of such interfaces, and the reactivity of PPV systems when exposed to 
ultraviolet light and oxygen. Models of the infrared optical processes in the thin film 
composites further assist in distinguishing between chemical and optical effects. While 
the bulk degradation of the PPV derivatives is relevant to the stability of 
electroluminescent polymer-based devices, it is also important to note that interfacial 
reactions occur to a considerable extent in these systems. At the polymer-metal interface, 


electrochemical reactions take place that lead to the oxidation of the metal and the 
formation of gas, which escapes by rupturing the surface of the device. There is also the 
possibility that metal ions may diffuse into the polymer film, driven by the electric field. 
There is still much to be understood about the degradation of these devices, but the IR 
techniques, complemented with those of other surface analyses, are useful tools, as we 
have demonstrated for polyimide films. 


The authors thank The Electronics Packaging Program at MIT for support of the 
PI metal interface investigations, and Ian Parker and Floyd Klavetter of Uniax, Corp. for 
providing MEH-PPV samples. 


1. K.M. Anetsberger, in Metallized Plastics 1: Fundamental and Applied Aspects, 
edited by K.L. Mittal (Plenum Press, New York, 1989), p. 29. 

2. J.H. Burroughes, D.D.C. Bradley, A.R. Brown, R.N. Marks, K. Mackay, R.H. 
Friend, P.L. Burns, and A.B. Holmes, Nature 347, 539 (1990). 

3. J.R. Salem, F.O. Sequida, J. Duran, W.Y. Lee, and R.M. Yang, J. Vac. Sci. 
Technol. A4, 369 (1986). 

4. S.P. Kowalczyk, Y. Kim, G.F. Walker, and J. Kim, Appl. Phys. Lett. 52, 375 

5. R. Haight, R.C. White, B.D. Silverman, and P.S. Ho, J. Vac. Sci. Technol. A6, 
2188 (1988). 

6 . D.Y. Shih, J. Parasczak, N. Klymko, R. Flitsch, S. Nunes, J. Lewis, C. Yang, J. 
Cataldo, R. McGouey, W. Graham, R. Serino, and E. Galligan, J. Vac. Sci. 
Technol. A7, 1402 (1989). 

7. D.Y. Shih, N. Klymko, R. Flitsch, J. Parasczak, and S. Nunes, J. Vac. Sci. 
Technol. A9, 2963 (1991). 

8 . M.C. Burrell, P.J. Codella, J.A. Fontana, J,J. Chera, and M.D. McConnell, J. Vac. 
Sci. Technol. A7, 55 (1989). 

9. T. Strunkus, C. Hahn, and D. Frankel, J. Vac. Sci. Technol. A9,1272 (1991). 

10. MJ. Goldberg, J.G. Clabes, and C.A. Kovac, J. Vac. Sci. Technol. A6, 991 

11. S.C. Freilich and F.S. Ohuchi, Polymer 28, 1908 (1987). 

12. J.W. Bartha, P.O. Hahn, F. Legoues, and P.S. Ho, J. Vac. Sci. Technol. A3, 1390 

13. R.M. Tromp, F. Legoues, and P.S. Ho, J. Vac. Sci. Technol. A3,782 (1985). 

14. G.M. Adema, I. Turlik, L.T. Hwang, G.A. Rinne, and MJ. Berry, IEEE Trans. 13, 

15. J.M. Burkstrand, J. Appl. Phys. 52,4795 (1981). 

16. N.J. Chou and C.H. Tang, J. Vac. Sci. Technol. A2, 751 (1984). 

17. A.R. Rossi, P.N. Sanda, B.D. Silverman, and P.S. Ho, Organometallics 6, 580 


18. P.S. Ho, in Principles of Electronic Packaging , edited by D.P. Seraphin, R.C. 
Lasky, and C. Li (McGraw Hill, New York, 1989), p. 811. 

19. H. Satou, H. Suzuki, and D. Makino, in Polyimides , edited by D. Wilson, H.D. 
Stenzenberger, and P.M. Hergenrother (Chapman and Hall, New York, 1990), p. 

20. J.P. Lu, D.Volfson, F. Trusell, and K.F. Jensen, in Chemistry at Polyimide-Metal 
Interfaces: in situ FTIR Studies of Polymer Curing Processes and Thermal 
Stability , 1993 (Materials Research Society), p. 518. 

21. J.L. Koenig, Spectroscopy of Polymers (American Chemical Society, Washington, 
D.C., 1992). 

22. W.C. Stewart, J. Leu, and K.F. Jensen,, 1989 (Materials Research Society), p. 


23. D.S. Dunn and J.L. Grant, J. Vac. Sci. Technol. A7, 253 (1989). 

24. K. Kelley, Y. Ishino, and H. Ishida, Thin Solid Films 134, 271 (1987). 

25. R.G. Greenler, J. Chem. Phys. 44,310 (1966). 

26. W.R. Salaneck and J.L. Bredas, Synthetic Metals 67,11 (1994). 

27. D.L. Allara, A. Baca, and C.A. Pryde, Macromolecules 11, 1215 (1978). 

28. A.N. Parikh and D.L. Allara, J. Chem. Phys. 96, 927 (1992). 

29. J.P. Lu, P.V. Nagarkar, D. Volfson, F. Trusell, K.F. Jensen, and S.D. Senturia', J. 
Vac. Sci. Technol. in press (1995). 

30. M.M. Richter, F.R. Fan, F. Klavetter, A.J. Heeger, and A.J. Bard, Chemical 
Physics Letters 226, 115 (1994). 

31. H. Nishino, G. Yu, A.J. Heeger, T.A. Chen, and R.D. Rieke, Synthetic Metals 68, 
243 (1995). 

32. C. Zhang, S. Hoger, K. Pakbaz, F. Wudl, and A.J. Heeger, J. Electronic Materials 

33. H. Vestweber, J. Pommerehne, R. Sander, R.F. Mahrt, A. Greiner, W. Heitz, and 
H. Bassler, Synthetic Metals 68,263 (1995). 

34. F. Papadimitrakopoulos, M. Yan, L.J. Rothberg, H.E. Katz, E.A. Chandross, and 
M.E. Galvin, Mol. Cryst. Liq. Cryst. in press (1995). 

35. M. Yan, L.J. Rothberg, F. Papadimitrakopoulos, M.E. Galvin, and T.M. Miller, 
Physical Review Letters 73,744 (1994). 



*AT&T Bell Laboratories, Murray Hill, NJ 07974 

**Institute of Materials Science, University of Connecticut, Storrs, CT 06269 


The chemical and electronic properties of aluminum/poly(p-phenylenevinylene) (PPV) 
interfaces were studied in situ using x-ray photoelectron spectroscopy (XPS). It was observed 
that the aluminum atoms react with the oxygen-containing groups present as impurities 
on the surface of PPV to form Al-O-C linkages. The Al atoms also interact with the tt- 
system of the polymer as indicated by changes in the valence band. Contrary to to recent 
suggestions (Ettedgui et al 1 ) the relation between surface oxygen content and band bending 
is not straightforward, as shown by deposition on PPV surfaces prepared by two different 
synthetic routes. 


One of the factors affecting the stability and reliability of electroluminescent devices 
using organic materials as light emitters is the metal/organic interface. In the case of PPV, 
aluminum has been suggested for use as the electron-injecting electrode. Recently, there 
have been several experimental and theoretical studies of the interface between aluminum 
and conjugated polymers as well as small molecules. 1-5 

Ettedgui et al. 1 studied Al/PPV interfaces using XPS. They concluded that the onset 
of polymer band bending (i.e. shift of the Cls peak to higher BE with Al coverage) depends 
on the oxygen surface concentration (a parameter dependent on sample preparation). They 
observed the onset of band bending with as little as 1 A coverage for 5% surface oxygen, and 
as much as 30 A for 10% surface oxygen. 

In this paper we report results of in-situ XPS measurements during Al deposition on 
PPV surfaces. From the C Is, O Is and Al 2 p core levels as well as the valence band we 
get information about the chemical and electronic properties of the Al/PPV interface as it 


The PPV films were prepared by converting the precursor at 200°C in an inert atmo¬ 
sphere. The precursor was synthesized using two different routes. The first one has been 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

described elsewhere, 6 and the second one is proprietary. The surface concentration of oxy- 
gen of the films was around 4 atom % when measured at 10° take-off angle using XPS. The 
exposure of the samples to air was limited to a few minutes during sample loading. During 
sample preparation and transfer care was taken to avoid the simultaneous exposure of the 
samples to both oxygen and light in order to prevent photo-oxidation of the polymer found 
to occur when PPV is exposed to light in the presence of oxygen. 7 

The XPS experiments were performed using a Scienta 300 spectrometer at Lehigh Uni¬ 
versity. Details about the instrument and the metal deposition can be found elsewhere. 8-9 

Binding Energy (eV) 

Binding Energy (eV) 

Figure 1. C Is XPS spectra of (a) PPV films at 10° and 90°, and (b) PPV films at 10° as a function of Al 


Fig. la shows the C Is spectrum of a PPV film (1st synthetic route) acquired at 
two take-off angles. Besides the main peak at 284.8 eV originating from the aromatic and 
aliphatic parts of the molecule, there is more structure at high binding energy, which is 
better shown in the tenfold expanded spectra. The peak at 286.5 eV corresponds to carbon 
single-bonded to oxygen, which could come from either hydroxyl or methoxy groups, while 
the peak at 288.2 eV corresponds to carbonyl carbons. The higher relative intensity of 
the carbonyl peak at 10° photoelectron take-off angle indicates that the surface is richer in 
carbonyl groups. The two broad peaks observed at 291.5 eV and 294.5 eV are the tt — tt* 
shake-up peaks originating from the phenylene and vinylene groups, respectively. The loss 
of the latter at 10° indicates the lack of vinylene groups in the near surface region, which is 
consistent with the increased concentration of oxygen impurities (such as carbonyl, methoxy, 
and hydroxy substituents on the vinyl group) at the surface. This is also supported from 
the surface composition data calculated by integrating the C Is and O Is peaks. At 10 , 
with a sampling depth of approximately 15 A , the fraction of oxygen is 4.1 atom %, while 


at 90° with a 90 A sampling depth, the fraction is only 1.6 atom % oxygen. It should 
be noted here that these values for the atomic composition represent a convolution of the 
actual concentration profile and a weighting function that decays exponentially with depth. 
More detailed angle resolved experiments 1 0 show that the actual oxygen concentration on 
the surface is approximately double the concentration measured at 10° take-off angle. 

The C Is spectra after Al metal deposition of 0, 2, 4, 8, 12, and 20 A are shown in for a 10° take-off angle. There are no significant shifts in the main peak that would 
indicate band bending. It is observed that at the lowest coverage of 2 A the carbonyl peak 

Binding Energy (eV) 

Binding Energy (eV) 

Figure 2. (a) 10° Al 2p, and (b) valence band XPS spectra of a PPV film as a function of Al coverage. 

Fig. 2a shows the Al 2 p spectra as a function of coverage. At the lowest coverage of 2 
A there is a mixture of oxide and metallic aluminum on the surface. The oxide peak comes 
from Al atoms that have reacted with the carbonyl and other oxygen containing groups of 
the surface. As the coverage increases, the metal-to-metal oxide ratio increases indicating 
that the oxide is located at the A//PPV interface and not at the metal vacuum interface (as 
would be the case if the metal was oxidized by residual oxygen in the vacuum chamber). 

Recent theoretical calculations and UPS experiments 4 have shown that the valence band 
spectrum of PPV has two distinct bands, one around 2 eV and the other around 4 eV. These 
are associated with the vinylene and phenylene groups of the polymer chain, respectively. 
The valence band spectrum of our sample acquired using XPS at 10° and 90° take-off angles 
is shown in Fig.2b. Also shown is the same region after a 2 A Al metal deposition. It is 
observed that, at 10° the sharp band at 4.1 eV (b) associated with the phenylene group 
disappears. At 90°, on the other hand, the phenylene peak is only slightly attenuated even 
after a 2 A Al deposition. The broad vinylene peak around 2 eV (b) is only observed at 
90°, in agreement with the previous conclusions that the top surface is depleted of vinylene 
groups due to substitution by oxygen containing groups. 



The C Is core level spectra of PPV, converted under the conditions described above, 
show that the surface of the film contains oxygen impurities in the form of carbonyl, methoxy, 
and hydroxy substituents on the vinylene group. The hydroxy and methoxy substituents 
are products of side reactions during the preparation of the soluble PPV precursor, while 
the carbonyl substituents are products of oxidation during the thermal conversion of the 
precursor to PPV. 7 The higher concentration of carbonyl groups on the surface than in the 
bulk, as evidenced by the angle resolved spectra, makes these groups the primary targets of 
reaction with the vapor-deposited Al atoms. At the lowest Al metal coverage examined (2 
A) the carbonyl peak disappears due to the reaction with the metal atoms, with Al — 0 — C 
the most probable product. The 0 Is and Al 2 p spectra also support this conclusion. The 0 
Is peak shifts to lower binding energy where metal oxide binding energies are found, while 
the Al 2 p peak shows a mixture of Al metal and Al - O. Reaction of the Al atoms with 
hydroxy or methoxy groups is more difficult to discern, since the shift in the binding energy 
would be small compared to the original value. 

Further evidence of interaction of the metal with the polymer comes from the valence 
band spectra, where the signature peak of the phenylene peak disappears at a coverage 
of 2 A. Theoretical modeling and experiments with model compounds 5 have shown that 
when Al metal atoms interact with a pristine PPV surface, the result is reaction with the 
vinylene group and disruption of the conjugation. In the present case, the oxygen impurities 
immediately react with the metal atoms. However, the results suggest that the surface 
phenylene groups also take part in the reaction with the deposited metal atoms at the early 
stages of deposition. This discrepancy with theoretical predictions and experimental data on 
oxygen free surfaces is due to the presence of oxygen impurities on the surface of the sample. 
These impurities deplete the near surface region of vinylene groups and leave the phenylene 
groups available for interaction with the Al atoms. 

Following a 2 A Al deposition, the phenylene peak in the PPV valence band spectrum 
disappears at a 10° take-off angle. However, it is clearly present in the 90° spectrum (Fig.5). 
This observation indicates that the reaction is limited to the near surface region, and no 
significant diffusion of the metal atoms has occurred. The lack of metal atom diffusion into 
the bulk of the polymer is indicative of the high reactivity of the metal atoms with the 
surface oxygen impurities. The high reactivity results in cluster growth of the Al film, which 
is reflected in small shifts in the Al 2 p binding energy. These results have been explained 
in detail elsewhere. 9 Similar shifts have also been observed during titanium deposition on 

self-assembled monolayer surfaces terminated with groups of varying reactivity. 

No band-bending was observed in the PPV C Is spectra even at the highest Al coverage 

(20 A). This is in disagreement with previously published work by Ettedgui et al. 1 who 
observed a gradual shift of the C Is peak of approximately 0.5 eV. They correlated the 
onset of band-bending with the oxygen content of the surface, claiming that higher surface 
oxygen content results in the formation of a thicker oxide layer that delays (or even prevents) 


the onset of band bending to higher metal coverage by preventing the metal from interacting 
directly with the polymer. We have presented direct evidence (i.e. valence band spectra) 
that Al does interact with the polymer, and even with oxygen-free groups such as the phenyl 
groups. Despite the fact that our films were not converted in-situ in the UHV system (as 
Ettedgui et al. 1 did with their samples), the oxygen content of our films is 4 atom %, which 
is lower than the 5 atom % reported by Ettedgui et al 1 However, no significant or consistent 
shift was observed, although we noticed a slight broadening of the C Is peak. In addition, 
the energy resolution was much better in our case judging from the FWHM of the C Is peak 
ofPPV (0.9 eV vs. 1.5 eV). 

Figure 3. (a) C Is, and (b) Al 2 p XPS spectra of a PPV film (alternate synthetic route) at 10° as a function 
of Al coverage. 

Preliminary results on PPV prepared using a proprietary alternate synthetic route have 
shown that band-bending and surface oxygen concentration are not related in a straightfor¬ 
ward manner. In particular, deposition on the PPV surfaces has resulted in immediate C Is 
band bending* (Fig. 3a) even though the surface oxygen concentration of the new sample 
was the same as the old. The new route however differs from the old one in that it provides 
PPV films free of carbonyl groups. This observation suggests that the kind of oxygen present 
in the film and not only its total concentration is what determines band-bending. This ob¬ 
servation indicates the important role carbonyl groups play in the electronic properties of 
the interface, in addition to the dramatic effect they have on the luminescence of the film 
itself. 6 

* Note that the Al 2p spectrum (Fig.3b) does not shift at all, ruling out charging as the cause for the 
C Is shift. It is also noteworthy that metallic Al appears only at the last deposition step, although band 
bending is directly correlated to interaction of the polymer with the metal. 



Our results show that, during the formation of the Al/PPV interface, oxygen-containing 
groups (primarily carbonyls near the surface) are the primary reactive sites for the deposited 
metal atoms, resulting in the formation of Al — O — C linkages with the surface. Contrary 
to theoretical predictions, Al atoms do interact with the phenylene groups of the polymer 
molecule as changes in the valence band suggest. The high reactivity of the oxygen-modified 
PPV surface results in the immobilization of the deposited metal atoms, preventing surface 
and bulk diffusion and promoting cluster growth of the metal overlayer. Our results also 
suggest that the relationship between the onset of band bending and surface oxygen content 
is more complex than previous reports 1 suggested. 


The authors would like to thank Sehwan Son of AT&T Bell Laboratories for the syn¬ 
thesis and preparation of the PPV films (second route). 


l E. Ettedgui, H. Razafitrimo, K. T. Park, Y. Gao, and B. R. Hsieh, J. Appl. Phys. 75, 7526 

2 Y. Gao, K. T. Park, and B. R. Hsieh, J. Appl. Phys. 73, 7894 (1993). 

3 P. Dannetun, M. Boman, S. Stafstrom, W. R. Salaneck, R. Lazzaroni, C. Fredrickson, J. L. 
Bredas, R. Zamboni, and C. Taliani, J. Chem. Phys. 99, 664 (1993). 

4 P. Dannetun, M. Logdlund, M. Fahlman, M. Boman, S. Stafstrom, W. R. Salaneck, R. 
Lazzaroni, C. Fredrickson, J. L. Bredas, S. Graham, R. H. Friend, A. B. Holmes, R. Zamboni, 
and C. Taliani, Synth. Metals 55-57, 212 (1993). 

5 C. Fredrickson, R. Lazzaroni, J. L. Bredas, P. Dannetun, M. Logdlund, and W. R. Salaneck, 
Synth. Metals 55-57, 4632 (1993). 

6 F. Papadimitrakopoulos, K. Konstadinidis, T. M. Miller, R. Opila, E. A. Chandross, and 
M. E. Galvin, Chem. Mater. 6, 1563 (1994). 

7 F. Papadimitrakopoulos, M. Yan, L.J. Rothberg, H.E. Katz, E.A. Chandross and M.E. 
Galvin, Mol. Cryst. Liq. Cryst. 69, 663 (1994). 

8 K. Konstadinidis, P. Zhang, R. L. Opila, and D. L. Allara, Surf. Sci. accepted (1995). 

9 K. Konstadinidis, F. Papadimitrakopoulos, M. Galvin, and R. L. Opila, J. Appl. Phys, in 
press (1995). 

10 F. Papadimitrakopoulos, K. Konstadinidis, and M. E. Galvin, Macromol. submitted 


Part V 

Primers for Interface Preparation 


Y. M. Tsai*, F. J. Boerio*, and Dong K. Kim** 

♦Department of Materials Science and Engineering, University of Cincinnati, Cincinnati, Ohio 

**The Goodyear Tire & Rubber Co., Akron, Ohio 44309-3531 


Plasma polymerized acetylene films contained mono- and di-substituted acetylene 
groups, aromatic groups, and carbonyl groups which resulted from reaction of residual free 
radicals with oxygen when the films were exposed to the atmosphere. There was some evidence 
for formation of acetylides in the interphase between the films and the substrates. Reactions 
occurring in the interphase between the plasma polymerized films and natural rubber were 
simulated using a model rubber compound consisting of a mixture of squalene, zinc oxide, 
carbon black, sulfur, stearic acid, diaryl-p-diphenyleneamine, and N,N-dicyclohexyl- 
benzothiazole sulfenamide (DCBS). Zinc oxide and cobalt naphthenate reacted with stearic acid 
to form zinc and cobalt stearates. The stearates reacted with the benzothiazole sulfenamide 
moiety of DCBS and with sulfur to form zinc and cobalt accelerator complexes and 
perthiomercaptides. The complexes and perthiomercaptides reacted with squalene and the 
plasma polymer to form pendant groups which eventually disproportionated to form crosslinks 
between squalene and the primer. Migration of double bonds during reaction of the model 
rubber compound with the films resulted in formation of conjugated double bonds in squalene. 


Recently we have been interested in the use of plasma polymerized acetylene films as 
primers for rubber-to-metal bonding. The breaking strength of miniature lap joints prepared 
using natural rubber as an “adhesive” to bond together steel substrates coated with thin films of 
plasma polymerized acetylene was equal to that of similar joints prepared from brass substrates 
[1]. FTIR spectra of the as-deposited films were characterized by bands related to mono- and di- 
substituted acetylene, methyl, and methylene groups. Peaks related to aromatic structures were 
observed in positive SIMS spectra. Results from AES showed that the oxide on the substrate 
surface was partly reduced during deposition [1,2]. 

An important question regarding the use of plasma polymerized films as primers for 
rubber-to-metal bonding concerns reactions in the interphase between natural rubber (NR) and 
the primer. These interphases are difficult to characterize because they are buried and because 
the reactions which occur are complex. Several investigators have simulated reactions occurring 
in the interphase between NR and brass utilizing model compounds in which NR is replaced by a 
low molecular weight organic compound having a similar structure [3-6], We have reported 
preliminary results in which a model compound containing squalene instead of NR was used to 
simulate reactions in interphases between NR and plasma polymerized acetylene films [1]. The 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

purpose of this paper is to report new results obtained using infrared spectroscopy to characterize 
reactions between plasma polymerized acetylene films and the model rubber compound. 


1010 cold-rolled steel (CRS) sheets 0.5 mm thick were obtained from The Goodyear Tire 
and Rubber Company (Akron, Ohio). The as-received steel substrates were mechanically 
polished, rinsed in distilled water and acetone, and blown dry with nitrogen. 

Plasma polymerization and etching were carried out in a tubular reactor which was 
inductively coupled to an RF (13.56 MHz) power supply. A detailed description of the reactor 
has been presented elsewhere [7]. Prior to plasma polymerization, the polished substrates were 
etched in an argon plasma for 10 minutes. Plasma polymerized films of acetylene were then 
deposited on the substrates using argon as the carrier gas. The thickness of the films was 
determined by using a Rudolph Research Model 436 ellipsometer to examine polished steel 
substrates before and after deposition of the films. 

Reflection-absorption infrared (RAIR) spectra of plasma polymerized films were 
obtained using a Perkin-Elmer 1800 Fourier-transform infrared spectrophotometer and a 
reflection accessory from Harrick Scientific Co. An angle of incidence of 80® was used to obtain 
the spectra. RAIR was also used to examine primed steel substrates after reaction with a model 
rubber compound consisting of squalene (100 parts per hundred or phr), zinc oxide (10 phr), 
carbon black (10 phr), sulfur (5 phr), stearic acid (2 phr), and N,N-dicyclohexyl-benzothiazole- 
sulfenamide (DCBS), cobalt naphthenate, and diaryl-p-diphenylene-amine (each 1 phr). 
Polished steel substrates primed with plasma polymerized acetylene films were immersed into a 
stirred mixture of these materials at a temperature of 150°C. During the reaction, the mixture 
was purged with nitrogen to reduce oxidation. At times between 1 and 100 minutes, substrates 
were removed from the mixture, rinsed with hexane, dried, and examined using RAIR. 

During the reaction, aliquots were removed from the mixture using a pipette, placed in a 
glass bottle, and allowed to stand overnight so that the insoluble components would settle to the 
bottom of the bottle. A few drops of the liquid were placed onto a KBr pellet and examined by 
transmission infrared spectroscopy to determine the effect of the reaction on the structure of 
squalene. Raman spectra were obtained from several aliquots using a Bio-Rad FTS-60A FTIR 
system equipped with FT/Raman accessory. 


The RAIR spectrum of an as-deposited plasma polymerized acetylene film (thickness 
-63.5 nm) on a polished steel substrate is shown in Figure 1A. Peaks characteristic of 
hydrocarbons were observed near 2960, 2928, 2870, 1450 and 1375 cm'^. The strong band near 
3295 cm -1 was assigned to CH stretching in monosubstituted acetylene groups (R-C=C-H) [8]. 
Weak bands related to -C=C- stretching in mono- and di-substituted acetylene were observed 
near 2103 and 2210 cm” 1 , respectively [8], A band assigned to C=C stretching was observed 
near 1600 cm” 1 . The band near 916 cm” 1 was related to the CH 2 wagging mode of vinylidene 
groups. A weak band near 1950 cm” 1 was assigned to C=C=C stretching. A weak band possibly 
related to acetylide species [9] was observed near 3250 cm -1 in spectra of films that were only a 


few nanometers in thickness. Bands near 3055, 3027, 1595, 1510, 1495, 758 and 700 cm'l were 
assigned to aromatic rings. The band near 3455 cm"l was 


Figure 1. (A) - Reflection-absorption infrared spectrum of an as-deposited plasma polymerized 
acetylene film with a thickness of 63.5 nm on a polished steel substrate. Spectra 
shown in (B), (C), and (D) were obtained after reaction of similar films with the 
model rubber compound for 15, 30, and 45 minutes, respectively. 

assigned to O-H groups while bands near 1715 and 1680 crn'l were assigned to C=0 groups 
resulting from the reaction of trapped radicals with atmospheric oxygen and moisture. 

Infrared spectra of plasma polymerized acetylene films on steel substrates after reaction 
with the model rubber compound for various times are shown in Figures IB-ID. After 15 
minutes, the intensity of the band near 3295 cm"l decreased greatly and new bands appeared 
near 1765, 1649, 1539, and 1512 cm"l. After 30 minutes reaction time, additional bands 
appeared near 1011, 1030, 1085, 1232, 1320, 1430, and 1551 cm - l. The band which appeared 
near 1539 cm"l was related to zinc or cobalt stearate which were formed by reaction of ZnO and 
cobalt naphthenate with stearic acid. Bands which appeared near 1011, 1030, 1085, 1232, 1320, 
1430, and 1551 cm'l after 30 minutes were all related to the benzothiazole sulfenamide moiety 

The band which appeared near 1649 cm"l after 15 minutes of reaction was related to C=C 
double bonds in squalene. Linning and co-workers observed a band near this frequency in 
infrared spectra of NR cured with sulfur either with or without accelerators and observed a 
similar band in infrared spectra of squalene reacted with sulfur [10]. We observed a strong band 
near 1667 cm - l in the infrared spectrum of squalene, indicating that squalene is mostly the trans 
isomer. We did not observe a band near 1649 cm'l in transmission infrared spectra of squalene 
or the model rubber compound after reaction for a few minutes. However, a band did appear 
near 1649 cm“l in transmission spectra of the model compound after reaction for 15 minutes. 
This band, which became stronger as reaction time increased, was most likely related to a C=C 
double bond with an attached sulfur or to conjugated double bonds [10]. 


We have also obtained Raman spectra of the model compound as a function of reaction 
time. For short reaction times, only one band near 1670 cm'l was observed. New bands, which 
were attributed to conjugated double bonds [11], were observed near 1646, 1640, 1628, 1600, 
and 1580 cm -1 in Raman spectra of the model compound after 90 minutes. These observations 
support the assignment of the band near 1649 cm'l in RAIR spectra of plasma polymerized 
acetylene films after reaction with the model rubber compound (see Figure 1) to conjugated 
double bonds and also support the conclusion that crosslinking occurred between squalene and 
the plasma polymerized film. 

RAIR spectra shown in Figure 1 provide some information regarding the crosslinking 
mechanism. As indicated above, the bands near 1011, 1030, 1085, 1232, 1320, and 1430 cm 
which appeared after 30 minutes (see Figure 1C) were all related to the benzothiazole 
sulfenamide moiety of DCBS. According to Chapman and Porter [12], the early stages of curing 
natural rubber with sulfur, an accelerator, and an activator involve formation of zinc stearate and 
then a zinc accelerator complex or a zinc accelerator perthiomercaptide complex. The zinc 
accelerator complex has the structure (I) while the zinc accelerator perthiomercaptide has 
structure (II) where X is an accelerator fragment and L is a ligand: 





XSS a ZnSS b X (II) 





Since the accelerator used here was N,N-dicyclohexyl-benzothiazole-sulfenamide (DCBS), X has 
the structure III: 





and the ligand L has the structure (IV): 

The zinc accelerator perthiomercaptide reacts with rubber to form accelerator perthiomercaptide 
pendant groups such as (V) which eventually disproportionate to form crosslinks such as (VI). In 
the current case, the DCBS moieties were most likely present as the zinc or cobalt accelerator 
perthiomercaptide (V) or an accelerator perthiomercaptide pendant group (VI). 



The appearance of bands related to the benzothiazole sulfenamide moiety of the DCBS 
accelerator after 15 minutes reaction between the plasma polymerized film and the model rubber 
and their gradual disappearance after about 30 minutes was consistent with the formation of 
pendant groups such as (V) and their disproportionation to form crosslinks such as (VI). 

The assignments of the bands near 1512 and 1765 cm" 1 are not known at this time. No 
bands were observed near these frequencies in the infrared spectra of any of the neat compounds 
in the model rubber. However, a band was observed near 1512 cm -1 in transmission infrared 
spectra of the model rubber compound after reaction for five minutes but the intensity of this 
band did not seem to change as a function of reaction time. This band is probably related to an 
as yet unknown product which forms quickly in the reaction mixture. Linning and co-workers 
observed a band near 1765 cm'l in infrared spectra of squalene after reaction with sulfur but did 
not discuss it [10]. This band is probably related to carbonyl stretching in ester groups in 
strained cyclic compounds or with electron withdrawing groups attached [13]. 

As the reaction between the plasma polymerized film and the model rubber compound 
progressed, there were significant changes in the spectra in the region between 3000 and 2800 
cm -1 (see Figure 1). After 15 minutes, the band near 2870 cm -1 , which was characteristic of the 
plasma polymerized film, began to disappear and a shoulder began to appear near 2852 cm"I. 
After 30 minutes, the band near 2852 cm"I was well resolved but the band near 2870 cm"* was 
not observable. The new band which appeared near 2852 cm"I was related to CH 2 stretching in 
squalene and its appearance provided additional support for “crosslinking” between squalene and 
the plasma polymerized primer. 



Plasma polymerized films of acetylene on polished steel substrates contained various 
functional groups, including mono- and di-substituted acetylene, aromatics, methyl, and 
methylene groups. The films also contained carbonyl groups which apparently resulted from 
reaction of residual free radicals with oxygen when the films were exposed to the atmosphere. 
Some acetylides may have formed in the interphase between the films and the steel substrates. 
When the plasma polymerized films were reacted with a model rubber consisting of a mixture of 
squalene (instead of natural rubber), zinc oxide, carbon black, sulfur, stearic acid, cobalt 
naphthenate, diaryl-p-diphenyleneamine, and N,N- dicyclohexylbenzothiazole sulfenamide, 
another interphase was created. Zinc oxide and cobalt naphthenate reacted with stearic acid to 
form stearates. The stearates reacted with the benzothiazole sulfenamide moiety of N,N- 
dicyclohexylbenzothiazole sulfenamide and sulfur to form rubber-bound pendant 
perthiomercaptide groups. Eventually the perthiomercaptides underwent disproportionation to 
form sulfidic crosslinks between the plasma polymerized acetylene films and squalene. 
Migration of double bonds in squalene during the reaction resulted in the formation of 
conjugated dienes and trienes. 


This research was supported in part by grants from the National Science Foundation and 
the Edison Materials Technology Center (EMTEC). The support of The Goodyear Tire and 
Rubber Co. is also acknowledged. 


1. Y. M. Tsai and F. J. Boerio, Surf. Interface Anal., accepted for publication, 1995. 

2. Y. M. Tsai, U. R. Aggarwal, F. J. Boerio, D. B. Zeik, S. J. Clarson, W. J. van Ooij and 

A. Sabata, J. Appl. Polym. Sci.: Applied Polymer Symposium 54, 3 (1994). 

3. W. J. van Ooij and A. Kleinhesselink, Appl. Surface. Sci. 4, 324 (1980). 

4. W. J. van Ooij, Rubber Chem. Technol. 51, 52 (1978). 

5. W. J. van Ooij, W. E. Weening, and P. F. Murray, Rubber Chem. Technol. 54,227 (1981). 

6. J. J. Ball, H. W. Gibbs, and P. E. R. Tate, J. Adhesion 32,29 (1990). 

7. D. B. Zeik, S. J. Clarson, C. E. Taylor, F. J. Boerio, W. J. van Ooij, and A. Sabata, 
presented at 205th National Meeting of the American Chemical Society, Denver, CO, 1993. 

8. N. B. Colthup, L. H. Daly, and S. E. Wiberley, Introduction to Infrared and Raman 
Spectroscopy. 3rd Edition, Academic Press, San Diego, CA, 1990, Ch. 6. 

9. C. C. Chang and R. J. Kokes, J. Catal. 28, 92 (1973). 

10. F. J. Linning, E. J. Parks, and J. E. Stewart, J. Res. NBS 68A, 499 (1964). 

11. J. R. Shelton, J. L. Koenig, and M. M. Coleman, Rubber Chem. Tech. 44, 904 (1971). 

12. A. V. Chapman and M. Porter, in Natural Rubber Science and Techn olo gy , A. D. 

Roberts, Ed., Oxford Science Publications, Oxford, UK, 1988, Ch. 12. 

13. N. B. Colthup, L. H. Daly, and S. E. Wiberley, Introduction to Infrared a nd Raman. 
Spectroscopy. 3rd Edition, Academic Press, San Diego, CA, 1990, Ch. 9. 



K.S. Macturk*, C.L. Schutte, C.R. Schultheisz, D.L. Hunston and M.J. Tarlov** 

Polymers Division, National Institute of Standards and Technology, Gaithersburg, MD 20899 
* National Research Council Research Associate; ** Process Measurements Division 


We investigated the role of coupling agents with respect to the relative durability of glass 
fiber/epoxy matrix composites exposed to water, which degrades both glass fibers and the 
fiber-matrix interface. Interface chemistry was tailored by coating fibers with mixtures of 
different coupling agents. Single-fiber fragmentation test results showed little decrease in the 
strengths of the interface and a slight decrease in fiber strengths upon exposure to water. 
XPS results showed expected variations in surface composition. Contact angle measurements 
demonstrated variations in surface hydrophobicity as coupling agent mixtures were changed. 


Many studies highlight the relationship between the fiber-matrix interface and composite 
properties. 1,2,3 The behavior of the interface upon exposure to hostile environments is of 
special concern. In this work we attempt to "engineer" the interface to achieve strength and 
durability. To this end, coupling agent hydrophobicity and reactivity with the matrix is varied 
in an effort to increase resistance to moisture attack on the strength of both the fiber and 
interface, as measured by a single-fiber fragmentation test. X-ray photoelectron spectroscopy 
(XPS) and contact angle measurements provide insight into the interface chemistry, which is 
related to the mechanical properties. The goal is to better understand the decrease of strength 
in glass fibers and the interface caused by water. Improved composite durability will allow 
wider use of these materials in several new markets including automotive applications, deep 
water oil production and transportation infrastructure. 

The single-fiber fragmentation test consists of applying tension to a dogbone sample of 
epoxy resin containing a single glass fiber. When the applied stress exceeds the failure stress 
of the fiber, the fiber begins to fragment. As the applied stress increases, fragmentation 
continues until the stress transferred across the interface via shear is no longer sufficient to 
fracture the fiber fragments. Completion of the fragmentation process is referred to as 
saturation. The average fiber fragment length at saturation (L s ) is related to the critical length 
(L c s=4L/3). The raw data consist of the number of breaks as a function of strain. 

The effective interfacial shear strength (x) is related to the apparent strength of the fiber 
at the critical length (<r f ), the critical length and the diameter (d) of the fiber fragment 4 

t a) 


The fragmentation test has been used to study the effect of various environmental 
conditions. 5,6,7,8,9 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

Equation (1) requires the strength of the fiber at the critical length in the resin be known. 
Traditionally, the fiber strength at the critical length is evaluated by tedious measurements 
of single fiber strengths at longer lengths extrapolated to the value of strength at the critical 
length. An additional complication results from loss of glass fiber strength due to 
moisture. 10 Also, it is difficult to simulate the conditions the fibers experience in the resin. 
The following approach is intended to circumvent these difficulties. 

The strength of glass fibers is limited by the presence of surface defects which act as stress 
concentrators. 11 The strength of the fibers is expressed using a two parameter Weibull 
probability model. The probability of survival (P s ) at stress (a) for a fiber of length L is 

P s = exp 



( 2 ) 

where cto is a scale parameter for a length L 0 and p is the shape parameter that is independent 
of length. Strengths of fibers are assumed to depend on critical flaws whose spatial locations 
follow a Poisson distribution. 12 The probability of survival at a given stress level is 

P s = exp[-A,(o)L] (3) 

where X(c) is a cumulative flaw density, or the number of critical flaws per unit length up 
to stress a. Equating the probabilities of survival, substituting the number of breaks n(a) in 
length L 0 for X(a), and taking the logarithm twice gives the relation between the 
experimentally accessible variables (n and a) and the Weibull parameters (o^ and p) 

ln[n(a)] = plna-pina 0 ( 4 ) 

The value of the stress is found by measuring strain in the dogbone during the test and the 
tensile modulus of the fiber. The Weibull parameters are found from a plot of ln[n(a)] vs. 
ln(a) (slope=P, intercept= -plna,,, Figure 1). The critical length fragment strength (<a,>) is 

< C/ >=a 0 (V4) 1 ' 1 ' (5) 

Thus, interface and fiber strengths are evaluated as a function of moisture exposure. 

Single-Fiber Fragmentation Test 

Single filaments of unsized E-glass (donated by Owens-Coming) with a diameter range of 
12-20 pm were coated with solutions containing mixtures of two different silane coupling 

* Certain commercial materials and equipment are identified in this paper in order to 
specify adequately the experimental procedure. In no case does such information imply 
endorsement by the National Institute of Standards and Technology, nor does it imply 
necessarily that the items are the best available for the purpose. 


agents. The first was n- 
(Cl 3 Si(CH 2 ) I7 CH 3 ) which is denoted OTS. 

It was chosen because it is hydrophobic. 

However, it has the disadvantage of being 
unreactive with the epoxy. The second, 

3-aminopropyltriethoxysilane, denoted 
APS, (EtOSi(CH 2 ) 3 NH 2 ) was chosen 
because the amine group reacts with the 
epoxy resin and it is relatively 

Solutions of silane mixtures were 
deposited on the fibers from a 0.1% by 
weight silane solution of a 50/50 mixture 
by volume of tetrahydrofuran (THF) and 
ethanol (0.72 moles THF/mole ethanol). 

Three silane ratios were used, 100% 

APS, 100% OTS, and a 50%/50% 
mixture of each by volume (2.6 moles 
APS/mole OTS). Silanes were added to 
the solvent solution and mixed for one 
hour at room temperature. Fibers were 

dipped in 100 ml of the silane solutions for 2 minutes. The fibers were then removed from 
the solution and hung in a graduated cylinder to dry overnight. 

Single-fiber fragmentation test samples were prepared using the procedure of Drzal. 13,14 
Dogbone samples were made by aligning a single fiber in a mold, casting DGEBA epoxy 
resin (Epon 828, Shell) containing 14.5 phr metaphenylenediamine as a curing agent, and then 
curing the mold in an oven at 75 °C for 2 hrs followed by 2 hrs at 125 °C. Prior to 
conditioning the dogbones in water the sample ends were coated with a 5 minute epoxy to 
reduce transport of water along the fiber. The samples were conditioned in 75 °C distilled 
water for 2 to 10 weeks. Samples were removed from the 75 °C water and placed in room 
temperature distilled water for one day prior to testing. 

The fragmentation test procedure was similar to that described in an earlier paper. 14 The 
fiber was examined with an optical microscope to ensure suitable fiber alignment. The initial 
number of breaks, if any, were counted, and the diameter of the fiber was measured. Two 
marks were placed on the sample surface approximately 1 cm apart to measure strain in the 
sample during the test. Next, the sample displacement was increased. After 10 minutes the 
number of breaks, if any, was recorded and the sample strain measured. The process was 
repeated until the number of breaks in the sample reached saturation. Three to five samples 
were tested for each condition. 

Figure 1 Natural logarithm of the number of 
breaks vs. the natural logarithm of stress. The 
slope is P and the intercept is -(3 In (Xq. 

Silicon Wafer Sample Preparation 

In order to characterize the silane coatings with XPS and contact angle measurements more 
easily, flat plate samples were prepared. Single crystal silicon wafers of (111) orientation 
with a thickness of 575 ± 50 pm (Silicon Sense) were used to simulate the glass surface. 
Wafers were cut to 1.6 cm by 1.1 cm, rinsed with THF followed by distilled water and 


allowed to dry overnight. Next, the 
wafers were dipped in the silane solution 
as described in the previous section for 
the fibers. 

XPS Analysis 

XPS data were acquired using a VG 
ESC A Lab system. Excitation was 
provided by an A1 anode operating at 240 
Watts. The analyzer was operated in the 
retarding mode at a pass energy of 100 

Contact Angle Measurements 

We measured advancing and receding Applied strata (*) 

contact angles of deionized water on the Figure 2 Number of breaks vs. applied strain 
treated silicon wafer surfaces using a for a dry and water conditioned sample, 
contact angle goniometer (Rame-Hart, 

Model 100). The values of the contact 

angles were measured three times in three different areas of the surface resulting in a 

reproducibility of ± 3 °. 

A dynamic contact angle analyzer (Cahn, Model DCA-312) was used to measure the 
contact angle of deionized water on the silane coated glass fibers. A platform speed of 100 
qm/sec was used. Seven to eight fibers of each sample type were placed on a piece of tape 
and submerged simultaneously to produce sufficient signal. Three runs were done for each 
sample type for a reproducibilty of ± 2 °. 


Raw data for two 50 % APS samples 
are shown in Figure 2. The horizontal « 
shift between dry and wet (75 °C, 2 * 

weeks) samples indicates more fiber g 
breaks at lower applied strain due to jj 30 

degradation of the fiber. The vertical 5 

change in the total number of breaks 5 
relates to changes in the strength of both -5 20 
fiber and interface. 3 

Results for the interfacial shear 2 1Q 
strength as a function of time in 75 °C ~ 
distilled water are given in Figure 3. As 
expected, the dry 100 % APS coupling 0 
agent, which reacts with the epoxy, 
produces the strongest interface. The 100 
% OTS (0% APS) gives the weakest Figure 3 Interfacial shear stress vs. time in 75 
interface. The 50% APS results are °c distilled water. 

0 500 1000 1500 2000 

Hours In 75 *C Watsr 


between the 0 % and 100 % values. Q 

Figure 3 also shows a slight decrease in 
interfacial strength for the 0 % APS 
samples aged in water. The interface £ 2 ‘ 5 

strengths of the 100 % APS samples “ 

remain constant over times of 1800 | 2.0 

hours. * 

Fiber strengths for different coatings -g 1 B 
are compared in Figure 4 which shows 5 
the fiber strength for a length of 15 mm | 

versus time in 75 °C water. For the 100 & 

% OTS samples, the fiber strength is • 
fairly constant as a function of time in c °- 5 
water. In contrast, for both the 50% and 
100 % APS samples, the fiber strength 0.0 
appears to decrease as time in water 
increases. There are two possible 
explanations which, taken independently Figure 4 Fiber strength at a length of 15 mm 
or collectively, may explain this behavior. V s. time in 75 °C distilled water. 

First, the more hydrophobic OTS may 
produce a "barrier" coating which 

protects the fiber from water damage. Second, poor stress transfer during swelling of the 
OTS samples may result in less stress corrosion of the fiber. 

The values of advancing and receding contact angles of deionized water for the three 
mixtures of coupling agent on both fibers and wafers are shown in Figure 5. These results 
support the idea that the 100 % APS coating is more hydrophilic while the 100 % OTS 

coating is more hydrophobic. The 
contact angles for the treated fibers and 
silicon wafers are similar and show the 
same trends. These preliminary data 
suggest that under similar reaction 
conditions the resulting coatings on the 
fiber and wafer are chemically similar. 
Preliminary ellipsometric analyses of £ 
coated Si wafers yield thickness values in £ 
the range of 2.5 nm. These results < 

suggest similarly thin coatings on the 8 

fiber. Work is ongoing to investigate this 
relationship more fully. 

Preliminary XPS results indicate the 
surface chemical composition changes 
with the mixture of coupling agent 
deposited on the surface. As expected, 
the intensity of the nitrogen peak grows 
with increasing percentage of APS. The 
nitrogen increase correlates well with the 
increase in the interface strength of dry 

% Amino Slluna 

Figure 5 Advancing and receding contact 
angles for glass fibers and silicon wafers vs. % 
amino coupling agent. 


samples and is consistent with the lower contact angle of the 100 % APS samples, since the 
hydrophilicity should increase with the amount of nitrogen present. 


Preliminary results showed that interfacial shear strength was controlled by chemical 
composition of coupling agent mixtures deposited on the fiber surface. The 100 % APS gave 
the strongest interface and best retained its strength over time in 75 °C distilled water. The 
hydrophobic 100 % OTS coating seemed to reduce the loss of fiber strength over time in 
water. Contact angle results demonstrated that the surface hydrophobicity changed with 
different silane mixtures. XPS results indicated the surface composition varied with the 
mixture of coupling agent. Work is continuing to better understand the relationship between 
surface composition and mechanical property changes upon aging. This type of information 
will lay the groundwork for increased use of structural composites in applications where they 
are exposed to moisture. 


1. JJ. Lesko, R.E. Swain, J.M. Cartwright, J.W. Chin, K.L. Reifsnider, D.A. Dillard, 
J.P. Wightman, J. Adhesion 45, 43 (1994). 

2. E. Mader, K. Grundke, H.-J. Jacobasch, G. Wachinger, Composites 25, 739 (1994). 

3. A.M. Serrano, I. Jangchud, R.K. Eby, KJ. Bowles, D.T. Jayne, Mat. Res. Soc. 
Proc. 305, 105 (1993). 

4. A. Kelly, W.R. Tyson, J. Mech. Phys. Solids 13, 329 (1965). 

5. V. Rao, L.T. Drzal, Polymer Composites 12, 48 (1991). 

6. M.J. Fowlkes, W.K. Wong, Polymer 28, 1309 (1987). 

7. L.T. Drzal, M.J. Rich, M.F. Koenig, J. Adhesion 18, 49 (1985). 

8. A.T. DiBenedetto, P.J. Lex, Polym. Eng. and Sci. 29, 543 (1989). 

9. X.S. Bian, L. Ambrosio, J.M. Kenny, L. Nicholais, A.T. DiBenedetto, Polymer 
Composites 12, 333 (1991). 

10. R.J. Charles, J. Appl. Phys. 29, 1549 (1958); J. Appl. Phys. 29, 1554 (1958). 

11. A.A. Konkin, in Handbook of Composites. Vol. LStrong Fibers, edited by W. 

Watt and B.V. Perov (Elsevier Science, New York, 1985), p. 255. 

12. H.D. Wagner, A. Eitan, Appl. Phys. Letters 56, 1965 (1990); see also B. Yavin, 
H.E. Gallis, J. Scherf, A. Eitan, H.D. Wagner, Polymer Composites 12, 436 (1991). 

13. L.T. Drzal, M.J. Rich, J.D. Camping, W.D. Park, in Proc. of 35th Ann. Tech. 

Conf.. Reinforced Plastics , (Composites Institute, 1980), p.l. 

14. C.L. Schutte, W. McDonough, M. Shioya, M. McAuliffe, M. Greenwood, 
Composites 25, 617 (1994). 




*Sandia National Labs, Dept. 1815, P.O. Box 5800, Albuquerque, NM. 87185-0367 


Functionalized block copolymers were synthesized as adhesion promoters using Ring- 
Opening Metathesis Polymerization (ROMP). They were designed for glass/epoxy and 
copper/epoxy interfaces. The former contained triethoxysilane groups in the first block and 
secondary amine groups in the second. The latter contained imidazole groups in the first and 
amine groups in the second block. These block copolymers were shown to form ordered 
monolayers on the respective glass and copper surfaces using neutron reflectivity. Adhesion 
measurements showed an enhancement of adhesion after application of these block copolymers. 


In the manufacturing of printed wiring boards, often adhesion is a concern between a 
glass weave and an epoxy matrix, and between an epoxy matrix and a copper substrate. The goal 
of this work is to develop novel functionalized block copolymers to promote adhesion at 
inorganic substrate/polymer interfaces. The idea is that one block will be functionalized in order 
to react with the inorganic substrate, and the other block will be able to react with the polymer 
matrix. We envision several potential advantages of functionalized block copolymers over small 
molecule coupling agents, including greater control over the structure of the interphase region 
and enhanced adhesion through both entanglement and covalent bonding between the block 
copolymer and the polymer matrix. 

Our program involves four key elements: the synthesis of suitable functionalized block 
copolymers, characterization of the conformation of the copolymers at the interface by neutron 
reflectivity, characterization of the degree of bonding by spectroscopy, and measurement of the 
mechanical properties of the interface. In this paper we discuss block copolymers designed as 
adhesion promoters for the glass/epoxy and copper/epoxy interfaces. For the glass/epoxy 
interface we have synthesized diblocks with one block containing triethoxysilane groups to bond 
to the glass, and the other block having secondary amine groups to bond to the epoxy. We have 
also made a triblock copolymer where the middle block is hydrophobic in order to promote 
moisture resistance. For the copper/epoxy interface we have made a diblock with one block 
containing imidazole groups to bond to copper and a second block containing the secondary 
amines, again to bond with the epoxy matrix. Below we describe the synthesis of the block 
copolymers by living, ring-opening metathesis polymerization (ROMP) 1,2 and the first 
characterization data obtained by NMR spectroscopy, small-angle x-ray scattering (SAXS), and 
neutron reflectivity. We also report here some preliminary adhesion measurements of these 


Mat. Res. Soc. Symp. Proc. Vol. 385 ° 1995 Materials Research Society 


Neutron reflectivity experiments were performed using the POSY II reflectometer at the 
Intense Pulsed Neutron Source at Argonne National Labs. Small-Angle x-ray scattering was 
performed using a Rigaku instrument with a 1.54-A Cu Ka rotating-anode point source, Charles 
Supper double mirror focusing optics, and a Nicolet two-dimensional detector. 



All reactants were purchased from Aldrich Chemical Company. Tetrahydrofuran was 
used as the polymerization solvent, and was vacuum distilled from a sodium/benzophenone ketal 
solution immediately prior to use. The molybdenum based initiator ((2,6-diisopropyl-phenyl- 
imido)neophylidine-molybdenum-bis-t-butoxide) 3 was purchased from Strem and used without 
further purification. All polymerizations were performed under an inert atmosphere in a Vacuum 
Atmospheres dry box. 


Triethoxysilyl norbomene (1) was purchased from HULS and was dried by distilling 
from and storing over sodium hydride. The monomer was filtered through activated neutral 
alumina immediately prior to use. 

5-(n-methyIimidazoIe)-2-norbornene (2): dicyclopentadiene (63.Og, 0.477 mol) and 
freshly distilled allylimidazole (75.Og, .694g) were heated to reflux neat for 16 hrs. After cooling 
down, the solution was vacuum distilled at 0.1 torr. After taking a cut from 50-100°C, 65.3g 
(54%) of the pure product came over at 105-110°C. This monomer was then dried by distilling 
from and storing over calcium hydride, and filtering through a bed of activated neutral alumina 
immediately prior to use. 

5-(t-butylaminomethyl)-2-norbornene (3): step 1: allyl-t-butvlamine . Allylbromide 
(140g, 1.16mol) was slowly dripped into a flask containing 500 mis. of mechanically stirred t- 
butylamine. This exothermic reaction was contained using a condenser fitted onto the flask. At 
the end of the addition, the solution was stirred overnight, and the precipitates were filtered off 
and washed with ether. The filtrate was distilled at atmospheric pressure with the final product 
coming over at 105°C. step 2: dicyclopentadiene (14.0g, .llmol) and allyl-t-butylamine (above, 
60.Og, .53mol) were heated neat in a high pressure reactor at 185°C for 15 hrs., and then cooled 
to room temperature and the solution distilled under vacuum at 0.1 torr. The pure product was 
collected at 36°C. This monomer was dried by distilling from and storing over calcium hydride 
and filtering through alumina immediately prior to use. 

5-(octaoxymethyl)-2-norbornene (4): sodium hydride (4.50g, .187mol) and 120mls dry 
thf were placed in a three-neck flask fitted with a condenser and addition funnel. Iodooctane 
(40.0g, .167mol) was then added to the flask with a nitrogen purge. The solution was slurried, 
and a solution of norbomene-2-methanol (21.7g, .175mol) in 40mls dry thf were placed in the 
addition funnel. The slurry was heated to 50°C and then the norbomene-2-methanol solution 


slowly dripped into the slurry with stirring. After complete addition, the solution was stirred for 
30 minutes and then allowed to cool to room temperature. The solvent was removed under 
reduced pressure, and the resulting cloudy oil was distilled at 0.1 torr, with the product coming 
over at 120°C (a yield was not recorded for this monomer). The monomer was further purified 
by distilling from and storing over sodium hydride, and filtering through alumina immediately 
prior to use. 

Partially deuterated forms of the above monomers were made identically to the reaction 
schemes above by starting with a partially deuterated dicyclopentadiene. 

Block copolymers 

Four block copolymers were made and are shown below. They are designated by the 
monomer repeat unit number. 

Block Copolymers for the glass/epoxy interface Block Copolymers for the copper/epoxy interface 

Typically lg of polymer was formed in 15 mis. dry thf by stirring the appropriate amount of the 
molybdenum initiator with the appropriate amount of the first monomer, then upon complete 
consumption, the appropriate amount of the second monomer, etc. After the polymerization of 
the final block, the living chain ends were terminated by the addition of benzaldehyde. These 
block copolymers were characterized by proton NMR, which showed complete consumption of 
all monomer in most cases. The 1-3 and 1-4-3 block copolymers were used straight from the thf 
solution, whereas the 2-3 and 2-4-3 block copolymers were precipitated from acetonitrile, dried, 
and redissolved in chloroform for future use. 

Each block in all of the polymers had a targeted molecular weight of 15,000 g/mol. By 
NMR, the only difference was that the immidazole blocks were less than 15,000 g/mol: typically 
they ended up being 10,000 g/mol. All the other blocks were assumed to be the targeted 
molecular weight. One additional diblock of the 2-3 system was made that had a molecular 
weight of 10,000 / 40,000 g/mol (to see the effect of lengthening block 3). 

In all cases, a designation of d that immediately follows the number of the block indicates 
that the block is selectively deuterated over the other blocks in the polymer. These partially 
deuterated blocks contained 4 to 5 deuteriums per repeat unit. 

Block copolymer treatment of surfaces 

For the glass/epoxy interface the diblock (1-3) or triblock (1-4-3) was dissolved in 
isopropanol at a concentration of 0.02 g/ml, and clean silicon wafers were submersed in the 
solution for several hours. A small amount of tin octoate was added to promote hydrolysis of the 
ethoxysilane groups. The wafers were removed and rinsed with excess isopropanol, and dried. 


For the copper/epoxy interface, the diblock (2-3) was dissolved in either thf or methanol at the 
same concentration as above, and a silicon wafer coated with a copper layer was submersed in 
the solution for several hours. It was then removed and rinsed with excess solvent, and dried. 

Adhesion sample preparation and testing 

Lap sheer samples were made by coating glass slides or copper coupons with the 
appropriate block copolymers, then applying a 1 in epoxy resin on top, then curing for 2 hrs. at 
170°C. Yield strengths were measured by pulling these samples on an Instron. Floating roller 
peel samples were prepared for copper epoxy interface by treating a copper foil with the block 
copolymer (2-3), then curing a layer of epoxy on top, then measuring the strength required to 
peel the copper from the epoxy. 



The monomers used in this study are readily available through Diels-AIder reactions of 
the appropriate vinyl or allyl compounds with cyclopentadiene. The monomers are thus 
relatively inexpensive. The polymerization of these monomers appears to be efficient and 
quantitative as observed by *H NMR (figure 1). In figure la, a representative monomer spectrum 
is shown. With all monomers, proton signals are seen between 5.8-6.4 ppm. which are 
associated with the protons on the carbons of the highly strained double bonds. Figure lb shows 
the NMR spectrum of the 1-4-3 triblock copolymer taken straight from the polymerization 
solution. This spectrum indicates a complete disappearance of all signals between 5.8-6.4 ppm., 
and the appearance of a broad polymer peak, typical of polynorbomenes, between S.2-5.6 ppm. 

Gel Permeation Chromatography (GPC) has proved quite difficult for these block 
copolymers due to the presence of the amine functionalities. These amine groups cause the 
polymers to stick onto GPC columns. This phenomena has been observed by other researchers. 
There are other arguments for the existence of a living polymerization, and thus formation of 
block copolymers. First of all, it is well documented that the initiator used here is able to 
polymerize most norbomene-type monomers in a living fashion 4 ' 7 . Second, it is known that bulk 
block copolymer materials phase separate in a well-ordered fashion on a meso-scale , and this 
results in well defined scattering peaks observed in small-angle x-ray scattering (SAXS). Figure 
2, showing the SAXS pattern for the 1-4-3 triblock, demonstrates several scattering peaks, 
indicating the presence of meso-phase separation brought on by a block copolymer. All four 
block copolymer samples discussed here show typical phase separation scattering peaks, which 
can only be brought about by the presence of block copolymers. Therefore all of these materials 
have the desired block copolymer structure. 

Characterization of adsorbed films bv neutron reflectiv itY 

It is important in these systems to determine what is the conformation of the block 
copolymers on the substrate to which they are adsorbed: whether one block is preferentially 
adsorbed or the blocks are mixed, and whether they form monolayers or multiple layers. To 


Figure 1. Typical l H NMR of a norbomene monomer (top) and of the 1-4-3 
triblock copolymer (bottom). 


h Cl/nm3 

Figure 2. Small-angle x-ray scattering pattern of triblock 1-4-3. 

address this question we prepared block copolymer samples with partially deuterated amine 
blocks (block 3) for the silicon surfaces, and partially deuterated imidazole blocks (block 2) for 
the copper surfaces, and examined adsorbed films of these polymers by neutron reflectivity. 
With this technique the neutron refractive index profile normal to the surface is obtained with 
~5A resolution. The selective deuteration allows one to distinguish between the blocks. 

For the silicon surfaces, figure 3 shows the neutron reflectivity pattern for an untreated 
silicon surface (a), the 1-3 diblock copolymer treated silicon surface (b), and the 1-4-3 triblock 
copolymer treated surface (c). The reflectivity patterns suggest several things: first of all, the 
only way to interpret the data is to say that there are well-ordered monolayers adsorbed onto the 
silicon surface in both the diblock and triblock case. In the diblock case (b) the trace is shifted 
down and to the right compared with the untreated surface. This implies that the amine block is 
the block sticking up away from the surface. In the triblock case (c) the trace is shifted up and to 
the left, indicating the amine block is sticking down onto the silicon substrate, and the 
triethoxysilyl block is sticking up away from the surface. The latter case is counter intuitive, but 
can be explained by taking into account the solvent that was used for the adsorption. The 
hydrophobic middle block 4 of triblock 1-4-3 does not like the isopropanol environment, and 
most likely creates micelles in solution. This micellization may hide block 1 in the center, and 
thus only block 3 would be on the outer shell of the micelle, thus it would be the only block 
available to adsorb onto the surface. 

Results for the diblock copolymer onto copper (with targeted block lengths of 15K-15K) 
adsorbed from a solution in methanol are shown in figure 4a. The main effect of the adsorbed 
copolymer on the reflectivity is to shift the fringes to lower q. This is consistent with the 
imidazole block (block 2) selectively adsorbing to the surface, as shown in figure 4b. The curves 
in figures 4b-d were calculated using the known atomic compositions of the blocks and assuming 
a 30A thickness for the layers formed by each block, which is consistent with the size of the shift 
observed in figure 4a. Thus, these data suggest that the copolymer adsorbs in a monolayer film 
onto the copper, and that the blocks form separate layers with the imidazole block (block 2) 
adsorbed selectively to the surface (by comparison to the theoretically calculated curve shown in 
figure 4b). 

Adhesion tes ting 

Lap sheer samples were made for glass surface by adsorbing the block copolymer 1-3 
onto two microscope slides, and curing a layer of epoxy between the two slides. When pulled on 
the instron, all of the samples failed in the glass slide rather than in the epoxy layer or the 
interface, thus only qualitative results can be reported at this time. It was observed that untreated 
glass slides had poor epoxy wetting characteristics; the epoxy would bead up when applied to the 
surface. However, when treated with the block copolymer, the epoxy formed uniform layers 
without beading up. Another test was to place the post-cured lap sheer sample in boiling water 
for several days to weaken the epoxy layer. It was observed that untreated lap sheer samples fell 
apart in boiling water whereas block copolymer treated samples did not. Therefore there is some 
enhancement of adhesion due to the application of the block copolymer. 

Lap sheer samples for copper coupons treated with the 2-3 diblock were more 
straightforward. Failure always occurred near the interface, and thus quantitative results could be 
obtained. We discovered that untreated (microetch only) copper samples had an ultimate yield 
strength of 1250 lb/in 2 , brown oxide samples (industry standard chemical roughening) had a 



q (A'*) 

Figure 3. Neutron reflectivity data for: a) untreated silicon surface, b) diblock 
copolymer l-3d treatment (d stands for partial deuteration of block 3), 
and c) triblock copoymer l-4-3d treatment. 


Figure 4. Neutron reflectivity for: a) actual data of adsorbed block copolymer 2-3, b) theoretical 
calculation for block 2 selective adsorption onto copper, c) theoretical calculation for 
block 3 selective adsorbtion onto copper, and d) for blocks 2 and 3 randomly mixed 
as a single layer adsorbed onto copper. 


strength of 1670 lb/in 2 , and block copolymer treated samples had a strength of 1720 lb/in 2 . This 
represents a 40% increase in adhesion over no treatment at all, and a comparable adhesion to that 
of the brown oxide treatment. 

Floating roller peel tests were also conducted on copper surfaces with a cured epoxy 
layer. We observed that peel strengths for untreated copper were 1.55 lb/in., those for the brown 
oxide treatment were 8.2 lb/in., and for the dibiock 2-3 treatment we obtained peel strengths of 
3.09 lb/in. Thus in the case of the peel tests we were able to double the strength over untreated 
samples, but were not able to obtain the strength observed by the brown oxide treated samples. It 
is known that the brown oxdex treatment promotes a roughening of the surface, which increases 
the surface area significantly. If this factor is taken into account, perhaps our materials are 
comparable, on a per unit surface area basis, to strengths obtained from brown oxide treatments. 


In this paper, we have demonstrated the ability to make diblock and triblock copolymers 
using a living ROMP technique, which are capable of bonding to silicon and copper surfaces 
depending on the functionalities. The existence of a block rather than random copolymers was 
substantiated by the SAXS patterns of these materials. Upon selectively deuterating one block, 
we were able to show by neutron reflectivity that in all cases, both on silicon surfaces and on 
copper surfaces, they form ordered monolayer structures. We demonstrated qualitatively that 
adhesion is enhanced on glass surfaces by treating them with the appropriate block copolymer. 
We also observed that treating copper surfaces with their appropriate block copolyemers 
increased both the lap-sheer yield strength, and the peel strength of these samples. Future work 
will include varying the number and type of reactive sites on the polymer chain to determine the 
effect on adhesion. 


We wish to thank Dr. Anuj Bellare at M.I.T. for acquiring the small-angle x-ray 
scattering data, and Dr. John Emerson at Sandia Labs for doing the peel adhesion testing. This 
work was supported by the U.S. Department of Energy under contract # DE-AC04-94AL85000, 
and by the National Consortium of Manufacturing Sciences (NCMS). 


1. R. H. Grubbs and W. Tumas, Science, 243, 902 (1989). 

2. R. R. Schrock, Acc. Chem. Res., 23,158 (1990). 

3. R. R. Schrock et al, J. Am. Chem. Soc., 112, 3875 (1990). 

4. R. R. Schrock et al, Macromolecules, 24, 4495 (1991). 

5. R. R. Schrock et al, J. Am. Chem. Soc., 113, 6899 (1991). 

6 . R. S. Saunders et al, Macromolecules, 25,2055 (1992). 

7. R. S. Saunders et al, Macromolecules, 24, 5599 (1991)(and references contained therein). 




Russian Academy of Sciences, Silicate Chemistry Department, Odoevsky Str., 24/2, 

St.-Petersburg, 199155, Russia 


The present work was aimed at the development of new organosilicate polymeric composite, 
based on polydimethylphenylsiloxane/polyurethane (PDMPS/PU) miscible blend, filled with 
silicates and metal oxides. Considerable improvement of mechanical and corrosion-protective 
properties due to introduction of polyurethane has been observed. Coatings with sufficient 
thermostability (up to 300° C) have been obtained in the case of 20% polyurethane content, 
related to amount of binders. The effects coating heat treatment temperature and curing 
conditions on adhesion of metal/composite interface have been studied. Surface energy 
characteristics of this coating have been obtained and were correlated with its microstructure, 
determined from Scanning Electron X-ray Microprobe Analyses. Recently developed 
composite appeared to show increased durability in atmosphere operation conditions. 


Composite based on polyorganosiloxane - silicate - oxide systems find wide-spread use in 
modem engineering [1-4]. As their principal and the most substantial components are polymers, 
having organic groups, and silicates, they are referred to as organosilicate materials or 
polymeric composites. Strong bonds, including chemical bonding, proved to occur between 
these main components during states of making [5] and heat-setting [6]. Its composition and 
structure afford complex of valuable technical properties for coatings, hermetics, adhesives and 
high-temperature resistant glassplastic binders over a wide temperature range. 

Although polyorganosiloxane is one of the most widely used polymers due to its high 
resistance to heating, oxidation, radiation, humidity as well as for fine electric insulating 
characteristics and inertness to many common reagents, combining it with other polymers 
broadens the potential areas of polyorganosiloxane utilization even further. As is known, such a 
combination might be a co-condensation product, obtained by heating partially condensed 
polyorganosiloxane with an organic polymer that contains a considerable proportion of reactive 
groups, or a simple mixture. It will be noted that properties of coating based on polymeric blend 
are defined generally by the degree of compatibility of its components [7]. 

Examination of these systems is of great theoretical and practical interest because it may lead 
to formulation of new material receiving in accordance with rule of additiveness a new property 
or improved complex of known ones [8]. 

The present paper describes the development of new organosilicate polymeric composite with 
excellent adhesion to metals, bending and impact resistance as well as improved weathering 
resistance, based on PDMPS/PU miscible blend. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 


The materials used were as follows: commercial polydimethylphenylsiloxane with 1% OH- 
end groups content of following formulation {[C 6 H 5 SiOj 5 ] {37 [(CH 3 ) 2 SiO]} n (KO-921) 
and commercial linear prepolymer based on polyether, polytetrahydrofuran, and 
2 ,4-toluylendiisocyanate with 5.2- 6.3% isocyanate groups content (SKU-PFL). Commercial 
Diamet, aromatic diamine, 4,4’-methylen-bis(o-chloro-aniline), for PU component and 
tetrabutoxytitan for PDMPS were in use as curing agents. 

Adhesion was scored on a four-point scale in accordance with GOST 15140-78 by means of 
grid nicks. 

Bending tests were performed following the procedure described in GOST6806-73. 

In order to determine impact resistance of films and coatings we used custom scientific 
instrument falling weight impact tester U-l (GOST 4765-79) and samples on aluminium 

Data in coating hardness were obtained by conventional method (GOST 5233-67) on 
pendulum device M-3. 

To define polymer distribution Scanning Electron X-ray Microprobe Analyses of PDMPS/PU 
films were carried out with microanalyzer “Camebax” of French firm Setaram. Blend with 
component ratio 80:20, doped with TBT, was spread on Teflon substrate and cured at normal 
temperature. At least three independent measures were performed on internal and external 
surfaces of each sample. 

Critical surface tensions were defined by Zisman’s method [9]. Contact angle measurements 
were made on horizontal microscope. Compounds of ethyleneglicol homologous series and its 
solutions with surface tension less than 50*10 J/m were employed as test fluids. 

Corrosion resistance of coatings was defined in accordance with GOST 9.074-77 in apparatus 
of artificial weather (IP-1-3). 


Over a wide range of PDMPS/PU concentration ratios at normal temperature these blends 
appeared to form transparent solutions in common solvent and transparent films, deposited 
from these solutions; in other words, this polymeric system demonstrated kinetical stability. 
Rheological study proved this system to be pseudocompatible; the blend rheological properties 
were quite similar to those of the original polymers. 

Because of original polymer properties and organosilicate composite purpose blends with 
dominant PDMPS content were chosen as a subject of further investigation. 

Table I represents properties of polymeric films (cured under addition of tetrabutoxytitan 
(TBT) and heating at 120° C during 3 hours) in comparison with those for original polymers. 
Only 120° C was found to be enough for network structure formation of a binder controlled by 
means of oil-and-gasoline resistance testing. 

The effect of heating at 150° C on PDMPS/PU films colour changes was observed. However, 
after 20-hours exposition under 200° C the adhesion of PDMPS/PU films to steel and 
aluminium bases was equal to original, while PU film adhesion decreased to 3-4 points. Blends 
with 20% content of PU showed high water-repellency and sufficient thermostability, and 


Table I. Physical-mechanical Properties of Polymeric Films. 

Blend composition, PU:PDMPS 






Adhesion, point 





Bending strength, mm 





Impact strength, N'cm 





Contact angle, degree 





Impact strength after heating at 250° C, N'cm 





at 200° C 

Critical surface tension, J/m 2 , x 10' 3 



improved mechanical properties, therefore, 20% PU content can be considered as the most 
favourable. These films were studied on Scanning Electron X-ray Microprobe Analyzer. 

Silicon element content in top and bottom layers of coating appeared to be 12.7 and 9.0%, 
respectively, indicating nonuniformity of coating over its thickness and giving the evidence of 
preferable location of PU at the boundary with substrate and concurrent saturation of external 
layer with PDMPS. It is such structure of coating to explain increased adhesion and high water- 
repellency of films, based on combined binder. 

Due to high sensitivity of PU even to small amounts of moisture, fillers with low moisture- 
keeping-ability were used for reinforcement of this polymeric matrix. New organosilicate 
composite, marked OS-51-PU, was compared with its analogue OS-51-03-1 (without PU) and 
commercial OS-51 -03 (differing from OS-51 -03-1 in its silicate part). 

The setting of coatings was carried out at 120 °C in addition of curing agents TBT, typical 
for polyorganosiloxanes, and Diamet, general used for PU curing. 

Differences in composition and curing conditions were reflected by coating properties and 
energetical level of their surface, assessed by means of Zisman’s critical surface tension (y c ). 
For OS-51-03-1, OS-51-PU (in presence of Diamet) OS-51-PU (in absence of Diamet) 
coating y c 10 ' 3 was equal to 37.0, 45.6, 37.1 J/m 2 , respectively. Almost the same values of 
y c for OS-51-03-1 and OS-51-PU (in absence of Diamet) confirmed obtained earlier data on 
polymer distribution in coating matrix. Diamet dopping caused quick setting of PU interfering 
will the natural orientation in coating volume. 

Table II represents data on polymeric composites properties. 

Data have been published indicating that introduction of PU results in considerable 
improvement of mechanical properties, especially bending strength, without essential change 
in heatresistance, however,some decreasing of hardness for OS-51-PU coatings is defined 
primarily by the presence of PU component too . 


Table II. Mechanical Properties and Heat-resistance of Polymeric Composites. 

Sample description 

composite curing 




strength, mm 
thickness, pm 
50 200 

in per 

ance, °C 









120° C, 







3 h 

















120 °C, 3 h 






Data presented in Table II were utilized for evaluation of quality reserve R in accordance 
with the following equation [10]: 

R=I kj 


X i0 



, ( 1 ) 

where kj - contribution of property “i” in comprehensive rating of coating quality (the 
summary of k t equal to 1), x io and Xj max - the actual and maximum value of property “i”, 
respectively, and “n” is number of properties. 

It is a matter of fact that property contribution is dependent on coating purpose. In 
atmospheric conditions operating coating is affected by several varying factors (oxidizing air, 
moisture, lighting, temperature, ets.) and changes in properties. According to method of expert 
qualimetry for atmosphere-resistant coating the contribution of each property (adhesion, 
hardness, flexibility) in complex quality rating is equal to 0.127, 0.042, 0.037, respectively 
[11]. Quality reserve evaluated in accordance with equation (1) is equal to 0.53, 0.55 and 0.86 
for OS-51-03, OS-51-03-1 and OS-51-PU coating, respectively. 

OS-51-PU coating also compared favourably with OS-51-03 and OS-51-03-1 coatings in 
corrosion- protective properties. Corrosion resistance was estimated after accelerated testings 
simulated atmospheric conditions of temperate climate regions by means of quantitative 
method of appearance rating (Fig.l). According to this method [12] all types of coating failure, 
causing the changes of decorative (Ad) and protective (Ap) properties during testing, relative 
estimation of property change degree and contribution of each type of failure to the general 
characteristic of coating state are considered. 


Exposure time (hours x 10 3 ) Exposure time (hours x 10 3 ) 

Fig.l. Changes of corrosion-protective ( A) and decorative ( B) properties with 
testing time for coatings 

OS-51-03 (1 and 1’); OS-51-03-1 (2 and 2’); OS-51-PU (3 and 3’) 

It is seen that in the case of OS-51-PU the induction period of operation is longer than in 
other cases (1 year), at the time when its state changes slightly those of OS-51-03 and OS-51- 
03-1 change markedly, during the period of 20 months OS-51-03 coatings reach its limit while 
OS-51-03-1 and OS-51-PU coatings conserve their resource of operational properties. 

The results show that both induction time and quality reserve increase from base OS-51-03 
coating to OS-51-PU one. Long service life (no less than 5 years in tropical regions) for OS-51- 
03 was proved by numerous accelerated tests and long-term operation time in conditions of any 
climate zone [1]. The longer induction period and the higher coefficient R for OS-51-PU 
coating, in comparison with OS-51-03, permit to expect more prolonged durability. 
Computation according to the procedure for forecasting of coating durability [13] based on data 
of expert qualimetry estimated this merit to be equal to 1.6. 


1. N.P.Kharitonov, V.A.Krotikov, V.V.Ostrovsky. Qr ganosillkatnye compozitsii . (Nauka, 
Leningrad ,1980), p.91. 

2. Qr ganosilikatnve i kremnivorganicheskiev materialv v praktike stroitelnvch, 
protivokorrosionnykh. zashitno-dekorativnykh, rempntnykh i restovratsionnykh rab.o.t edited 
by V.A.Krotikov, (Znanie, Leningrad, 1991). 

3. G.S.Buslaev, V.V.Sergeeva, Lakokrasochnye matherialy i ikh primeneniye, 2, 3 (1994). 

4. G.V.Belinskaya, I.B.Peshkov. N.P.Kharitonov . Zharostoikava izolvatsiva obmotochnvch 
provodov . (Nauka, Leningrad 1978), p.159. 

5. V.A.Krotikov, N.P.Kharitonov, in Neorganicheskiye i organosilikatnve pokrvtiya . edited by 
M.M.Shultz (Nauka, Leningrad, 1975), p.374-383. 

6. V.A.Krotikov, N.P.Kharitonov, L.V.Filina, et al., Neorganicheskiye materialy, 6, (2), 362 


7. V.E.Guf, V.N.Kuleznev. Struktura i mechanicheskive svoistva polimerov. (Vysshaya 
shkola, Moskva1972). 

8. W.Berger. Sitzungsber. der Acad, der Wiss. der DDR, Math.-Natur.-Techn., 15 , 5 (1982). 

9. W.A.Zisman, Ind. Chem., 55, N10, 19 (1963). 

10. V.V.Verholantsev et al., Lakokrasochnye materialy i ikh primeneniye, 6 , 20 (1979). 

11. K.A.Krishtoff, et al., Lakokrasochnye materialy i ikh primeneniye, 5, 35 (1979). 

12. GOST 9.407-84. Unified system of corrosion and aging protection. Paint coatings. Method 
of appearance rating. 

13. M.I.Karyakina, N.V.Maiorova, Lakokrasochnye materialy i ikh primeneniye, 5, 41 (1985). 


Fart VI 

Interfaces and Composites 



Department of Materials Engineering, Drexel University, 32nd and Chestnut Sts., Philadelphia, 

PA 19104. 


Toray M40 graphite fiber / Epon 828 epoxy resin single fiber composites with both sized 
and unsized fibers were exposed to distilled water at 50°C and 100°C, 10% NaOH and HC1 
aqueous solutions at 50°C, and air at 100°C. Micro Raman spectroscopy was used to measure the 
strain and interfacial shear stress profiles as a function of environmental exposure. It was found 
that the degradation mechanism is primarily a mechanical failure of the fiber/matrix interface. 


It is well understood that the exposure of most fiber reinforced polymer composites to 
aqueous environments results in a degradation in their mechanical properties 1 * 2 . Degradation can 
occur in the matrix 2 , the fiber 3 , or the interface / interphase region 2 * 4 ' 7 (in this paper ’interface’ is 
used to describe the fiber/matrix interface and the interphase region). The mechanisms of 
degradation, however, are still not clear, and before fundamental modeling of failure and 
prevention of degradation can occur, these mechanisms must be understood and quantified. This 
paper focuses on interface / interphase degradation because it is a notoriously weak link in a 
composite, and a deterioration of interface adhesive properties affects the off axis mechanical 
properties and the toughness of the composite 1 * 8 . 

Interface degradation can be the result of a several mechanisms 2 ' 7 including a reduction in 
matrix elastic modulus (matrix plasticization), relaxation of compressive radial stresses, or bond 
failure at the interface. Bond failure can be the result of hydrolysis reactions 5 or mechanical tensile 
stresses at the fiber matrix interface 9 . To date there is no clear understanding of which mechanism 
dominates, primarily because this is matrix and fiber system dependent. In this study we 
demonstrate how micro Raman spectroscopy can be used to determine which degradation 
mechanisms prevail and prove that in the graphite / epoxy system studied, degradation is due to 
mechanical failure in the interface region. 


Using micro Raman spectroscopy, interface degradation was observed by monitoring the 
interfacial shear stress distribution along the fiber as a function of environmental exposure in both 
single fiber composites. In the following sections, the details of composite fabrication, 
environmental exposure, matrix modulus determination, and the use of micro Raman spectroscopy 
(MRS) to measure the interfacial shear stress are outlined. 

Composite Fabrication 

All composites had a matrix of Shell Epon 828 cured with metaphenylene diamine m-PDA 
(stoichiometric ratio 14.5phr). The matrix had the following properties: a glass transition 
temperature of 160°C, a coefficient of thermal expansion of 50*10' 6 K* 1 , and a modulus of 3.6 
GPa. The graphite fibers were Toray M40 sized and unsized with a modulus of 392 GPa, an 
average diameter of 6.5 pm, an average strain to failure of 0.7%, (sized) and a coefficient of 
thermal expansion in the axial direction of -1.2*10 -6 K' 1 . The sizing is an epoxy compatible sizing 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

and is a pure epoxy resin. 

Single fibers were placed into a dog bone shaped mold (ASTM D 1708-84) and the mold 
was filled with resin and curing agent. The composites were cured for 2 hours at 75°C followed 
by 2 hours at 125°C. The samples had a gauge length 50 mm long, 5 mm wide, and 2.5 mm thick. 
Single fiber composites were used to isolate the effect of the interface from fiber / fiber interaction 
and to control the alignment of the fibers. The effect of fiber / fiber interaction is currently under 
study and will be reported in a future publication. 

En vi ron m ental Exposure 

Composites were immersed in distilled water at 100°C and 50°C, and 10% aqueous 
solutions of HC1 and NaOH at 50°C. The weight gain was determined as a function of time. The 
level of saturation was calculated by dividing the weight gain at a given time over the maximum 
weight gain. Samples were removed after achieving different levels of saturation, namely 25%, 
50%, 75%, and 94-98%. Another set of samples was exposed to 100°C in air for 93.5 hours, the 
same time required to reach 94% saturation in boiling water. 

In addition to composite samples, cubes of matrix were placed in the 4 environments and 
the matrix volumetric strain and weight gain were recorded. The following summarizes those 
results: 1) in 100°C boiling water, the weight gain was linear with the square root of time and 
eventually reached a plateau at 3.28%, 2) in 50°C distilled water, two stage absorption behavior 
was observed and the maximum weight gain was 3.05%, 3) the aqueous solutions of HC1 and 
NaOH caused a weight gain of 3.1% and 2.44% respectively. 

Matrix Modulus Determination 

The modulus of the matrix was tested both before and after environmental exposure using 
dynamic mechanical analysis (DMA) and tensile testing. The DMA parameters were a heating rate 
of 5°C per minute and a frequency of 1 Hz under nitrogen atmosphere. The tensile tests were run 
at a strain rate of .25%/min. 

Strain Profiles and Interfacial Shear Stress Measurements 

The Raman effect is inelastic scattering of visible light. Monochromatic light is used to 
excite the sample to a virtual excited state (Figure 1). If the sample drops from the excited state 
into a normal vibrational mode, the resulting scattered light has lower energy than the incident 
light. In addition, the sample may start from an excited vibration mode and drop into the ground 
state and the resulting scattered light has higher energy than the incident light. These are called 
Stokes and anti-Stokes scattering respectively. When strain is applied to a crystalline sample, the 
interatomic distance changes, and thus the vibrational frequency of some of the normal modes 
change. This results in a shift in the Raman peak position (Raman frequency) with applied strain 
(Figure 1). In tension and slight compression, there is a linear relationship between the Raman 
peak shift and the applied strain. The slope of this line is the Raman Frequency Gage Factor - 
RFGF. Once the RFGF is known, an unknown strain in an embedded fiber can be calculated 

using equation 1 where ej is the strain, v 0 is the Raman peak position at zero strain, and Vi is the 
Raman peak position observed. 

8i = (Vi - V 0 ) / RFGF [1] 

The measurements were taken with an ISA Raman microprobe with a notch filter and a 
Jobin Yvon HR 640 spectrometer attached to a modified Olympus microscope. An argon ion laser 
(the 514.5 nm line) was used and the power at the sample was 2mW. A low power was used to 
prevent local heating effects. The spectra were recorded with a charge coupled device interfaced 
with a PC for data storage and manipulation. 


Monochromatic light Elastic + inelastic Scattering 


Fibers were calibrated in air and the Raman peak shift as a function of known applied strain 
determined. A single fiber 10 cm long was placed in a small loading device capable of ljjtm 
resolution in the linear deformation (a resolution of .01% in the applied strain). For Toray M40 
fibers, the peak assigned to the E2g mode shifts -10.4 cnrrV% strain, and the second order Aig 
peak shifts 25.14 cm-V% strain. The maximum error in determining the peak position was found 
to be ± 0.5cm-1 which is a error in the strain of ±0.04%. 

Tensile tests were conducted on unexposed and exposed composite dog bone samples 
The composites were loaded incrementally, and the strain in the fiber scanned at several applied 
strain levels. In addition, the residual strain due to fabrication and subsequent exposure was 

recorded. From the strain profiles, the shear stress, x, can be calculated using a force equilibrium 
in the fiber axial direction 11 . 

Ti = E*(D/4)*(d£i/dx) 

[ 2 ] 

where d£/dx is the change in strain with distance along the fiber, D is the fiber diameter, and E is 

the fiber modulus. The change in strain with distance, de/dx, was determined by fitting a curve 
through the strain points using a locally weighted least square error fitting routine and taking the 
derivative of the resulting curve. The maximum error in the shear stress determined trom the 

accumulated errors is ±5 MPa. . , 

From the interfacial shear stress (ISS) profiles, the stress transfer mechanism can be 
determined. For example, if the interfacial shear stress is highest at the fiber end and constantly 
decreasing to zero in the fiber middle, this is classic Cox model loading and indicative of an ideal 
interface 10 . If the interfacial shear stress is close to the matrix yield stress near the fiber end and is 
constant for some distance along the fiber length, this is indicative of matrix yielding limiting the 
stress transfer behavior 12 . If the interfacial shear stress is constant near the fiber end but a low 
value, this is indicative of a frictional stress transfer mechanism 12 . These three types of loading 

are illustrated in Figure 2. . , _ ^ . 

The strain profiles and interfacial shear stress were determined for as-prepared, 
hydrothermal, and thermally exposed samples at various levels of saturation. The term maximum 
ISS is defined as the maximum ISS observed at any applied strain level averaged over all the fiber 
fragments tested. 


Residual Strain 

The residual strain due to fabrication and environmental exposure was measured by 
observing the Raman frequency of the embedded fiber before applying external loading to the 
composite. This was compared to a fiber in air with zero applied strain. The residual strains are a 
summation of the compression strains during curing, compressive strains during cooling due to the 
differences in coefficient of thermal expansion, tensile strains due to matrix swelling 13 , and any 
strain relaxation that occurs at elevated temperature. It should be noted that axial strains measured 
which are the result of a volumetric strain such as differences in CTE and/or matrix swelling are 
indicative of radial strains of the same sign 9 . Therefore a tensile residual axial strain implies a 

tensile radial strain. , . . nK£? 

The residual strain due to fabrication in the as-prepared sized fiber composite was -0.15% 

strain and -0.2% in the unsized composites. These strains are the result of cooling from 125°C to 
room temperature. Adsorption of the various environmental media and the resulting matrix 
swelling caused the axial residual strains in the fiber to increase during exposure. The residual 
strains as a function of matrix weight gain (degree of saturation) are shown in Figure 3. All 


measurements were taken at room temperature. The fiber residual strain in samples exposed to 
50°C distilled water increased up to 0.5%. Samples exposed to boiling water showed only a 
+0.1% strain at 94% saturation. This low value is probably the result of sttain relaxation that took 
place during exposure. As the matrix swelled and placed the fiber in tension, the temperature was 
close enough to the glass transition temperature to allow the strains to relax. Hence, the 
composites exposed to boiling water had relaxed to a lower strain during exposure than the 
composites exposed to 50°C. This resulted in a lower strain upon cooling. Thus, the exact strain 
induced during exposure is unknown. . 

Clear evidence that the fiber experienced strains higher than 0.7% (the failure strain of the 
fiber) during exposure to boiling water is that the fiber failed during exposure to saturation levels 
of 50% or greater. This was observed for all environments at saturation levels of 50% or greater. 

In 10% aqueous solutions of HC1 and NaOH, the maximum residual strain was about 
0.25% (Figure 3) but shows a trend similar to the samples exposed to 50°C distilled water. The 
lower residual strains are the result of lower matrix swelling during exposure. 

■ 50°C Water 
• 10% NaOH 
A Boiling Water 
□ 10% HC1 

Figure 3. Axial residual strain in the fibers as the result of weight gain during exposure to 
various aqueous environments. The arrow indicates, the relaxation that probably 
occurred during exposure for the samples exposed to boiling water. 

Strain Profiles and Interfacial Shear Stress in Unexposed Composites 

During the tensile test on the unexposed composite with sized fibers, the first fiber 
fragmentation was observed at 0.8% applied strain. This is consistent with a nominal strain to 
failure of 0.7% plus the -0.15% residual strain in the fiber. The fiber strain profile at this applied 
strain level is shown in Figure 4a. Note that because this is a long fragment, only half of the 
fragment is shown. The fiber strain builds from zero at the fiber end to a strain equal to the matrix 
strain and eventually stabilizes. At this strain level, the ISS profiles for most of the fragments 
show a plateau between 35 and 40 MPa which was the maximum ISS observed for all strain 
levels. This high and constant ISS value is indicative of stress transfer limited by the matrix yield 
stress 12 . At 1.0% applied strain interfacial failure occurred near the fiber end as indicated by the 


lower (20 MPa) but constant ISS value. 

In the unsized fiber composites, the fibers failed at 1.2% strain. Given the residual strain 
of -0.2%, this is a nominal failure strain of 1.0% which is significantly higher than the sized 
fibers. The strain profile and ISS at 1.2% applied strain are shown in Figure 4b. Notice that the 
maximum ISS is almost 40 MPa and is constant near the fiber end for about 100|im. This is very 
similar to the composite with sized fibers indicating that the sizing had little effect on the ISS 
profile. Behavior similar to the sized fiber composites is also observed at higher applied strain 

The similarity in behavior of the sized and unsized fibers is somewhat surprising. 
Although the main purpose of sizing is to protect the fiber, it has also been known to increase the 
interfacial shear stress because of the improved wetting between matrix and fiber and thus 
improved bonding during curing. In this system, the maximum ISS is not affected by the sizing. 
It has been suggested that this is because in both cases, the maximum ISS is limited by the shear 
strength of the fibers' outer layers. This matter is under investigation, but preliminary results 
show that interface failure occurred in the matrix region. 

Strain and Interfacial Shear Stress Profiles for the Exposed Composites 

Figure 5 shows a comparison of the ISS at 0.8% for the composites exposed to boiling 
water up to 95% of saturation for both the sized and unsized fibers. The results for the two fiber 
systems are similar: fiber failure occurred during exposure, the maximum ISS dropped to 20 MPa, 
and the ISS is still constant near the fiber end. This low but constant ISS is indicative of a 
frictional stress transfer mechanism, implying that the interface bond failed 12 . The mechanism of 
failure/degradation is discussed in the next section. 

Only sized fiber composites were placed in 50°C 10% aqueous solutions of NaOH and 
HC1 and distilled water. Table I shows the maximum ISS as a function of saturation for all the 
samples tested including exposure to 100°C air for the same amount of time it took to reach 95% of 
saturation in water. The maximum ISS only dropped to 30 MPa for these two environments. 
Note also that the matrix strain and fiber residual strain at saturation are lower in both these 
systems compared to composites exposed to 50°C water.. 

Mechanism of Degradation 

The mechanism of degradation observed in this composite system could be due to matrix 
plasticization, chemical bond failure, or mechanical bond failure. Matrix plasticization reduces the 
matrix modulus which would reduce the interfacial shear stress according to the shear lag model 10 . 
In order to measure any change in matrix modulus due to exposure both tensile tests and dynamic 
mechanical analysis was performed on the pure matrix. The DMA results (Figure 6) show that 
there is no significant decrease in the modulus (although a decrease in glass transition temperature 
was observed). The modulus obtained from tensile tests also did not change. It could be argued, 
however, that the matrix in the region near the interface has somewhat different starting properties 
and a different response to environmental exposure than the bulk matrix. Thus macroscopic testing 
does not suffice to rule out plasticization. The observation that both the sized and unsized fiber 
composites behaved the same, however, is strong supporting evidence that matrix plasticization is 
not the cause of the reduction in interfacial shear stress. If matrix plasticization were the cause of 
the reduction in ISS, then the sized and unsized fibers which have different curing agent to resin 
ratios should have responded differently to hydrothermal exposure as documented by Drzal et al 2 . 

Therefore the degradation mechanism is either chemical or mechanical interface failure. 
Chemical bond failure would be the result of hydrolysis and should depend on the temperature of 
exposure as well as the pH of the solution. The temperature did not affect the degradation in ISS 
and both an acidic and alkaline aqueous solution reduced the amount of degradation. This implies 
that chemical degradation is not a primary degradation mechanism. The most compelling evidence 
is for mechanical degradation of the interface. Figure 7 shows the maximum ISS vs. the matrix 
swelling strain. All the samples, sized and unsized and for each exposure condition, fit on the 



Exposed / Unsized 


same curve. This strongly implies that interface degradation occurs when the radial strain in the 
interface due to matrix swelling is large enough during exposure to cause failure of the mterfacial 
bonds. This results in the frictional stress transfer mechanism (a low but constant ISS) observed 

after exposure. . , . . .. 

A second possibility is that the radial tensile strain causes an apparent reduction in the 
maximum ISS because the residual strain reduces the shear stress necessary to cause yielding. 
This would explain the slow reduction in the maximum ISS from 40 MPa to 30 MPa as a function 
of matrix strain. The pressure dependence (or normal stress dependence) of the yield stress is well 
established 14 , and in addition, using a von Mises type criterion, a radial tensile strain would reduce 
the shear stress required to cause yielding 14 . This hypothesis is under investigation. 

The discrepancy in the literature about which degradation mechanism predominates 2 ; 7 may 
depend on the kinetics of chemical vs. mechanical degradation. In addition, the mechanism of 
matrix plasticization versus a mechanical failure of the bonds is also system dependent. Clearly 
some matrices are more susceptible to matrix plasticization than others 2 *^. 

Table I. Summary of weight gain, matrix strain, fiber residual strain, and interfacial shear stress 
(ISS) for samples tested under a variety of environmental conditions. 


Maximum ISS (MP) 

Figure 6. The storage modulus obtained using dynamic mechanical analysis for Epon 828 

m-PDA cured epoxy resin as a function of exposure to aqueous environments. All 
samples were tested after achieving 95% or greater degree of saturation. 

■ 50°C Water 
□ 10% HC1 

# 10% NaOH 
A Boiling Water 
p-r-, Sized 
m Boiling Water 

Figure 7. The maximum interfacial shear stress vs. matrix swelling strain due to exposure to 
several aqueous environments. 


A pplication to Bulk Composites 

The axial and radial tension in the fiber interface region due to matrix swelling is strongly 
affected by the volume fraction of fibers. At some point the matrix swelling will cause radial 
compressive strains instead of tensile strains at the interface because of the constraint of 
neighboring fibers 9 . This will suppress the mechanical degradation mechanism. It is well 
documented, however, that there is an inhomogeneous distribution of fibers in a composite and 
that in joints and comers where there is a low volume fraction of fibers, the current degradation 
mechanism will apply. Therefore, this single fiber work is relevant to bulk composite failure in 
regions with a high volume fraction of resin. The volume fraction at which the radial strains will 
become compressive instead of tensile is under investigation. 


1. The sizing on Toray M40 fibers does not significantly affect the maximum interfacial shear 
stress or the mechanism of degradation in boiling water. 

2. The mechanism of degradation in Toray M40 / Epon 828 - mPDA cured matrix composites 
is primarily mechanical failure of the interface due to matrix swelling. This was proved by 
showing that matrix plasticization did not occur and that the degradation was independent 
of the pH of solution and the temperature of the environment. Degradation depended 
strongly on the matrix swelling strain. 

3. The presence of HC1 and NaOH decreases the matrix swelling of the matrix by about 25%. 
This reduces the amount of interfacial degradation. 


The authors are grateful for funding from the Office of Naval Research, Materials Division. 


1. G.S. Springer, Environmental Effects on Composite Materials Vol. II. , edited by G.S. Springer 
(Technomic Publishing, Lancaster, PA, 1984), chapter 1. 

2. L.T. Drzal, M. J. Rich, M. F. Koenig, J. Adhesion, 18, 49 (1985). 

3. C. Lhyman and J.M. Schultz, J. Mater. Sci., 18, 29 (1979). 

4. D.A. Biro, G. Pleizier, Y. Deslandes, Comp. Sci. Tech., 46 , 293 (1993). 

5. U. Guar, B. Miller, Polymer Composites, 11, 217 (1990). 

6. M.S. Amer, M.J. Koczak, C. Galiotis, L. S. Schadler, Adv. Comp. Lett. 3, 17 (1994) 

7. D.H. Kaelble, PJ. Dynes, L. Maus, J. Adhesion, 8, 121 (1976). 

8. I.M. Daniel, G. Yaniv, G. Peimanidis, J. Engineering Materials and Technology, 110 169 

9. F.W. Crossman, R.E. Mauri, W J. Warren in Advanced Composite Materials Environmental 
Effects. ASTM STP 658 . edited by J. Vinson (American Society for Testing and Materials, 
Philadelphia, PA, 1978) pp.205-220. 

10. H.L. Cox, Brit. J. Appl. Phys., 3, 72 (1952). 

11. N. Melanitis, C. Galiotis, P.L. Tetlow, C.K.L. Davies, J. Comp. Mat. 26, 574 (1992). 

12. L. Dilandro, A.T. Dibenedetto, J. Groeger, Poly. Comp., 9, 209 (1988). 

13. H.T. Hahn, J. Comp. Mat. 10, 266 (1976). 

14. I.M. Ward and D.W. Hadley, An Introduction to the Mechanical Properties of Solid 
Polymers . (John Wiley and Sons, New York, 1993), pp.214-244. 




*Toin University of Yokohama, Department of Materials Science and Technology, Yokohama, 

**Maeta Concrete Industry Ltd., Central Research Laboratory, Yamagata, Sakata, Japan 


Here we investigated the interactions of phenol resin precursor and calcium aluminates, in 
relation to a recently innovated cement based material having a high flexural strength of more 
than 120MPa. An anhydrous phenol resin precursor was used as the binder and water was not 
contained in the initial composition. 

The method of processing consists of mixing of the cement, the phenol resin precursor, and 
small amounts of N-methoxymethyl 6-nylon and glycerol under high shear. Addition of the 6- 
nylon and glycerol was necessary to produce viscoelastic cement paste through a twin roll mill. 
Setting of calendered sheets takes place during the heat curing at 200°C. 

The best combination for high flexural strength among all the cements tested is the mixture of 
calcium aluminate cement and the resole type of phenol resin. Resulting outstanding affinity 
suggested specific interactions between the phenol resin and calcium aluminate. 

We here propose the interaction evidence of phenol moiety-calcium aluminate, based on the 
experimental data of differential scanning calorimetry and conduction calorimetry. 


It has been well-established that cement hardened in the presence of sufficient amount of 
water can be used to form architectural cement materials. Ordinary hardened cement material, 
however, shows flexural strength of less than 20MPa, and therefore, enhancement of the 
flexural strength of the cement has been a long standing problem in the field of cement 

Recently we innovated a new type of composite material which consists of phenol resin and 
calcium aluminate cement and which shows excellent flexural strength (>120MPa). 

The flexural strength of the innovated material is unusually high, compared to ordinary 
cement based materials and even to various types of polymer cement composites. The 
combination of resole type phenol resin and calcium aluminate cement display unique properties 
relative to several phenol/powder combinations attempted. Based on these observations, we have 
assumed certain specific interactions between the phenol resin and calcium aluminate cement to 
occur during hardening. Of further interest is that the present composite material is fabricated by 
an essentially anhydrous compositions. The necessary water for hardening is furnished by the 
precursor during thermal curing between 130°C-170°C, which we considered as the 
fundamental principle on which the innovation rests [1] [2] [3], The main purpose of the present 
report is to discuss the specific affinity between calcium aluminate and phenol moiety. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 



Calcium aluminate cement (r.d. 3.01) used is mainly composed of 54.3wt% AI2Q3, 37.0wt% 
CaO, 4.5wt% SiCh and 1.5wt% FesO (here after this cement is abbreviated as CAC). The main 
mineral constituent of CAC is monocalcium aluminate (CA), but CA 2, Ci 2A7, C 2 AS and a AI2Q3 
have also been identified by X-ray diffraction. 

For differential scanning calorimetry analyses, pure monocalcium aluminate (CA) was also 
used to identify the most favorable mineralogical component of CAC. The pure CA was prepared 
by reacting equimolar amounts of CaCCb and AI 2 Q 3 powder at 1450°C. 

To confirm the affinity between CAC and phenol resin, ordinary portland cement, alumina 
and silica powder were also used. Ordinary portland cement (r.d. 3.16) is mainly composed of 
5.3wt% AI 2 G 5 , 64.4wt% CaO, 22.0wt% SiCb and 3.0wt% Fe 2 Cfe (here after this cement is 
abbreviated as OPC). Alumina powder (r.d. 3.95) contains 99.6wt% AhCb. Silica powder (r.d. 
2.63) contains 92.0wt% Si02 and 6.0wt% AI 2 O. 

Commercially available resole type phenol resin precursor was used as the main binder. The 
precursor is essentially anhydrous and soluble in methanol, and contains 58.0wt% of nonvolatile 
matter. Specific gravity is 1.06 and viscosity is 250cps. 

N-methoxymethyl 6-nylon was incorporated to modify the phenol resin and to develop the 
plasticity of the paste. Glycerol was used as a plasticizer. 


General method of preparing specimens is nearly equal to the technology of macro-defect 
free cement developed by Birchall and coworkers [4] at Imperial Chemical Industries in the early 
1980’s. Powder-resin paste was mixed under high shear and calendered through a twin roll mill 
to obtain the flexible sheets. The calendered sheets were allowed to cure at 200°C for 18 hours. 
The mix proportions are given in Table I. 

Table I. Mix proportions. 

Phenol resin 

Mix proportions by weight 


Cement, , 


Solid substances 


solution of 
phenol resin 







(vo! %) 






















ai 2°3 







Si °2 






The cured sheets were tested for flexural strength. The specimen size was 25 mm wide, 200 
mm long and 2mm thick. Half of each specimen was subjected to a three point bend test in which 
the span-depth ratio was 40. The remaining part of each specimen was placed in water at 20°C. 
After 7 days immersion in water, the flexural strength was again measured in the same manner. 


Differential Scanning Calorimetry (DSC) analyses were made on several powders (CAC, 
CA, AhCb, SiOz) +phenol resin combinations to understand their interactions during hardening. 
Each powder was mixed with phenol resin in resin-powder ratio of 0.5, then the paste was 
vacuum dried at 30°C for 24 h. The resulting lumps were ground and pressed into an aluminum 
crucible. The DSC analyses were carried out on about 10 mg of the sample at a heating rate of 
10°C per minute. 

The influence of phenol resin on the normal hydration of cement was studied by conduction 
calorimetry as described by Rodger etal. [5], CAC, CAC-resin composite (Mix No.l) and SiOz- 
resin composite (Mix No.4) were analyzed. Both CAC-resin and SiOz-resin composites were 
ground and sieved to a size of less than 53 pim, and then used in calorimetry. After mixing the 
powder and water, the heat evolution curves were recorded at 20°C using a conduction 
calorimeter. The water-powder ratio was 0.5. 


Influence of Basic Components on Strength and Water Resistivity 

Fig. 1 shows the original flexural strength and residual strength after 7-day water 
immersion. The original flexural strengths of non-cement based specimens are half of the CAC 
based specimen, although the total porosity and maximum pore size of the microstructure are 
found to decrease [6], The flexural strength of the CAC based specimen is higher than that of 
the OPC based specimen. After 7 days in water, the flexural strength of both non-cement 
based specimens decreases by about 60%, while CAC based specimen increases by 9%. 

CAC OPC A1 2 0 3 Si0 2 

Fig. 1. Original flexural strength and residual strength after 7-day water immersion. 

In the OPC based specimen, the reduction of flexural strength appears after water immersion, but 
the loss in strength is only 30%. 

The most striking feature of above result is its water resistivity depending on the base 
material. Thus, it is easily inferred that the water resistivity is not an individual property of 


hardened resin itself. It is supposed that the phenol resin and cement have mutual influence and 
affinity related to their bond properties. There are two hypotheses to account for strength loss of 
the non-cement based specimens. These may be summarized as follows: 

1. The interparticle bond strength of the non-cement based material is less than that of the 
cement based composite due to the absence of a strong interphase region [6]. Thus, the 
mechanical breaking of weak van der Waals bonds occurs, in the same manner of low 
fired clay bricks when water ingresses into the structure. 

2. The system of the non-cement based material exists in an unstable state because the 
organic-inorganic interaction does not occur. 

Differential Scanning Calorimetry Analyses 

Generally, the hardening of the phenol resin precursor used here takes place at 130 - 170°C 
by the reaction of methylene bridge formation. Thus, thermal analyses would give more insight 
as to the mutual influence of phenol resin and cement: that is, the system is measured as a 
function of temperature. 

Fig. 2. DSC thermograms. 

Curve 1 in Fig. 2 shows a DSC thermogram of the phenol resin precursor. The sharp 
endothermic peak around 162°C is seen, corresponding to a hardening reaction of the resin. In 


the case of CAC+phenol resin (curve 5), an endothermic peak is not seen and may be 
overlapped by a large broad exothermic peak around 171 °C. Curve 4 of CA+phenoI resin is 
essentially the same as curve 5. This indicates that CA is the component of the cement which 
seems to be responsible for any possible interactions. On the other hand, in the cases of Si02 and 
AI2Q5 (curves 2 and 3) which are inert fillers, neither a sharp endothermic peak nor broad 
exothermic peak is seen. The difference between these peaks seems to depend on the stabilization 
in each system. From the above results, it is suggested that there is an interaction between 
calcium aluminate and phenol resin. 

Conduction Calorimetry 

Fig. 3 shows the integral heat evolution curves obtained from a hydrating mix of CAC, 
CAC-resin composite (Mix No.l) and SiCh-resin composite (Mix No.4). The curve of CAC 
shows a sharp increase in the amount of heat evolution at about 15 hours, thus indicating the heat 
of hydration. The presence of phenol resin in CAC results in lack of such a sharp increase, 
however the amount of heat evolved increases gradually as the time proceeds. In case of SiCh - 
resin composite, the amount of heat evolved remains nearly constant up to the test termination at 
168 hours. 

0 42 84 126 168 

Hydration Time (hour) 

Fig. 3. Integral heat evolution curves for CAC, CAC-resin composite 
and Si 02 -resin composite. 

These results suggest that CAC becomes less sensitive to water in the presence of phenol 
resin. This may be due to the following reasons. 

1. The presence of phenol resin retards cement hydration as described by Vipulanandan and 
Krishnan for water soluble phenol [7] or the hydration ability of cement is overtaken by 
the phenol resin. 

2. A hard phenol resin layer which may exist around the cement particles obstructs the water 
ingress into the cement particles, thereby preventing the cement hydration. 



The knowledge gained from this work suggests that the resole type phenol and calcium 
aluminate interact to make a strong composite with high water resistivity. Differential scanning 
calorimetry shows difference in hardening between calcium aluminate and inert fillers. The large 
exothermic peaks in both calcium aluminate and calcium aluminate cement confirmed the 
stabilization of the calcium aluminate-phenol system. In contrast, high water resistivity is only 
given by calcium aluminate cement-phenol composite. Thus, it appears that the development of a 
strong bond by interactions is essential for durability. However, this study indicates that further 
investigation should be carried out to understand the reactions during setting of the composite. 


The authors wish to thank Mr. Tokuhiko Shirasaka and Mr. Takayoshi Okamura (Central 
Research Laboratory, Chichibu Onoda Cement Corporation) for their assistance on chemical 
analyses and useful discussions. 


1) T. Kobayashi, G.K.D. Pushpalal, M. Hasegawa, Japanese Patent Application No. JP 
301514 / 92 to Maeta Concrete Industry Ltd., Japan (1992). 

2) T. Kobayashi, G.K.D. Pushpalal, M. Hasegawa, European Patent Application No. 
93307706 to Maeta Concrete Industry Ltd., Japan (1993). 

3) T. Kobayashi, G.K.D. Pushpalal, M. Hasegawa, U. S. Patent Application No. 93307706 
to Maeta Concrete Industry Ltd., Japan (1993). 

4) J.D. Birchall, A.J. Howard, K. Kendall, "Cement Composition and Product," U. S. Patent 
No. 4410366(1983). 

5) S. A. Rodger, S. A. Brooks, W. Sinclair, G. W. Groves, D. D. Double, Journal of Material 
Science 20 , 2853 (1985). 

6 ) G.K.D. Pushpalal, N. Maeda, T. Kawano, T. Kobayashi, M. Hasegawa, T. Takata, 
Properties and Flexural Failure Mechanism of a High Strength Phenol Resin-Cement 
Composite, Submitted to 8th International Congress on Polymers in Concrete, Belgium 
(July 1995). 

7) C. Vipulanandan and S. Krishnan, Cement and Concrete Research 23, 792 (1993). 




University of Michigan, Macromolecular Science and Engineering Program, Ann Arbor, MI 48109 


Ceramic fiber spinning is critical to the manufacture of fibrous monolith ceramics, and 
understanding the interactions between the polymer matrix and ceramic particle filler is necessary to 
predict the flow properties of these systems. In this study, poly (methyl methacrylate) (PMMA) 
was filled with octadecanol-coated Stober silica, and the rheological behavior of the filled polymers 
investigated at various filler volume fractions. The rheological behavior of these materials was 
studied in both dynamic and steady-state experiments. The time required for filled PMMA to reach 
steady-state behavior under constant shear stress was found to be long, on the order of an hour. 
The steady-state viscosities increased as expected with filler volume fraction, but did not correlate 
well with existing models. This is hypothesized to be a result of matrix-filler surface interactions, 
and will be investigated further in future work. 


Dispersion flow behavior is by no means a new field. From Einstein’s work in 1905 1 through 
Tanaka and White’s theory 2 , much work has been done to correlate flow behavior to component 
material properties. Significant successes have been achieved in predicting steady-state viscosity 
for Newtonian fluids filled at low volume fractions with non-interacting particles 3 ' 4,5 . Other 
research has extended these findings, both empirically and theoretically, to non-Newtonian fluids, 
interacting particles, and higher volume fractions 6,7 . What is missing, however, is an investigation 
of the time-dependent properties of these filled systems. Transient, time-dependent behavior is 
usually neglected in fluid dynamics treatments since in most fluids this behavior is limited to a very 
short times after flow is induced. This is not the case for filled polymers, however. Batch 
processing times for filled polymers vary greatly, but are typically no longer than 60 minutes 8 . Our 
research, however, has shown that poly (methyl methacrylate) filled with silica can take more than 
60 minutes to achieve steady-state flow behavior. This would indicate that one cannot assume 
steady-state behavior for most processing times. 

This paper describes the results from a non-polar ceramic filler in poly(methyl methacrylate), a 
strongly polar polymer. The overall study will include three different surface polarity fillers and 
three different polymers in various relative volume fractions. The ceramic filler used was colloidal 
silica with octadecyl groups chemically grafted to the surface. The volume fraction of filler was 
varied between 20 and 50 percent, typical for high solids applications such as ceramic fiber 
spinning 9 . 



The materials used in this study were colloidal silica and poly(methyl methacrylate) (PMMA). 
The silica was made by condensation reaction of tetraethyl ortho silicate (TEOS) catalyzed by 
ammonium hydroxide in ethanol, in a technique developed by Stober 10 . The reactant and catalyst 
ratios were determined from previous work by Davis 11 and Bogush 12 . The spherical particles were 
characterized using scanning electron microscopy, and their size was determined to be 0.45 Jim in 
diameter, with a standard deviation of 0.07 p,m. Octadecanol was then grafted to silanol surface 
sites through an esterification reaction at 180°C to give a uniform non-polar surface 13 . Besides 
providing a non-polar surface on the ceramic, the octadecanol provided steric stabilization, 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

preventing the silica particles from irreversibly aggregating when removed from ethanol. TTie 
grafted chains were too short, however, to result in entanglements or physical cross-links . Thus, 
the particles still behaved as independent components within the polymer matrix. 

The PMMA had a number average molecular weight of 25,000. It was purchased from 
Polysciences as 200 Jim beads, and was used as received. 

Filled Polymer Preparation 

The coated silica and polymer beads were allowed to dry for 72 hours in an 85°C ventilated 
oven, and then were stored until use in a desiccator. The silica was mixed into the PMMA in a 
Brabender Plasticorder operating at 200°C and 60 rpm. To further eliminate adsorbed water on the 
polymer, the PMMA was melted and mixed in the Plasticorder for 20 minutes at 200°C prior to the 
additionof the filler. No degradation of the polymer matrix was observed during mixing. 

Rheological Characterization 

All experiments were performed using a Bohlin CS 50 controlled stress rheometer. A cone and 
plate geometry (40 mm diameter, 4° angle) was used to provide a constant stress field within die 
sample during testing. Tests were run at 200°C to ensure that the sample was fluid enough ror 
measurable results, given instrument stress limitations. Samples were allowed to equilibrate at 
200°C for 60 minutes prior to testing, to eliminate any residual stresses incurred during processing. 
It was determined that the instantaneous viscosity and dynamic moduli of samples changed hnearly 
with time after an equilibration period of 1 hour. This indicates that there could be up to a 60% 
variation in sample properties over experiment run time (Figure 1.). Since the increase is linear for 
all parameters, the effects of thermal history on viscoelastic properties may be eliminated by 
normalizing the plots to a constant thermal exposure. For the purposes of this paper, however, 
only data that has similar residence times at elevated temperatures will be compared. 

Time (seconds) 

Figure 1. Variation in storage modulus with time at processing 
temperature. The sample used is 35% (vol) silica at a shear stress of 50 Pa 
and a frequency of 0.005 s' 1 . The temperature is 200°C. 

Four types of tests were performed: stress sweep at constant frequency, frequency sweep at 
constant stress, stress ramp, and creep compliance. The first type, stress sweep, gave a general 
idea of the linear viscoelastic region for these materials. Frequency sweeps were used to probe the 
sample structure without perturbation. Yield stresses for these materials were determined using a 


60 second stress ramp. Creep compliance tests were used both to determine the time-dependent 
behavior of the system and to determine the steady-state viscosity for the material. 


Despite the fact that few filled ceramic processing applications operate within these materials’ 
linear viscoelastic regime, these experiments were designed to remain within this region for 
comparison with existing work and future modeling purposes. The extent of the linear viscoelastic 
region was determined by oscillatory experiments in which the frequency was held constant, and 
the shear stress varied. The storage and loss moduli were measured by these experiments. A low 
shear stress plateau was expected in both moduli, followed by a gradual decrease. The onset of the 
decrease in moduli is generally considered to correlate to the end of the linear viscoelastic region. 
The data acquired from our materials, however, showed relative maxima in these moduli, with the 
shear stress corresponding to the maxima increasing with filler volume fraction. Increasing storage 
modulus is indicative of structure formation, and this sort of behavior has not been previously 
reported, nor does it fit linear viscoelastic theory. Further experiments suggested that the behavior 
is an artifact of the long stress relaxation times of these materials. The samples were still re¬ 
forming their equilibrium structure when the tests began, following a 2400 second equilibration 
time after loading. Longer equilibration times are thus desirable, but the properties of the materials 
changed with time at the operating temperature of 200°C. Ultimately we selected an equilibrium 
time of 3600 seconds, long enough for most equilibrium structure to be re-formed, but not long 
enough to significantly distort the viscoelastic properties by thermal degradation. 

The frequency sweep experiments produced the expected results. As also noted by Sideridis 15 , 
the moduli increased monotonically with frequency and filler volume fraction, and the behavior 
agreed with the literature. 

Although there is controversy regarding the existence of a yield stress 16 , stress ramp 
measurements were used to determine an “apparent yield stress” for the materials studied. As 
expected, and predicted by Tanaka and White, the yield stress increased with increasing filler 
concentration (Table 1.). The pure polymer exhibited no yield stress. 

Table 1. Yield stress measurements for filled PMMA 

Volume Fraction Filler 


233 Pa 


333 Pa 


1580 Pa 

The creep compliance experiments represent our first attempt to quantify the highly time- 
dependent behavior of these materials that affected the stress sweep experiments. Repeated creep 
tests indicated that a shear stress of 20 Pa was within the linear viscoelastic region for both 0.20 
and 0.35 silica volume fraction samples. Difficulty was encountered in determining the linear 
viscoelastic region for the 0.50 volume fraction sample because an applied stress of 20 Pa resulted 
in strains at the lower limit of instrument resolution. In such situations it is common to consider the 
limit of instrument resolution as the edge of the linear viscoelastic region. All creep compliance 
experiments were therefore performed at a shear stress of 20 Pa. 

The compliance curves are shown in Figure 2. Steady-state for viscoelastic liquids is evidenced 
by a linear compliance curve (constant slope). As can be seen, none of the materials tested is 
approaching this condition rapidly, even after 2400 seconds. The shape of the curves, however, is 
nearly identical, indicating that all three samples have similar flow mechanisms. The scatter evident 
in the 0.5 volume fraction sample is a result of instrument noise; the shape of the curve is 


Figure 2. Creep compliance behavior for silica-filled PMMA. Samples 
were run at a shear stress of 20 Pa and a temperature of 200°C. 

Besides time-dependent behavior, compliance curves also can give steady shear viscosity data. 
This is the most commonly reported parameter of filled systems, and the most often modeled. The 
steady shear viscosity is the inverse slope of the compliance asymptote. The compliance curves are 
not yet at equilibrium, but the steady-state viscosities can be estimated from the last few points on 
the graph. Since the slope is continually decreasing toward the asymptote, the estimated viscosities 
will be slightly low, but they will be sufficient to compare several of the theories found in the 
literature. The estimated viscosities and curve fits are found in Figure 3. The viscosity theories 
investigated were proposed by Mooney 3 , 

Krieger and Dougherty 4 , 




( 2 ) 


and Frankel and Acrivos 6 . 

7} r = 1 + - 
lr 8 



In all cases, 0 is filler volume fraction, 0 m is maximum packing fraction, and K is an 
empirical constant. In Figure 3., K is set at 2.5 (after Einstein 1 ), and 0 m at 0.63 
(random packing of monodisperse spheres). 

Figure 3. Approximate steady-state viscosities for silica filled PMMA. 

Data extrapolated from compliance curves and fit with existing viscosity 

Even though the estimates for steady-shear viscosity are inherently low, all of the models examined 
predicted viscosities that are even lower. There are several explanations for this. The equations of 
Mooney and Krieger-Dougherty were developed based on the original model of Einstein. While the 
pairwise interactions postulated in these models certainly exist in all systems, at high filler volume 
fractions, many particles interact simultaneously in a cage or cell interaction model. This was the 
approach taken by Frankel and Acrivos, however, and their equation also could not describe our 
data. Some characteristic of our system must be not be taken into account by these models. 

It is our belief that the discrepancy between our observed viscosities and those predicted by 
these models is a result of surface interactions between matrix and filler. These interactions, if 
present, would result in the formation of a stable structure that would resist deformation and flow, 
and would be evidenced by elevated steady-state viscosity. If this is correct, it would follow that 
increasing the surface polarity of the fillers would result in even stronger structure formation, and 
correspondingly higher viscosities. This will be investigated in future work. 


While these models were all developed for Newtonian matrices, this is not likely to be the most 
significant factor contributing to the discrepancy between their predictions and our data. Polymers 
are inherently non-Newtonian, but the PMMA molecules used in this work are only 250 repeat 
units long, and the deviation from Newtonian behavior is negligible for the stress levels used in 
these experiments. 

There are many other predictive theories, all applicable in certain situations, which were not fit 
to this data. The models of Einstein 1 and Batchelor 5 are applicable at far smaller filler volume 
fractions than are investigated here. More detailed models exist, namely those by Kemer 17 and 
Tanaka and White 2 , but in the absence of extremely detailed information on both components (e.g. 
reciprocal Debye length and Hamaker constant) these were not applicable. The typical use of the 
Tanaka and White expression is empirical, with the unknown quantities acting as empirical 
parameters, but with only four data points, the number of adjustable parameters would nearly 
ensure a fit. 


The results of this work indicate that there is a need for further study in two areas of dispersion 
flow behavior, namely time-dependent properties and matrix-filler interaction effects. The 
application of any steady state theory to practice will be limited by transient time-dependent 
behavior, which has been shown in this work to persist for process-length times, at least for one 
system. Steady state viscosity approximations were compared to three existing theories and the 
data were not adequately described by any one theory. Since approximation errors would only 
have increased the discrepancy, it is evident that some phenomena exists in our system that did not 
affect those on which the models were developed. This phenomena has been hypothesized to be 
matrix-filler polar surface interactions, and will be studied in future work. 


We gratefully acknowledge the financial support of the National Science Foundation under 
Grant CTS-9058078. 


1 A. Einstein, Ann. Physik 17, 459 (1905). 

2 H. Tanaka and J.L. White, J. Non-Newtonian Fluid Mech. 7, 333 (1980). 

3 M. Mooney, J. Colloid Sci. 6, 162 (1951). 

4 I.M. Krieger and T.J. Dougherty, Trans. Soc. Rheology 3, 137 (1959). 

5 G.K. Batchelor, J. Fluid Mech. 83, 97 (1977). 

6 N.A. Frankel and A. Acrivos, Chem. Eng. Sci. 22, 847 (1967). 

7 D.G. Thomas, J. Colloid Sci. 20, 267 (1965). 

8 D. Kovar and G.A. Brady (personal communication). 

9 Ibid. 

10 W. Stober, A. Fink, and E. Bohn, J. Colloid Interface Sci. 26, 62 (1968). 

11 K.E. Davis. Sedimentation and Crystallization of Hard-Sphere Colloidal Suspensions: Theory 
and Experiment . Ph.D. thesis, Princeton University, 1989. 

12 G.H. Bogush, M.A. Tracy, and C.F. Zukoski, J. Non-Crystalline Solids 104, 95 (1988). 

13 R.D. Badley, W.T. Ford, F.J. McEnroe, and R.A. Assink, Langmuir 6, 792 (1990). 

14 L.J. Fetters, D.J. Lohse, D. Richter, T.A. Witten, and A. Zirkel, Macromolecules 27 (17), 

15 E. Sideridis, Comp. Sci. Tech. 27, 305 (1986). 

16 D. De Kee and C.F. Chan Man Fong, J. Rheol. 37 (4), 775 (1993) 

17 E.H. Kemer, Proc. Phys. Soc. London B69, 808 (1956). 



Barry J. Bauer, Catheryn L. Jackson, and Da-Wei Liu 

National Institute of Standards and Technology, Gaithersburg MD 20899 


Interpenetrating polymer networks have been synthesized by performing sol-gel 
chemistry and conventional organic polymerizations in mixtures of the monomers. The organic 
polymers were acrylates, and the inorganic phase was Si0 2 formed by hydrolysis of 
orthosilicates. Polymerizations were conducted at a variety of relative rates, and the chemistry 
was designed to allow different amounts of grafting between the components. The morphology 
was characterized by transmission electron microscopy and small angle neutron and x-ray 
scattering. Wide variations in morphology were observed depending on the polymerization 
conditions, ranging from grossly phase separated to dendritic to finely divided structures (at a 
100A size scale). The phases ranged from mixtures of the two components to relatively pure 
phases. The interface between the phases ranged from very narrow to relatively broad. 


Recently, organic-inorganic composites have been synthesized by forming the Si0 2 phase 
through sol-gel chemistry using interpenetrating polymer network techniques [1-7]. The 
polymerization scheme for the most general case is as shown below: 

Si(OR) 4 +2H 2 0 + Monomer + Solvent —> 

Si0 2 + 4ROH + Polymer + Solvent 

Two independent reactions can take place; the sol-gel reaction will form an inorganic 
component and the organic reaction will form an organic polymer. The resulting material is 
an interpenetrating polymer network (IPN) made up of organic/inorganic phases. If the solvent 
is a monomer, shrinkage is reduced since it becomes incorporated in the sample as a nonvolatile 
polymer. "Non-shrinking" composites can be prepared if the "R" group can polymerize with 
the organic monomer, since no weight loss occurs due to elimination of volatile components 
from the reaction mixture [6]. 

IPNs of two organic polymers are traditionally formed by independent polymerizations 
that give a sample that is microscopically phase separated, but macroscopically uniform [8]. 
The two polymerizations may be either sequential or simultaneous. At the beginning of the 
reaction, the two monomers or the monomer-polymer combination form a single phase, but 
phase separation occurs during polymerization and a finely divided morphology results. Grafting 
between the components of IPNs can also greatly affect the point of the phase separation and 
the nature of the resulting morphology. Studies of IPNs made from polymers that form 
miscible blends have shown that while the crosslinking reaction, without grafting between the 
components, can strongly promote phase separation [9], the opposite is true when grafting takes 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

place [10]. The two-phase morphology formed with IPN chemistry is often significantly 
different from that of a simple physical mixture of the two components. In particular, co- 
continuous morphologies can yield materials with bulk mechanical properties that are 
dominated by the minor component. In contrast, conventional silica filled polymers are made 
by physically mixing preformed silica into polymers, which tends to produce properties that are 
either dominated by the matrix polymer, or are an average of the properties of two components. 

For silica filled polymers formed by swelling an organic polymer or network in orthosilicate 
monomers and subsequently polymerizing by the sol-gel method [1-4], the silica is usually more 
finely dispersed than a physical mixture but the morphology often contains discrete particles. 

In this paper we explore the morphologies of organic-inorganic IPNs synthesized with 
the Si0 2 phase made by sol-gel chemistry and the organic phase made from poly(2- 
hydroxyethyl acrylate) (HEA). The synthetic variables include the relative reaction rates 
(sequential vs. simultaneous polymerizations) and grafting. Small angle x-ray scattering 
(SAXS), small angle neutron scattering (SANS), and transmission electron microscopy (TEM) 
are used to learn how synthetic variables affect the resulting structures. 


The ungrafted HEA-series was prepared by adding a water/HF mixture to 
tetraethylorthosilicate (TEOS) in the presence of an organic monomer, 2-hydroxyethyl acrylate 
(HEA), and benzoyl peroxide. The concentration of benzoyl peroxide was kept constant but 
the amount of HF was varied by a factor of 70 so that the relative rates of the two 
polymerizations covered the range of vinyl > sol-gel to sol-gel > vinyl. All samples were 
thermally polymerized at 70°C until fully reacted, and then dried in vacuum at 100°C to remove 
the ethanol produced in the reaction. The grafted B-series was prepared by using equal weights 
of HEA and Tetrakis(2-acryloxyethoxy)-silane (TAEOS) and polymerizing as in the case of the 
TEOS system, also with a varied HF concentration. More synthetic details are presented 
elsewhere [11-12]. 

Small angle x-ray scattering was performed at the NIST Polymers Division SAXS 
facility [13-14]. Small angle neutron scattering (SANS) was performed at the 30 meter NSF 
instrument at NIST [13,15]. Thermogravimetric analysis (TGA) was used to determine the Si0 2 
content of the specimens and the decomposition temperature profile. Approximately 5 mg of 
the sample was heated at 10 °C per minute from 30 to 800 °C in an air atmosphere. For the 
transmission electron microscopy (TEM) studies, ultra-thin sections were cryomicrotomed using 
a diamond knife at temperatures from -20 °C to -80 °C. The sections were imaged on a Philips 
400T at 120 kV [13]. The contrast between the silica containing phase and the polymer was 
sufficient for imaging and no staining was required. 


We will first present a discussion and analysis of the scattering data. A double 
logarithmic plot of SAXS or SANS intensity, I, versus the scattering vector, q, for the HEA- 
series and the B-series, is shown in figure 1. The scattering vector is defined as q = 
47t/A,(sin(0/2)), where 0 is the scattering angle and X is the wavelength. The variation in these 
series are the catalyst ratio and grafting. Both SANS and SAXS data are shown in arbitrary 
intensity units and are shifted vertically for clarity. Samples HEA-1 and HEA-2 have a constant 


power law over a considerable q range, with a slope of approximately -3.1. At higher scattering 
angle, the slopes in the Porod region are all near 9log(I)yQlog(q) = -4, indicating strong phase 
separation into domains with little mixing of the phases. Samples HEA-4 and HEA-5 have 
power law scattering < -4, suggesting that broader interfacial areas exist. The B series have 
power laws of dlog(I)/3log(q) = -2.4 for sample B-2 and -2.0 for sample B-7, indicating 
extensive mixing of the two components and weak phase separation. For example, a limiting 
power law of -2 is seen in the case of miscible polymer blends where intimate mixing of the 
two components occurs [19]. In addition, sample B-2 has an extended power law range down 
to q = .004 A ' 1 and sample B-7 has very weak scattering at low-q with a weak peak at .035 A 1 . 

Small angle scattering 
from two phase materials is 
generally characterized as having 
contributions from the domain 
structures of the two phases, the 
interfacial regions between the 
domains, and mixtures of the two 
components within the phases. 

In cases where contributions 
from mixtures of components 
within the phases is negligible, 
data analysis in the Porod region 
can give characteristic phase size 
and interfacial thickness [16-18]. 

The scattering from a two phase 
mixture with volume, V, 
interfacial area. A, volume 

fraction of one component, <(>, F i gure i. sans and saxs from poIyhea/sio., ipns. 
and interfacial thickness, T has 

an asymptotic form as shown in equation 1. In this form the scattered intensity, I, is in both 
numerator and denominator, so that absolute intensity calibration is not necessary. The surface 
to volume ratio is related to characteristic phase sizes, with 1, and 1 2 being the average chord 
size of phases 1 and 2, with equal to 4<J>j V/S. [18]. T is a characteristic interfacial thickness 
characteristic of the zone of mixing between the two phases. 

--glng) - S/v exp ( -T 2 g 2 ) 

r g *n<i)dQ (1) 


Table I list the catalyst concentration and Si0 2 content for the various samples as well 
as the values of the V/S and T calculated from the SAXS and SANS characterization. The B 
series samples never approach a region with a -4 power law, so a Porod analysis is impossible 
and values of V/S and T cannot be determined. For the HEA series, the samples with the 
lowest HF catalyst levels showed power law scattering over a wide q range. For samples where 
the vinyl polymerization is relatively fast compared to the sol-gel reaction, the domain sizes are 
largest and negligible interfacial thicknesses are measured. As the HF concentration increases, 
the domains become smaller and the interfaces become broader. 



HF cone 

Si0 2 


V/S (A) 

T (A) 

TEM (A) 









gross phase 




















fine texture 







fine texture 




no Porod 



mottled, fine 






single phase 

Table I. Sol-Gel IPNs 

Sample HEA-1 has gross phase separation early in the polymerization forming large 
white clumps of polyHEA. It has been noted that polyHEA is not soluble in TE0S/H 2 0 
without a cosolvent [1], so when the polyHEA is formed before the gelation of the TEOS, gross 
phase separation occurs. TGA of different parts of the sample gave Si0 2 contents between 0.10 
and 0.23 g/g demonstrating the large size scale of the phase separation. SANS and SAXS 
techniques cannot see phases this large and can only see an average of structures within the 
phases. Scattering techniques that probe a larger size scale such as light scattering were not 
attempted, but the opaque, white appearance of the samples strongly suggest large domain sizes. 

For the lower sol-gel rates in the HEA series, a dendritic morphology is observed by 
TEM as discussed below, and the power law scattering of fractals is present. For the higher 
sol-gel rates in the HEA series, there is a much smaller size scale present (larger V/S ratio) and 
broader interfacial areas are present. The gelation occurs first as in any sol-gel reaction in a 
good solvent with the organic phase then forming from the polymerization of the "solvent". 
In samples with covalent bonds between the organic and inorganic components, the low limiting 
power law suggests less complete phase separation, and in the case of B-7, a "copolymerization" 
giving a single phase seems evident. 

The morphology of the organic/inorganic IPNs was also studied by TEM. For HEA-1, 
gross phase separation has occurred at an early stage of the polymerization and the sample is 
very non-uniform as is seen in figure 2. For HEA-2, a dendritic structure with fairly diffuse 
edges is seen with a size range of Si0 2 rich areas from 200-2000 A as is seen in figure 3. The 
TEM of HEA-3 has a similar structure to HEA-2, but the edges of the domains of Si0 2 which 
have formed appear rounder. A much different morphology of very finely dispersed Si0 2 was 
observed for both HEA-4 and HEA-5. A TEM micrograph for a typical region of HEA-4 is 
shown in figure 4. The grafted B-series was less distinctive in character by TEM. B-2 has a 
mottled appearance of ~ 0.1pm scale superimposed on a finer texture. The texture of B-7 is 
very uniform over a large area, and is only slightly greater than that observed for a pure organic 
polymer containing no Si0 2 as is seen in figure 5. 


ms mm 

Wm i 

‘ r 

l , > ^ »* ■ i "'It' ^ 

*'/*i, r ,'i .-;■ 


i'V “**f , 1 ^ ? 1 



sk§1 ■ , 


: ; Sfe 

Figure 4. TEM of HEA-4. Figure 5. TEM of B-7. 


A wide variety of morphologies have been prepared from the same starting materials by 
controlling the relative rates of polymerization and the amount of grafting between the 
components for the acrylate/Si0 2 IPNs described. These results are consistent with a similar 
series of organic-inorganic IPNs synthesized from an epoxy organic phase [20]. 

SANS and SAXS results for the two-HEA samples with the lowest sol-gel catalyst show 
a power law scattering over a wide angular range which is consistent with the dendritic 
structures seen in TEM. Porod analysis of the samples with the lower HF catalyst content 


suggest sharp boundaries between the phases. The two samples with the highest sol-gel catalyst 
concentration have broader interfacial areas. The volume to surface ratio shows large sizes for 
the three samples with the low catalyst concentration and small sizes for samples with high 
catalyst concentration. These indicate two morphological types, one in which there are large 
dendritic structures with sharp interfaces, and another in which there are small structures with 
large interfacial areas. This is qualitatively in agreement with the TEM results. The B-series 
of samples contain monomers that have covalent bonds between the sol-gel and vinyl 
components. The phase separation in these samples is considerably weaker, as seen by both 
scattering and TEM. 


1. B. M. Novak, Adv. Mat., 5, 422 (1993), and references within. 

2. J. E. Mark, Y.-P. Ning, C.-Y. Jiang, W. C. Roth, Polymer, 26, 2069 (1985). 

3. C. J. T. Landry, B. K. Coltrain, M. R. Landry, J. J. Fitzgerald, V. K. Long, J. Polym. Sci., 
26B, 3702 (1993). 

4. H. H. Huang, B. Orler, G. L. Wilkes, Macromolecules, 20, 1322 (1987). 

5. B. M. Novak, C. Davies, Macromolecules, 24, 5481 (1991). 

6. M. W. Ellsworth, B. M. Novak, Chem. Mat., 5, 839 (1993). 

7. M. W. Ellsworth, B. M. Novak, JACS, 113, 2756 (1991). 

8. L. H. Sperling, Intreoenetrating Polymer Networks, edited by D. Klempner, L. H. Sperling, 
and L. A. Utracki, (Advances in Chemistry Series 239, (1994), p. 3. 

9. B. J. Bauer, R. M. Briber, C. C.Han, Macromolecules, 22, 940 (1989). 

10. B. J. Bauer, R. M. Briber and B. Dickens, Interpenetrating Polymer Networks , edited by 
D. Klempner, L. H. Sperling, and L. A. Utracki, (Advances in Chemistry Series 239, 1994), p. 

11. B. J. Bauer, C. L. Jackson, A. I. Nakatani, and J. D. Barnes, submitted for publication in 
Chemistry of Materials. 

12. C. L. Jackson, B. J. Bauer, J. D. Barnes, PMSE Preprints, 71, 85 (1994). 

13. Certain commercial materials and equipment are identified in this paper in order to specify 
adequately the experimental procedure. In no case does such identification imply 
recommendation by the National Institute of Standards and Technology nor does it imply that 
the material or equipment identified is necessarily the best available for this purpose. 

14. J. D. Barnes and F. Mopsik, 46th Annual Technical Conference Proceeding, Society of 
Plastics Engineers, 12, 1179 (1988). 

15. H. J. Prask, M. Rowe, J. J. Rush, I. G. Schroeder, J. Res. Natl. Inst. Stand. Tech., 98, 1 

16. G. Porod, Kolloid Z., 124, 83 (1951). 

17. G. Porod, Kolloid Z., 125, 51 (1952). 

18. Roe, R.-J., Encyclopedia of Polymer Science and Engineering. (Wiley Interscience, New 
York, 1988) v. 17, p. 981-9. 

19. C.C. Han, B. J. Bauer, J. C. Clark, Y. Muroga, Y. Matsushita, M. Okada, Q. Tran-Cong, 
T. Chang, I. C. Sanchez, Polymer, 29, 2002 (1988). 

20. B. J. Bauer, D.-W. Liu, C. L. Jackson, and J. D. Barnes, submitted for publication in 
Polymers for Advanced Technologies. 



Colorado School of Mines, Golden, CO 80401, USA 


Improving the processing and formability of ceramic components prior to firing (as green 
bodies) requires an enhanced understanding of how the polymeric binder components function. 
We report on the role of surface energetics on the structure of the copolymeric monolayers 
formed via adsorption from solution. Also, results on the effects of surface energetics on the 
kinetics of the adsorption are reported. A silicon wafer with an oxide layer is used as the surface 
and adsorption takes place from toluene. Surface energetics are varied by treatment of the oxide 
surface with a series of silane coupling agents which contain either amine, epoxide, or vinyl 
functional groups. The block copolymers used consist of relatively short polyethylene oxide) 
(PEO) blocks and much longer polystyrene (PS) blocks. Ellipsometry is used to determine the 
grafting density, a (chains/nm 2 ), and Fourier Transform Infrared spectroscopy is used to 
investigate the copolymer on the surface. It is seen that the time required to reach equilibrium 
increases as the strength of the interaction between the copolymer and the surface increases. 
Also, the diblock copolymers appear to obey the scaling laws proposed by Marques and Joanny 
on all the surfaces studied. ( i.e., o«l/ N A> when the copolymer is symmetric or moderately 

symmetric and aocl/p 2 , when the copolymers are asymmetric, where Na is the number of 
segments of the adsorbing block and P is the ratio of the size of the nonadsorbing block to that 
of the adsorbing block.) 


Polymer coils terminally attached to a solid surface constitute an interface of particular 
interest For example, they are used to enhance the biocompatibility of artificial implants, in 
affinity chromatography and in electrode modification applications. They are also used in the 
stabilization and flocculation of colloidal particles such as those used as precursors ( “green 
bodies”) to the sintering of ceramics. 1 It is important to understand the polymer-inorganic 
interface problem to improve the processibility and formability of the “green body.” 

Many studies have addressed diblock copolymer behavior at the solid-liquid interface. 2,3 
Scaling laws have been proposed for three different regimes of symmetry by Marques and 
Joanny. 4 However, as pointed out by Guzonas et al, the predicted scaling behavior in two of the 
three regimes is similar and hence they suggested a slightly different classification based on the 
value of p; Copolymers with p > N J 5 are identified as highly asymmetric (Tail regime), those 
with 1 < P < N®' 5 as moderately asymmetric and those with p < 1 as symmetric (Head 
regime). 5 The crossover at N®‘ 5 is indicative of crossover from the head regime ( in which the 
size of the adsorbing block dictates the surface coverage) to the tail regime ( in which the 

* Author to whom correspondence should be addressed 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

excluded volume interactions of the nonadsorbing blocks dictates the surface coverage). Our 
previous studies have focused on the adsorption behavior of block copolymers on to a native 
oxide layer of the silicon wafer. 6,7,8 Silane coupling agents are extensively used to enhance the 
adhesion of organic polymers to inorganic surfaces. Hence, an understanding of the adsorption 
process of copolymers onto oxide surfaces modified by silane coupling agents should aide 
material processing. 

In this study, we report on the adsorption behavior of diblock copolymers onto 
chemically modified surfaces. The block copolymers used consist of relatively short 
poly(ethylene oxide) (PEO) blocks and much longer polystyrene (PS) blocks. On polar surfaces, 
the PEO preferentially adsorbs to the surface, whereas the PS block remains dangling in 
solution. Surface energetics are varied by treating the surface with a series of silane coupling 
agents. Adsorption takes place from toluene, a good solvent, onto different surfaces. Fourier 
Transform Infrared spectroscopy (FTIR) is used to investigate the nature of the chemical 
structure on the surface. Preliminary work in developing a methodology to obtain quantitative 
information from the FTIR spectroscopy is also presented. 


The copolymers used in this study were obtained from Polymer Laboratories (Amherst, 
MA) and have the characteristics shown in Table 1. For these samples, both regimes of 
symmetry are covered (i.e. copolymers with p<N^ lie in the so called symmetric to 
moderately symmetric regime and copolymers with p>N^ lie in the highly asymmetric 
regime). All of the polymer samples have polydispersity indices (M w /M n ) less than 1.15. 











PS( 1.7k)-PEO(0.3 k) 




PS( 145k)-PEO(2k) 




PS(3 59k)-PEO(2k) 







P S (64k)-PEO(0.5 k) 






Table 1: Material characteristics of the copolymers studied. 

The silicon wafers were obtained from Silicon Source Inc. (Phoenix) and treated as 
described previously. 2 The film thickness of the oxide layer was independently measured using 
ellipsometry prior to further treatment. The silane coupling agents used were obtained from 
Dow Coming Co. (Midland, MI). The three coupling agents employed in order to give different 
surface energetics were Z-6020® (Diamino functionality), Z-6040® (Epoxide functionality) and 
Z-6075® (Vinyl functionality). The method of application of the silane coupling agent to the 
wafer has been described by Plueddemann et al. 9 After application, the thickness of the silane 
film was independently measured using ellipsometry. The average thickness of the epoxide 


films was determined to be 14.0 ± 3.0 A, the amine films to be 22.0 ± 3.0 A and the vinyl films 
to be 13 ±3.0 A. HPLC grade toluene (Malinckrodt Inc.) was used after filtering three times 
through 0.2 micron Whatman filters. The adsorption experiments were all conducted at 
solution concentrations of 1.00 ± 0.01 mg/ml and the temperature was controlled at 25.0 ± 
0.1°C. Low angle laser light scattering (Wyatt Technologies DAWN B) was used to ensure 
that there was no formation of micelles at the concentrations used in the experiments. A 
specially modified rotating analyzer ellipsometer (Gaertner Scientific ) was used to measure the 
adsorbed amounts. 10 A Bio-Rad FTS-40 FTIR spectrometer with a narrow-band-pass mercury 
cadmium telluride (MCT) detector was used to obtain the spectra of the films. The spectra were 
collected in transmission using at least 512 scans at a resolution of 2 cm' 1 . The native wafer is 
subtracted out to obtain the spectra of the film. 

The relationship of the measured ellipsometric angles (\|/ and A) to the film parameters of 
a multilayer stack is given by: 


e“tan V = -*- = F(n k ,d k ) (1) 

where and represent the indices of refraction and the thicknesses of each layer present 
(denoted by subscript k). Measurement of the two independent quantities, q/ and A, allows for 
the solution of two unknowns, a layer thickness, di, and a refractive index, ni. In this study the 
adsorbed layer thicknesses and refractive indices are determined assuming a homogeneous layer 
for the thin films. The adsorbed amount of copolymer proves insensitive to the model assumed. 2 
The adsorbed amount is calculated from: 

A = d i c . = d i( n i- n o)/( d "/ dc ) # (2) 

where ii] represents the refractive index of the adsorbed layer and di its thickness, n 0 represents 
the refractive index of the copolymer solution and (dn/dc)^ is the specific refractive index. 

Knowing the adsorbed amount, the grafting density, ct, is calculated from Equation 3. 

ct (chains / nm 2 ) = - ^” 8/ in \ N (chains / mol) x 10 -18 (m 2 / nm 2 ) (3) 

M w (mg/mol) AV 

Another important variable in the scaling description of copolymer adsorption is the asymmetiy 
ratio, p, given by: 

P = a'(N B /N A ) 0 ' 6 (4) 

where a' is the ratio of the size of a single nonadsorbing block segment to a single adsorbing 
block segment and is taken as 0.87 for PS-PEO block copolymer. 

In FTIR spectroscopy, the integrated area of the C-H antisymmetric stretch (2920 cm' 1 ) 
can be used to characterize the film on the surface. The calibration procedure involves the 
gravimmetric determination of the mass of the polystyrene on spun-coated wafers. In FTIR 
transmission of thin films on a substrate, a sine wave interference pattern is generated (see 
Figure 4 below). The thickness of the cast film can be determined by the relationship: 

d = (l/2nAv) (5) 


where d is the thickness of the film, n, its refractive index and Av is the spacing between the 
maxima of the interference fringes in wavenumbers. 


Results from the study are seen in Figure 1, which shows that the grafting density of the 
symmetric to moderately symmetric diblocks scale with the reciprocal of the head size 
(ctocI/Na). This appears to be true for the different surfaces studied, (i.e. the scaling behavior 
appears to be similar for the amine, epoxide and vinyl surfaces.) Note that as N A becomes 
larger, the adsorbed amounts converge to similar values. 

For highly asymmetric diblocks, the surface density scales with the reciprocal of the 
square of the asymmetry ratio (a oc l/ft 2 ) as shown in Figure 2. This is again true for all the 
different surfaces studied. Thus, the scaling laws proposed by Marques and Joanny for diblocks 
in the symmetric to moderately symmetric regime as well as in the highly asymmetric regime are 
apparently obeyed on all the surfaces. It is noted that as l/p 2 becomes smaller that the 
adsorbed amounts converge to similar values. This is to be expected because as p becomes 
large, it is the excluded volume interactions between the nonadsorbed blocks of the copolymer 
molecules which govern the surface coverage. 

Figure 3 demonstrates the kinetic behavior of one of the diblocks (PS(145k)-PEO(2k)) 
on the different surfaces. It is seen that as the strength of surface interaction increases, as 
measured by a greater adsorbed amount, that equilibration takes longer. That is, in moving from 
strong bonding to weak bonding, the system equilibrates much faster. 

Figure 1: Scaling behavior for symmetric to moderately symmetric diblocks 
(P < N®* 5 ) on Native (•), Epoxide (A) and Amine (♦) 


Figure 2: Scaling behavior for highly asymmetric polymers (p > N^ 5 ) on Native 
(A), Epoxide (•) and Amine (♦) surfaces 

Figure 3: Adsorption kinetics for diblock PS(145k)-PEO(2k) on native (A), 
epoxide (O) amine (♦) and vinyl (□) surfaces. 


A typical FTIR transmission spectrum for a thick polystyrene film on a silicon wafer is 
shown in Figure 4. From the interference fringes seen in the thick films, we can directly 
determine the thickness of the PS film using Equation 5. However for thin films, it is difficult to 
obtain measurable interference. In this case, the film thickness may be measured indirectly. A 
calibration curve of the film thickness determined using Equation 5 against the corresponding 
integrated areas of the C-H stretch is shown in the inset of Figure 4 . The relationship is linear 
and hence can be extrapolated to thin films if performed with high enough precision. For one of 
the diblocks (PS(145k)-PEO(2k)), the magnified spectrum in the C-H stretch region is shown in 
Figure 5. For the above diblock, the thickness of the diblock film is determined by FTIR to be 
148.0 ±3.0 A. This contrasts with the thickness for the copolymer film determined directly 
using ellipsometry, which is 100.0 ± 5.0 A. Also, a curve of the integrated area, in absorbance 
units, of the C-H stretch against the mass of PS on the surface can be generated. The mass of 
PS on the wafer can be determined gravimetrically by the difference in dry wafer weights before 
and after the spin coating process. A calibration of this nature would permit the determination of 
the amount of polymer adsorbed onto the surface directly. 


The present experimental study lends itself to some interesting conclusions regarding the 
use of coupling agents at the polymer-inorganic interface. First, in terms of maximizing the 
surface coverage of the block copolymers it is apparent that changes in copolymer architecture 
are more important than the acheivable surface energetics modifications. It is to be noted, 
however, that the twin-handle of varying copolymer architecture and surface energetics allows a 
rather detailed control of copolymer grafting density. Coupling agents can therefore be used to 
modify chain grafting densities without changing copolymer architecture. 

The diblock materials are found to obey the scaling laws proposed by Marques and 
Joanny for each of the different surfaces studied (i.e. native oxide surface, the epoxide surface 
and the diamino surface). In the symmetric to moderately symmetric regime the grafting density 
scales with the reciprocal of the head block size (ct oc 1/N^) and in the highly asymmetric regime, 

the grafting density scales with the reciprocal of the square of the asymmetry ratio (a ocl/p 2 ). 
However as l/(3 2 decreases, the adsorbed amounts approach the same value. This is on account 
of the fact that as the copolymer becomes highly asymmetric, the surface coverage is dictated 
only by the excluded volume interactions of the nonadsorbing blocks and hence surface 
chemistry is less important. A similar argument may be made at the extreme of the tail regime in 
which we have moderately symmetric or symmetric copolymers with large adsorbing block sizes. 
For these copolymers, it is the size of the adsorbing block which dictates the surface coverage. 
Successful usage of coupling agents to modify the properties of the diblock copolymer - 
inorganic interface is thus predicated on the appropriate selection of chain architecture. 

Figure 3 illustrates that the strength of surface interactions plays an important role in the 
kinetics of the adsorption process. As the strength of surface interaction decreases, the 
equilibration time for the adsorption process also decreases. Highly interacting systems, such as 
the native oxide surface are known to possess long-lived non-equilibrium states. 12. This aides 
in understanding the time required to synthesize the polymer-ceramic interface which is an 
important component in the pre-processing stages of ceramic manufacture. 


Figure 4: Typical FTIR spectrum for a thick polystyrene film and calibration 
curve for integrated area of C-H stretch and thickness of PS film 

Figure 5 : FTIR spectrum of the diblock PS(145k)~PEO(2k) in the region of the 
C-H stretch. 


It is also demonstrated through Figure 4 that it is possible to use FTIR spectroscopy to 
determine the thickness as well as equilibrium adsorbed amount of polymer on the surface. For a 
thin polymer film, the integrated area of the C-H stretch correlates with the thickness of the 
polymer film on the surface. 

Hence, we have elucidated the effect of specific surface interactions on the structure and 
dynamics of the formation process of the polymer-ceramic interface. This is important since it 
involves the formation of the “green body” and understanding the structure and kinetics of this 
process would help in optimizing material charaterictics. 


We are grateful to Professor Thomas Furtak of the Physics Department at the Colorado 
School of Mines for discussions regarding the application of FTIR spectroscopy to monolayer 

This work is supported by the Petroleum Research Foundation of the ACS under grant 
No. 27654-G7 and by the NSF under grant CTS-9410081. 


(1) Bohlein-Mauss, J.; Sigmund, W.; Wegner, G. et al. Adv. Mater., 4, 973, 1992 

(2) Motschmann, H.; Stamm, M.; Toprackcioglu, C. Macromolecules , 24,3681, 1991 

(3) Halperin, A.; Tirrell, M.; Lodge, T.P. Adv. Poly. Sci., 100, 31, 1992 

(4) Marques, C.M.; Joanny, J.F. Macromolecules , 22, 1454, 1989 

(5) Guzonas, D.A.; Boils, D.; Tripp, C.P.; Hair, M L. Macromolecules, 25, 2434, 1992 

(6) Pai-Panandiker, R. S.; Dorgan, J.R. Macromolecules , In Press (June 1995) 

(7) Dorgan, J.R.; Stamm, M.; Toprackcioglu, C. et al Macromolecules, 26, 5321,1992 

(8) Stamm, M.; Dorgan, J.R. Coll, and Surf., 86, 143, 1994 

(9) Plueddemann, E.P. 27th Annual SPI Technical Conference, Paper 21 -B (1972) 

(10) Pai-Panandiker, R.S.; Dorgan, J.R. Rev. Sci. Instr., 66, 1112, 1995 

(11) Culler, S.R.; Mckenzie, M.T.; Fina, L.J.; Ishida, H.; Koenig, J.L Appl Spec, 38, 791, 

(12) Chakraborty, A.K.; Adriani, P.M. Macromolecules, 25, 2470, 1992 

(13) Chakraborty, A.K.; Adriani, P.M. J. Chem. Phys., 98, 4263, 1993 


Part VII 

Surface Preparation 


Departments of Chemistry and Chemical Engineering, Mississippi State University, 
Mississippi State, MS 39762 


Ex-PAN carbon fiber surfaces, oxidized to varying degrees by HN0 3 , have been 
characterized by NaOH, dye and HC1 uptake, ion scattering spectroscopy (ISS), and angle- 
resolved X-ray photoelectron spectroscopy (ARXPS). Subsequent treatments with 
tetraethylenepentamine to introduce amino groups or epichlorohydrin to introduce epoxy groups 
have been thoroughly characterized. The efficiency of using these surface functions to bond 
polymers onto fiber surfaces has been investigated using model anhydrides and isocyanates (for 
amines) and diols and diamines (for epoxides). The fraction of surface-bound groups which 
react drops with an increase in molecular size of the species being grafted. Crosslinked 
elastomeric polyurethane layers with designed modulus values and thicknesses have been 
bonded to these fibers. Composites have been prepared (epoxy matrices). The impact strengths 
and interlaminar shear strengths (ILSS) were studied as a function of the interphase modulus 
and thickness. Impact strengths increased (even at 500-1500 A thicknesses). ILSS values 
depend on interphase modulus. 


Poor bonding at the fiber/matrix interface degrades the properties of composite materials. 
Improvements in this bonding increases the interlaminar shear strength (ILSS) but, most often, 
decreases the impact strength. 1 * 3 Impact strength and toughness can be increased by adding a 
second dispersed rubbery phase to the matrix. 4 ’ 5 Several theoretical studies proposed inserting 
a thin elastomeric layer at the fiber/matrix interface (a concentric interphase layer coated on the 
fiber) to improve fracture toughness and impact strength. 6 * 10 

Carbon fibers have been coated with thermoplastic polymers to serve as interphases 
using dip coating, 11 ’ 12 interfacial polycondensation 1 * and electropolymerization. 14 * 17 Block 
copolymers containing polyisoprene and styrene/maleic anhydride segments were coated on 
carbon fibers by filament winding. 18 Thin polyimide 19 and polyamide 20 coated carbon fibers 
have been used in epoxy matrix composites. However, the strength of the fiber/interphase bond 
or the interphase matrix bond has not been specifically addressed or chemically designed in 
these studies. Strong interface/fiber adhesion is required to achieve high ILSS. 

Unlike previous studies using polymeric interphases, we have reported 21 ' 23 the first 
studies where carbon fibers have been specifically functionalized with -OH, -COOH, and amino 
groups to react with interphases and the quantity of these functions were determined. Then 
crosslinked elastomeric polyurethane interphase layers of defined thicknesses and elastic moduli 
were bonded to these carbon fiber surfaces followed by construction of carbon fiber epoxy 
matrix composites containing these elastomeric interphase layers. 21 * 23 These systems exhibited 


Mat. Res. Soc. Symp. Proc. Vol. 385 *1995 Materials Research Society 

higher impact strengths than identical composites without elastomeric interphases and, in some 
cases, no loss of ILSS occurred. 21 ’ 22 The defined functionalization of carbon fibers with -OH, 
-COOH, amino, and epoxy 24 groups will be summarized here together with their reaction 
efficiency at the surface with model compounds. 24,25 Also, the effect of defined elastomeric 
interphase layers on impact strength (IS) and ILSS will be discussed. 


Oxidation and Binding Amino Groups. Ex-PAN based carbon fibers (Celion G30-500, 
from BASF, Inc. and Thomel T300 from Amoco, Inc.) were oxidized in 70% wt. nitric acid at 
115 °C for varying times. These oxidations greatly increased the surface quantity of acidic 
phenolic hydroxyl and carboxylic acid functions. Acidic surface functions were quantitatively 
determined by NaOH uptake experiments. HC1 uptake was used to follow amount of basic 
functions on the fiber (e.g., pyridine type nitrogen, etc.). Next these HN0 3 -oxidized fibers were 
treated at 190-200 °C with tetraethylenepentamine (TEPA) to bind amino groups to the surface. 
TEPA forms amide functions with the carboxyl groups. Loop structures can also form (see 
Scheme 1). The quantity of acid groups (see Table I) increased from 3.6 /*eq/g (as-received) 
to 49 jaeq/g (after 90 min of oxidation). The quantity of surface basic groups was below 0.5 
«eq/g until after TEPA treatment. The amino group surface concentrations for TEPA-treated 
fibers were much higher, ranging from 4 fj& q/g for as-received fibers to 55 /ieq/g for fibers 
which had been oxidized 90 min. 

Phenolic hydroxyls can't react with TEPA and they remain on the surface. Some or the 
carboxylic acid groups are likely to be unreacted. In related studies oxidized fiber surfaces 
were treated with SOCl 2 (forming very reactive -COC1 groups) and then with TEPA. All 
carboxyl groups were converted to amides. The remaining acidic functions were phenolic 
hydroxyls. They constituted 35-40% of the acid groups introduced on HN0 3 oxidation (-COOH 
groups 65-60%). Combining this information with (1) the decrease in acid groups (Table I) and 
(2) the increase in amino groups during the reaction with TEPA gave a ratio (amine added: 
COOH consumed) - 2.6. Thus, about 37% of the grafted TEPAs are looped. 

Table I. Surface Functions Introduced on Carbon Fibers Upon HN0 3 
Oxidation (70% wt, 115 °C) and TEPA Treatment (190-200 °Q 

HN0 3 Oxidation Time (min) 






Initial Acid Groups (ueq/g) 






Initial Basic Groups (ueq/g) 






Acid Groups after TEPA 



Treatment («eq/g) 




Basic Groups after TEPA 



T reatment(oeq/g) 




“Since it is increasingly difficult to add successive protons to grafted TEPA amino sites 
(due to the build up of charge-charge repulsion) correction factors were carefully derived 
from experiments with the reaction products of TEPA and phenyl isocyanate. The 
detailed reasoning and experiments are presented elsewhere. Correction factors of 
1113 and 1 212 are required for adding a third and fourth proton, respectively, during 
titration to a -CONH-CH 2 CH 2 NHCH 2 CH 2 NHCH 2 CH 2 NH-CH 2 CH 2 NH 2 fragment. 


Scheme 1 





HNQ 3 

115 °C 

10 , 







190-200 °C 
-H 2 0 






-|-c—OCH 2 CH—ch 2 
—OCH z CH — CH 2 — Cl 
-C — OCH 2 CHCH 2 CI 
o OH 




Oi II 

— C—OCH-jCH—CH 2 


-OCH 2 CH — CH 2 

—C — OCH 2 CH — ch 2 

T11 \/ 

Binding Epoxide Groups. Epoxide functions were successfully bound onto HN0 3 - 
oxidized fibers by a two stage process (Scheme 1). First, epichlorohydrin was reacted in the 
presence of benzyltrimethylammonium chloride to form surface bound a-chloropylene glycol. 
Then NaOH-promoted ring closure in epichlorohydrin under anhydrous conditions formed the 
corresponding surface-bound glycidyl esters and ethers over a several day period. NaOH was 
coated on fibers from step 1 and then the fibers were refluxed in epichlorohydrin. More than 
50% of the acidic functions (on fibers starting with 60 ^eq/g of acid groups) were converted 
to glycidyl groups without any polymer formed. A maximum of about 34 yueq/g of epoxy 
functions were introduced. 25 As the surface acid group concentration increased the efficiency 
(% of acid groups grafted) of epoxide grafting decreased. The first reliable noninterfering 
method to quantitatively analyze epoxy groups on carbon fiber surfaces was developed using 
HBr ring-opening and a bromide selective electrode. 25 Until now carbon fiber surface 
epoxidation had been unsuccessful. The previous literature as well as several unsuccessful 
routes in our lab have been discussed. 25 

Reaction of Oxidized Carbon Fibers with Model Isocyanates. Thornel T300 carbon 
fibers were oxidized in 70% HN0 3 for 60 min at 115 °C. The resulting fibers, having 61.5 
fie, q/g of acidic groups, were reacted for 3 h at 80 °C in xylene, chlorobenzene and DMF with: 
1,6-diisocyanatohexane (mwt. 168), a tetraethyleneglycol oligomer capped with TDI, mwt. 920 
(PET-95D) and with a prepolymer tetraethyleneglycol oligomer terminated with isophorone 
diisocyanate (XAPC-722, mwt. 1444). The terminal isocyanate groups were capped as 
urethanes by treatment with ethanol. The grafting efficiencies (G e , % of acidic surface 
functions which reacted with the isocyanates) decreased sharply as the size of the diisocyanate 
increased. G e values fit: G e = KM a where M = mol. wt., K = 2750, a = -0.812. Furthermore, 
the grafting efficiency was highest in DMF. Apparently, the larger molecules shield other 
surface functions. Isocyanates form urethanes when reacted with phenolic hydroxyls and 
amides (via loss of CO 2 ) when reacted with carboxyl groups. 



Mwt. 168 

G e 


Mwt. 920 

G e 


Mwt. 1444 

G e 

= 18% (xylene); 37% (chlorobenzene); 40% (DMF) 

= 9.6% (DMF) 

= 3.7% (xylene); 6.3% (chlorobenzene); 7.5% (DMF) 


-C0 2 
2. EtOH 

O O 

|C ^|-c—N-^H^NCOE t 
--OH 6 



0 O 

Celion G30-500 fibers were reacted with neat phenylisocyanate at 90 °C after oxidation 
for various times in HN0 3 . The grafting efficiencies were about 30% as the surface acid group 
concentration on the fibers was varied from 12.5 to 47.1 ^oeq/g (see Table II). 

Table II. Consumption of Acidic Surface Functions on Oxidized 
Celion G30-500 Fibers by Phenylisocyanate at 90 °C 

HN0 3 Oxidation Time (min) 





Initial Acidic Groups 




Acidic Groups After PhNCO 




Grafting Efficiency G e (%) 




Grafting Acetic Anhydride and Phenylisocyanate to Surface-Bound Amines. Celion 
G30-500 fibers, grafted with TEPA, had surface amino concentrations from 4.7-45.5 /*eq/g. 
They were reacted in neat acetic anhydride (75 °C, 3 h) and neat phenyl isocyanate (90 °C, 3 
h) followed by washing and soxhlet extraction in acetone (3 h). The grafting efficiencies 
ranged from 50-64% for phenylisocyanate and 49-69% for acetic anhydride. 

Starting Amine Concentration (jutq/g) 4.7 18.7 31.3 45.5 

G e (Phenyl Isocyanate) (%) 51 64 50 54 

G e (Acetic Anhydride) (%) 49 69 51 63 

Reactions of Surface-Bound Amino Groups with Epqxv Systems of Varying Molecular 
Weights. TEPA-treated Celion G30-500 fibers were reacted at 80 °C for 19 h with four model 
epoxides in xylene with phenol as the catalyst: (a) styrene oxide (mwt. 120), (b) EPON 828 
(mwt. 378), (c) EPON 834 (mwt. 500) and EPON 1001 (mwt. 1000). The grafting efficiencies 
of the epoxy models dropped sharply as molecular weight (size) increased (Table III). The 
values of G e for epoxy grafting fit the equation: G e = K x M a , where K and a were constants 
and M = mol. wt. (e.g., plots of log G e versus log M were linear) K = 25700, a = -1.256. 

ARXPS/ISS Surface Characterization of Nitric-Acid Treated Carbon Fibers. ARXPS 
show the surfaces of as-received BASF (Celion G30-500) carbon fibers are appreciably 
oxidized. The primary carbon/oxygen functionality is C-OH (63% of total oxidized carbon 
species). Nitric-acid oxidation increases the O/C concentration ratio from 24% (as-received) to 
33% (90-min exposure to HN0 3 at 115 °C). The ratio of (carbonyl+carboxyl) groups to 


hydroxyl groups increases from 0.52 to 1.3 over this range of HN0 3 oxidations. ISS reveals 
significant sodium and calcium concentrations on the as-received fibers. After exposure to 
HN0 3 (60 min, 115 °C), the outermost atomic layers are comprised only of carbon, oxygen, 
and nitrogen. 

Table III. Grafting Epoxy Model Systems to TEPA-Treated Carbon Fibers. 
Surface Grafting Efficiencies Vs. Molecular Weight 

Model Reactant 

Mol. Wt. 

Initial Amine 
Functions (weq/g) 

Amt. Grafted 

G e (%) 

Styrene Oxide 





EPON 828 





EPON 834 





EPON 1001 





Composites Containing Elastomeric Polyurethane (PIT) Interphases. Continuous fiber 
EPON 830 matrix composites (14 plys) were made from carbon fibers coated with cured 
polyurethane interphase layers from 500 to 2500 A thick. The Vf=56%. The ILSSs were 
measured by three-point bending (ASTM D2344-67) and Izod impact strengths (IS) by ASTM 
D256-72. Three PU resins were selected with varying stiffness: PU 504/317 (E=138MPa, 
elongation 29%), PU 504 (E = 7.9 MPa, elongation 290%) and PU 722 (E = 0.5 MPa, 
elongation 900%). When the interphases were constructed on TEPA-treated fibers the 
impact strengths of the fibers increased markedly relative to identical composites without 
these interphases. The IS values increased by 1.5 times (PU 504/317), 1.4 times (PU 504) and 
2.2 times (PU 722). Remarkably, most of the increase in IS was achieved with interphase 
thicknesses of only 1000 to 1500 A! 

The stiffer interphases (PUs 504/317 and 504) gave little or no decrease in the 
composites' ILSS. For 504/317 the ILSS were unchanged versus composites with no interphase 
while those with 504 had ILSS values that were 90-95% of composites with no interphase. 


1. L. T. Drzal, M. J. Rich, M. F. Koenig and P. F. Lloyd, J. Adhesion, 16, 133 (1983). 

2. B. D. Agarwal and L. J. Broutman, Analysis and Performance of Fiber Composites 

(John Wiley & Sons, New York, 1980), p. 33. 

3. R. V. Subramanian and A. S. Crasto, Polym. Composites, 7, 201 (1986). 

4. J. F. Gerard, Polym. Eng. Sci., 28 (9), 568 (1988). 

5. J. N. Syltan and F. J. McGarry, Polym. Eng. Sci., 13, 29 (1979). 

6. V. A. Matonis and M. C. Small, Polym. Eng. Sci., 9 (2), 90 (1969). 

7. L. J. Broutman and B. D. Agarwal, Polym. Eng. Sci., 14, 581 (1974). 


8 . 

S. D. Gardner, C. U. Pittman, Jr. and R. M. Hackett,/. Comp. Mat., 27 (8), 830 (1993). 

9. S. D. Gardner, C. U. Pittman, Jr. and R. M. Hackett, Comp. Sci. Technol. , 46, 307 

10. B. Y. Low, S. D. Gardner, C. U. Pittman, Jr. and R. M. Hackett, Comp. Sci. Technol., 
52, 589 (1994). 

11. J. R. Dauksys, J. Adhesion, 5, 211 (1973). 

12. L. D. Tryson and J. L. Kardos, 36th Ann. Conf. on Reinforced Plastic Comp. Inst. SPI, 
2-E (1981), p. 1. 

13. J. H. Crammer, G. C. Tesoro and D. R. Uhlmann, Ind. Eng. Chem. Proc. Res. Dev., 21, 
185 (1982). 

14. R. V. Subramanian and A. S. Crasto, Polym. Composites, 7, 201 (1986); S. Dajardin, 
et al., J. Mater. Sci., 21, 4342 (1986). 

15. J. P. Bell, J. Chang, H. W. Rhee and R. Joseph, Polym. Composites, 8, 46 (1987); H. 
W. Rhee and J. P. Bell, ibid., 12, 213 (1991). 

16. J. Chang, J. P. Bell and R. Joseph, SAMPE Quarterly, 18 (3), (1987). 

17. A. S. Wimolkiatisak and J. P. Bell, J. Appl. Polym. Sci., 46, 1899 (1992) and Polym. 
Composites, 10, 162 (1989); J. Chang, J. P. Bell and S. Shkolnik, J. Appl. Polym. Sci., 
34, 2015 (1987). 

18. J. F. Gerard, Polym. Eng. Sci., 28 (9), 568 (1988). 

19. M. Kodama, I. Karino and J. Kobayashi, J. Appl. Polym. Sci., 33, 361 (1987). 

20. T. Skourlis, T. Duvis and C. D. Papaspyrides, Composites Sci. Technol, 48,119 (1993). 

21. C. U. Pittman, Jr., G. R. He, B. Wu, L. Wang, S. D. Gardner, Proceedings of the First 
International Conference on Composite Engine ering fICCE/U, D. Hui, Editor, Aug. 28- 
31 (1994), New Orleans, LA, pp. 403-404 and The Fifth International Conference on 
Composite Interfaces (\CCl-V). Molecular Interactions and Structure, Ishida, Ed., June 
20-23 (1994), Chalmers University of Technology, Goteborg, Sweden 

22. G. R. He, PhD Thesis, Mississippi State University, 1994. 

23. Wei Li, MS Thesis, Mississippi State University, 1995. 

24. Z. Wu, MS Thesis, Mississippi State University, 1994. 

25. Z. Wu, C. U. Pittman, Jr. and S. D. Gardner, Carbon, in press (1995); ibid, submitted 



Materials Science and Engineering Department, University of Pittsburgh, Pittsburgh, PA. 
15261 . 


Polymers tethered by one end onto a solid surface are referred to as polymer "brushes". 
We consider brushes composed of copolymers that contain both A and B monomers. The A 
monomers are compatible with the surrounding solvent, while the B sites are solvent- 
incompatible. The solvent incompatibility causes the B sites to associate into domains or clusters 
within the layer. We use Monte Carlo computer simulations and self-consistent field calculations 
to determine the effect of copolymer architecture on the structure of the polymer brush. In 
particular, we alter the copolymer sequence distribution (the arrangement of the A and B 
monomers along the length of the chain) and determine how both the vertical and lateral 
morphology of the brush are effected by these variations. The results provide guidelines for 
controlling the size and shape of the B domains, and consequently, the morphology of the 
tethered layer. 


A polymer "brush" consists of a dense layer of chains that are anchored by one end onto 
a solid surface. The behavior of brushes has come under considerable scrutiny in the last several 
years 1. Interest in these systems stems from the fact that the tethered chains can be used to 
control the properties of the underlying substrate. For example, tethering chains onto a non¬ 
reactive surface can facilitate the adsorption of a subsequent layer or coating. The brush, 
sandwiched between the solid surface and the outer coating, acts as a "primer" or glue between 
the two layers. To take full advantage of these surface-modifying properties, we must enhance 
our understanding of how the composition of the tethered chains affects characteristics of the 

brush. . . , , 

In this paper, we probe the effects of chain composition by examining brushes that 
contain different AB copolymers. The A sites are considered to be compatible with the 
surrounding solvent, while the B sites are solvent-incompatible. The solvent incompatibility 
drives the B sites to undergo a mutual attraction. We will refer to these interacting sites as 
"stickers". (The A sites, on the other hand, are non-interacting.) We first determine the behavior 
of brushes where the B sites, or stickers, are located just at the tops of the chains We then 
investigate brushes where the stickers are distributed along the length of the tethered polymers 
4 . Specifically, in the first study we use Monte Carlo computer simulations to model the 
behavior of self-avoiding, grafted polymers that contain attractive sites on the free ends 2 . As 
we demonstrate below, the mutual attraction between these ends causes the chains to aggregate 
into clusters. However, the interaction not only alters the structure, but also enhances certain 
properties of the layer. In particular, the attraction effectively "pulls" the ends out of the bulk of 
the brush: now, all the ends lie exposed near or at the top of the layer. (In the absence of this 
interaction, the ends lie dispersed and buried within the brush.) Consequently, they can readily 
react with subsequent polymer chains or an additional coating. Thus, adding the attractive end- 
groups provides an effective means of improving the adhesive properties of the layer. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

In the second investigation, we combine Monte Carlo simulations and numerical self- 
consistent field (SCF) lattice calculations to determine the properties of brushes in which the 
stickers (and non-interacting sites) are distributed along the length of the each chain 4- Here, we 
alter the sequence distribution in these AB copolymers to determine how both the vertical and 
lateral morphology of the brush are affected by such variations. The results provide guidelines 
for controlling the size and shape of the B domains, and consequently, the morphology of the 
tethered layer. 


In our Monte Carlo simulations 2, we consider systems of monodisperse, grafted 
copolymers. The chains are modeled as self-avoiding random walks on a three dimensional, 
cubic lattice. Each chain is anchored at one end to the Z = 1 plane, which represents the surface 
of a solid substrate. We use the bond fluctuation model to simulate the motion of the 
polymers. We introduced an attractive interaction, A, between the sticker sites. This parameter 
exerts an effect through the Metropolis algorithm movements that reduce the energy of the 
system are accepted with unit probability and movements that increase the energy of the system 
are weighted by a Boltzmann factor. 

The size of the lattice in our simulations was 64 x 64 x 156 sites, the largest dimension 
being along the Z direction. Periodic boundary conditions were imposed in the X and Y 
directions, while the Z direction was bounded by impenetrable hard walls at Z = 1 (the grafting 

surface) and Z = 156. The system was equilibrated by running the simulation for at least 2xl0 5 
Monte Carlo steps, where each Monte Carlo step involved N |^ n x (N beads )^ attempted 
moves. (The parameter refers to the number of chains in the system and refers 

to the number of sites or beads on each chain.) The chain length was 36 segments for all the 
simulations. In the first part of the study, the stickers are located on the free ends of the chains 
and the number of these terminal stickers is varied. The results were averaged over three to five 
independent runs for each system. 

In the second part of the study, we use these Monte Carlo simulations and self- 
consistent field theory to determine the effect of copolymer sequence distribution. Our self- 
consistent field method is based on that developed by Scheutjens and Fleer 8, which in turn is 
based on a Flory-Huggins approach, combining Markov chain statistics with a mean-field 
approximation. The equations in this model are solved numerically and self-consistently. The 
self-consistent potential is a function of the polymer segment density distribution and the Flory- 
Huggins interaction parameters. The effect of randomness on the properties of copolymers was 
first incorporated into the SCF model by van Lent and Scheutjens 9 Here, each monomer has a 
finite probability of being either an A or B site. Thus, their model introduces an annealed 
randomness, where the ensemble of sequence distributions in the system is allowed to fluctuate. 
Consequently, it is difficult to isolate the effect of systematically varying the sequence 
distribution through this model. 

We adopt a different approach to specifying the randomness of the chains 4. We 
consider each polymer as a separate molecule and fix its sequence distribution at the start of the 
calculation. This models a quenched or frozen randomness, where each chain retains its 
sequence distribution throughout the calculation. For all the SCF results presented in this paper, 
each system involved specifying the sequence distribution of 2500 individual chains and all the 
results were averaged over five different realizations of these systems. In addition, the Flory- 

Huggins % parameter characterizing the A-B interaction was fixed at XAB = TO. The B-solvent 
interaction was described by setting %BS = 1-0, while the A-solvent interaction was given by 


XAS = 0-0- The calculation was performed for a chain length of 100 units, keeping the grafting 
density fixed at 15%. 

In both the Monte Carlo and SCF calculations, the sequence distribution of each chain is 
set as follows. We let the variables Pa and Pb denote the fraction of A and B monomers in the 
molecule, respectively. Given a site A on the chain, the conditional probability that the next site 
is a B monomer is given by the parameter Pa->B- We define an order parameter, f, through the 
following equation 10: 

PAPA-*B = (PAPB) 1/2 f (1) 

where f lies between 0 and 1. For all these calculations, we chose PA = PB = 0.5. In this case, 
f -* 0 corresponds to a diblock copolymer, f = 0.5 represents a purely random chain and f = 1 
corresponds to an alternating chain. 


Effect of the number of terminal stickers 

The structure of the grafted layer depends on the interplay of a number of variables, 
such as the number of stickers, the grafting density, the chain length and the attractive energy 
between the stickers. In this paper, we probe the effect of the first two parameters. (The effects 
of chain length and sticker attraction in the case where the grafted chains contain a single 
terminal sticker is described in reference 3.) We start by determining the effect of the number of 
terminal stickers on the morphology of the layer. In particular, we studied systems that had 3,5 
and 9 functional groups on the free end of each chain. The grafting density (the number of 
chains per unit area of grafting surface) was fixed at 12%, or a total of 122 tethered chains. The 
interaction energy between the stickers was set equal to -1.2 kT. 

We calculated p (z), the density profiles of all the monomers in the brush as a function 
of Z. The density profiles are calculated by counting the total number of monomers present in a 
particular plane in the Z direction. This sum is then normalized by setting £ p (zi) = 1, where i 
designates the specific plane. The density profiles for the various examples are shown in Figure 
I, where we also plotted the curve for the case of no stickers. 

In these plots, the oscillatory behavior near Z = 1 arises from our choice of the bond 
fluctuation algorithm to model the grafted layer and has been observed in other simulations of 
polymer brushes using this technique H. The region 5 < Z < 30 represents the bulk of the 
brush, while the Z > 30 domain represents the top of the layer. The density profiles for the 3,5, 
and 9 sticker polymers show strikingly different characteristics. The 3 sticker system exhibits a 
density profile that is identical to the unperturbed, no sticker case. Relative to the 3 sticker 
example, the density profile for the 5 sticker case shows an enhancement in the polymer density 
near the top of the layer. When there are 9 stickers on the ends of the chains, the density profile 
in the bulk region remains fairly constant, but lies lower than the corresponding values for the 3 
and 5 sticker cases. There is, however, a significant peak near the top of this brush. 

The origin of the relatively high polymer concentrations near the top of the layer in the 5 
and 9 sticker scenarios can be understood by examining Figure 2, which displays the density 
profiles for just the sticker sites. As can be seen, the ends with 5 and 9 stickers are confined to a 
narrow region of Z, which lies near or at the top of the layer. On the other hand, the density 
profile for the ends with 3 stickers is effectively spread over the entire range of Z values. Thus, 


the distinct features in Figure 1 are due to the localization of the end sites at the top of the layer 
in the instances of 5 and 9 sticky ends. 

0 20 40 60 


Fig. 1 Normalized density profiles for all the monomers in the brush. 

0 20 40 60 


Fig.2 Normalized density profiles for just the sticker sites. 

The graphical output from the simulation reveals that the ends are not only localized at 
the top of the layer, but the stickers are also aggregated into distinct clusters In the case of 3 
stickers, there are no signs of clustering. Clustering begins to occur in the case of 5 stickers, 
while the 9 sticker example shows the most pronounced cluster formation. 

Of particular interest is determining how the average height of the layer is affected by the 
presence of the stickers. The average layer height is calculated as the first moment of the density 
profile. The 9 sticker case displays the greatest layer height 2. This phenomenon can be 
understood by referring to the density profiles in Figure 1. Due to the formation of the clusters 
in the 9 sticker example, the polymer density is depleted within the bulk of the brush and 
enhanced near the top of the layer. In effect, the mutual attraction between the stickers pulls the 
ends out of the lower depths and draws them to the outer surface. Thus, there is an enhanced 
contribution from polymer segments that lie in the Z > 30 range, with a commensurate decrease 
in contributions from the Z < 30 region. Consequently, there is an increase in the average layer 
height in the case that displays the most pronounced clustering. 

Effect of Grafting Density 

Another parameter that controls the structure of the layer is the grafting density. As the 
grafting density is decreased, the chains are less stretched and the height of the brush decreases. 
Variations in grafting density should also exhibit a pronounced effect on the morphology of the 
clusters that are formed. To investigate this phenomena, we fixed the number of stickers at 9 
(where we know that clustering occurs) and varied the grafting density of the brush. Grafting 

densities of 12%, 8% and 5% were analyzed. Figure 3 shows the density profiles p (z) of all 
the monomers in the brush for these three grafting densities. As indicated by the pronounced 
peaks at the top of the layers, clusters form in all these systems. However, the height at which 
the clusters are located decreases as the grafting density is decreased. In other words, the 
clusters form closer to the surface at the lower grafting densities. 


Fig.3 Normalized density profiles for all the monomers in the brush for the three grafting 
densities: 12%, 8% and 5%. 


The variations in grafting density also affect the lateral spacing between the clusters. We 
found that the clusters lie further apart at lower grafting densities. Decreasing the grafting 
density increases the average distance between the polymer chains. As the distance between the 
polymers is increased, the chains must undergo considerable lateral stretching in order for the 
stickers to associate. As a result, the clusters lie at lower heights and are more spread out. We 
also observed that the average layer height decreases with the decrease in grafting density. 


F.ffect of Varying the Sequ ence Distribution 

In order to study the effect of sequence distribution on the properties of polymer 
brushes, we ran the simulations for three particular realizations of the order parameter f: f = 0.1, 
0 5 and 1 0 The first case (f = 0.1) corresponds to a polymer brush composed of "blocky 
copolymers, the second (f = 0.5) to a brush consisting of purely random copolymers and m the 
third case, the brush is made up of alternating copolymers. 

We first discuss the results obtained from our Monte Carlo simulations, where the 
surface coverage was fixed at 15% (corresponding to Nchain - 153). In Figure 4, we plot the 
normalized density profiles for the solvophobic component of the brush. The plot clearly shows 
an enhancement of the solvophobic component density, pg(z), near the grafting surface. This 
indicates a "layering" transition, but the reason for this behavior is, at first glance, not entirely 
obvious. As the placement of the monomers along the chain length is random, there should be 
no preferential plane (in the Z direction) along which clustering occurs. 

Fig. 4 Normalized density profiles for the solvophobic component. Results obtained from the 
Monte Carlo simulations. 


We believe that this "layering" is a result of chemical heterogeneities that stem from the 
statistical manner in which we add A and B monomers while growing the copolymers. As a 
result, the brush is composed of some A-rich polymers and some B-rich polymers, even though 
the overall composition of the brush is symmetric (equal numbers of A and B monomers). 
Consequently, when the system is exposed to a selective solvent, the B rich copolymers 
contract to avoid the solvent. This results in an enhancement of the B monomers near the 
grafting surface. This effect is most pronounced in the blocky copolymer brush as finite size 
effects are strongest here. For the random and the alternating brush, the block length (which 
scales as 1/f) is too small to cause any noticeable aggregation in the Z direction. We note that 
these compositional fluctuations are always present for finite length polymers and disappear 
only in the limit of infinitely long chains ^. 

The overall density profile of the brush, however, should be independent of f as the 
brush is compositionally symmetric. This is shown in Figure 5 where the total density profiles 
(for the A and the B monomers) are plotted for the three values of f. (The slight enhancement of 
density for the blocky copolymer brush arises from the statistical fluctuation described above.) 

0 20 40 60 


Fig.5 Normalized density profiles for the entire brush. Results obtained from the Monte Carlo 


To check our Monte Carlo result, the overall density profile of the brush in.the Z 
direction was calculated by the SCF method and is plotted in Figure 6. The density profiles for 
the three f values are now identical, reflecting the compositional symmetry in our system. 
However, there is still a "layering" effect present in this calculation as the B monomers 
segregate to the surface (Figure 7). This indicates that the "layering" transition is not an artifact 
of the Monte Carlo simulations. 

Fig. 6 Normalized density profiles for the entire brush. Results obtained from the SCF 


To investigate the formation of lateral structures in the polymer brush we define an order 
parameter following the treatment of Lai 13: 

¥ (x,y) = < Pa (x,y) - Pb ( x >y)> . 

( 2 ) 

where Pa is the number of A sites in the Z direction for each point (x,y) and similarly, pg is the 
number of B sites in this direction. The angular brackets denote an ensemble average in the 
Monte Carlo simulation. The order parameter, V (x,y), for the three values of f is plotted in 
Figures 8(a)-(c). The darker regions in these plots denote the B-rich domains, while the lighter 
areas are the A-rich regions. The variations in the order parameter are clearly evident from these 
plots. The block copolymer brush shows the largest variations, with large, clearly defined 
domains of B monomers existing in the system. This behavior is reminiscent of the rippling 
phase seen in the grafted, incompatible homopolymers 13,14 The fluctuations in the order 
parameter are reduced for the random copolymer brush and disappear for the alternating 
copolymer brush. 


Fig. 8 Plots of the two dimensional order parameter, 'P (x,y). In (a) f = 0.1, in (b), f = 0.5 
and in (c), f = 1.0. The darker regions are the B-rich areas, while the A-rich areas are 
the lighter regions. 


In order to quantify these fluctuations, it is necessary to determine a characteristic length 
for the B domains. Unfortunately, finite size effects limit our ability to accurately extract these 
values from our simulations. In spite of this limitation, it can be clearly seen that the copolymer 
sequence distribution controls both the size and the distribution of these B-rich domains (Figs. 
8(a) -(c)). 


In conclusion, we showed how the presence of attractive end-groups can drive grafted 
chains to aggregate into clusters and thereby, alter the structure of polymer brushes. Using 
computer simulations, we demonstrated how the structure of the layer can be tailored by varying 
the number of functional groups (or the block length of stickers) and the grafting density. 

The findings presented here show qualitative agreement with recent results by Li and 
Balazs 3. They developed a Hamiltonian for grafted polymers that contain a single sticker at the 
end of each chain. Through the model, they varied the length of the chain and the sticker-sticker 
attraction. When the attraction between these end-groups was sufficiently strong to overcome 
the stretching energy of the chains, the stickers aggregated into finite clusters, with the chain 
ends localized near the top of the layer. Thus, both models point to the ability of the sticker sites 
to control the lateral and surface properties of the brush. 

We note that the formation of clusters within polymer brushes was also predicted to 
occur when the layer is immersed in a sufficiently poor solvent 15-18 . Here, the chains 
associate into local bundles to minimize polymer-solvent contact. This behavior alters the lateral 
properties of the grafted layer, however, the distribution of chain ends in the poor solvent case 
is much broader 15,19 than presented here or in reference 3. 

There are significant advantages to having the majority of the end-groups lie near the top 
of the layer (as seen in Fig. 3). Specifically, this structure can enhance the adhesive or wetting 
and lubricating properties of the surface. Thus, our results can provide guidelines for fabricating 
coatings and films with these enhanced characteristics. 

We also examined the case where the stickers, or solvophobic B sites, are distributed 
along the length of the chains. By combining SCF calculations and Monte Carlo simulations, 
we could visualize both the vertical and lateral structure of the brush. Our findings show that the 
overall morphology of the brush is dependent on the sequence distribution of the component 
copolymers. These results provide guidelines for controlling the size and shape of the B 
domains, and consequently, tailoring the brush for specific applications. 


A.C.B. and D.G. gratefully acknowledge financial support from ONR, through grant 
N00014-91-J-1363. A.C.B. and R.I. thank the DOE for financial support through grant DE- 
FG02-90ER45438. A.C.B. and M.F. thank the NSF for financial support through grant DMR- 
9407100 and the Hoechst Celanese Corporation. 


1. S. T. Milner, Science 251, 905 (1991). 


2. D. Gersappe, M. Fasolka, A. C. Balazs and S. H. Jacobson, J. Chem. Phys 100, 

1970 (1994). 

3. W. Li and A.C. Balazs, Molecular Simulations 13, 257 (1994). 

4. D. Gersappe, M. Fasolka, R. Israels and A. C. Balazs, submitted, Macromolecules. 

5. I. Carmesin and K. Kremer, Macromolecules 21, 2819 (1988). 

6. H.P. Deutsch and K. Binder, J. Chem. Phys. 94, 2294 (1991). 

7. N. Metropolis , A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, and E. Teller, J. 
Chem. Phys. 21, 1087 (1953). 

8. G. Fleer, M. A. Cohen Stuart, J. M. H. M. Scheutjens, T. Cosgrove and B. Vincent, 
Polymers at Interfaces . (Chapman and Hall, London, 1993). 

9 . B. van Lent and J. M. H. M. Scheutjens, J. Phys. Chem. 94, 5033 (1990) 

10. A. C. Balazs, I. C. Sanchez, I. R. Epstein, F.E. Karasz and W.J. MacKnight, 
Macromolecules 18, 2188 (1985); C. Yeung, A. C. Balazs and D. Jasnow, 
Macromolecules 25 , 1357 (1992). 

11. P. Lai and K. Binder, J. Chem. Phys. 95, 9288 (1991). 

12. A. R. Nesarikar, M. Olvera de la Cruz and B. Crist, J. Chem. Phys. 98, 7385 (1993). 

13. P. Y. Lai, J. Chem. Phys 100, 3351 (1994). 

14. J. F. Marko and T. A. Witten, Macromolecules 25, 296 (1992); G. Brown, A. 
Chakrabarti and J. F. Marko, Europhys. Lett. 25, 239 (1994). 

15. P. Lai and K. Binder, J. Chem. Phys. 97, 586 (1992). 

16. C. Yeung, A.C. Balazs and D. Jasnow, Macromolecules 26 , 1914 (1993). 

17. K. Huang and A.C. Balazs, Macromolecules 26, 4736 (1993). 

18. G. Grest and M. Murat, Macromolecules 26, 3108 (1993). 

19. R. Israels and A.C. Balazs, preliminary results from self-consistent field calculations on 
end grafted chains with stickers. 




*Department of Chemical Engineering, University of Colorado, Boulder CO 80309-0424 
**Department of Applied Mathematics, Research School of Physical Sciences and Engineering, 
Australian National University, Canberra AUSTRALIA. 


We describe the flow and filtration control imparted to membranes through the adsorption of 
polymer brushes onto the interior pores. The brushes exhibit a negative Poisson’s ratio, i.e. they 
swell under shear, and as a result behave as sensors and valves controlling the flow and filtration 
through the pore. The valve-behavior of brushes adsorbed onto cylindrical pores displays the 
same constant discharge control but also exhibits a critical shear rate for brush swelling. 


The control of liquid flow is accomplished conventionally by a system of sensors and 
valves: the sensors measure the flow and the valves are adjusted if this flow is not the desired set 

point. Recently, Sevick & Williams [1] proposed a novel microvalve which does not involve any 
movable parts or any external feedback control loop. The microvalve consists of a conduit lined 
with a polymer brush, i.e., a collection of end-grafted polymers absorbed onto the conduit 
surface. Figure 1. The polymers are grafted so densely that they overlap strongly and are forced 
to stretch away from the surface forming an elastic layer. Recent experiment [2,3] and theory 
[4,5] demonstrated that a polymer brush has a negative Poisson's ratio, i.e. it swells under an 
applied shearing force, and that there is little flow penetration into the brush [6], The valve-like 
response of two sandwiched brushes to an applied pressure drop was shown theoretically and can 
be simply explained as follows. An applied pressure differential results in laminar flow which 
imposes a shear stress on the brush, causing brush expansion. This brush swelling reduces the 
cross-sectional area for solvent flow, and thereby readjusts the prevailing shear stress. Both the 

shearing force and the brush swelling are determined self-consistently, resulting in a non-linear 

pressure-discharge relation. 




FIGURE 1: (A) The good-solvent planar 
brush microvalve. The planar polymer 
brushes, each of height H 0 , are grafted to 
opposing planes of length L, width Z, and 
separation W « Z. (B) When a good 
solvent flows across the planes, the brushes 
swell to a height H > H 0 . The flow 
penetrates the brushes only weakly so that 
the cross sectional area for flow is reduced. 
The brush swelling produces a closing of the 
flow area and a dramatic reduction in 
discharge results. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

To clearly demonstrate this, consider two opposing brushes where the grafting planes are 
separated by W and lined with polymer brushes of height H. The fluid penetrates only weakly 
into the brushes and consequently the pressure-driven flow is laminar and through a channel of 
effective height W-2H 

Q = ^Z(W-2H) 3 /(12ti) (1) 

where Q is discharge or volumetric flow rate, LxZ is the dimension of the grafting planes, rj is 
the solvent viscosity and AP is the pressure drop. The novel feature of this assembly is that H is 
a function of the pressure drop; this is particularly important as the discharge depends upon the 
third power of (W-2H). Hence, even a weak dependence of H on AP provides a strong non¬ 
linear AP dependence on the discharge. 

This expandable brush lining might be an attractive agent for modifying membranes to 
achieve specific flow and filtration properties. By adsorbing brushes into the interior pore walls, 
each pore becomes dynamic, responding to pressure fluctuations and providing microvalve 
action. Depending upon the grafting density and polymer chain length, the brush response in the 
pores will limit or maintain the flux through the membrane despite an increasing applied 
pressure. Moreover, the effective pore size depends upon the applied pressure drop, allowing the 
filtration properties of the membrane to change with the flow rate. This flow and filtration 
control is the subject of this proceedings; however, other brush effects are also envisioned. For 
example, a polymer brush adsorbed on the exterior, downstream surface forms a directional 
valve or diode [7]. The brush spills over the pore aperture, forming a gating valve. Downstream 
flow is impeded until a critical pressure is achieved, and upstream flow is blocked. 

In this proceedings, we describe flow and filtration properties of brush-lined membranes. 
In the following section, we review the single, planar brush-valve and in Section III we extend 
this analysis to cylindrical pores. In Section IV we consider briefly the role of pressure drop 
upon filtration properties and how this is role is modified by a distribution in the membrane pore 


In this section we review briefly the description of the planar brush-valve behavior, 
originally detailed in Sevick & Williams. Based upon the previous equilibrium arguments, we 
find an equilibrium expression for the brush height as a function of shearing force acting at the 
brush surface. However, in the brush-valve, the shearing force is NOT a direct or controllable 
variable, but instead depends upon the applied pressure drop. We construct a continuum 
mechanical expression describing the dependence of the surface shearing force upon the applied 
pressure drop for any given brush height. Solving the equilibrium and mechanical expressions 
simultaneously with Equation (1) provides a simple prediction for the discharge as a function of 
applied pressure drop. This is detailed in the remainder of this section. 

Each tethered chain in a brush can be pictured as a connected sequence of blobs [8]. The 
blobs are of radius £ = Rj/ 2 / L 3 ^ 2 where Rp is the Flory radius and L is the end-to-end distance 
of the chain. Within each blob the monomers are locally correlated as in a Flory chain. For 


distances larger than the blob size, the blobs act as hard spheres. The number of blobs per chain 
is given by N b = (L / R F ) . The brush is an assembly of chains end-grafted onto the grafting 
plane with density 1 / d 2 where d is the distance between grafting points. The grafting density is 
assumed to be large, i.e. the chains overlap. When the brush assembly is sheared, the free-chain- 
ends are displaced distance X in the x direction, the chain is tilted, and the end-to-end distance L 
is increased, The number of blobs per chain, N b increases with L further enhancing the 
excluded volume interaction between blobs. As a consequence the brush will swell: the height of 
the brush changes from H 0 to H. This is the mechanism for brush swelling proposed by Barrat 
[3]. To analyze the expected degree of swelling and chain tilt with applied shear, the free energy 

of a chain in the brush assembly can be expressed [1,3] as 


F chain = |kT-^ + kT^N^" 1 - F„X (2) 

2 N b £ z 

where the volume per chain is X) = d 2 H. The first term represents the stretching penalty of a 
Gaussian chain of blobs [9] that opposes brush swelling and shear, the second term is the 
excluded volume interactions amongst blobs that promote brush swelling, and the final term is 
the work performed by the shearing force F|| which tilts the chains, displacing the free chain-end 
distance X in the x direction. Here H 0 is the unsheared brush height H 0 = (2/5) 1 ^d 2 ^Rp^. 

We now construct the equilibrium and continuum mechanical equations necessary for 
predicting microvalve behavior. Minimizing the free energy under shear yields two 
equations, OF / de = 0 and OF / OA = 0, where e = X / H 0 and A = H / H 0 describe x- 
displacement and swelling under shear. These two equations are combined to give an explicit 
equilibrium relation between the dimensionless shearing force, / = F[£ 0 / kT, and swelling A [1] 

/ = |4- 3/2 (A 3 -l) 1,2 (2-A 3 )' 3/4 (3) 

which has the form shown by filled circles in Figure (2). This is the equilibrium relationship 
describing the brush height for an applied shearing force. The mechanical equation, relates the 
shearing force acting 

FIGURE 2: The brush swelling, A = H/H 0 , 
versus the dimensionless shear force per 
chain, / = F|£ 0 / kT, according to the Barrat 
mechanism [3] for planar grafting plane 
(solid line with filled circles) and for 
concave grafting surface with Hq/R = 0.78 
(filled squares) and Ho/R = 0.80 (asterisks). 
At very large forces, the planar brush height 
grows to a maximum of 125% of the 
unsheared brush height. The maximum 
brush swelling decreases with increasing 


at the elastic brush surface to the applied pressure drop and is next derived. As shown in [5] the 
flux through the brushes is insignificant in comparison to the discharge through the brush-free 
portion of the conduit. Nevertheless, the pressure-driven flow through the brush will place a 
shearing force along the contour of the chains. We sum the shear acting along the chain contour 
in approximate manner, assume that force is wholly localized at the tip of the chain, and is equal 
to F||. The pressure within the root and bulk of the brush, i.e. all but the tip-blob, has a force 
P (H - E,)H, 2 where P' is dP/dL. In the tip region, the velocity gradients are important and 

the force is ^ 2 a = ^ 2 r|| 

= -s (W - 2H)P . The total force, assumed localized at the tip is 

F|[ = ~WP s (4) 

where varies with the swelling, A, according to 

f = A- 3 ' 2 (2-A 3 ) 3/4 (5) 


and is much smaller than W. Since there is a reduction of blob size with brush expansion and 
shear, the shearing force required to maintain the expanded brush height is reduced. Equations 
(4) and (5) form the mechanical relation between f and A [1] which, taken together with the 
equilibrium / vs. A relation, Equation (3), provide the necessary information to construct 
dimensionless discharge, q, versus dimensionless pressure, p, descriptions. The q vs. p behavior 
depends upon P = W/H 0 , the ratio of the conduit size to the no-shear brush height, and is given 
by filled circles in Figure (3). For very small planar brushes, the pressure discharge behavior is 
nearly identical to that of conduits without a polymer liner, Figure (3a). Planar brushes which 
penetrate significantly into the conduit demonstrate a discharge or volumetric flux that is 
constant over order of magnitude changes in pressure drop, Figure (3b). Planar brushes which 
fill over 80% into the conduit will exhibit cut-off valve behavior, Figure (3c). 

The dimensionless pressure p = WP ^ / kT is an important parameter in valve design. 


The teim kT / is the osmotic pressure in the brush [8] and also the elastic modulus of the 
brush [10]. As the interesting valve behavior occurs for values of p on the order of 1, (Figure 3), 
the operating range of P must be roughly kT / . In other words, the valve effects occur 

when the applied force per unit area of the brush equals the osmotic pressure [1]. Loosely 
grafted brushes are "soft", i.e., they have large blob sizes £ 0 , small moduli, and are sensitive to 
small fluctuations in pressure. In contrast, more densely grafted brushes have smaller blob sizes, 
larger elastic moduli, and consequently, do not respond as dramatically to pressure fluctuations. 
The appropriate pressure gradient P = kT / (w^o) is clearly very sensitive to the blob size £, 0 , 
and hence to the grafting density. To examine some numerical examples we take W ~ so 
P = kT/ For a weakly grafted brush -1000 A and P 1 ~4 x 10 7 Pa/m ~ 10 2 atm/m. For a 
membrane of thickness 1000A this corresponds to a pressure drop of 10' 5 atm. However, for a 
more densely grafted brush with blob size = 100 A and again with L = 1000 A, the pressure 
drop is much larger, 0.1 atm. 


0.01 T ^ 

FIGURE 3: The dimensionless discharge q 
versus the dimensionless pressure p for values 
of P=Hq/(W/ 2) for planar brush (filled circles) 
and P =Ho/R for cylindrical brushes (asterisks) 
for various flow regimes. Expressions for p 
and q are below. 

(a): p = 0.74. The cross sectional area for 
flow is initially large. At low pressures the 
height of the brush is not changed 
significantly by the flow. The discharge is 
thus proportional to the pressure. At higher 
pressures, the force on the planar brush is 
large and it expands to its maximum height of 
2 1/3 H 0 but the cross sectional area for flow is 
large and the discharge is again proportional 
to the pressure. In contrast, the cylindrical 
brush swells to reduce the cross sectional area 
sufficiently to render an almost constant and 
lower discharge. Its maximum swelling is less 
than that of the planar brush, (b): P = 0.78. 

The cross sectional area for flow is reduced 
from (a). The discharge in the planar brush 
valve grows initially and levels off to a nearly 
constant discharge. In comparison, the 
cylindrical brush valve does not swell with 
initial increase in pressure until a critical 
pressure is achieved beyond which swelling 
occurs. Increased pressure further swells the 
brush, decreasing the flux. 

(c): P = 0.80. The planar brush exhibits cut¬ 
off behavior, i.e., the brush swells continually 
as pressure is increased from 0. The flux 
increases until a critical pressure beyond 
which the swelling dramatically reduces the 
cross sectional area for flow and hence the 
discharge. In contrast, the cylindrical valve 
does not swell as pressure is increased from 0 
and flow is standard poiseuille flow. At a 
critical pressure, the discharge is maximum 
and the brush begins to swell, reducing the 
cross sectional area for flow and the discharge. 
At larger pressures, the increase in discharge 
with pressure and the decrease in cross 
sectional area balance in such a way that 
discharge is nearly constant 


° 0 25 50 75 100 

PLANAR BRUSH-VALVES q = 6Qq^ / (kTW 2 z) p = WP / (2kT) 

CYLINDRICAL BRUSH-VALVES q = 16Qn5* < (^kTR 3 ) p = DP£ 3 / (kT) 



The microvalve behavior of a polymer brush adsorbed onto the interior wall of a 
cylindrical conduit should differ from that of the planar brush-valve in two ways. Firstly , the 
discharge, Q, through a cylindrical pore depends pore dimension, radius R, to the fourth power, 
rather than the third power. As such the cylindrical brush-valve should be even more sensitive to 
pressure fluctuations than the planar brush-valve of the previous section. This is easily seen 
through comparison of Equation (1) with that of discharge through cylindrical conduit: 

Q =—rc(R -H) 4 / (8r|) (6) 

Again, since brush height, H, varies with the shearing stress or applied pressure drop, the 
discharge Q is nonlinear in AP. Secondly, we might expect that the elastic moduli of a 
cylindrical brush is larger than that of a planar brush with the same grafting density, and hence 
the cylindrical brush-valve is less sensitive to small pressure fluctuations. In the spirit of the 
Alexander -deGennes model where each chain is stretched equivalently and the ends of the chain 
are located at the tip of the brush, the volume per grafted chain is smaller for convex grafting 
surfaces than for planar ones. Consequently, the larger excluded volume interactions enhance 
chain stretching, resulting in larger brush heights. This result is available from the free energy 
expression, Equation (2), where the volume per unit chain, d is now (fl - H 2 / 2Rjd 2 and H</R, 
the ratio of brush height to pore radius, is the curvature factor. The blob size varies along the 
chain, being largest at the grafting root and smallest at the crowded brush tips. However, for 
simplicity we complete our brush description using an average blob size H, 0 and with no-shear 
and under shear, respectively. The no-shear brush height, H 0 , is found from minimization of the 

free energy with respect to the chain's end-to-end distance or brush height: 

r -il/3 

H 0 = (2/5) 1/3 d- 2/3 R5/3. 

1 - 3H 0 / 2R 
(1-H 0 /2R) 2 


Thus, larger curvature factors provide more highly stretched chains, smaller blob sizes, larger 
elastic moduli, and brush-valves that are less sensitive to small pressure fluctuations. 

To compare the sensitivity of the cylindrical and planar brush valves, we construct the 
equilibrium and mechanical expressions that are analogous to Equations (3) and (4). The height 
of the brush under shear is found from minimization of the free energy with respect to £ and A 
and provides an equilibrium f vs. A relationship in the form of two equations which reduce to the 
planar brush-valve case in the limit Hq/R —»0. The brush height as a function of applied 
shearing force is shown in Figure (2) for a cylindrical brush for P = Hq/R = 0.78 (squares) and 
0.80 (asterisks). Note that with cylindrical brushes, the larger the no-shear brush height, the 
smaller the maximum stretching. More importantly, with larger brushes there exists a critical 
shearing force required for the onset of brush swelling. It is interesting to note that Klein’s 
surface force measurements exhibited a critical shearing force [2]. The mechanical / vs. A 


relation is found by summing the shearing force acting along the contour of the chain and 
assuming that it is wholly localized at the tip. The shearing force acting upon chain root and 
bulk is identical to that found in the planar geometry; however, the shear acting on the tips 

differs- a = = ri—(— (r 2 - r 2 ) . The total shearing force localized at the brush tip is 

| 3r | 3r^4rp >) 

F||=tp'(R + H)!; 2 (8) 

As the blob size £ decreases with swelling, the shearing force required to sustain brush swelling 
decreases. If the brush height is comparable to the pore dimension, then the shearing force is 
larger in the cylindrical case than that required in the planar brush case. This is expected as the 
blobs decrease in size from root to tip and consequently, the brush is more "stiff" requiring larger 
shearing force to affect and maintain expansion. 

Combining the equilibrium / vs. AP, Figure 2, the continuum mechanical expression, 
Equation (8), and Equation (6) provides dimensionless discharge versus pressures predictions. 
Figure (3) shows p vs. q for three different brush heights, (3 = 0.74,0.78, and 0.80 where the 
planar brush exhibits linear, constant-discharge, and cut-off valve behavior and the cylindrical 
brush exhibits nearly constant-discharge after a critical pressure beyond which valve effects 


Membrane pores are unlikely to be uniform in size and instead possess a distribution of 
pore sizes. By adsorbing polymer brush and applying a pressure drop, the "effective" pore size 
distribution will vary as a function of pressure drop. For example, upon adsorption of a chain of 
a given size, the smallest pores will have very small channels for flow which will decrease 
dramatically with shear (see Figure 3c). At the same time, larger pores will be relatively 
unaffected (Figure 3a) while intermediate pores will show constant discharge behavior over a 
range of pressure drops (Figure 3b). In this way, the filtration behavior of the brush-lined 
membrane is "dynamic", behaving as different membranes by simply varying the pressure drop 
across the membrane. Currently, one of us (FAB) is constructing discharge and effective 
"porosity" descriptions of membranes with various pore size distributions and with constant 
grafting density. It should be noted that grafting density of physi-sorbed chains may vary from 
pore to pore, depending upon the pore curvature. There exists no theory nor experimental results 
which detail this. 


In this proceedings, we have reviewed our description of planar brush valves, extended 
the description to cylindrical pores and discussed briefly how these predictions will impact the 
flow and filtration control of membranes with adsorbed polymer brushes. These descriptions 
were prompted by recent surface force measurements made by Klein [2] which were interpreted 
in terms of negative Poisson’s ratio of polymer brushes. Several theories explaining this behavior 
have been put forth, including our extension to a novel application, polymer brush microvalves. 


In order to construct microvalves, these descriptions must be demonstrated experimentally. To 
this end we have initiated measurements of discharge versus pressure drop across PVP-PS lined 
track-etched polycarbonate membranes. However, additional studies are needed. First, brush 
swelling under shear must be demonstrated directly using ellipsometry or neutron scattering. 
Secondly, we must understand and quantify the brush adsorption on the membrane surface; for 
example, how is grafting density affected by pore size for different chain architectures, i.e. for 
block copolymers with different anchor block size? These experiments are necessary for 
verifying and constructing brush-lined membranes with flow and filtration control. 

[1] Sevick and D.R.M. Williams, Macromolecules 27, 5285 (1994). 

[2] J. Klein, D. Perahia, and S. Warburg, Nature 352, 143 (1991). 

[3] J. Klein, Pure Appl.Chem. 64, 1577 (1992). 

[4] J.-L. Barrat, Macromolecules 25, 832 (1992). 

[5] V. Kumaran, Macrotnolecules 26,2464 (1993). 

[6] J. Klein, Y. Kamiyama, H. Yoshizawa, J.N. Israelachvili, G.H. Fredrickson, P. Pincus, L.J. 
Fetters, Macromolecules 26, 5552 (1993). 

[7] E.M. Sevick and D.R.M. Williams, Advances in Porous Materials, MRS Symposium, in 

[8] P.G. deGennes, Scaling Concepts in Polymer Physics (Cornell University Press: Ithaca NY 

[9] P. Pincus, Macromolecules 9, 386 (1976). 

[10] (a) G.H. Fredrickson, A. Ajdari, L. Leibler, and J.-P. Carton, Macromolecules 25, 282 
(1992). (b) D.R.M. Williams, Macromolecules 26, 5096 (1993). (c) S.A. Safran and J. Klein, 
J.Phys.Paris II 3, 794 (1993). 


Scanning conduction microscopy: a method of probing 
abrasion of insulating thin films on conducting substrates 

J. T. Dickinson* and K. W. Hipps** 
*Department of Physics 
**Department of Chemistry 

Washington State University, Pullman, WA 99164-2814 


The use of Scanning Force Microscopy (SFM) to probe wear processes at 
interfaces is of considerable interest. We present here a simple 
modification of the SFM which allows us to make highly spatially resolved 
measurements of conductivity changes produced by abrasion of thin 
insulating films on metal substrates. The technique is demonstrated on 
fluorocarbon polymer thin films deposited on stainless steel substrates. 


The scanning force microscope (SFM) is a useful tool for examining the consequences 
of tribological wear, particularly on inherently flat surfaces such as single crystals. However, on 
moderately rough surfaces, changes in topography due to tribological loading are often difficult 
to interpret in terms of wear processes. We have examined the wear of an important class of 
coatings, namely fluorocarbon thin films deposited on stainless steel, using a simple 
modification of standard SFM techniques. Metal-coated silicon nitride tips are used to probe 
current flow between the tip and the conducting substrate. Simultaneous topography and 
conduction images are acquired. During wear (performed outside of the SFM), localized 
thinning of the film and exposure of bare metal are easily and unambiguously detected by this 
method on size scales less than 50 nm. We describe the technique 1 and present results on two 
types of fluorocarbon thin films. This technique is similar in spirit to the use of combined 
STM/SFM imaging 2 , and spatially resolved potentiometry used recently to image potentials of 
metallic structures on integrated circuits. 3 Point-by-point measurements of semiconductor 
conductivity have also been made using an SFM with conducting tips in efforts to probe 
semiconductor doping profiles, 4 and to provide spatially controlled potentiometry for thin-film 
structures. 5 


Film topography was characterized with a Digital Instruments Nanoscope III scanning 
force microscope (SFM) operated in the contact mode. SFM scans were acquired at scan sizes 
ranging from 200 x 200 nm 2 to 100 x 100 (im 2 and at tip velocities from 0.2 to 200 (im/s. A 
simple modification of the SFM allowed simultaneous, spatially resolved measurements of film 
conductivity. A schematic of the experimental arrangement is shown in Fig. 1. Scanning 
conduction microscopy (SCM) measurements were made using commercial triangular Si 3 N 4 
SFM tips (115 p.m from tip to base) 6 that were sputter coated with 300 A of Au or Ag at room 
temperature. Both sides of the cantilever were coated to minimize cantilever deflection due to 
stresses in the metal coating. The SFM was operated in the contact mode, using a nearly 
constant compressive force, typically 30-60 nN. Assuming a tip radius of 50 nm, the pressure 
applied by the tip is about 20 MPa, well below the yield stress of most materials. Forces were 
determined from the measured displacements and quoted force constants for the commercial 
cantilever beams used in this experiment. The sample was mounted on double stick tape to 
insulate it from ground. Electrical contact to the conductive cantilever was made through the 
substrate. The circuit was completed by a 30-pm wire spot welded to the conducting cantilever 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

Fig. 1. Schematic diagram of scanning conduction experimental arrangement. 

Current measurements were performed with a Keithley 602 electrometer whose output 
was amplified with a Tektronix 3A9 differential amplifier. The amplifier output was digitized 
in phase with the acquisition of the SFM image, providing a SCM image that corresponded to 
the topographical image acquired simultaneously. For improved time resolution, current 
measurements were occasionally acquired at higher sampling rates during individual scans with 
a LeCroy 6810 digitizer. A bias voltage (typically, 1 volt) was applied to the sample through a 
limiting resistor (Riimiting, typically 1-10 MQ) to prevent current-induced damage to the metal 
coating of the SFM tip when metal-to-metal contact was made. When the SFM tip is over a 
region where the polymer film is thick, the effective tip-to-substrate resistance is high (>10 12 12) 
and little current flows through the circuit. When the tip crosses a bare region of the substrate, 
the tip-to-substrate resistance falls to low values (< 10 5 12) and the resistance of the circuit is 
approximately equal to that of the limiting resistor. Regions of intermediate conductivity are 
also evident. A typical SCM image is a 3-dimensional plot of the current magnitude vs position. 

The useful lifetimes of the conducting tips were limited by wear of the metal coating and 
ranged from 10-30 scans. High currents (> pA) associated with low limiting resistances or high 
bias voltages, however, dramatically shortened tip lifetimes. Both positive and negative bias 
voltages produced nearly equal currents with no apparent nonlinearities, suggesting ohmic 
conduction. Little effort was made to electrically shield the sample; pickup or perhaps ground 
loops generated small but detectable modulations in the current signal which are visible in the 
images presented here. 

Samples of polytetrafluoroethylene (PTFE) coated stainless steel were prepared in our 
laboratory by laser ablation deposition. 7 ' 9 PTFE films, -200 nm thick, were produced by 
focusing the output of a Lambda Physik EMG 203 MSC laser operating at 248 nm (KrF) 
through a quartz window onto a PTFE substrate (Goodfellow Chemicals) mounted in vacuum at 
a pressure of 10* 4 Pa. The laser was operated at a pulse repetition rate of 10 Hz and a fluence of 
4 J/cm 2 per pulse. The ablation products were collected on a stainless steel substrate held at 340 
°C. TEM and XRD analysis confirmed that the resulting 300-nm thick films were semi¬ 
crystalline PTFE. Stainless steel exposed by wear processes would be oxide coated. 

Film abrasion was performed by manually drawing a cotton swab dampened with 
acetone across the surface. The applied normal force was approximately 2 N over a nominal 
contact area of 3 mm 2 . Although not well controlled, the wear produced by successive linear 
strokes was indeed cumulative and served as a simple means to demonstrate the technique. 



Figures 2(a)-2(e) show 10 x 10 |im 2 SCM images of a PTFE film on stainless steel prior 
to and following varying amounts of abrasion. The scans were acquired at a tip velocity of 20 
pm/s with a 100-mV bias voltage applied through a 10 MO limiting resistance; a full-scale 
reading on the electrometer corresponds to 10 nA. As shown in Fig. 2(a), the nonabraded films 
are basically insulating. Sampling several regions of the film, we typically observed one or two 
submicron-sized regions of enhanced conductivity [similar to the small region in the foreground 
of Fig. 2(a)] in approximately 20% of the scans. These features were reproducible upon 
repeated scans. The large area of no current indicates that there is no evidence of noise due to 
discharges that might be generated by contact electrification. 10 - 12 Spatial resolution, as 
determined by the smallest patches of conductivity on otherwise non-conducting films, is as 
small as 20 nm. We suspect that this resolution is significantly influenced by the structure of 
the conducting patch (steep changes in conductivity) and the morphology of the conducting 
coating at the tip (i.e., perhaps only a limited portion of the tip is participating). One would 
expect that small modulations in conductivity would result in much lower resolution. 

Figure 2(b) shows a typical SCM scan after a small amount of abrasion (5 strokes of the 
swab). Some conduction through the film is evident in the SCM scan. All of the conduction 
zones shown show currents well below the value for tip-to-bare metal contact (10 nA). The 
total area of the conduction zones (regions exhibiting currents above the noise level) 
corresponds to 0.6% of the scanned area. 

Further abrasion (a total of 8 strokes) increases the area of the conduction zones, as 
shown in Fig. 2(c). The total area of conduction zones is now 2% of the scanned area. A few of 
the conduction zones display full scale current readings (10 nA) but most display intermediate 
values. All of the conduction zones in Fig. 2(c) are a micron across or less, with the majority 
being much smaller. 

Additional abrasion (a total of 16 strokes) further increases the fractional area of the 
conduction zones, as shown in Fig. 2(d); the conduction zones now occupy approximately 6% 
of the surface. Most of the conduction zones still display less than full scale current readings. 

Wiping the surface with the cotton swab 50 times produces large conduction zones, as 
shown in Fig. 2(e). Much of the surface now displays full-scale current readings. About 30% of 
a random sample of 10 x 10 pm 2 scans on this surface showed large, highly conducting patches. 
The remainder of the scans displayed large numbers of small, highly conducting regions over 
most of the scan, similar to the upper third of Fig. 2(e). 

SFM topographs acquired simultaneously with these particular SCM scans generally 
showed little correlation with the regions of conduction. The surface relief actually dropped 
progressively with abrasion, perhaps due to the removal of particulates (a common problem with 
thin films grown by pulsed laser deposition). In Fig. 2(f) we show the SFM topograph 
corresponding to the SCM scan of Fig. 2(e); the highly conducting zone appears to be in a 
region which tends to be elevated. 

Because our abrasion was not well controlled, quantitative analysis of the data is not 
particularly meaningful. Nevertheless, the fraction of the scanned area occupied by conduction 
zones is a clean, monotonically increasing function of the “extent of abrasion” (number of 
strokes). The conducting area fraction increased rapidly over the first several strokes, and 
subsequently displayed a slower increase proportional to the square of the number of strokes. 
An improved abrasion apparatus under construction will allow for controlled wear under more 
technologically interesting conditions. 

The influence of the bias voltage on conduction was examined over a range of 10-1000 
mV. Basically, the images remain invariant when the current scales are normalized. Over this 
limited range of tip bias, the measured current is a linear function of bias. Higher biases 
allowed for the detection of intermediate conducting regions that were in the noise at lower bias 
voltages. Typical resistances of the tip + film in these intermediate conductivity regions were 
10-30 MQ. Conduction occurred with both plus and minus bias voltages. 

The current at a fixed position can be examined as the tip approaches and leaves the 
surface. Figure 3 shows simultaneous force (a) and current (b) measurements as the tip 
approaches a region of intermediate conductivity on an abraded PTFE film (275 mV tip bias; 


Current in A) 

Time (msec) Time (msec) 

Fig. 3. Simultaneous force and current measurements during tip approach and contact [(a) 
force, (b) current] and withdrawal and detachment [(c) force and (d) current] on a region 
of intermediate conduction. 

full scale current corresponds to 100 nA). Note that (1) the onset of measurable current (labeled 
A) occurs after contact when the force has reached -15 nN; (2) the current continues to rise as 
the force increases (A to B), but then (3) stops rising at B, at a force of ~50 nN, even though the 
force continues to increase. Force (c) and current (d) measurements made as the tip is 
withdrawn are shown in Figs. 3(c) and 3(d). The current does not start to drop until C, when the 
force has dropped to ~25 nN, and decreases continuously until the force is very nearly zero (D). 
Note that adhesion produces an attractive force between the tip and surface as tip is withdrawn 
further; nevertheless, no current is observed after D even though there is obviously a connection 
between the tip and surface up to the instant of detachment. This physical connection could be 
due to direct polymer-tip adhesion, perhaps dominated by electrostatic forces. 12 

On high current patches (“metal-to-metal”), slightly higher forces were often required to 
produce measurable currents, with onsets at ~20 nN; maximum currents were reached at forces 
of ~65 nN. During tip retraction the current begins to drop at ~60 nN and reaches zero at ~55 
nN. Even higher adhesive forces were observed in the final stages of tip withdrawal from the 
high conductivity regions, again implying continued physical contact without measurable 
current. This higher adhesion may be due to an intervening water layer (which would have poor 
conductivity), consistent with the hydrophilic nature of an oxidized metal surface. 

With slight differences due to the hysteresis during make vs break, the force dependence 
of the current is nearly the same for these two regions, which display different levels of 
conduction. However, the current fluctuations were noticeably higher for the intermediate 
currents. We suggest that this is due to the conduction through thinned polymer, generally 
involving a hopping mechanism, which would tend to be “noisier”. 1 3 Conduction from the tip 
to the stainless steel may also be complex, but definitely displays lower amplitude fluctuations. 

Numerous efforts are currently under way to produce stable conducting cantilevers 
which would facilitate SCM studies, e.g., by Thomson and Moreland. 14 Coatings employing 
higher cohesive energy metals such as W and Rh may provide more durable, longer-lasting tips, 
with greater tolerance for higher contact forces and higher currents. Doped diamond tips, if 


sufficiently conducting, could be very robust and allow a variety of material systems to be 
studied. In principle, diamond tips would allow for the study of nm-scale wear of very hard 
coatings on metals and semiconductors. 


Sputter deposited metal coatings on commercial S 13 N 4 tips offer sufficient conductivity 
to allow SCM imaging and thereby provide a spatial probe of wear. Scans of undamaged 
polymer films normally showed no detectable conductivity with applied forces up to 100 nN, 
although some scans revealed small patches of conductivity. Thinned regions produced by wear 
exhibited intermediate conduction, and exposed, “bare” metal with high conductivity was 
readily detected. Damage accumulation with increased tribological loading was readily 
apparent in the evolution of these features. In some cases, useful correlations can be made 
between simultaneously acquired AFM topography and SCM scans. The high spatial resolution 
observed in the conduction scans is probably due to the nature of the damaged surface (e.g., 
exposure of very small, conductive regions to the metal substrate), and perhaps only limited 
portions of the tip participate in the current path. Compressive forces of at least ~20 nN were 
required to establish significant conductivity in both the low and high conductivity regions. A 
plateau in current occurred at applied compressive forces of about 60 nN. Upon retraction of 
the tip from the surface, measurable hysteresis in the current was observed, probably due to 
contact maintained by adhesive forces. Scanning at higher compressive forces and/or applied 
bias produced measurable conductivity in larger areas without any obvious change in 


We would like to thank X. D. Wang for his assistance during early stages of this 
experiment. This work was supported by grants from the National Science Foundation under 
the Surface Engineering and Tribology Program CMS-9414405, the Division of Materials 
Research DMR-9201767, and an Instrumentation Grant DMR-9205197. 


1. J. T. Dickinson, L. C. Jensen, K. H. Siek, and K. W. Hipps, to appear in Rev. Sci. Instrum. 

2. L. A. Wenzler, T. Han, R. S. Bryner, and T. P. Beebe, Rev. Sci. Instrum. ££, 85 (1994). 

3. M. Anders, M. Miick, and C. Heiden, J. Vac. Sci. Technol. A £, 394 (1990). 

4. J. Snauwaert, L. Hellemans, I. Czech, T. Clarysse, W. Vandervorst, and M. Pawlik, J. Vac. 
Sci. Technol. B 12(1), 304 (1994). 

5. M. Anders, M. Mueck, and C. Heiden, J. Vac. Sci. Technol. A &, 394 (1990). 

6 . Digital Instruments, Inc., Santa Barbara, CA 

7. Graciela B. Blanchet and S. Ishmat Shah, Appl. Phys. Lett. £2 1026 (1993). 

8 . G. B. Blanchet, C. R. Fincher, Jr., C. L. Jackson, S. I. Shah, and K. H. Gardner, Science 

9. Wenbiao Jiang, M. G. Norton, Lancy Tsung, and J. T. Dickinson, J. Mater. Res. IQ, 1038 

10. Sunkyo Lee, L. C. Jensen, S. C. Langford, and J. T. Dickinson, J. Adhesion Sci. Technol. 2, 
1 (1995). 

11. L. Scudiero, J. T. Dickinson, L. C. Jensen, and S. C. Langford, J. Adhesion Sci. Technol. 2, 
27 (1995). 

12. R. G. Horn and D. T. Smith, Science 256 . 362 (1992). 

13. J. T. Dickinson, S. C. Langford, and L. C. Jensen, J. Mater. Res. £, 2921 (1993). 

14. R. E. Thomson and J. Moreland, “Development of highly conductive cantilevers for atomic 
force microscopy point contact measurements,” to be submitted. 




* Department of Mechanical and Materials Engineering and 

** Department of Physics, Washington State University, Pullman, WA 99164 


Thin films of polytetrafluoroethylene have been formed by the pulsed-laser deposition 
technique. The structure of the films was found to be dependent upon the substrate temperature 
during deposition. At substrate temperatures from room temperature to 200°C the films were 
determined, by transmission electron microscopy and X-ray diffraction techniques, to be 
amorphous. Films formed at higher substrate temperatures were found to contain both amorphous 
and crystalline components. The data for the crystalline component is consistent with it being 
highly ordered with the long helical molecular chains aligned parallel to the film-substrate interface 
plane. The maximum amount of crystalline material occurred when the substrate temperature was 
close to the melting temperature of the polymer. 


Most applications for polymeric materials have been based on bulk forms of the material. 
The possibility of depositing thin films of polymers is of interest and novel methods for polymer 
film deposition are important because many polymers with desirable physical properties, e.g., 
high-temperature durability, are intractable. Such polymers cannot be processed by conventional 
solution or thermal techniques, and hence are not available as thin coatings. The applications for 
polymer films are driven, in part, by the electronics industry because of the desire for 
miniaturization and the integration with other materials and technologies. For these applications a 
dry processing method is often required. Many techniques have been used for the deposition of 
thin films of polymeric materials and several examples are given in Table I. 

The first reported use of the pulsed-laser deposition (PLD) technique for the formation of 
polymer films was by Hansen and Robitaille [1] and we believe that this technique has 
considerable potential in this area. PLD utilizes the material ablated from a target as the source of 
deposited material. Several groups have demonstrated that laser pulses in the UV part of the 
electromagnetic spectrum can be used to ablate and etch away the surface layers of a polymer 
[e.g., 2-6]. However, it is only fairly recently that efforts have been directed towards the 
synthesis of polymeric materials in thin-film form using the PLD technique. 

In studies of the formation of polytetrafluoroethylene (PTFE) films via plasma-assisted 
polymerization it has been demonstrated that the presence of radical and ionic species was a 
necessary requirement for polymerization. The species that have been proposed as major reactive 
participants include: CF, CF 2 , CF 3 , CF 3 +, C 3 F 5 +, F, and C. Several groups studying the laser 
ablation of polymers have identified the presence of these and other species, in addition to the 
monomer, emitted as a result of UV laser irradiation of PTFE targets [e.g., 5,6,7]. Thus, laser 
irradiation of PTFE produces all the essential species for polymerization and film growth, 
although not necessarily in the optimum proportions. The successful use of the PLD technique for 
the formation of PTFE films was first reported by Blanchet and Shah [ 8 ]. In this present article, 
the morphology of laser-deposited PTFE films is reported. 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

Table I. Deposition Techniques for Polymer Films 

_ Polymer _ Deposition Method _ Reference 

polyaniline vacuum evaporation 9 

methylmethacrylate-tetramethyltin plasma polymerization 10 

polythiophene electrochemical polymerization 11 

epoxy-amine chemical vapor deposition 12 

polymethylmethacrylate photodecomposition 13 

polytetrafluoroethylene pulsed-laser deposition 8,14 

poly(p-phenylene) vacuum evaporation 15 

polytetrafluoroethylene ion beam sputtering 16 

polytetrafluoroethylene rf sputtering 17 

polyphenelyne sulphide _ pulsed-laser deposition _ 18 


The experimental details of the deposition process used in our laboratory have been 
reported in detail elsewhere [14]. Briefly, the PTFE films were formed by focusing a Lambda 
Physik EMG203 excimer laser operating on KrF (X = 248 nm) onto a 1.0-mm-thick PTFE target 
supplied by Goodfellow Corporation. The laser fluence was varied between 3 and 9 J/cm 2 , the 
pulse duration was 20 ns, and the pulse repetition rate was 30 Hz. The films were deposited in 
vacuum (base pressure 10’ 6 - 10" 7 Torr) onto either glass microscope slides, 3-mm-diameter 
carbon-coated copper grids, single crystal NaCl, or single crystal KBr. The latter were used for 
characterization of the films by infra-red (IR) spectroscopy. The carbon-coated copper grids and 
the NaCl substrates were used to facilitate fabrication of samples for examination using 
transmission electron microscopy (TEM): thin films deposited onto NaCl were floated off in 
deionized water and collected on a copper grid. Films were deposited onto glass substrates for 
characterization using X-ray diffraction (XRD). 

The substrates were mechanically clamped onto the stainless steel plate of a small resistive 
heater and were positioned parallel to the target. The distance between the target surface and the 
substrate was ~ 5 cm. The temperature of the substrate heater was monitored by a thermocouple 
embedded into the heater block at a position just behind the substrate. All the temperatures 
reported here are those of the substrate heater as measured by the thermocouple. It should be 
noted that the actual substrate temperature may be lower than that of the substrate heater. If it is 
assumed that the only thermal losses are radiative from the surface of the substrate, the actual 
substrate temperature, at a measured value of 350°C, has been calculated to be ~10°C lower. 


XRD patterns recorded from the target material showed several peaks consistent with the 
poly crystal line nature of bulk PTFE. XRD patterns recorded from PTFE films deposited onto 
glass substrates at a heater temperature of 350°C exhibited a single sharp peak at a 26 value of 
18.09°, implying diffraction from the {100} planes of PTFE, which have a spacing of 0.490 nm. 
Crystalline PTFE has a hexagonal unit cell at 25°C with lattice parameters a = 0.566 nm and c = 
1.95 nm [19]. The crystal lattice consists of the periodic arrangement of the long axes of the 
helical molecular chain repeat units. The repeat unit for the hexagonal modification is 15 CF 2 
groups. The presence of a single peak in XRD patterns recorded from the films indicates that the 
crystalline regions are oriented with the long helical chains of the PTFE molecule parallel to the 


film-substrate interface plane. XRD patterns recorded from PTFE films deposited on substrates 
heated to i 200°C showed no sharp diffraction maxima. 

A qualitative examination of the degree of crystallinity of the laser-deposited PTFE films, 
as determined by the intensity of the (100) peaks in the XRD pattern, found that it varied with the 
substrate temperature used during deposition. From the plot shown in Figure 1 it can be seen that 
the maximum amount of crystalline material occurs at a heater temperature of 350°C. For the data 
shown in Fig. 1 the films were all of an equivalent thickness. 

Figure 1. Change in intensity of the (100) peak as a function of substrate heater temperature 
during film deposition. 

Electron diffraction patterns recorded from thin PTFE films deposited onto NaCl 
substrates confirmed that films formed on substrates heated to < 200°C were amorphous. The 
diffraction patterns showed two broad diffuse rings as illustrated in Figure 2(a). Electron 
diffraction patterns recorded from films deposited onto NaCl substrates heated to 350°C showed a 
series of discrete rings, as shown in Figure 2(b), indicating that these films contained a crystalline 
phase arranged in a random ("polycrystalline") manner. 

It is suggested that the reason why the crystallinity of the films changes with the substrate 
temperature during deposition, is that at substrate temperatures close to the melting point of PTFE 
there may be sufficient molecular mobility to allow orientation of the molecular chains into a 
crystalline configuration. It has been observed that crystallization of PTFE granules heated on 
NaCl substrates was enhanced at temperatures very near the melting temperature [20]. The ability 
to change the amount of crystalline material in the laser-deposited PTFE films by modifying the 
substrate temperature may have a significant impact on the applications of these films. 

TEM images of thin PTFE films deposited using a laser fluence of 3J/cm 2 at a heater 
temperature of 350°C indicated that the films contained both amorphous and crystalline regions. 
Two types of crystalline region were observed: small (~ 5 nm diameter) crystallites and larger 
crystalline regions ~ 60 nm in diameter. Both types of crystalline region were found to occur 
within an individual film. Figures 3 and 4 show high resolution electron micrographs of the 
smaller and larger crystalline regions, respectively. The long-range crystalline perfection of these 


regions can be clearly seen in the high resolution images. The plane spacing in both images was 

measured to be 0.283 nm. This spacing corresponds to that of the {l20}-type planes in the 
hexagonal PTFE crystal structure. 

Figure 2. Electron diffraction patterns recorded from thin PTFE films deposited onto NaCl 
substrates heated to (a) 200°C and (b) 350°C. (Note: different camera constants were 
used for the two patterns.) 

Figure 3. High resolution electron micrograph of a small crystalline inclusion in a PTFE film. 


Figure 4. High resolution electron micrograph of part of a large crystalline region. 

Wavenumber 2000 

Figure 5. FTIR spectra recorded from PTFE films deposited onto KBr disks at 200°C (light line) 
and 350°C (dark line). 

Selected-area electron diffraction (SAED) patterns recorded from the crystalline region 
shown in Fig. 4 showed only discrete reflections indicating the mono-crystalline nature of the 
inclusions [141. All the reflections could be indexed as arising from PTFE. The presence of the 

{00/}-type reflections in the electron diffraction patterns is consistent with the long molecular 
chains lying in the film-substrate interface plane and supports the earlier observation that the 
crystalline component in these films is highly textured. 

Some preliminary results from in situ heating experiments performed on the amorphous 
films in the TEM indicated that the amorphous phase was stable up to temperatures of 280°C; the 
maximum temperature used in this present study. Further studies in this area are in progress. 

Transmission IR spectra recorded from laser-deposited films on KBr substrates showed 
the chemical similarity between the amorphous and semi-crystalline variants of the PTFE films 
produced in this study. The spectra showed two strong absorption peaks at 1211 cm' 1 and 
1152 cm* 1 which correspond to the asymmetrical and symmetrical CF 2 stretching modes. Also 
resolved were the PTFE rocking and bending modes at 508 and 554 cnr 1 and the CF 2 wagging 
and chain stretching modes at 625 and 639 cm 1 . 


In conclusion, thin PTFE films have been formed by pulsed-laser deposition. The films 
were found to consist of a mixture of amorphous and crystalline components—the relative 
proportions of which depended upon the substrate temperature during deposition. XRD and TEM 
studies have shown that the crystalline component of the films is highly ordered with the 
molecular chains lying parallel to the film-substrate interface plane. Films formed at substrate 
temperatures <, 200°C appear to be amorphous. 

The authors would like to thank Professor K.W. Hipps for his help with the IR analysis. WJ 
acknowledges the support of the Materials Science Program at Washington State University. 


1. S.G. Hansen and T.E. Robitaille, Appl. Phys. Lett. 52, 81 (1988) 

2. P.M. Goodwin and C.E. Otis, J. Appl. Phys. 69, 2584 (1991) 

3. B.J. Garrison and R. Srinivasan, J. Appl. Phys. 57, 2909 (1985) 

4. S.V. Babu, G.C. D'Couto, and F.D. Egitto, J. Appl. Phys. 72, 692 (1992) 

5. J.T. Dickinson, J.-J. Shin, W. Jiang, and M.G. Norton, J. Appl. Phys. 74, 4729 


6. J.T. Dickinson, J.-J. Shin, W. Jiang, and M.G. Norton, Nucl. Instrum, and Methods 
B91, 672 (1994) 

7. G.B. Blanchet, C.R. Fincher, Jr., C.L. Jackson, S.I. Shah, and K.H. Gardner, 
Science 262, 719 (1993) 

8. G.B. Blanchet and S.I. Shah, Appl. Phys. Lett. 62, 1026 (1993) 

9. S.C.K. Misra, M.K. Ram, S.S. Pandey, B.D. Malhotra, and S. Chandra, Appl. Phys. 

Lett. 61, 1219(1992) 

10. M. Hori, S. Hattori, T. Yoneda, and S. Morita, J. Vac. Sci. Technol. B4, 500 (1986) 

11. B.D. Malhotra, N. Kumar, and S. Chandra, Prog. Polym. Sci. 12, 179 (1986) 

12. S. Tatsuura, W. Sotoyama, and T. Yoshimura, Appl. Phys. Lett. 60, 1661 (1992) 

13. S. Deshmukh and E.W. Rothe, J. Appl. Phys. 66, 1370 (1989) 

14. W. Jiang, M.G. Norton, L. Tsung, and J.T. Dickinson, J. Mater. Res. 10, 1038 

15. T. Asakura and N. Toshima, Jpn. J. Appl. Phys. 33, 3558 (1994) 

16. J.S. Sovey, J. Vac. Sci. Technol. 16, 813 (1979) 

17. L. Holland, H. Bierman, and S.M. Ojha, Thin Solid Films 35, L19 (1976) 

18. S. Kale, M. Swaminathan, and S.B. Ogale, Thin Solid Films 206, 161 (1991) 

19. E.S. Clark and L.T. Muus, Z. Kristall. B 117, 119 (1962) 

20. S. Yamaguchi and M. Tsuji, J. Mater. Res. 8, 2942 (1993) 



Fraunhofer-Institute for Surface Engineering and Thin Films, 

Bienroder Weg 54 E, D-38108 Braunschweig, F.R.G. 


A novel low temperature process for titanium nitride (TiN) deposition by means of an 
electron cyclotron resonance (ECR) plasma CVD process was applied to poly(tetrafluoro- 
ethylene) (PTFE). The organometallic compound tetrakis(dimethylamido)titanium 
(TDMAT) introduced into the downstream region of a nitrogen ECR plasma was used as a 
precursor for TiN deposition at 100°C. 

The thin TiN films (thickness 15-30 nm) act as interlayers to activate the electroless deposi¬ 
tion of copper followed by an electroplating process. Prior to the deposition of the interlayer, 
the samples were treated on a biased susceptor with argon ions to enhance the adhesion of 
the TiN interlayer. This metallization procedure avoids the use of toxic and pollutive etching 
agents and yields adherent copper layers on PTFE. 

The maximum adhesion of the metal film on PTFE was established to be 13 N/mm^. As 
shown by atomic force microscopy (AFM), TiN grains were formed on the fluoropolymer 
surface. Film composition was investigated by secondary ionization mass spectrometry 


Fluoropolymers are of great interest in the electronic industry due to their merits like low 
dielectric constants, low dissipation factors, high thermal stability, and high chemical 
resistance. The low surface energy of fluoropolymers and their chemical inertness leads to 
tremendous difficulties regarding the metallization of such polymers [1]. Thus, a surface 
modification of fluoropolymers is required to improve the adhesion [2]. Hazardous and 
pollutive etching agents like sodium naphtalenide or sodium in liquid ammonia, are 
necessary to activate the poly(tetrafluoroethylene) PTFE surface before metallization by 
electroplating [3]. These treatments lead to the removal of fluorine leaving unsaturated sites 

like C=C or C=C which can be oxidized to provide a higher surface energy for the 
subsequent metallization step. 


Mat. Res. Soc. Symp. Proc. Vol. 385® 1995 Materials Research Society 

Besides these conventional chemical surface treatments some work is done regarding non¬ 
equilibrium plasmas, ion beams, and radiation interactions with PTFE surfaces. Enhanced 
copper-PTFE adhesion was achieved by presputtering treatment prior to the sputtering of 
copper [4]. The presputtering with Ar + ions lead to an increase of the surface roughness. 
The same effect was also observed at x-ray damaged PTFE [5] and bombarding the PTFE 
surface with Xe atoms in the keV range. The latter method leads to the formation of PTFE 
filaments on the surface. 

Haag and Suhr [7] reported on improved adhesion of electroplated copper on plasma etched 
PTFE. Prior to the electroplating, a thin metallic layer was deposited employing a plasma 
enhanced chemical vapor deposition (PECVD) process. Organometallic compounds of Pd, 
Pt, Au and Cu were used in this process. 

We report here about the application of a novel low temperature titanium nitride (TiN) 
deposition process [ 8 ] as pretreatment for a subsequent electroplating process. 


Depositions were carried out using an ASTeX HPM/M ECR source mounted on a reaction 
vessel equipped with a substrate holder which is connected to a radio frequency 
(13.56 MHz) generator. A detailed description of the experimental set up was published 
previously [ 8 ]. The PTFE substrate (5x5 cm 2 , thickness 1 mm, purchased from Goodfel- 
low) was placed four inches below the ECR source flange. To achieve a smooth fluoro- 

polymer surface for SIMS investigations Teflon® AF (DuPont) coated silicon was used as 
the substrate. A flow of 15 seem [standard cm 3 min" 1 ] nitrogen, regulated by a mass flow 
controller, was introduced through the ECR cavity while tetrakis(dimethylamido)titanium 
(TDMAT) vapor was introduced through a gas ring into the downstream region between the 
plasma cavity and the susceptor. The TDMAT (purchased from Solvay Germany) was kept 
in a double glass wall container to allow precise temperature control. The temperature was 
thermostatically maintained at 25°C to produce a flow of 2.5 seem. The microwave power 
(MW) was set to 400 W and thepressure was 0.1 Pa. No external heating of the graphite 
susceptor was applied. The substrate temperature rose after a process time of 20 minutes to 

Prior to the deposition of TiN, the substrates were treated with an ECR argon plasma using 
15 seem Ar and a MW power of 400 W. The samples were biased to -100 V for this treat¬ 
ment by connecting the susceptor with a radio frequency (rf) generator operating at 
13.56 MHz. 

After the deposition of the thin TiN film, the teflon substrates were electroless coated with 
150 nm copper. The copper bath consists of an aqueous basic copper sulfate (CUSO 4 ) 


solution (50 mmol/1) and formaldehyde as reduction agent. This copper film was enhanced 
to a thickness of 20 pm by electroplating using an acidic copper electrolyte (1 mol /1 CUSO 4 ) 
and a current density of 10 mA/crn^. 

The adhesion was checked by the scotch tape test and a z-axis pull-out test [9]. In this test a 
progressive load was applied to a piston stuck on the copper surface until the sample failed. 
The joint was prepared with an epoxy resin at 100°C. 

AFM investigations were performed with a commercial ambient AFM device using silicon 
nitride pyramidal tips with a tip radius of 50 nm, tracking forces of about 150 nN, and a scan 
frequency of 2 Hz. Root mean square (RMS) roughness measurements were taken after 
linear background substraction and substraction of the long range roughness from 3x3 pm^ 


Figure 1 represents a scanning electron micrograph of a PTFE sample electroplated with 
copper. In Figure 2 the influence of an argon plasma pretreatment on the surface roughness 
and the adhesion strength of the electroplated copper is plotted. The argon plasma pretreat¬ 
ment leads to an increase of the adhesion strength. This effect can be explained by the 
increasing roughness of the PTFE substrates due to the bombardment with Ar + ions [4]. 
Samples without argon plasma pretreatment had already failed the scotch tape test. The 
failure of the metal/fluoropolymer joint always occured at the TiN/PTFE interface. An 
influence of the interlayer thickness on the adhesion strength could not be found. The 
measured adhesion force of the electroplated copper of 13 N/mm^ is in the range found by 
Na/NHg treated PTFE [10]. At a pretreatment time above 10 min the surface roughness 

decreases, this is presumably caused by the increase of the surface temperature of the PTFE 
sample on the susceptor. The increase of the surface temperature leads presumably to the 

formation of volatile C x F y species. These 

species act as monomers for the redeposition 
of a thin plasma polymerized fluorocarbon 
layer which inhibits the progress of the sur¬ 
face roughness. 

Figure 3 reveals the influence of the 
substrate material and the deposition time 
on the sheet resistance. Due to the increasing 
film thickness the surface resistance decrea¬ 
ses in a linear fashion. The deposition rate 
on Si 02 is fairly low and was measured to 

Fig. 1: SEM of PTFE surface electroplated 
with 20 pm copper after electroless deposi¬ 
tion of copper. 


Pretreatment time [min] 

Fig. 2: Influence of an argon plasma pretreatment 
(15 seem Ar, HF bias -100 V, MW power 400 W, 
TiN deposition time 20 min) on the surface rough¬ 
ness and the adhesion strength of electroplated cop¬ 
per on PTFE 

Deposition time [min] 

Fig. 3: Sheet resistance of TiN interlayer on PTFE 
and SiC >2 vs. deposition time 

be 1.5 nm/min . Figures 4 a-d repre¬ 
sent the surface morphology of an 
untreated PTFE, an argon plasma 
pretreated PTFE, a TiN coated PTFE 
sample, and an TiN coated SiC >2 sub¬ 
strate, respectively. The argon plas¬ 
ma pretreated sample in Figure 4b 
indicates in comparison to Fig. 4a the 
formation of small islands on the sur¬ 
face which cause the enhanced 
adhesion of the TiN interlayer. The 
TiN on SiC >2 (Figure 4 d) exhibit 
well shaped small crystallites where¬ 
as unpronounced crystallites were 
grown on PTFE (Figure 4c). The ob¬ 
served differences in the crystallite 
morphology and the sheet resistance 
indicate that the growth of TiN on 
PTFE is strongly influenced by the 
fluoropolymer. SIMS measurements 
were performed using Teflon AF 
coated silicon as substrates. A 15 nm 
thick TiN layer corresponding to a 
deposition time of 10 minutes 
revealed a composition of 

Ti 1.0 N 1.12 c 0.75°0.12 F 0.33’ 

indicating a high impurity content ofcarbon and fluorine. 

The etch rate of PTFE immersed in an ECR nitrogen downstream plasma was determined to 
be 80-90 nm/min using the described experimental parameter set. The ability of the ECR 
nitrogen plasma to etch PTFE leads to the formation of volatile C x FyN z species which are 

presumably redeposited as plasma polymers. This explains the high contents of carbon and 
fluorine in this thin, grey colored interlayer and the higher sheet resistance compared to 
films on an Si0 2 substrate. With increasing deposition time the sheet resistances on PTFE 

tend towards the numbers measured on Si02 as illustrated in Figure 3. The progressing 

growth of the TiN interlayer suppresses the plasma etching of PTFE. This interlayer 


Fig 4- AFM images (lxl ^m 2 ) of untreated PTFE (a), PTFE pretreated with Ar plasma, 
-100 V rf bias, time 10 min (b), subsequently TiN coated (deposition time 20 min) PTFE 
surface (c), and TiN (deposition time 20 min) on SiC >2 (d) 

activates the electroless deposition of copper using formaldehyde as the reducing agent in a 
copper bath . That means that no further activation with palladium, as necessary in today s 
electroplating of plastics, is required. A summary of the applied process steps is illustrated 
in Figure 5. 


Thin TiN films deposited on fluoropolymers in a low temperature plasma enhanced process 
can be used as interlayers for the electroplating of copper. To achieve an adherent metalliza¬ 
tion a pretreatment with argon ions at a substrate potential of -100 V is required. The ion 
bombardment increases the surface roughness leading to an enhanced adhesion of the inter¬ 
layer. The thin interlayer, with a thickness of 15-30 nm, activates the deposition of electro¬ 
less copper. A subsequent electroplating process yields a copper coating with an adhesion 


I. ECR argon plasma treatment 
10 min, rf bias -100 V 


HI. Electroless copper deposition 
150 nm Cu, 30 min 

II. ECR plasma deposition of 
TiN interlayer (10-20 min) 



IV. Electroplating of copper 
20 (im Cu, 10 mA/cm 2 


Fig. 5: Overview of the used process steps for the metallization of PTFE 

force of 13 N/mm 2 . The described combination of the plasma based pretreatment steps 
followed by an electroplating process is a promising approach to avoid pollutive and hazar¬ 
dous reducing agents used in today's metallization of fluoropolymers. 


The authors are indepted to K. Schiffmann and P. Willich for AFM and SIMS 
investigations, respectively. 


1. P. K. Wu, G.-R. Yang, X. F. Ma, and T.-M. Lu, Appl. Phys. Lett. 65, 508 (1994) 

2. E. Sacher, Progress in Surface Science 47,273 (1994) 

3. L. M. Siperko, and R. R. Thomas, J. Adhesion Sci. Technol. 3,157 (1989) 

4. C.-A. Chang, Appl. Phys. 51, 1236 (1987) 

5. D. R. Wheeler and S. V. Pepper, J. Vac. Sci. Technol. 20,442 (1982) 

6 . R. Michael and D. Stulik, J. Vac Sci. Technol. A 4, 1861 (1986) 

7. C. Haag and H. Suhr, Appl. Phys. A 47, 199 (1988) 

8 . A. Weber, R. Nikulski, C.-P. Klages, M. E. Gross, W. L. Brown, E. Dons, and 
R. M. Charatan, J. Electrochem. Soc. 141, 849 (1994) 

9. F. Garbassi, M. Morra, E. Occhiello, Polymer Surfaces {John Wiley & Sons, 
New York, 1994), p. 341 

10. H. Schonhom and F. W. Ryan, J. Adhesion 1,43 (1969) 




♦Department of Physical and Theoretical Chemistry, University of Bonn, Wegelerstr. 12, 
D 53115 Bonn, Germany 

♦♦Department of Chemistry, Hong Kong University of Science and Technology, Clearwater Bay, 
Kowloon, Hong Kong 


Cu, Ni, and Au were deposited with defined patterns and good adhesion by electroless 
plating, e-beam evaporation, and sputtering onto Teflon (polytetrafluoroethylene, PTFE), Teflon 
ET (FTFE-co-ethylene), Teflon FEP (PTFE-co-hexafluoropropylene) and Teflon PFA (PTFE-co- 
perfluoroalkoxy vinyl ether) surfaces. The polymers had been irradiated in a tetramethyl - 
ammonium hydoxide solution (TMAH) by a Nd:YAG laser at 266 nm and by an excimer laser at 
248 nm prior to the metal deposition process. Both, the treated and virgin polymer surfaces were 
characterized by x-ray photoelectron spectroscopy (XPS), secondary ion mass spectroscopy 
(SIMS) and Micro-Raman spectroscopy. The increased metal to polymer adhesion at the interface 
was found to be due to chemical changes and is in the order Ni > Cu s Au. 


Fluoropolymers possess a variety of exceptional properties such as a low dielectric 
constant, high resistivity, low coefficient of friction and high thermal and chemical stability. 
Metallization of such fluorinated polymers offers new applications in the area of electronic 
packaging, if satisfactory adhesion can be obtained. Poly(tetrafluoroethylene) (PTFE), better 
known under its trade names Teflon™ (DuPont) or Hostaflon™ (HOECHST), has a dielectric 
constant (e) of approximately 2.1. This is considerably lower than the dielectric constant of a-Si 
( e s 4.0), which is used as a standard material in the microelectronic industry. The high electrical 
conductivity of Cu lines plated on a Teflon sheet can result in the development of very fast circuit 
boards with relatively little capacitive coupling. However, the chemical inertness of FTFE is 
responsible for its poor adhesion to metals. Numerous surface modifications have been reported 
to obtain improved results. Chemical etching [1,2], ion bombardment [3-5], plasma exposure [6], 
x-ray and laser irradiation [7-9] have been used to increase the surface wettability and, hence, to 
improve the adhesion of the deposited metal film. 

Combinations of physical and chemical surface modifications were reported by Okoshi et 
al. [10]. The PTFE surface became hydrophilic after exposure to an ArF (193 nm) excimer laser 
in a diborane/ammonia gas atmosphere. Niino and Yabe used anhydrous hydrazine gas to modify 
fluorinated polymers with ArF excimer laser irradiation [11]. These processes require either a 
high vacuum or toxic gas handling. 

We have developed a new method for depositing Cu, Ni, and Au on fluorinated polymer 
surfaces in large areas or imagewise by exposing the polymer surfaces in aqueous TMAH 
solution to a KrF excimer laser or to the 4th harmonic of a Nd:YAG laser. Conventional 
electroless plating processes follow the activation step. Also, e-beam and sputtering induced metal 
deposition have been investigated. In the present paper, emphasis has been put on the analysis of 
the polymer surface using x-ray photoelectron spectroscopy (XPS) as well as static and time of 


Mat. Res. Soc. Symp. Proc. Vol. 385 ® 1995 Materials Research Society 

flight secondary ion mass spectroscopy (SSIMS, TOF SIMS). These methods provided 
information on the chemical modifications of the surface which are responsible for the better 
adhesion at the metal/polymer interface. 


Sample preparation and irradiat ion 

PTFE, FEP, PFA and ETFE were obtained from HOECHST as 150 pm and 300 pm thin 
films and were used as received. Irradiation of the samples was carried out using a pulsed 
Nd:YAG laser (Spectra Physics, GCR3 series) in TEMoo mode at 266 nm, or alternatively with 
an excimer laser (Lambda Physik, EMG 202 MSC) at 248 ran. The polymer sheet was mounted 
in a quartz cell covered with a thin fused silica window. The cell itself was filled with aqueous 
tetramethylammonium hydroxide solution (25%, Riedel de Haen) and positioned on a XYZ 
translating stage. All three axes were driven by computer-controlled high-resolution stepper 
motors with a speed ranging from 0.005 to 0.5 mm/s. Fused silica lenses and UV mirrors were 
used to guide, expand and focus the laser beam. Laser irradiation was conducted with an incident 
beam perpendicular to the surface of the polymer films. Throughout the entire time of the 
irradiation, the fluence was constantly monitored and ranged between 50 mJ/cm 2 and 120 
mJ/cm 2 . The samples were exposed to a total of 100 - 9000 laser shots at a repetition rate of 1 - 
10 Hz. 

Microstructuring was achieved by placing a stainless steel mask in direct contact with the 
polymer sheet. Both, mask and polymer sheet were then exposed in solution as described above. 

Metal deposition 

Electroless Cu, Ni, and Au plating was used to metallize the microstructured fluoropolymer 
samples. Detailed descriptions for metal plating bath compositions can be found elsewhere and 
were applied as cited [12-14]. E-beam evaporation and sputtering methods were also used. The e- 
beam evaporation system (Denton, DV-SJ/20C) was used for the deposition of Cu and Ni films 
onto the treated fluoropolymer surfaces. E-beam evaporation of the metal films was carried out at 
a pressure of lxlO 6 Torr. The thickness of the metal deposited was controlled by time and a 
profiler. Gold was sputtered by 1 keV Ar + ion bombardment at 4xl0 6 Torr. The growth of the 
film thickness was monitored using a quartz crystal microbalance. 

Sample characterization 

XPS, SSIMS and TOF SIMS data were aquired on a Perkin Elmer (PHI 5500/ 
5600/7200) surface analysis system. XPS experiments were performed using A1 Ka radiation 
(1486.6 eV). The spectra were aligned on the F(ls) peak to account for surface charging. An 
initial survey sweep was done at a pass energy of 93.6 eV. Detailed multi-plexed analysis of the C 
Is and F Is photoelectrons was done at a pass energy of 23.5 eV. Ion bombardment for SSIMS 
was accomplished using Xe + primary ions at 4 keV, TOF SIMS data were aquired with Cs + 
primary ion bombardment. Static analysis conditions were assured by ion fluences not higher than 
8 x 10 11 ions/cm 2 . Additional spectroscopic data were obtained by Micro-Raman spectroscopy 

Surface morphologies were investigated with a cold field emission SEM (JEOL) at 0.5 to 


2 kV to minimize charging effects. Atomic force microscopy (Topometrix) was used to scan the 
virgin and irradiated polymer surfaces to get qualitative data concerning the surface roughness. 
The receding and advancing water contact angles were measured at 25°C and 50- 60% relative 
humidity using a goniometer (Ramd-Hart). 


Laser irradiation of all four fluoropolymers in TMAH solution (TMAH) resulted in 
increased hydrophilicity of the polymer surface. Contact angle measurements showed that 
irradiation of PTFE with a Nd:YAG laser (266 nm) at 100 mJ/cm 2 and 10 Hz resulted in a 
decrease of the contact angle from 100° to 40°. after 1500 shots. The decreased contact angle 
results from to improved hydrophilicity. Longer exposure did not result in higher hydrophilicity 
of the PTFE surface. FEP, PFA and ETFE irradiated under the same conditions resulted in 
smaller changes of the contact angle. The values ranged from 100°. for the virgin polymer to 60° 
to 70°. for the exposed fluoropolymers. 

Characteristic peaks for C Is and FIs peaks are detected on the virgin polymer surface. 
The atomic ratio of fluorine to carbon was 1.98, close to the expected value of 2.0. Oxygen 
detection from surface contamination was negligibly small. XPS spectra of PTFE exposed in 
TMAH solution reveal the appearance of the N Is and O Is peaks in correlation with exposure 
time. After 300 to 1800 shots with 100 mJ/cm 2 the intensity of the carbon C Is peak increased 
gradually as well as the O Is and N Is peak intensities. However, the surface modification and 
formation of hydrophilic groups in the polymer backbone were also accompanied by 
defluorination as indicated by a decreased F Is peak. Figure 1 shows the C Is spectrum of PTFE 
exposed to 1500 shots at 266 nm, 100 mJ/cm 2 and 10 Hz in TMAH solution. Five peaks can be 
fit and are assigned to graphitic carbon at 284.6 eV, -CHF-CF 2 -, -£-OH, and -C-NH 2 at 286 eV, 
-CF 2 -O- and -£=0 at 287.5 eV, -£F 2 -CHF- and -£F at 289.5 eV and -CF 2 at 292 eV. Nd:YAG 
and Excimer laser exposure of PTFE at 266 nm and 248 nm, both at 100 mJ/cm 2 and 10 Hz, 
produced the same change of the surface composition. It should be added that none or little 

2M 232 296 28* 286 2M 

linding Energy (eV) 

Fig. 1: XPS C Is spectrum of PTFE irradiated in tetramethylammonium hydroxide solution 

coloring of the PTFE surface was observed during the first 1200 shots although good metal 
adhesion was already obtained after 900 laser shots. 


Applying the same conditions, the XPS survey and C Is spectra of FEP and PFA gave 
similar results. However, the degree of surface modification deviated from the PTFE results. In 
accordance with the change of contact angles, both fluoropdymers showed smaller O Is and N Is 
peak intensities and a degree of defluorination for FEP and PFA similar to PTFE. Exposure of 
ETFE at 10 Hz and 100 mJ/cm 2 resulted in strong graphitization. Although the survey spectra 
show an increased oxygen peak intensity, no nitrogen peaks could be detected. Curve fits of the C 
Is spectra reveal increased peaks at 284.6 eV due to graphitic carbon and at 286 eV due to CHF- 
CF 2 - and -£-OH groups. The peak at 292 eV of exposed ETFE was reduced by two third 
compared to the original peak of the virgin polymer. 

With SSIMS and TOF SIMS,' spectra were taken of both virgin and exposed 
fluoropolymer samples as shown in figure 2. The virgin PTFE spectrum displayed characteristic 
peaks at m/z = 12 for C+, m/z = 31 for CF+, m/z = 69 for CF 3 + m/z = 74 for C 3 F 2 + and m/z = 

Fig. 2: SSIMS Spectra of virgin and exposed PTFE 

93 for C 3 F 3 + Larger fragments with m/z = 131, 143, 155, and 181 were present as well and 
attributed to C 3 F 5 +, C 4 F 5 +, CsF 5 +, and CsF 7 + . SSIMS and TOF SIMS spectra of PTFE 
exposed to 1200 shots at 266 nm, 100 mJ/cm 2 at 10 Hz showed additional peaks at m/z = 28 for 
CH 2 N+, m/z = 29 for CHO+, and m/z = 30 for CH 4 N+ Further peaks were detected at m/z = 42 
for C 2 H 4 N+ or 02 ^ 0 +. Peaks at m/z = 58 were assigned to C 3 HsN + and at m/z = 59 to 
C 3 H 4 F+ Although the peak intensities are smaller than the characteristic PTFE peaks, they lead 
to the assumption that the polymer was partly reduced during the laser treatment and that 


hydrophilic groups were induced into the polymer backbone. 

SSIMS spectra of FEP and PFA indicated that both polymers were defluorinated and 
partially carbonized as is evident by an increased m/z = 12 peak intensity. No peaks for other 
surface modification were detected. Carbonization of ETFE surfaces was already very strong after 
300 shots at 100 mJ/cm 2 and 10 Hz. SSIMS spectra showed inreased peak intensities for m/z = 
12 for C + and m/z = 27 for C 2 H 3 4 *. No characteristic peaks for aromatic fragments were detected 
for increased unsaturation. 

AFM and cold field emission SEM were used to investigate the contribution of surface 
roughness to the metal/polymer adhesion. Figure 3 shows SEM micrographs of virgin PTFE and 
FEP as well as PTFE and FEP exposed in TMAH solution for 1200 shots at 266 nm, 10 Hz and 

Fig. 3: SEM micrographs of virgin (a) and exposed (b) PTFE and virgin (c) 
and exposed (d) FEP. — = 1 pm in (a) - (d). 

100 mJ/cm 2 . Laser irradiation of PTFE in TMAH at fluences of up to 100 mJ/cm 2 increased the 
surface roughness without producing holes or craters on the surface even in large scale. The 
pictures of FEP are characteristic for PFA and ETFE as well. The virgin FEP surfaces show 
almost no irregularities. However, on the exposed polymer surface spots of carbonized polymer 
can be seen. In case of exposed ETFE these spots are ten to hundred times larger than those on 
FEP and PFA surfaces. Micro-Raman spectra confirmed via a characteristic peak at 1600 cm* 1 
that the dark spots stem from fluoropolymer regions degraded to amorphous carbon. The polymer 
surface between the amorphous carbon spots were free of any graphitization. Surface roughness 
was increased for all three polymers by two to five times. The maximum height of the surfaces are 
03 fim. 

On the basis of these results we deposited metal films on the imagewise modified 
fluoropolymer surface. Figure 4 shows a copper pattern on PTFE deposited by electroless 
plating. Very strong adhesion was obtained for PTFE compared to FEP, PFA and ETFE. Scotch 
tape tests applied to metal films on PTFE could not remove the deposited Cu, Ni, and Au even 
after several applications. In general, adhesion of Ni was found to be stronger than that of Au and 
Cu. Metal films deposited by e-beam and sputtering methods resulted in slightly stronger 
adhesion compared to the films deposited by electroless plating. However, it can be assumed that 
this difference is due to the grainy nature of the films deposited by electroless plating methods. 


Fig. 4: Copper pattern on PTFE deposited by electroless plating. 

= 100 pm. 


The interactions between the tetramethylammonium hydroxide solution (TMAH) and the 
nuoropolymers under UV - laser irradiation lead to increased hydrophilicity. As demonstrated 
with XPS and SIMS, the effect especially on PTFE is not purely of a physical nature, but also 
due to chemical changes on the surface. It can be assumed that hydroxy and amino groups were 
induced into the polymer backbone, thus, increasing the metal/ polymer adhesion at the interface. 
Although showing amplified hydrophilicity, FEP, PFA and ETFE were mainly defluorinated and 
partially carbonized under laser irradiation in TMAH. 

Selective area metallization was possible by using a photographic contact mask during 
exposure of the fluoropolymers in solution. This method could be used for direct patterning of 
fluoropolymers, especially for FI FE. 


1. H. Schonhom and F.W. Ryan, J. Adhesion, 1,43 (1969) 

2. D.M. Brewis, R.H. Dahm and M.B. Konieezko, Angew. Makromol. Chem. 43, 191 

3. P. Bodo and J.E. Sundgren, J. Appl. Phys. 60, 1161 (1986) 

4. C.A. Chang, J.E.E. Baglin, A.G. Schrott and K.C. Lin, Appl. Phys. Lett., 51, 103 
d 987) 

C.A. Chang, Appl. Phys. Lett., 51,1236(1987) 

5. BJ.Tan, M. Fessehaie and S.L. Suib, Langmuir, 9,740 (1993) 

6 . S.L. Kaplan, E.S. Lopata and J. Smith, Surface and Interface Analysis, 20,331 (1993) 

7. D.R. Wheeler and S.V. Pepper, J. Vac. Sci. Technol., 20 (3), 442 (1982) 

8 . W.L. Perry, K.M. Chi, T. Kodas, M. Hampden-Smith and R. Rye, Appl. Surf. Sci., 

9. W. Jiang, M.G. Norton and J.T. Dickinson, Mat. Res. Symp. Proc., 304,97 (1993) 

J.T. Dickinson, J.J. Shin, W. Jiang and M.G. Norton, J. Appl. Phys., 74 (7), 472 (1993) 

10. M. Okoshi, M. Murahara and K. Toyoda, Mater. Res. Soc. Symp. Proc., 201,451 (1991) 

11. H. Niino and A. Yabe, Appl. Phys. Lett., 63 (25), 3527 (1993) 

12. H. Niino and Y. Yabe, Appl. Surf. Sc., 69,1-6 (1993) 

13. L.L. Gruss and F. Pearlstein, Plating and Surface Finishing, 70 (2), 47 (1983) 

14. Y. Okinaka, Plating, 58,914 (1970) 



IBM Corporation, Microelectronics Division, Endicott, NY 13760. 


UV/ozone cleaning is known to be effective for removing thin organic contaminants, 
but removal of silicon containing contaminants is questionable. Organo-silicon con¬ 
taminants, e.g., silicones, can result from a variety of integrated circuit chip and elec¬ 
tronic packaging fabrication processes. In this investigation, films of 
poly(dimethylsiloxane) (PDMS) on silicon substrates, with and without a gold coating, 
have been used to simulate such contamination up to a thickness of 50 nm. Although 
treatment consistently reduced the advancing Dl water contact angle, in some cases 
from a value greater than 100° to a value less than 5°, the hydrophilic nature of the 
treated surfaces was not due to complete contaminant removal, i.e., a significant amount 
of modified contaminant remained on the surface. High resolution x-ray photoelcctron 
spectroscopy (XPS) in the Si 2 p region suggest that O-Si-C bonds in the siloxane, ob¬ 
served prior to treatment, are converted to SiO x , where x is between 1.6 and 2. The time 
required to reduce the contact angle to a minimum value was greater for the thicker 
PDMS film samples. Deflection testing was used to evaluate the adhesion of an epoxy- 
based adhesive to intentionally-contaminated silicon chips, before and after UV/ozone 
treatment. Although PDMS contamination induced loss of adhesion between the chip 
and the adhesive, complete conversion to silicon oxides by UV/ozone treatment of con¬ 
taminants having a thickness of 5.0 nm has been demonstrated to restore adhesion to a 
value equivalent to that of uncontaminated silicon chip surfaces. 


Delamination of integrated circuit (IC) chips from polymer-based encapsulants and 
adhesives during accelerated reliability testing can result from contamination of the chip 
during device fabrication processes and wafer dicing. For example, residues of 
poly(dimethylsiloxane) (PDMS), a release agent on some dicing tapes, may be trans¬ 
ferred to the chips. Removal of such contaminants is imperative to assure reliably 
bonded interfaces [1]. 

Among the techniques that can be employed for removal of organic contaminants are 
plasma [2] and laser [3,4] processes. Plasmas used for this application are ionized gases 
at reduced pressures. Oxygen plasmas are most commonly used for cleaning of organics. 
Decomposition of organic materials is initiated by reaction with atomic oxygen from the 
plasma followed by desorption of volatile by-products. Organic removal can also be 
accomplished by exposure to laser radiation, particularly at wavelengths in the ultravi¬ 
olet (UV) region where most polymers are capable of absorbing radiation. Absorption 
is followed by polymer decomposition [3]. Both of these techniques utilize equipment 
that can require an appreciable capital investment. An alternative approach that uti¬ 
lizes reactive oxygen and UV radiation is UV/ozone cleaning. The apparatus for this 
technique is quite modest, usually consisting of a UV source, e.g., a low-pressure mer¬ 
cury vapor lamp, and a chamber to house the UV source and the sample(s). Cleaning 
is almost always performed in air at atmospheric pressure. In comparison with plasma 


Mat. Res. Soc. Symp. Proc. Vol. 385 c 1995 Materials Research Society 

and laser systems, UV/ozone cleaning equipment is relatively inexpensive. An overview 
addressing UV/ozone cleaning technology and its applications has been given by Vig [5]. 

The low pressure mercury vapor lamp with a quartz envelope emits strongly at two 
wavelengths, 184.9 nm and 253.7 nm. Oxygen molecules absorb strongly at 184.9 nm 
and dissociate to form atomic oxygen [5] that reacts with O 2 to form ozone. Ozone 
strongly absorbs at 253.7 nm and dissociates to form 0 2 and atomic oxygen. Both 
atomic oxygen and ozone can react readily with organic materials [6] forming polymer 
radicals. Absorption of UV radiation, if existent, can also lead to formation of polymer 
radicals, excited molecules, or, if the organic material's ionization potential is low 
enough, ions. Ultimately, the organic radicals react with atomic oxygen, molecular ox¬ 
ygen, or ozone to form low molecular weight, volatile fragments, like CO 2 and H 2 0, that 
desorb from the surface. UV/ozone cleaning processes are known to be effective for 
contaminants that are not too thick and its effectiveness is also limited to some types of 
organic materials, while removal of organo-silicon compounds is questionable. 

The effectiveness of a cleaning process can be evaluated by functional testing (c.g., 
adhesion measurements), chemical surface analysis (e.g., x-ray photoelectron 
spectroscopy [XPS]) and determination of changes in surface wetting (e.g., by contact 
angle measurements). Of course, functional tests are the most important measurement 
of the cleaning process, but these do not reveal whether the surface has been cleaned or 
simply modified, nor do they reveal the type of modification obtained. On the other 
hand, chemical analysis of the treated surface may indicate the type and degree of 
modification, but cannot always be correlated with adhesion. In addition, with most 
techniques of surface analysis, it is difficult to distinguish between silicon in the con¬ 
taminant and elemental silicon or silicon oxides in the underlying substrate. Contact 
angle measurements, the simplest of the techniques listed for evaluation of cleaning 
processes, can quantitatively reveal surface modification with monolayer sensitivity. 
The rate of reduction of contact angle with UV/ozone cleaning has been used to char¬ 
acterize the level of organic contamination on IC chips [!]. However, the wettability of 
a surface is not necessarily indicative of cleanliness, nor is it a predictor of good adhe¬ 
sion. Wetting and adhesion depend not only on surface contamination, but also on 
surface chemical composition (relative concentrations of atomic species), chemical 
bonding environments (e.g., oxidation and oxidation states), and roughness. In addition, 
the presence of a weak boundary layer layer that docs not adhere well to the bulk ma¬ 
terial, even though hydrophilic in nature, will typically exhibit poor practical adhesion 
to deposited materials. This paper reports on the effectiveness of UV/ozone cleaning in 
removing silicon-containing contaminants and the effect of the cleaning process on ad¬ 
hesion between IC chips and epoxy-based adhesives. 


Experiments were performed in a Uvocs, Inc., model T0606B UV/ozone cleaning 
system. The UV source is a 6" X 6" low-pressure mercury vapor grid lamp with a quartz 
envelope. Samples were placed 5.7 mm from the lamp envelope. 

Films of PDMS were spin-coated from solution onto gold-coated silicon wafers (pol¬ 
ished) having a diameter of 125 mm, using a Headway Research Inc. model 
EC101D-R485 photo-resist spinner. Spin speed was 4000 rpm for 30 seconds. Solutions 
were prepared by diluting Huls Petrarch Systems PDMS, silanol terminated 
(MW = 4200), with Burdick and Jackson high purity grade tetrahydrofuran (THF). 
Thickness of the PDMS film was controlled by varying the concentration, by volume, 
of the PDMS in the THF solvent. Concentrations in THF included 0.05, 0.1, 0.2, 0.4, 
0.6 and 2.0%. After spinning, films were dried by allowing the solvent to evaporate in 
air. Silicon wafers having a crystalline orientation of (100) were used so that smaller 


pieces of reproducible size and shape could be obtained by cleaving along orthogonal 
planes. Unless otherwise indicated, samples used for experiments with deposited con¬ 
taminants were about 1.0 cm square. Gold films on the silicon wafers had been vapor- 
deposited to a thickness of 78 nm. The gold barrier between the silicon wafer and the 
contaminant film allowed differentiation between the silicon in the contaminant and the 
silicon in the wafer during chemical analyses. Thin contaminant coatings were invisible 
to the unaided eye. Since both the thin contaminant films and the gold scratch easily, 
marks visible in the gold layer indicated areas of damage to the invisible contaminant 
coatings, so that these areas could be avoided during analyses. The gold surface also 
served as a reference layer for determining the thickness of PDMS films obtained from 
solutions having various concentrations in THF using Rutherford backscattering 
spectroscopy (RBS). The effects of UV/ozone treatment on the PDMS films were found 
to be more reproducible when the gold was cleaned for ten minutes in the UV/ozone 
chamber prior to spin-coating. Samples for adhesion testing were prepared by dipping 
TC chips into a 0.2% solution of PDMS in THF and drying in air. The chips were then 
bonded to a plate of nickel-plated copper using an epoxy-based adhesive. 

Advancing DI water contact angles on treated and untreated samples were measured 
with a Rame-Hart, Inc., model A-100 goniometer with optical protractor, using a sessile 
drop technique and drop volumes between 1 |d and 2 pi. For samples exposed to 
UV/ozone treatment, measurements were made immediately following treatment. Con¬ 
tact angles were recorded within 30 seconds from initial application of the drop. Since 
the PDMS thickness was uniform over the entire wafer to which it was applied, separate 
samples were used for each data point. Results obtained in this manner were found to 
be very consistent and reproducible. 

XPS analysis was performed in a PHI-5500 Multiprobe spectrometer equipped with 
a hemispherical analyzer using non-monochromatized AlK a rays for excitation. Survey 
and high-resolution spectra were collected with pass energies of 158 and 11.8 eV, re¬ 
spectively. Binding energies were referenced to the hydrocarbon peak at 284.6 eV. 
High-resolution XPS spectra in the C Is and Si 2 p regions were used to determine the 
contributions due to different chemical environments, and to follow them as a function 
of UV/ozone treatment. 

The RBS instrument used for measurement of PDMS film thickness consisted of a 
tandem accelerator to produce a He2 + ion beam of 2.1 McV. The typical geometry of 
this apparatus has been described elsewhere [7]. The thickness of PDMS films deposited 
at known concentrations (from 0.4 to 2.0%) in THF was estimated by the relative shift 
of the underlying gold layer, calculated using spectral simulation described by Doolittle 
[8]. Thickness at concentrations less than 0.4% were interpolated (including a data 
point at zero thickness for a concentration of zero percent). 

Adhesion of chips to the adhesive was measured using a deflection testing method. 
A force is applied at a controlled speed normal to and on the reverse side of a thin me¬ 
tallic plate to which the chip has been bonded. The plate is supported at its edges al¬ 
lowing it to deflect. The maximum force required to initiate separation is an indication 
of the adhesion at that interface. 


PDMS film thicknesses, measured using RBS, are shown as a function of the 
polymer's concentration in THF in Figure 1. The relationship between concentration 
and thickness is linear. The extrapolated value of thickness at 0.1% is near 2.5 nm. 

Samples were positioned such that the surface was 5.7 mm from the lamp and located 
within 2.5 cm of the geometrical center of the lamp. The rate of PDMS modification is 
proportional to the thickness of the contaminant as shown in Figure 2. The hydrophilic 


nature of the treated surfaces is not necessarily indicative of complete PDMS removal. 
Although removal of contaminants will typically result in reduced contact angles, mod¬ 
ification (like oxidation) of the contaminant films can have a similar effect, since 
oxidation of most polymer surfaces tends to increase the surface free energy. Mech¬ 
anisms leading to reductions in contact angles for samples with different thicknesses of 
PDMS contamination have been inferred from results of XPS analyses and will be dis¬ 
cussed in more detail below. 

Data are also shown for the as-received gold surface containing only adventitious 
contamination, or contamination that might result from the metallization process. Rate 
of reduction in contact angles for gold-coated samples washed in THF (not shown here) 
were nearly identical to that exhibited for as-received gold-coated samples, indicating 
that the solvent does not contribute to wafer contamination or cleaning rates. 

Atomic compositions obtained from XPS survey spectra are shown for the surfaces 
of the as-received gold, and for samples of PDMS having 2.5 nm thickness, before and 
after exposure to UV/ozone treatment for 10 minutes, in Table I. The gold surface ex¬ 
hibits some level of adventitious carbon and oxygen. In fact, contact angles on 
UV/ozone-cleaned gold surfaces have been observed to increase from 2° to 20° after 
sitting in air for 10 minutes following UV/ozone cleaning of the gold. The appearance 
of gold on the 2.5 nm thick PDMS-coated sample is consistent with the sampling depth 
of XPS, typically 3 to 5 nm. The concentration of carbon decreases following UV/ozone 
treatment, however a large increase is observed in the level of oxygen. Silicon is still 
detected. The signal due to gold does not increase significantly with treatment. These 
results suggest that although the UV/ozone treatment reduced the contact angle from a 
value greater than 100° to a value less than 5°, a significant amount of modified con¬ 
taminant remains on the surface. 

High resolution XPS spectra in the Si 2 p region are shown for PDMS samples before 
and after ten minute UV/ozone exposure in Figure 3. The peak at 102.4 eV, observed 
prior to treatment, can be attributed to O-Si-C bonds in the siloxane and the peak at 
103.8 eV, observed following treatment, corresponds to SiO x , where x is between 1.6 and 
2 [9]. For a thin PDMS film treated to a minimum contact angle, all of the silicon that 
remains on the surface is converted to an oxide. Because of the relative concentrations 
of silicon and oxygen shown in Table I, it appears that some oxygen must also be asso¬ 
ciated with carbon residues. UV/ozone treatment does not completely remove PDMS, 
but modifies the film considerably. 


0 2 4 6 8 10 12 14 

Treatment Time (min) 

Figure 1. Thicknesses of PDMS, ob- Figure 2. Contact angles for PDMS 
tained using RBS, as a function of con- Films of various thickness (concentration 
centration in THF. in THF) and as-received gold. 


Table I. Atomic percent compositions of selected test surfaces. 






As-received Au 





2.5 ntn PDMS, untreated 





2.5 nm PDMS, 10 minute treatment 





Table II. Deflection test peak force values (Newtons) for adhesion 
between intentionally-contaminated IC chips and adhesive. 


No UV/Ozone 

10 Min. UV/Ozone 

Uncontaminated Chip 



PDMS-Contaminated Chip (1) 



PDMS-Contaminated Chip (2) 



Conversion of organo-silicon materials to silicon oxides is a phenomenon that is well 
documented for exposure to oxygen plasma environments [1,10,11]. In particular, 
polysiloxanes have been known as negative e-beam resists since the early 1970s [10]. 
Oxygen reactive ion etching of silicon-containing polymers results in an initial thickness 
loss and a gradual slowing of polymer erosion until etching ceases. During etching, it 
is believed that silicon-containing monomer diffuses to the polymer surface where it is 
converted to Si0 2 and functions as an increasingly effective etching barrier [2]. 

Several IC chips were coated by dipping into solutions of 0.2% PDMS in THF. Of 
these, several received UV/ozone treatment. Samples were then bonded to nickel-plated 
copper plates and deflection testing was performed to measure the adhesion between the 
chip and the bonding adhesive. Samples that had not been coated with PDMS were also 
tested. Results are shown in Table II. PDMS contamination induced loss of adhesion 
between the chip and the adhesive. UV/ozone treatment restored adhesion to a value 
equivalent to that obtained for samples without contaminant. Hence, even though the 
contaminant was not removed, thorough conversion to oxides restored adhesion. 

Figure 3. High resolution 
XPS spectra in the Si 2 p 
region for PDMS samples 
before (a) and after ten 
minutes of UV/ozone ex¬ 
posure (b). 



UV/ozone cleaning has been shown to be effective in restoring adhesion between 
contaminated IC chips and an epoxy-based adhesive. Although complete removal of 
silicon-containing contamination was not possible, adhesion improvement was observed 
for films on which the organo-silicon contaminant is completely converted to silicon 
oxides and the original thickness of the contaminant was on the order of 5.0 nm. 


The authors acknowledge helpful discussions, insights, guidance and assistance re¬ 
lated to product, materials and equipment, and analyses provided by the following: R. 
Fusi, I. Memis, T. Wu, A. Quinn (all of IBM Corporation), C.I. DeJesus (Rensselaer 
Polytechnic Institute), and E. Lasky (Uvocs, Inc.). 


1. G.S. Ganesan, G. Lewis, and H.M. Berg, Int. J. Microcircuits Electron. Packag., 17 
(2), 152 (1994). 

2. F.D. Egitto, V.Vukanovic, and G.N. Taylor, in Plasma Deposition, Treatment, and 
Etching of Polymers , edited by R. d'Agostino, (Academic Press, Inc., San Diego, CA, 
1990), pp. 321-422, and references therein. 

3. R. Srinivasan and B. Braren, in Lasers in Polymer Science and Technology: Appli- 
cations, Volume III, edited by J.-P. Fouassicr and J.F. Rabck, (CRC Press, Inc. Boca 
Raton; FL, 1990), pp. 133-179. 

4. H.K. Park, C.P. Grigoropoulos, W.P. Leung, and A.C. Tam, IEEE Trans. Compon. 
Packag. Manuf. Technol. A, 17, 631 (1994). 

5. J.R. Vig, in Treatise on Clean Surface Technology , edited by K.L. Mittal, (Plenum 
Press, New York, 1987), pp. 1-26. 

6 . J.F. Rabek, in Comprehensive Chemical Kinetics: Volume 14, Degradation of 
Polymers, edited by C.H. Bamford and C.F.H. Tipper, (Elsevier Scientific Publishing 
Company, Amsterdam, The Netherlands, 1975), pp. 423-538. 

7. P.F. Green, C.J. Palmstrom, J.W. Mayer, and E.J. Kramer, Macromoleculcs, 18, 501 

8 . L.R. Doolittle, Nucl. Instrum. Methods, B9, 344 (1985). 

9. J. Li, J.W. Mayer, L.J. Matienzo, and F. Ernmi, Mater. Chem. Phys., 32 (4), 390 

10. E.D. Roberts, J. Electrochem. Soc., 120, 1716 (1973). 

11. M.J. Owen and P.J. Smith, J. Adhes. Sci. Technol., 8 (10), 1063 (1994). 



Alsheh, D., 57 
Amer, MaherS., 155 

Balazs, Anna C., 201 
Bargon, Joachim, 239 
Bauer, Barry J., 179 
Bike, Stacy G., 173 
Boerio, F.J., 125 
Booth, Glyn, 195 
Bruce, Frank A., 213 
Byrd, E., 57 

Chaput, Cyril, 49 
Chen, Fang, 3 
Cumpston, B.H., 103 
Czandema, A.W., 11 

Dickinson, J. Thomas, 221, 227 
Dietz, A., 233 
Dorgan, John R., 185 
Drevillon, B., 27 
Dwight, David W., 33 

Egitto, F.D., 245 

Fasolka, Michael, 201 
Fuerniss, S.J., 245 

Galvin, M., 117 
Gardner, Steven D., 195 
Gersappe, Dilip, 201 
Gu, Weiqun, 33 

Hasegawa, Masaki, 167 
He, Guoren, 195 
Herdt, G.C., 11 
Hipps, K.W., 221 
Hiraoka, Hiroyuki, 239 
Hirayama, S., 91 
Hunston, D.L., 131 

Israels, Rafel, 201 

Jackson, Catheryn L., 179 
Jensen, K.F., 103 
Jiang, Wenbiao, 227 
Jung, D.R., 11 

Kampe, Stephen L., 33 
Kaufmann, P.M., 43 
Kent, Michael S„ 137 
Kim, Dong K., 125 
Klages, C.-P., 233 
Kobayashi, Tadashi, 167 
Koczak, Michael J., 155 
Konstadinidis, K., 117 
Krasil'nikova, Larisa N., 147 

Lackritz, Hilary S., 3 
Langer, R., 43 
Latsch, Stefan, 239 
Libera, M., 65 
Lichtenstein, P. Ross, 33 
Liu, Da-Wei, 179 
Lu, Guo-Quan, 33 
Lu, J.P., 103 

Macturk, K.S., 131 
Madras, Cynthia, 71 
Maeda, Naomi, 167 
Matienzo, L.J., 245 
McDonough, W., 57 
McNamara, K.„ 43 
Miaoulis, loannis N., 71 
Mooney, D.J., 43 

Nakamura, Y., 91 
Nguyen, T., 57 
Norton, M. Grant, 227 

Opila, R., 117 

Pai-Panandiker, Rahool S., 185 
Papadimitrakopoulos, F., 117 
Patel, A., 65 

Pittman, Jr., Charles U., 195 
Pockelmann, R., 233 
Poncin-Epaillard, F., 27 
Pushpalal, G.K. Dinilprem, 167 

Rivard, Charles-Hilaire, 49 
Rostaing, J.C., 27 
Rucker, Derek 173 

Sano, K., 43 

Saunders, Randall S., 137 
Schadler, LindaS., 155 
Schultheisz, C.R., 131 
Schutte, C.L., 131 
Seiler, J., 57 
Selmani, Amine, 49 
Sevick, Edith M., 213 
Singamsetty, C., 195 
Spalik, J., 245 
Subrahmanyan, Suchitra, 3 
Suzuki, Y., 91 

Takata, Tomonori, 167 
Tarlov, M.J., 131 
Tchouppina, Svetlana V., 147 
Theil, Jeremy A., 97 
Tsai, Y.M., 125 

Vacanti, J.P., 43 
Vallon, S., 27 


Wang, Lichang, 195 
Watanabe, Y., 91 
Weber, A., 233 
Wentworth, S., 65 
Williams, David R.M., 213 
Willis, B.G., 103 
Wong, Peter Y., 71 
Wu, Biahua, 195 

Wu, H. Felix, 33 
Wu, P.K., 79 
Wu, Zhihong, 195 

Yahia, L'Hocine, 49 

Zukas, W., 65 



abrasion, 221 
acrolein, 3 
activation energy, 71 
adhesion, 27, 33, 57, 79, 245 
strength, 91 

adhesive block copolymers, 137 
adsorption, 185 
alumina substrate, 91 
aluminum, 117 

atomic force microscopy, 221 

band bending, 117 
block copolymers, 137 

calcium aluminates, 167 
carbon fiber surfaces, 195 
cement paste, 167 
chemical vapor deposition, 233 
composites, 57, 131 
contamination, 245 
copolymer sequence distribution, 201 
copolymeric monolayers, 185 
copper, 239 

coupling agents, 131, 185 

differential scanning calorimetry, 65, 167 
durability, 131 

elastomeric polyurethane layers, 195 
electroplating, 233 
ellipsometry, 27, 185 
environmental interface, 155 
epichlorohydrin, 195 


matrix interface, 33 
reinforced composites, 33 
filtration, 213 
flexural strength, 167 
fluoropolymers, 233, 239 
Fourier transform infrared spectroscopy, 103 

gas phase, 3 
gold, 239 

hexamethyl disiloxane, 97 
hydrophylic groups, 239 

integrated circuit, 245 

interface, 79, 117, 131 

interfacial water, 57 

interlaminar shear strength, 195 

interpenetrating polymer networks, 179 

interphase, 65, 125 


modulus, 195 

mechanical properties, 147 

metal/composite interface, 147 
metal/polymer systems, 79 
model rubber compound, 125 
Monte Carlo computer simulations, 201 

nickel, 239 

nondestructive testing, 33 
nuclear magnetic resonance, 27 

optical ceramic, 71 

organosilicate polymeric composite, 147 
oxygen plasma, 91 

phenol resin precursor, 167 
photopolymerization, 3 
plasma, 27, 233 

enhanced chemical vapor deposition, 97 
polymerized primers, 125 
poly(dimethyl phenyl siloxane), 147 
poly(dimethyl siloxane), 245 
poly(methyI methacrylate), 173 
poIy(phenylene vinylene), 103, 117 
poly(tetra fluoroethylene), 227, 233, 239 
polycarbonate, 27 
polyimide, 103 
thin film, 91 
bond, 71 

brushes, 201, 213 
coating, 57 

interactions, 11, 103 
interfaces, 11 

polyurethane elastomers, 195 
pore diffusion, 213 
pull test, 91 

pulsed-laser deposition, 227 

Raman spectroscopy, 155 

relaxation time constant, 71 

ring opening metathesis polymerization, 137 

rubber to metal bonding, 125 

scanning conduction microscopy, 221 
self-assembled monolayers, 11 
self-consistent field calculations, 201 
siliceous substrate, 57 
silicon dioxide, 97, 173 
SIMS, 239 

single fiber composite, 155 
step coverage, 97 
sticking probability, 97 
stress relaxation, 71 

energetics, 185 

second harmonic generation, 3 

tetraethylene pentamine, 195 


thin films, 221, 227 
titanium nitride, 233 

transmission electron microscopy, 65, 227 
transverse tensile strength, 33 
tribology, 221 

vibration damping, 33 

water, 131 
wear, 221 

XPS, 79, 117,239 

XPS/1SS of metal-polymer interfaces, 11