Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "Physics of self-sustained oscillations in the positive glow corona"

See other formats


Physics of self-sustained oscillations in the positive glow corona 



(N 

o 

(N 

00 



Oh 
I 



C/2 

o 

Oh 



> 

m 

rn 
O 
(N 



X 



Sung Nae ChcQ 

Micro Devices Group, Micro Systems Laboratory, Samsung Advanced Institute of Technology, 
Samsung Electronics Co., Ltd, Mt. 14-1 Nongseo-dong, 
Giheung-gu, Yongin-si, Gyeonggi-do 446-712, Republic of Korea. 
(Dated: 5 July 2012) 

The physics of self-sustained oscillations in the phenomenon of positive glow corona is presented. The dynam- 
ics of charged-particle oscillation under static electric field has been briefly outlined; and, the resulting self- 
sustained current oscillations in the electrodes have been compared with the measurements from the positive 
glow corona experiments. The profile of self-sustained electrode current oscillations predicted by the presented 
theory qualitatively agrees with the experimental measurements. For instance, the experimentally observed 
saw-tooth shaped electrode current pulses are reproduced by the presented theory. Further, the theory correctly 
predicts the pulses of radiation accompanying the abrupt rises in the saw-tooth shaped current oscillations, as 
verified from the various glow corona experiments. 



I. INTRODUCTION 

Consider a point particle with charge ^ > near the con- 
ducting sphere of radius a and fixed at potential V > 0, which 
is illustrated in Fig. [T] The force acting on the charged point 
particle is given b>^ 



F = ^fl 



q 



r' A%eo{r^-a^) 



R, 



(1) 



where R is the particle's position vector, £„ is the vacuum per- 
mittivity, and r is the radial length from sphere's center. In Eq. 
([T]i, the force becomes negative in the limit r approaches a and 
becomes positive in the limit r goes to infinity. Between these 
two limits, there is a point of unstable equilibrium at which 
the force vanishes. Such location is identified by r = a + //) 
in Fig. [T] which point represents the borderline between re- 
gions C and D. In region C, the particle is repulsed from the 
sphere whereas, in region D, the particle is attracted to the 
sphere. Why? Well, nothing surprising here. The positive 
point particle induces negative charges at the sphere's surface; 
and, the force between the two is always attractive. This at- 
tractive force dominates in region D, and the point particle is 
attracted to sphere there. 

The same physics can be applied to describe the behavior 
of a charged point particle between the plane-parallel plates, 
which is illustrated in Fig. |2] At distances close to the anode, 
the charged point particle is attracted to the anode's surface 
whereas, for all other distances between the plates, the par- 
ticle is repulsed in the direction of the parallel plate electric 
field, Ep, which field is present even in the absence of the 
charged-paiticle. Consequently, the space between the plates 
is divided into regions C and D, where the location of unstable 
equilibrium is at distance lo from the surface of the anode, as 
indicated in Fig. |2] 

In both cases, the charged-particle dynamics is pretty bor- 
ing. The charged point particle ends up adhering to the sur- 
face of the anode when it is in region D whereas, when it is in 
region C, the particle gets repulsed in the direction of the ap- 
plied electric field. The problem becomes interesting when a 
charged-paiticle with structure is considered. Unlike the point 
particle, the structured particle can be polarized under exter- 



Region D 



Region C 




Cathode is grounded at infinity 

Figure 1: (Color online) Charged point particle near a conducting 
sphere of radius a, fixed at voltage V > 0. In region D, the particle is 
attracted to the sphere whereas, in region C, the particle is repulsed 
from the sphere. The electric field, E, is in the radially outward di- 
rection. 



nally applied electric field, such as E and E^ in Figs. [T]and 
|2] respectively. The resulting depolarization field formed in- 
side of the structured particle redistributes the negative bound 
charges to the particle's upper hemisphere surface and leaves 
the particle's lower hemisphere surface depleted of the nega- 
tive bound charges. Such redistribution of the bound charges 
inside of the structured particle is schematically illustrated in 
Figs. [3ja) and[3lb). As a consequence of the depolarization 
field inside of the structured particle, the particle is repulsed 
from the surface of anode in region A due to the Coulomb re- 
pulsion arising between the negative charges induced at the 
anode's surface and the negative bound charges formed at the 
surface of the particle's upper hemisphere, as illustrated in 
Fig. Oa). Such repulsive force decays rapidly outside of the 
region A. In region B, the force acting on the particle is domi- 
nated by the Coulomb attraction between the particle's excess 
positive charge, q> 0, and the negative charges induced by 
it at the surface of the anode. This force is eventually over- 




• q>0 
lc> Id \ jr^ 

Region C 

E„ 



Cathode 



Vr 



V, 



be eliminated by setting 02 = OC • m^^. The K^ and Kt, rep- 
resent the dielectric constants of the insulating shell and the 
space between the plates, respectively. 

The potential in regions Mi, M2, and M3 of Fig. |4]are ob- 
tained by solving the Laplace equation, V^V =0, with appro- 
priate boundary conditions. Solutions Vi , V2, and V3 are given 
byi 



Vi=VL + a+Ep{h-s), r<a, 



V2{r,e) = VL + P+Epih-s + jrcosO) 

X o^yEpCo&d 
r r^ 



a < r <b, 



Vi(r,e) ^VL + Ep{h- s + rcos.B) 
[y {b'^ - a^)-b^]Ep COS. e 



(2) 



(3) 



-C, r>b, (4) 



Figure 2: (Color online) Charged point particle inside a plane- 
parallel conductors separated by h. In region D, the particle is at- 
tracted to the anode whereas, in region C, the particle is attracted to 
the cathode. 



where C is a constant, is spherical polar angle defined in 
Fig. m r is radial length, Ep is parallel-plate electric field. 



E„ 



Vt-V,, 



-, Vr>Vi; 



(5) 



whelmed by the Coulomb repulsion in region A, and the whole 
process gets repeated, thereby resulting in a charged-particle 
oscillation in region D. 

The discussed charged-particle oscillation mechanism is 
intrinsic to the phenomenon of glow corona. ^~^ The self- 
sustained pulsing in the positive glow corona (also referred 
to as DC glow discharge) can be qualitatively explained from 
the aforementioned, oscillating, charged-particle dynamics. 
In this paper, I shall show that the self-sustained pulsing in 
the positive glow corona involves the kind of charged-particle 
oscillation mechanism discussed in Figs. |3ja) and |3jb). To 
accomplish this, I shall, first, briefly outline the dynamics of 
charged-particle oscillation under constant electric field. The 
result is then used to predict the current oscillations in the 
electrodes. This prediction of electrode current oscillations is 
compared with the results from the various glow corona ex- 
periments. 



II. CHARGED-PARTICLE OSCILLATION IN CONSTANT 
ELECTRIC FIELD 

The dynamics of charged-particle oscillation is discussed 
by considering a model configuration illustrated in Fig. 
where a core-shell structured particle has a conductive core of 
radius a and an insulating shell of thickness b — a. The con- 
ductor core has a surface free charge density of a\ and the in- 
sulator has a surface charge density of 02 ■ The surface charge 
density on the insulator has been introduced purely for math- 
ematical generalization. In the final expression, this term can 



and, constants a, j3, 7, A, and v are defined as 



a = 



'fl)(7i a^a\+b^a2 



/3 = 



teoK-2 beQK-i 

a{2b-a)o\ a^O\+b^a2 



bSQK2 



beoKi 



3K3b^ 



{K2 + 2K3)b^+2{K2-K3)a^'' 



(6) 



2a{b — a)ai a^(7i+b^a2 



eoK-2 



eoK-3 



The electric displacement in region M3 is obtained by com- 
puting 



D3(r,0) = -eoK-3Vy3(r,0), 



where 



d Id 



1 



dr r do rsin0 d(j) ' 

Cr = e;r sin cos ^ + Cy sin OsiiKJ) + 67 cos , 
eg = e.v cos 9cos(p + Cj cos sin ^ — e, sin 9 , 
e,j, — —exsm(j) + Gycos^ . 



(a) 



(b) 



c 
g 
'en 



I Coulomb repulsion 

Region A E 



Anode 



^ 



° I 

5> ''D 



m 



q>0 



O 

o 

X3 



E. 



Id > 2h 
Region B 



O 



Ia 




'■D l-A 



Region C 
Cathode is at the below (not shown here) 

Figure 3: (Color online) (a) In case of a structured charged-particle, the region D in Fig. [2]is further divided into two regions A and B. The 
particle is repulsed from the anode in region A due to the Coulomb repulsion between the image charge at the anode's surface and the negative 
bound charges formed at the surface of particle's upper hemisphere. The surface of charged spherical particle is an equipotential surface; and, 
any excess surface charges there are uniformly distributed over it. Without confusion, the contributions from such excess charges, unifoiTnly 
distributed over the surface of particle, are indicated by a big "+" symbol at the particle's center to distinguish them from the positive bound 
charges associated with the induced depolarization field, which are indicated by small "+" at the particle's lower surface, (b) The polarized 
charged-particle is attracted to the anode in region B due to the Coulomb attraction arising between the particle's excess charge q and the image 
(or induced) charge associated with it at the anode's surface. The width of region A is identified by Ia and the width of region B is given by 
Id — Iaj where Ip is the borderline between regions D and C. 



It can be shown that e, component of D (r, 0) in region M3, 
evaluated at the anode's surface, is^ 



J)3--,{x,y,s) ^e^eoK-s 



'3[y{b^~a^)-b^]EpS^ 



^ (x2+3;2+.92)V2 

(x2+3;2 +,2)3/2 PJ' 

At the anode's surface, the e^ component ofD{x,y,z) suffers 
a discontinuity, 

and, the surface charge density, there, is 



in region M3, evaluated at the cathode's surface, is 

3[y{b^-a^)-b^]Ep{s-hf 



:ezeoK-3 _ 

x^+y^ + {s-h) 

v{s-h)~[Y{b^-a^)-b^] Ep 



5/2 



x'^+y^ + {s-hY 



3/2 



At the cathode's surface, the 67 component of D {x,y,z) suffers 
a discontinuity, 

ez-^3;z(x,y,s-h) = aiip; 

and, there, the surface charge density is 



(yiup{p,s) = -eoK-3 



3[Y{b^-a^)-b^]EpS^ 



, (p2 + .2)5/2 

vs- [Y{b^-a'^)-b^]Ep 

(p2 + .2)3/2 



Eph 



wherep = y^x^ +y^. Similarly, the e,- component of D(x,3',z) 



CJ,/p(p,s)=eoK-3' 



3[Y{b^-a^)-b^]Ep{h-sf 

Tfi 



p^ + {h-sf 
v{h-s)+ [Y{b^-a'^)-b^]E^ 



p^ + ih-sf 



3/2 



-E„ 




e^ is out of the page 

Figure 4: (Color online) Configuration showing a core-shell struc- 
tured charged-particle between a DC voltage biased plane-parallel 
conductors. 



The force acting on the particle is 



Ft- = Fi + F2 — e^mg, 



(7) 



where m is particle's mass, ^ = 9.8 m ■ s ^ is gravitational con- 
stant; and, Fi and F2 are 



-CzTtVS 



P GiupPldpi 



(p2 + .2) 



3/2' 



F2 = e,7zv {h - s) / 
Jo 



(yupPidpi 



pl + {h-s) 



3/2' 



Here, F 1 is the force between the charged-particle and the im- 
age charge induced by it on the surface of the anode. Sim- 
ilarly, F2 is the force between the charged-particle and the 
charge it induces at the cathode's surface. For the parallel 
plate system which is microscopically large, but macroscopi- 
cally small, Eq. (ITJ becomes^ 

4 1^2 [h^sf s^ 



(8) 



where 



[h-s) I 



m^-K[a^ (Pm,l - Pm,2)+b^Pm,2\ 



with p,„ 1 and p,„ 2 representing mass densities of the particle's 
core and shell, respectively. 



The force Fj- {s) of Eq. ([8]l is plotted using the following 
parameter values: 

K-2 = 6, K-3 = l, 

0=1.5 /im, h=\ mm, 

b — a = 4nm, 

Vl = OV, (9) 

O2 = OC • m^2 (i e., insulator), 

p,„,i= 2700kg -m-^, 

p,„,2 = 3800kg -m-^, 

where the choice of 4nm for the shell's thickness and a dielec- 
tric constant of A:2 ^ 6 are typical for alumina nanoparticles. ^1^ 
To compare the cases which involve the positively and the 
negatively charged particles, the surface charge densities of 
(7i = 0.012c • mr^ and 0\ = -0.012C • roT^ have been con- 
sidered. For the force dependence on the applied parallel 
plate electric fields, VV = OV and Vj — 500V have been con- 
sidered for comparison. The VV = OV corresponds to the 
case where a charged-particle is placed between the grounded 
plane-parallel plates. With V^ = OV and h = 1mm, the an- 
ode voltage of Vt = 500 V corresponds to the parallel plate 
electric field of Ep — 500 kV • m^' . The results are shown in 
Fig. |5ja). As expected, when both plates are grounded, the 
charged-particle is attracted to the nearest grounded plate. For 
instance, in the region where s < 0.5 mm, the force Fj- (s) is 
positive whereas, for s > 0.5 mm, the force becomes nega- 
tive. Right at the midway between the plates, i.e., s ~ h/2, 
the net force on the charged-particle vanishes. Why? This 
is because, at s = h/2, the particle is equal distance away 
from the surfaces of the anode and the cathode plates; and, 
in such situation, the charges of opposite polarity are induced 
at the surfaces of the anode and the cathode with equal magni- 
tudes. Finally, the presence of the parallel plate electric field, 
||Ep|| > 0, offsets the force to the negative axis for positively 
charged particles. The resulting offset in the force gives rise 
to the oscillatory behavior of the positively charged particle in 
vicinity of the anode. 

What happens to the force. Ft (s) , when the particle is 
negatively charged? Equation (jSJ has been plotted using the 
same parameter values defined in Eq. (|9) with the charge den- 
sity of (7i — — 0.012 c ■ m^2 for a negatively charged particle. 



The results are shown in Fig. |5jb), where Ep — OkV • m ' 
and Ep = 500 kV • m^^ have been considered for compari- 
son. As with the case of positively charge particle, the neg- 
atively charged particle is attracted to the closest electrode 
when Ep = OkV • m^'. At the midway between the plates, 
i.e., s — h/2, the negatively charged particle feels no net 
force due to the fact that charges of opposite polarity are in- 
duced at the facing surfaces of each plates with equal magni- 
tude. At the presence of the parallel plate electric field, i.e., 
Ep = 500 kV • m^ ' in Fig. |3b), the force offsets to the positive 
axis. The resulting offset in the force gives rise to the oscilla- 
tory behavior of the negatively charged particle in vicinity of 
the cathode. 

The physical charged-particle with structure cannot pene- 
trate into the surfaces of the anode and the cathode plates, of 
course. Therefore, the parameter s in Figs. [Sfa) andjSjb) are 
restricted to a domain b < s < h~ b, where b is the charged- 



(a) 



z 



o 

Oh 



0.2 

0.15 

0.1 

0.05 



-0.05 

-0.1 

-0.15 

-0.2 



" 


17 n l^\7/rv^ 


' 


p -''son kV/m 










iV. 






\ 




^A 


v_ 










■ \ 









0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 
Position, s [mm] 

(b) 





0.2 




0.15 


z 


0.1 

0.05 




o 

y 


-0.05 


^ 


-0.1 




-0.15 




-0.2 



\ 














\ 

■ X 














\ 












""• N 


.V 












■ ', 












"~A', 














: : 1 




,^i 


= 
son 


kV/ir 
kV/ir 


1 — 





:...- 




h 






i i i ll 



0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 
Position, s [mm] 

Figure 5: (Color online) Plot of Ft (s) , Eq. (d, for Ep = OisV ■ m"' 
(a) The case of a positively charged parti- 



and Ep = 500kV-m-i 



,-2 



cle, where C7i = 0.012C ■ m .(b) The case of a negatively charged 



-2 



particle, where Oi = — 0.012C -mT . In both (a) and (b), all other 
parameter values are as defined in Eq. (|9). The anode is located at 
the left side. 



particle's radius and h is the gap between the anode and the 
cathode plates. The case of s — b corresponds to the situa- 
tion where the particle is in contact with the anode's surface; 
and, the case of s = h — b corresponds to the situation where 
the particle is in contact with the surface of the cathode. For 
this reason, the magnitude of the force acting on the charged- 
particle remains finite sX. s = b and s = h — bm Figs. |3a) and 
Hb). 



A. Positively charged particle with structure 

1. Constituent forces 

It is worthwhile to discuss the constituent forces of Fj-. 
When particle is sufficiently charged, the gravitational force 
becomes negligible; and, Eq. ^ can be expressed as 

Fr = f 1 1 + f 1 2 + fl 3 + f2 1 + f2 2 + f2 3i 

^j — ^ -^ — ,; > 

Fi F2 



where the constituent forces of Fj are 

,2 



fl.l =^z 



fl.2 = e,-: 



7r£oK-3V"- 

n£i^Kiv[Y{b^-a^)-b^]Ep 
4*3 



fi3 = -e:,n£(iK^vEp\ 
and for F2 , its constituent forces take on the form given by 



f 



2,1 



f2.2 = Cj 



'4{h-sf 
neoK3v[Y{b^-a^)-b^]Ep 



A{h-sy 
f23 = -e.,TzeoKT,vEp. 

Since v > and 1 > 7 > 0, one finds 

yip^ -a^)-b^ <0; 

and, the previous constituent force terms of Fi behave as 



fi,i 
whereas for F2 

f2,l- 



1 „ 

Cz^, fl.2 






1.3 



C^iS n 



1 



I2,2 ■ 



f 



2,3 



-e^Ep. 



{h-sY ' {h-sY 

Physically, fij (f2,i) represents an attractive force between 
charged particle and its image charge at the surface of an- 
ode (cathode). The f 1 3 (f2,3) represents the usual force on 
charged-paiticle by parallel plate electric field, Ep . The other 
force term, fi_2 (f2,2), arises as a consequence of a structured 
particle that becomes polarized under Ep. Such force vanishes 
in the absence Ep. 

So, what gives rise to charged-particle oscillation? The 
F2 cannot generate oscillations because f2,i, f2.2, and f2,3 
are all directed in the same direction. The Fi , on the other 
hand, contains constituent forces with opposite directions 
that compete one another; and, such terms generate oscil- 
lations. For instance, at distances very close to the anode, 
Fi « fi,2 '^ —tvEpS^^; and, such force is responsible for the 
Coulomb repulsion in region A of Fig. Oa). This force de- 
cays rapidly with distance. Consequently, outside of the re- 
gion A, the contributions from f 1 2 become negligible. In re- 
gion B, e.g.. Fig. Ob), the force on the particle is dominated 
by Fi « fi.i '^ e^i^2 This force attracts the charged-particle 
back to the anode. It is this "push-pull" competition between 
fl.l and fi.2 that gives rise to charged-paiticle oscillation in 
region D, as illustrated in Fig. |6] 

The width Ia of region A increases with Ep as a conse- 
quence of fi,2 ~ —t,EpS^^. If region A widens by A/^, the 
width of region B decreases by the same amount. This is 
because the width Id of region D remains fixed for a given 
charged-paiticle of constant excess charge q. Consequently, 
the particle's oscillation frequency increases with Ep in region 
D. Such is schematically illustrated in Fig. |6] The new path 1 , 
which is the plot of Zd{t) versus time corresponding to the 
case of higher Zip, has a higher oscillation frequency than the 
old path 1 . 



Old path 1 




Regions Fj~ fji New path 1 




Ft = F2 



Region C 



Figure 6: (Color online) Schematic illustration of positively charged, 
structured, particle oscillating in vicinity of the anode. The dominant 
constituent force terms are shown in regions A and B. No oscillation 
mode exists near the cathode, region C, for a positive particle. The 
old path 1, new path 1, and the path 2 represent the schematic plot of 
Zd {t) versus time, where time is the horizontal axis. The new path 1 
corresponds to a case where Ep is higher than the one in old path 1. 



2. Potential energy 

By definition, the force F is defined as the negative gradient 
of the potential energy function U, 

F = -Vf/. 

For a one dimensional force, such as the one in Eq. dH), this 
implies 



(a) 



(b) 



r , s / — Positive particle 



• m 



2i 

a> I 



E„ 



,-./2'*ref 



Q. 



'"l' *>-e 



.® 



0) 



0) 

c 



Negative particle 



ryS 



Figure 7: (Color online) Illustrates the integral path in a line integral 
for a given parallel plate electric field Ep . (a) The case of positively 
charged particle, (b) The case of negatively charged particle. 



That said, Eq. ^ is inserted for Fj- in Eq. jTOl i to yield 

;reoK-3V 



U{s] 



V V 

72 



(/,-.r 



+ 



{y{h'^-a"')-b"'\Ep 
mg / (is, 

"'■Vref 



- 8£p \ ds 



(11) 



where Ep > 0. Equation (fTTI) is evaluated utilizing the follow- 
ing integral formulas; 



FrW 



dU_ 

' ds 



where U = U {s) . The one dimensional potential energy func- 
tion, U (s) , becomes 



Uis) 



Ft ■ Cyds, 



(10) 



■«ref 



ds ~ s — s, 



ref 5 



■Sref 



^ 1 _ 1 1 



Sref ■ 



Smf S 



where Si-^f is the reference point in which U (si-ef) <U {s) . For 
instance, at the presence of the parallel plate electric field, Ep, 
the potential energy of a positively charged particle may be 
computed by integrating the line integral along the path illus- 
trated in Fig. |3a). On the other hand, the potential energy 
for a negatively charged particle subjected to the same paral- 
lel plate electric field can be computed by integrating the line 
integral along the path illustrated in Fig. lUb). Consequently, 
the presence of electric field, Ep, as well as its directions and 
the polarity of the charged-particle affect the choice of .s^ief in 
the line integral of Eq. (fTOl i. For such reason, I shall only 
consider the case where 1 1 Ep 1 1 > for the evaluation of U (s) . 



' 24, 2.2' 



■«ref ■ 



^ 1 11 

ds = 



iref (h-sf '" h-s h-s,ef' 



rds 



s,,iih~sf 2{h-sf 2{h-s,,ff' 



The result is 



Uis) 



4 ] s h — s 2s^ 

[Y{b^~a^)~b^]Ep ^ 



2{h- s) 



- ^EpS > + mgs 



TteoK^v j V V ^ [Y{b^ -a^) -b^]Ep 

4 I Si-ef h - Sref 

[y{b^-a^)-b^]Ep 



2^2 



2{h- iref ) 



8EpSref>-mgs,ef, (12) 



where Ep > 0. The parameter s is related to the parameter Zd, 
which is defined in Fig. H] by 

s = Zd + b and s = Zd- (13) 

Utihzing this definition, Eq. ( fT2b can be rewritten as 

4 [Zd + b h-Zd — b 
[y{b^-a^)-b^]Ep [y{b^-a^)-b'^]Ep 
2{zd + bf 2{h-zd-bf 



+%Ep{zd + b) \+mg{zd+b) 
Tze^KiV \ V V [yip^ -a^) -b^\Ep 



4 1^ iref h - iref 

2 (/! - iref)^ 



2?2 



Sfipio ? - mgS^ef, 



where Ep> Q.\ shall set iref at the midway between the par- 
allel plates, 



and the U (zd) becomes 



(14) 



Uizd) 



4 [zd+b h-Zd-b 
[y{b^-a^)-b^]Ep [y{b^-a^)-b^]Ep 
^ 2(zd + bf 2(h~zd~bf 

+8Ep{zd+b) \+mg{zd+b) 



—^[v + Eph^)~-mgh, 



(15) 



where Ep > 0; and, the explicit expressions for Ep, y, and v 
are defined in Eqs. ^ and (|6]i: 

Ep = ^—^, Vt>Vl; 
n 

3K3b^ 



V = 



{K2 + 2K3)b^+2{K2-K3)ai' 

2a{b — a)a\ a^Oi+b^Oi 



Equation ( fTsT i is plotted for a positively charged particle in 
which (7i = 0.012C • m^^ and all other parameter values are 
same as defined in Eq. (|9]l. 



K-2=6, K-3 = l, 

a=1.5^m, h—\mm, 

b — a = 4nm, 

Vl = OV, 

C72 = OC • roT^ (i.e., insulator), 

p„,i=2700kg-m-3, 

p,„,2 = 3800kg-m-3. 



For the plot, the anode voltages of Vj = IkV, Vt — 2kV, 
and Vt = 3kV are considered for comparison. For Vl = OV 
and h = I mm, these anode voltages correspond to the par- 
allel plate electric field strengths of Ep = 1 MV • m^\ Ep = 
2MV • m^', and Ep = 3MV • m^^, respectively. The results 
are shown in Figs. [HJa) and[8jb), where the "potential well" 
corresponding to each Ep are formed near the anode. The 
depth of the "potential well," wherein the charged-particle can 
have oscillatory solutions, depends on the magnitude of the 
applied parallel plate electric field. Because the physical par- 
ticle cannot penetrate into the surface of the anode, the param- 
eter Zd cannot be negative valued. When Zd — 0/xm, the parti- 
cle is right on the anode's surface; and, this restricts the height 
of the potential well at Zd — 0/xm to a finite value, which can 
be verified from Fig. |8jb). In the case of Ep = \ MV • m^' , 
the width of the potential well is approximately ^ 250 /im 
and the particle is restricted to 0/xm < z^ < 250 /im for os- 
cillations. The width of the potential well decreases with the 
applied parallel plate electric field. For instance, in the case 
of Zip — 2MV • m^^ , the potential well width is approximately 
'^ 125 jim and the particle is restricted to jUm < Zd ^125 /im 
for oscillations. Physically, this corresponds to the narrowing 
of the positive glow region with increased Ep. 



3. Dynamics in nonrelativistic regime 

The particle's equation of motion, in the nonrelativistic 
limit, is obtained by solving 



e.^ms — Fj-, 



(16) 



where s = cfis/df- is the particle's acceleration. Insertion of 
Eq. ^ for ¥t in Eq. ( fT6l ) yields 



ne^K-iV I V 



yib'-a'^-b'lEp 



4m [s2 {h_sf 



eoK-2 



eoK'3 



\y(b^-a^)-b^]Ep 1 



8 



0.30 
0.25 
0.20 
0.15 
0.10 
0.05 
0.00 
-0.05 
-0.10 
-0.15 
-0.20 
-0.25 
-0.30 



^ -0.02 n 

£; -0.04 -; 

i? -0.06 -1 

„ -0.08 -: 

£? -0.10 - 

g -0.12 - 

1 -0.14 - 

2 -0.16 - 

o 

°- -0.18 Lj 



(a) 



1 ! 1 ! ! 
























































,-.-^-' 








--^ 
























/.' 


1 \;l\//.-vi 






T ■ F -^ MV/m 


.: 


i'--'" ■ r ^ \JI\//rv^ 


., 




/' 






, 



0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 
Position, Z.I [mm] 



(b) 



E = 1 MV/m 
gp = 2 MV/m 



50 100 150 200 

Po.sition, Zj [|J.ni] 



250 



300 



Vt = IkV and Vt = 2kV corresponds to parallel plate elec- 
tric field sti-engths of Ep = I M V • m" ' and fi^ = 2 M V • m" \ 
respectively. The results are shown in Fig. |9] The dependence 
of oscillation frequency on Ep is consistent with the argument 
discussed previously in Fig. |6] 




2 4 6 8 10 12 14 16 18 20 
Time, / [|ls] 

Figure 9: (Color online) Oscillating positively charged nonrelativis- 
tic particle with structure in vicinity of the anode. The plot of zd (?) 
has been obtained from Eq. dlVt using the initial conditions speci- 
fied in Eq. l |18t and the parameter values specified in Eq. (|9) with 
CTi — 0.012C • m^ . In the plot, the anode is located at Zd = Oflm. 



Figure 8: (Color online) (a) Plot of the potential energy function, 
U (zd) of Eq. ([BJ, for £,, = 1 MV -m"' , fip = 2MV ■ m"' , and Ep = 
3MV ■ m^' . The particle has a charge density of CTi = 0.012C ■ m^^ 
and all other parameter values are as defined in Eq. (|9]. (b) Enlarged 
plot of U (zd) for domain < Zd < h/2, where h= I mm. In the plot, 
the anode is located at Zd = Oflm. 



where e^ has been dropped for convenience. Utilizing the re- 
lations in Eq. (fTsl i. this can be rewritten as 



Zd ^ 



neoKiV 



Am 


\{Zd + bf 

-a')-b']Ep 


[Zd + bf 

'y{b^-a^)-b^]Ep 



{h-zd-bf 



{h - Zd 



8£„ 



(17) 



Equation (fTTT i is solved via Runge-Kutta method. For the pa- 
rameter values, I shall use the same values specified in Eq. 
(|9]l with (7i — 0.012C ■ m^^. For the initial conditions, I shall 
choose 



Zrf (0)^30 Aim and 2^(0) =0. 



(18) 



The initial condition for the particle's position has been cho- 
sen from the consideration of the potential energy function 
illustrated in Fig. |8j for instance, Zd{0) ~ 30/xm is within 
the potential well illustrated in Fig. |8jb)- The anode voltages 
of Vj- = 1 kV and Vj ~ 2 kV have been considered for com- 
parison. For Vl — OV and h — 1mm, the anode voltages of 



4. Dynamics in relativistic regime 

In the relativistic generalization, the equation of motion for 
the charged-particle is obtained by solving 



d 
dt 



1- 



'.^2 



(19) 



where c = 3 x lO^m • s^' is the speed of light in vacuum and 
s — ds/dt is the charged-particle's speed. Insertion of Eq. (|8) 
for F7 in Eq. ( fT9b . and after some rearrangements, yields^ 



■2 \ 3/2 / I 






4m 



s"- {h-sf 



[y{b'^-a^)-b^]Ep 

[Y{b'^-a^)-b^]Ep 
[h-sf 




(20) 



where e, has been dropped for convenience. The Zd parameter 
defined in Fig. |4]is related to the parameter s by 

s = Zd + b, s = Zd, s = Zd; 



and the expression in Eq. (|20] | becomes 



Zd 



c^l I 4m \{za+bf 






[y{b^-a^)-b^]Ep 



{h-Zd-bY 



[Y{b^~a^)~b^]Ep 
{h-Zd~bf 



{Zd + bf 



'8E„ 



(21) 



Equation (ISTT i describes the charged-particle's motion at all 
speed ranges. 



B. Negatively charged particle with structure 

1. Constituent forces 

For negatively charged core-shell structured particle, ne- 
glecting the gravity, Fj of Eq. ([8]) gets modified asZ 

F7' = ni.i+ni,2 + ni.3+n2,i+n2^2 + n2,3, 



N2 




Cathode 



where the constituent forces of Ni are 



Figure 10: (Color online) Illustration of negatively charged, struc- 
tured, particle oscillating in vicinity of the cathode, i.e., the regions 
A' and B' . The dominant constituent force teims are shown in regions 
A' and B' . No oscillation mode exists near the anode, i.e., the region 
C'. The width of the region A' is identified by l^i, and the width of 
the region B' is given by /£>/ — /^/ , where I pi is the borderline between 
the regions B' and C' . The path 1 and path 2 represent the schematic 
plot of Zd (t) versus time, where the horizontal axis is the time. 



ni.i =e,-:- 



TzeoK^v' 



4^2 



1 



ni.2 



neoK3\v\\'Y{b^~a'^) -b'^\Ep 

ni.3 = e,7tEQK3 \v\Ep^ t,Ep\ 
and the constituent forces of N2 are given by 






2 



n2,i 



TteofCsV 
-e,- =■ ^ - 

neoK-i\v\\y{b^- 



1 



{h-sY 



-b'\En 



"2.2=6,- 

A{h-sy 
n2,3 = e,7i;£oK-3 \v\Ep - t^Ep. 



(/,- 



Here, Ni is the force between charged-particle and its image 
charge formed at the anode's surface whereas N2 corresponds 
to the force between charged-particle and its image charge 
contribution at the cathode's surface. The notations (Ni,N2) 
are introduced to distinguish from (Fi,F2) of the positive 
charged-particle case. In the case of negative charged-particle, 
the force N2 gives rise to oscillations; and, the particle oscil- 
lates in vicinity of the cathode, as schematically illustrated in 

To validate the argument illustrated in Fig. [TO] Eq. (fTTI i is 
evaluated for a negatively charged particle, Oi = — 0.012C • 
rcT^, using the following initial conditions: 



Zd{Q)^h~2b 



I 

-Zd 



and z,/(0)=0, 



(22) 



where z'y = 30jUm. To compare the result against the posi- 
tively charged particle situation discussed in Fig. |9] the par- 
ticle's initial position has been assigned such that there is a 



gap of 30 /im between the particle's lower surface and the 
cathode's surface. The choice of z^ = 30 /im in Eq. (l22l) en- 
sures such criteria. The parallel plate electric field strength of 
Ep ^ 2MV-m^^ is chosen. For all other parameter values, the 
same values from Eq. (|9]l are used. That said, the result is plot- 
ted in Fig. [TT] where it shows the oscillation frequency and the 
oscillation amplitude identical to the positively charged par- 
ticle case corresponding to Ep = 2MV -m^' in Fig. |9l This 
time, however, the charged-particle oscillates near the cathode 
instead of the anode, as it is negatively charged. 




2 4 6 8 10 12 14 16 18 20 
Time, t [|ls] 

Figure 11: (Color online) Oscillating negatively charged nonrela- 
tivistic particle with stiiicture in vicinity of the cathode. The Zd (t) 
of Eq. Ml\ has been plotted using the initial value conditions de- 
fined in Eq. ( I22t and the parameter values defined in Eq. (|9) with 



CTi = -0.012 C • m-2 and Vt = 2kV (or Ep - 
ode plate is located at Zd = 0;Um. 



:2MV-m-'). The an- 



10 



2. Potential energy 

The potential energy for a negatively charged particle sub- 
jected to the parallel plate electric field, E^, is obtained from 
Eq. ([Tol l utilizing the line integral path illustrated in Fig. |2lb), 



U{s) 



Fj- • e.,ds 



*ret 



or 



t/ = - / Fr • t,ds, 



(23) 



where, for convenience, Si^f and s have been replaced by ri 
and r2 illustrated in Fig. [TJb), respectively. Equivalently, Eq. 
I can be rewritten as 



U 



Ft ■ e^ds, 



(24) 



where the upper and the lower integral limits have been re- 
versed. Multiplication of the both sides of Eq. (l24l l by a neg- 
ative one yields 



-U 



Ft ■ e^ds. 



(25) 



The right side of Eq. (I25l l can be obtained by making the 
following replacements in Eq. (fT2] i. 



^^ref 



ri and s — > r2, 



which yields 

nr2 



Ft ■ ^zds 



KEoK^V 



[y{b^-a^)-b^]Ep 



+ 



4 I r2 h-r2 

2{h-r2f 
neoKsv j V v [Y{b^ -a^) -b^]Ep 



8Epr2 > - mgr2 



4 I ri h — r\ 

[y{b^-a^)-b^]Ep 



2r? 



?<Epr\ > +mgri. 



2(h-riY 
Insertion of Eq. ( |26] l into Eq. (l24l i yields 



(26) 



U = -- 



r2 h - r2 



[y{h 



' a')-b^]Ep 



8£„r2 



2rl 



mgr2 



2{h-r2Y 

4 I ri h — ri 
[Y{b^-a^)-b^]Ep 



lr\ 



2{h-ny 



8£„ri 



- mgri . 



(27) 



For a negatively charged particle, ri — s and r2 = s^gf, and, 
thus, Eq. (IZTT i becomes 



C/(.) 



;reoK'3V 



[Y{b^-a^)-b^]Ep 



[Y{b^-a^)-b^]Ep 
2?2 



2{h- Sref) 



V 



7reoK"3V I V 

4 I ,? /i - .? 



8EpSM > - mgSref 

[y{b'-a')-b']Ep 
2*2 



2 (/i - .9) 



2 



-8£„i 



-ings. 



(28) 



In terms of the z^/ parameter, i.e., s = Zd + b, Eq. ( 1281 ) becomes 



f/(^</) 



4 1 iref /! - ^ref 

[y{b^-a^)-b^]Ep 



2s^ 



■cf 



2 (/! - iref ) 

7reoK-3V J V 



V [Y{b^-a^)-b^]Ep 



[Y{b^-a^)-b^]Ep 



Zd + b h-zd-b 2{zd + bf 

EEp{zd+b)\ 



2{h-Zd-bY 

+ mg [zd + b) . 

As with the case of the positive particle, I shall set Sref at the 
midway between the parallel plates, i.e., Eq. (fl4l l. 



*ref ^ 



and, this yields 


V 


^^^'^- 4 \zd + bU- 
[y{b^-a^)-b'^]Ep 
2{zd + bf 


-Zd-b 

'y{b^-a^)~b^]Ep 
2{h-zd-bf 



+EEp{zd + b) }+mg{zd + b) 



neoK^v 



v + Eph^)--mgh, 



(29) 



where the resulting expression is, form wise, identical to the 
one in Eq. ( fTST i, i.e., the result corresponding to the case of 
positive particle. 

Equation ^^ is plotted for CTi = -0.012Cm"2 with all 
other parameter values same as defined in Eq. (|9), 

K-2=6, K-3 = l, 

a— 1. Suva, h=lmm, 

b — a = 4nm, 

Vl = OV, 

C72 = OC • voT^ (i.e., insulator), 

p„,,i=2700kg-m-3, 

p„,,2 = 3800 kg -m-^ 



11 



For the plot, the anode voltages of VV = 1 kV, Vt = 2kV, and 
Vt = 3kV are considered for comparison. For V^ = OV and 
/i = 1 mm, these anode voltages correspond to Ep = I MV ■ 



m ^ , Ep — 2MV • m ', and Ep = 3MV • m ', respectively. 
The results are shown in Fig. [12] 



(a) 



0.30 
0.25 
0.20 
0.15 
0.10 
0.05 
0.00 
-0.05 
-0.10 
-0.15 
-0.20 
-0.25 
-0.30 



E = 1 MV/m 
E =2 MV/m 
E = 3 MV/m 



0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 
Position, z^ [mm] 



(b) 



-0.02 

-0.04 

-0.06 

-0.08 - 

-0.10 

-0.12 

-0.14 

-0.16 

-0.18 



1^: 



1 MV/m — 

2 MV/m 



700 750 



800 850 900 950 
Po.sition, Zj [|J.m] 



1000 



Figure 12: (Color online) (a) Plot of the potential energy function, 
U (zd) of Eq. ^, for Ep = 1 MV -m"' , fi^ = 2MV ■ m"' , and Ep = 
3MV-m^'. The particle has a charge density of CT] = — 0.012 C-m^^ 
and all other parameter values are as defined in Eq. (|9]. (b) Enlarged 
plot of U (zd) for domain h/2 <Zd < h, where h= 1 mm. In the plot, 
the anode is located at Zd = 0/im. 



From the physical arguments based on the constituent 
forces, the oscillatory solutions for a negatively charged par- 
ticle occur only in the domain Zd > h/2 in Figs. [T2ja) and 
Fig. [T2l b). The results show the potential well minima occur- 
ring in vicinity of the cathode side of the electrodes. Because 
the physical charged-particle cannot penetrate into the surface 
of the cathode, the parameter Zd in Fig. [T2jb) is bounded by 
< Zd < h — 2b. The case of Zd — h — 2b coiTesponds to a 
situation in which the charged-particle is in physical contact 
with the cathode's surface. There, the height of the potential 
well is finite and that criteria limits the width of the potential 
well, wherein the negatively charged particle can have oscil- 
latory solutions. For instance, in the case of Ep — 1 MV • m^' 
in Fig. [T2jb), the width of the potential well is approxi- 
mately ~ 250 /xm; and, the oscillatory solutions exist approx- 
imately for 745 lJ.m<Zd < 996.99 jim. Similarly, for the case 
of Ep = 2MV • m"' in Fig. Ob), the width of the potential 
well is approximately ^ 125 /xm; and, the negatively charged 
particle is expected to have oscillatory solutions in a domain 



^ 872jtim < Zd < 996.99 /xm. For the negatively charged par- 
ticle, the potential well, wherein the oscillatory solutions ex- 
ist, gets formed in vicinity of the cathode. The width of such 
potential well decreases with increased Ep. Physically, such 
property corresponds to the narrowing of the negative glow 
region with increased Ep . 



C. Dipole radiation 

1. Nonrelativistic regime 

Oscillating charged-particle radiates electromagnetic en- 
ergy. The power of such radiation, in the nonrelativistic limit, 
is given by Larmor radiation formula. 



Prad it) 






(30) 



where Zd is the nonrelativistic charged-particle acceleration 
defined in Eq. ([TtI i. and Qt is the effective charge carried 
by the particle^ 



Qr^AneoKiV. 



(31) 



Insertion of Eqs. ( [TtI i and dSTT i. respectively for 'id and Qt, 
into Eq. ( l30l l yields 



Prad{t) = 



8;reoK'f v^ / ;reoK-3V 



3c3 

V 






4;« 



{Zd^bf 



[y{b^-a^)-b^]Ep 



{h-Zd-t 



[zd + bY 



[Y{b^-a^)-b^]Ep 
(h-Zd-bf 




(32) 



where Zd is the solution to the Eq. ( fTTb . 

I shall use the following parameter values to evaluate 

Prad (f ) : 



K2 = 6, K-3 = l, 

fl = 25nm, h=10nm, 
b — a~ 2nm, 

yr = i6kv, yi = ov, 

C7i = 100C-m-2, ^^^> 



(72 = OC • m (i.e., insulator), 
p,„.i^2700kg-m-3, 
p,„,2- 3800kg -m-^. 



The corresponding charged-particle motion is obtained by 
solving Eq. ( [TtI i via Runge-Kutta method. For the initial con- 
ditions, I shall choose 



Zd(0) = ljtim and z^(0)=0. 



(34) 



Illustrated in Figs. [T3l a) and [T3l b) are the results of z^ (f ) and 
Prad (0 J respectively. The first sharp turning point in the plot 
of Fig. [T3l a) has been enlarged and is shown in Fig. [T3T c). 



12 



■2 20 



L 

43 



(a) 



(b) 






180 






# 


160 


s 






140 






a. 


120 


, ■ 


ino 


OJ 




# 


80 


o 




ft 


60 






rt 


40 






^ 


70 


Oi 








100 



150 200 250 
Time, t [ps] 



400 



50 100 



150 200 250 
Time, t [ps] 



300 350 400 




44.2 44.4 44.6 44 
Time, t [ps] 



44 45 

Time, t [ps] 



Figure 13: (Color online) (a) Plot of Zd (t) , Eq. l ll7t . using the parameter values defined in Eq. ( 133) and the initial conditions defined in Eq. 
l l34t . (b) The corresponding Larmor radiation power computed from Eq. J32t . (c) The first sharp turning point in (a) has been zoomed for a 
detailed view, (d) The first pulse in (b) has been zoomed for a detailed view. In the plot, the anode plate is located at Zd = 0/im. 



where it shows the particle rebounding approximately at a dis- 
tance of lOnm from the anode plate electrode's surface. The 
first pulse of the emitted Larmor radiation power has been en- 
larged and is shown in Fig. [T3l d). 

Is (Ji — lOOC • m^^ experimentally an attainable surface 
charge density? The answer to this question is yes. In fact, for 
systems wherein nanoparticles are deliberately ionized in con- 
trolled manner, the surface charge density of (7i — lOOC ■ m^^ 
is more reasonable than the one used in Eq. (|9), i.e., Gi = 
0.012Cm^^. Physically, (Ji — lOOC-m^^ corresponds to a 
case wherein each aluminum atoms in the volume of radius 
r = a contributing approximately one electron in the ioniza- 
tion process. This can be illustrated as follow. The mass of 
spherical aluminum core is 



nic = -TZa' p,ny. 



and, the mass of a single aluminum atom is given by 



culated as 



mai 






where m„, is the molar atomic weight and Na ~ 6.022 x 
lO^^mol^' is the Avogadro constant. The total number of 
aluminum atoms inside the volume of radius r — a can be cal- 



A^, 



al 



lUc 4NA7Za^Pm.l 



rriai 3m,v 

The aluminum core of radius r — a carries a surface charge of 

For a macroscopic particle, Qsc would be solely contributed 
from the atoms near the surface. However, for nanoparticles, 
the idea of "surface charge" becomes vague because it's not 
just those atoms near the surface, but the atoms in the entire 
volume of nanoparticle that contribute to Qsc- In that sense, 
Qsc should be more appropriately coined as the "total charge" 
caiTied by the nanoparticle, albeit it is still defined in terms 
of the surface charge formula, Qsc = 4-na^ai . That said, how 
many electrons must be removed from each aluminum atoms 
in order for the core to have net positive charge in the amount 
of Qsc — Ana^O\l The answer to this question is 



A^. 



Qsc 



37taimH 



qeNal qeNATlap^ 



m.l 



where qg « 1.602 x 10^'^C is the fundamental charge mag- 
nitude. For an aluminum atom, m„ « 26.98 g and the number 
of electrons to be removed per aluminum atom is 



13 



N,Kl.24 



N. 



1, 



where the greatest integer value has been taken for Ne, as 
there cannot be 1.24 electrons, of course. This result im- 
plies that, on the average, each aluminum atoms in the core 
loses one electron during the ionization process in the case of 
0\ = lOOC ■ rcT^ and a — 25 nm. 

The anode voltage of Vj = 16kV has been carefully cho- 
sen such that electrical breakdown does not occur between the 
parallel plates. Zouache and Lefort have demonstrated that 
by choosing a composite material for electrodes, for instance, 
composite material of 60% silver and 40% nickel, the DC bias 
voltage across the two electrodes can be as high as 3.85 kV 
at plate gap of 1 /im in vacuum before electrical breakdown 
takes place. ^^ In terms of electric field strength, this corre- 
sponds to Ep — 3.85GV • m^^ At plate gap of h = lO/im 
and the cathode grounded, the anode voltage of Vj = 16kV 
corresponds to £„ — 1.6 GV -m^^ which is much less than 



-1 



£,, = 3.85GV-m 



2. Relativistic regime 

The Larmor radiation formula, Eq. (l30i . is only valid for 
particle speeds that are small relative to the speed of light. 
In the relativistic generalization, the total power radiated by 
oscillating charged-particle is given by the Lienard radiation 
formulaji 



Prad (0 



8;reoK'3V^ 
3^3 



-3 



7^ 



(35) 



where Zd is the particle's acceleration associated with the rel- 
ativistic force, Eq. ( fT9] l. With the explicit expression for 'id 
inserted from Eq. (l2ll . the Lienard radiation formula of Eq. 
I becomes 



Prad (0 



7reoK-3V 
Am 



V 

{h-zd-bf 



[Y{b^~a^)-b^]Ep 



{zd+bY 



[y{b^-a^)-b^]Ep 



[zci + bf 



{h - Zd 




(36) 



which expression is identical to Eq. (BZb with an exception 
that Zd is now obtained from the relativistic dynamics, i.e., 
Eqs. ([Till or dZB- 



III. MECHANISM FOR SELF-SUSTAINED 
OSCILLATIONS IN THE POSITIVE GLOW CORONA 

The typical configurations for the electrodes in the glow 
corona apparatus are illustrated in Fig. [14] When the po- 
tential difference between the rod shaped electrode and the 



plate electrode are sufficiently large, but not large enough to 
cause an electric arc, a glow occurs near the surface of the rod 
shaped electrode. For the case in which the rod shaped elec- 
trode is an anode, the glow corona is referred to as the pos- 
itive glow corona whereas, if the rod shaped electrode is the 
cathode, the glow corona is referred to as the negative glow 
corona by convention. The schematics of the positive and the 
negative glow corona apparatuses are illustrated in Figs. [TlT a) 
and [T4T b). respectively. In this paper, only the positive glow 
corona is discussed. 



(b) 



Anode plate 



V. 



Ill 

V£ Positive glow corona 



Negative glow corona 't 

HI 



Cathode plate 



Figure 14: (Color online) (a) Schematic illustration of positive glow 
corona apparatus. In the positive glow corona, a glowing light is 
observed in vicinity of the anode, (b) Schematic illustration of a 
negative glow corona. In the negative glow corona, a glowing light 
is observed in vicinity of the cathode. 



The first reported account with the glow corona is by 
Michael Faraday in 1838. However, only recently, people 
have begun to extensively investigate the phenomenon due to 
its potential applications in various scientific and engineering 
fields, such as semiconductor lithography, materials process- 
ing, plasma lighting, and so on. ^ ' Among the interesting prop- 
erties of the phenomenon of glow corona, the self-sustained 
electrode current oscillations in the positive glow corona is 
the most extensively investigated one. Such self-sustained os- 
cillations in the electrode current persists albeit the system as 
a whole is biased with a DC voltage across the electrodes. To 
this date, the basic underlying mechanism behind such self- 
sustained current oscillations remains unclear. ^~^ 

Morrow did a theoretical work in an attempt to explain 
the underlying physics behind the self-sustained current os- 
cillations in the positive glow corona.^ Qualitatively, his pre- 
dictions are consistent with the various experimental obser- 
vations by others. Despite the fact that plasmas are ion- 
ized gasesr^^ which may contain particles of all charged 
species (positive, negative, or neutral) of various sizes (atoms 
or nanoparticles). Morrow showed that the self-sustained os- 
cillations in the electrode current are predominantly due to the 
mobility of the positive ions in the gas. He has calculated that 
approximately 96% of the variations in the electrode current is 
due to the oscillations of the positive ions in the plasma. The 



14 



current in the electrode can thus be expressed as 

u 

where Qj.n and zj „ [t] are the effective charge and the ve- 
locity of the nth charged-particle, respectively. In this paper, 
there is only one charged-particle; and, the expression for the 
electrode current becomes 

l(f)-erz^(0, 

where Qt is the effective charge carried by the charged- 
particle and ij [t] is the charged-particle's velocity. For the 
core-shell structured charged-particle considered here, the ex- 
pression for Qt is^ 

Qt = 87Za{b-a)ai h47r(fl^c7i +/7^C72), 

which is a constant. Neglecting the constant terms, the current 
in the electrode becomes 



(a) 



I(0-Zd(f). 



(37) 



The z^/ (f ) has been computed for the nonrelativistic charged- 
particle whose position versus time plot and the associated 
Larmor radiation power are illustrated in Figs. [TSja) and 
[T3l b). The result is shown in Fig. [TSl a). where the first abrupt 
rise in the velocity has been zoomed for a detailed view in Fig. 
[TSl b). This result is compared with waveforms of experimen- 
tal discharge current measurement at the electrode for a posi- 
tive corona in nitrogen at 35 Torrby Akishev et air' Similarly, 
the result is also compared with the prediction by Morrow.^ 
Remarkably, the profile of Eq. (l37t . which is shown in Fig. 
[TSl a). closely resembles both experimental and theoretical re- 
sults by Akishev et al. and Morrow, respectively. For instance, 
the current in Eq. (l37T i has a saw-tooth shaped wave pro- 
file, which is qualitatively similar to the waveforms obtained 
by Akishev et al. and Morrow. Moreover, presented theory 
predicts pulses of radiation output occurring precisely at the 
point where the current rises abruptly. This can be checked 
from Figs. fTSl b) and [TSl a). where it shows pulses of radi- 
ation power coinciding with the abrupt rises in the charged- 
particle velocity versus time plot. Such characteristic is con- 
sistent with results obtained by Akishev et al. and Morrow. 
This shows that positive ion oscillations in the positive glow 
corona involve the kind of charged-particle oscillation mech- 
anism discussed in Fig. [3] 

The minor discrepancies in the electrode current waveforms 
between the result of this work and the experimental measure- 
ments by others can be attributed to the differences in the setup 
of the apparatus. ^"^ Illustrated in Fig. [T6l a) is the equivalent 
circuit diagram representation for a typical apparatus in the 
glow corona experiments. The setup for a typical glow corona 
experiment involves the ballast resistor Rj,, a shunt resistor TJj, 
and a stray capacitance C, from the external circuit. In the pos- 
itive glow corona experiment,^ the typical geometry for the 
anode and cathode electrodes are as illustrated in Fig. [T6l b) 
whereas, in a typical DC glow discharge experiments, the par- 
allel plate geometry, such as the one illustrated in Fig. [TSl c). 



> 



100 




\ 


^. 


(\ 


f\ 


50 




\ 


\ 


\ 


\ - 





^ 


\ 


\ 


\ 


\ - 


-50 


■\ 


\ 


\ 


\ 


\- 


100 


- \l 


\ 


\l 


\ 


v 



50 100 150 200 250 300 350 400 
Time, t [ps] 

(b) 



100 - 



"^ 


50 


1 





> 


-50 



-100 



40 



42 



44 46 

Time, ; [ps] 



48 



50 



Figure 15: (Color online) (a) The plot of velocity, zj (t) , correspond- 
ing to the charged-particle in Fig. [13] (b) The first abrupt rise in the 
velocity has been zoomed for a detailed view. 



is the typical configuration for the electrodes^ The presence 
of the ballast and the shunt resistors in the circuit keeps the 
anode at voltage Va {t) and the cathode at voltage Vc (0 ■ 

These configurations for the glow corona experiment. Figs. 
[T6l a)-fT6l'c). are compared with the model configuration con- 
sidered in this work. Fig. |4] The equivalent circuit diagram 
for the model illustrated in Fig. |4]is as shown in Fig. [T6jd). 
Unlike the typical setup in the glow corona (or DC glow dis- 
charge) experiments, the equivalent circuit diagram for the 
model considered here does not contain the ballast and the 
shunt resistors in the circuit. Absence of these resistors in the 
circuit keep the anode and the cathode voltages fixed, respec- 
tively, at Vt and Vl in Fig. |4] Contrary to this, the anode 
and the cathode voltages in a typical glow corona experiment 
are not fixed at some constant values due to the presence of 
the ballast and the shunt resistors. For instance, the voltages 
Va (t) and Vc{t) in Fig. \T6\ a) are not constants, but vary in 
time due to the dynamics of ionized atoms in the space be- 
tween the anode and the cathode electrodes. For this reason, 
the electrode voltage oscillation measurements from an exper- 
iment, in which the setup is equivalent to the one illustrated 
in Fig. [T6l a). cannot be used directly to test the theory pre- 
sented in this paper. However, the current oscillations in the 
electrodes are present in all of the configurations in Fig. [16] 
For instance, an oscillating ionized particle between the an- 



15 



Oscillating ionized atoms 
Light emission 



(b) 



Anode 



(a) 



y. 



c. 



R, 



^^)athode 



--VvW T 



V 



L,^ 



-A^ T — 



Anode 



/via/V . , vwuv* 



IflCathode 



(d) 



Vr 



Anode 



Cathode 



Oscillating 
charged-particle 



Figure 16: (Color online) (a) Typical experimental apparatus in the glow corona (or glow discharge) experiment, (b) Electrodes in the positive 
glow corona experiment, (c) Electrodes in the DC glow discharge experiment using parallel plates. The ballast resistor Ri, and the shunt 
resistor Rs are shown. The voltmeter is placed across the shunt resistor. The Q is a stray capacity of external circuit, (d) The equivalent circuit 
diagram corresponding to the model in this paper. 



ode and the cathode gives rise to an electrode current, Ii (f ) in 
Fig. \T6\ a) and I2 (f) in Fig. [T6l d). which oscillates in corre- 
lation to the motion of oscillating ionized particle. Such elec- 
trode current oscillations get induced in the circuit regardless 
of whether the ballast and the shunt resistors are present in the 
circuit or not. The electrode cuiTent oscillation measurements 
from Figs. [T6la)- fT6l c). therefore, can be used to test the the- 
ory presented here; and, this is essentially what was assumed 
in Eq. (|37]i. 

Since the geometry of the anode used in the experiment 
is different from the simple plate geometry assumed in the 
model adopted in this paper, the resulting waveforms of oscil- 
lating electrode cuiTents from this theory, I2 (f ) , and the ex- 
periment, Ii (f) , are not identical. Nevertheless, qualitatively, 
both Ii (f) and I2 (f) show the same characteristic behavior. 
This must be so because the basic mechanism behind the os- 
cillations in Ii (f) and I2 (f ) originates from the same physics. 



One such characteristic behavior is the presence of radiation 
output accompanying the abrupt rises in the Ii (t) and I2 (f) , 
which was discussed previously from Figs. \T3\ h) and flSl a). 

Besides the geometrical differences in the anode, the model 
treated here has only a single charged-particle in the space 
between the anode and the cathode whereas, in the glow 
corona experiments^^i^ the space between the electrodes is 
filled with an ionized gas, i.e., many ionized atoms. Despite 
these differences, the theory qualitatively reproduces the self- 
sustained current oscillations in the electrode, consistent with 
the results from the various glow corona experiments. Such 
result suggests that the phenomenon of self-sustained elec- 
trode current oscillations in the positive glow corona is a man- 
ifestation of the charged-particle oscillation discussed in this 
paper. 

The radiation power in Fig. [T3l b) is emitted at frequency of 
approximately lOGHz, which is not a visible light. How can 



16 



the pulses of light accompanying the saw-tooth shaped elec- 
trode current oscillations be explained? To answer this, typi- 
cal glow corona experiment involves gases. There, individual 
atoms can be highly ionized to oscillate at frequencies large 
enough to emit visible light. The plasma as a whole, how- 
ever, oscillates at much smaller frequencies because its dy- 
namics involves the collective motions of all constituent ion- 
ized atoms. This qualitatively explains the pulses of light ac- 
companying the self-sustained electrode current oscillations, 
which oscillates at much lower frequencies. 

As an extension of this theory, the self-sustained oscilla- 
tions in the negative glow corona. Fig. [I4b . can be quali- 
tatively explained from the negatively charged particles go- 
ing through an oscillatory motion in vicinity of the cathode, 
which is schematically illustrated in Fig. [TO] The neon lamp 
used in the Pearson-Anson relaxation oscillator is the clas- 
sic example of negative glow corona at work. When the neon 
bulb is biased with a direct current (DC) voltage, typically 
around 60 V, a glow gets formed around the cathode lead. No 
such glow occurs near the anode lead. It is unlikely that neon 
atoms inside the lamp are positively charged. Quantum me- 
chanical calculations show that it takes minimum electric field 
strength of approximately 4.8V-A (or 48 G V • m^ ' ) to strip 
an electron from a neon at temperature of 273°K.'^ In a typ- 
ical neon bulbs used in the Pearson-Anson relaxation oscilla- 
tors, the anode and the cathode leads are separated by a gap 
of just few millimeters. Assuming a gap of 1 mm between the 
electrodes, and a DC bias voltage of 60V, the electric field 
between the electrodes is 60 kV • m^'. This electric field is 
not large enough to ionize a neon atom. However, an electric 
field of 60kV • m^ ' at relatively warm temperature is sufficient 



to emit electrons from the surface of the cathode lead. Be- 
cause neon is highly electronegative, it attracts any free elec- 
trons nearby and becomes negatively charged.— Such case, in 
which a negatively charged neon atom oscillates in vicinity of 
the cathode, is qualitatively explained by the theory presented 
in this paper. 



IV. CONCLUDING REMARKS 

The self-sustained electrode current oscillations in the 
positive glow corona can be qualitatively explained by the 
oscillatory solutions which is quite naturally obtained from 
the associated electromagnetic boundary value problem. 
To demonstrate this, a simple, DC voltage biased, plane- 
parallel plate system with a charged-particle inside has 
been considered. The resulting oscillatory solutions for the 
charged-particle motion qualitatively explains the observed 
experimental results. The remarkable similarities in the 
waveforms of the self-sustained electrode current oscillations 
between the various experiment s^'^'^ and the prediction from 
this work indicate that the basic underlying mechanism 
behind the self-sustained oscillations in the positive glow 
corona involves the kind of push-pull mechanism discussed 
in Fig. |3] 

V. ACKNOWLEDGMENTS 

The author acknowledges the support for this work pro- 
vided by Samsung Electronics Co., Ltd. 



* Electronic address: sungnae.cho@samsung.com 

J. Jackson, Classical Electrodynamics - Third Edition, Ch. 2 (John 

Wiley & Sons, Inc., 1998). 
^ M. Goldman, A. Goldman, and R. Sigmond, Pure & Appl. Chem. 

57(9), 1353-1362(1985). 

3 R. Sigmond, J. Phys. IV France 7, C4-383 (1997) 

4 R. Morrow, J. Phys. D: Appl. Phys. 30, 3099 (1997). 

Yu. Akishev, M. Grushin, A. Deryugin, A. Napartovich, M. 

Pan'kin, and N. Tmshkin, J. Phys. D: Appl. Phys. 32, 2399 

(1999). 
^ N. Allen, M. Abdel-Salam, M. Bouflendj, I. Cotton, and B. Tan, 

lET Sci. Meas. Technol. 1(2), 103 (2007). 
■^ S. Cho, Phys. Plasmas 19(3), 033506 (2012). 
^ K. Tamura, Y. Kimura, H. Suzuki, O. Kido, T. Sato, T. Tanigaki, 

M. Kummada, Y Saito, and C. Kaito, Jpn. J. Appl. Phys. 42, 7489 

(2003). 

R. Sohal, G. Lupina, O. Seifarth, P. Zaumseil, and C. Walczyk, 



Surface Science 604, 276 (2010). 
'" N. Zouache and A. Lefort, IEEE Trans. Dielectr. Electr. Insul. 

4(4), 358 (1997). 
" T. Higashiguchi, H. Terauchi, N. Yugami, T. Yatagai, W. Sasaki, 

R. D' Arcy, P. Dunne, and G. O'Sullivan, Appl. Phys. Lett. 96(13), 

131505 (2010). 
'^ A. Bogaerts, E. Neyts, R. Gijbels, and Joost. van der Mullen, 

Spectrochimica Acta Part B 57(4), 609 (2002). 

T. Gyergyek, M. Cercek, M. Stanojevic, and N. Jelic, J. Phys. D: 

Appl. Phys. 27, 2080 (1994). 
''* T. Kuschel, B. Niermann, I. Stefanovic, M. Boke, N. Skoro, D. 

Marie, Z. Petrovic, and J. Winter, Plasma Sources Sci. Technol. 

20,065001 (2011). 
15 D. Brandon, Br J. Appl. Phys. 14, 474 (1963). 
'^ N. Islam and D. Ghosh, J. of Quantum Information Science 1, 135 

(2011).