# Full text of "Equation of state of charged colloidal suspensions and its dependence on the thermodynamic route"

## See other formats

Equation of state of charged colloidal suspensions and its dependence on the thermodynamic route o (N O a B o > in rn O X Thiago E. Colla, Alexandre P. dos Santos, and Yan Levin Instituto de Fisica, Universidade Fedaral do Rio Grande do Sul, CP 15051, 91501-970 Porto Alegre, RS, Brazil. The thermodynamic properties of highly charged colloidal suspensions in contact with a salt reservoir are investigated in the framework of the Renormalized Jellium Model (RJM) . It is found that the equation of state is very sensitive to the particular thermodynamic route used to obtain it. Specifically, the osmotic pressure calculated within the RJM using the contact value theorem can be very different from the pressure calculated using the Kirkwood-BufF fluctuation relations. On the other hand, Monte Carlo (MC) simulations show that both the effective pair potentials and the correlation functions are accurately predicted by the RJM. It is suggested that the lack of self-consistency in the thermodynamics of the RJM is a result of neglected electrostatic correlations between the counterions and colons. PACS numbers: I. INTRODUCTION In spite of the fundamental importance — both prac- tical and theoretical — the thermodynamic properties of charged colloidal suspensions are far from understood^"— . Even such basic question as the existence of a liquid-gas phase transition in these systems still remains a topic of debate^"—. The difRculty in describing the thermody- namics of charged colloidal suspensions is a consequence of both size and charge asymmetry between the different components of the system and the long-range nature of the Coulomb interactio n^i^i^° . To simplify the theoretical description one often uses the, so-called. Primitive Model (PM). In this model all charged components — colloidal particles, colons, and counterions — are treated explic- itly, while the solvent — usually an aqueous medium — is considered as a dielectric continuum. The interactions between the colloidal particles, the counterions, and the colons have both Coulomb and hard-core components. Image effects resulting from the dielectric discontinuities across the particle surface are usually neglected at the lowest order of approximation. Colloidal suspensions often contain salt. For theoret- ical description it is, therefore, convenient to work in a semi-grand-canonical ensemble in which the number of colloidal particles is fixed, while the concentration of salt is controlled by an externally imposed chemical potential. Physically this can be realized by separating the suspen- sion from a salt reservoir by a semi-permeable membrane transparent only to microionsiiii^. The large asymmetry between the colloidal particles and the microions, forces us to employ different approx- imations to account for the correlations among the var- ious components of suspension. The correlations among the microions can be described by a linear Debyc-Hiickel (DH) like theory. For dilute colloidal suspensions these correlations arc usually negligible. On the other hand, to account for strong colloid-ion and colloid-colloid interac- tions requires a full non-linear theory. One approach that has proven to be very useful for describing the non-linear correlations between the colloidal particles and the coun- terions is the concept of charge rcnormalizatioE>i*i^'i^. The idea is that strong electrostatic attraction between the colloidal particles and their counterions will lead to accumulation of counterions near the colloidal sur- face. These countcrions can be considered to be "con- densed" (strongly bound) to the coUoidal particle, ef- fectively renornializing its bare charge. For strongly charged colloidal particles the renormalized charge will, in general, be much smaller in magnitude than the bare chargoH. An alternative, but equivalent way of modeling col- loidal suspensions is to explicitly trace out the microion degrees of freedom in a semi-grand-canonical partition functioni^. This way the multi-component colloidal sus- pension is mapped onto an equivalent one-component system in which only the colloidal particles are explicit. This coarse-graining procedure defines the, so-called. One Component Model (OCM). In this approach, all the contributions coming from the traced-out microions arc implicit in the effective interactions between the col- loidal particlesi^. The apparent simplification over the original problem is only formal, since the effective inter- action between the colloidal particles now has a many- body character—""— and is state-dependen t ^^'^^ , further complicating the thermodynamic calculationsiSri^. For weakly charged colloidal particles, the effective interaction potential in the OCM takes a particularly simple form known as the Derjaguin-Landau-Overbeek- Verwey (DLVO) pair potential^^i^^. Pu{r) = \i Ze'' 1 + Kfl 2 _-« (1) where a and —Zq are the colloidal radius and charge, re- spectively. The inverse Debye screening length is k^ = y''4:TrXB{p+ + P-), where p+ and /?_ are the mean con- centrations of the monovalent counterions and colons in- side the suspension, and A^ = Pq^ /e is the Bjerrum length. Due to the global charge neutrality, p+ — p^ — Z p — 0, where p is the concentration of colloidal parti- cles. For strongly charged colloidal particles, the linear DLVO theory is not sufficient to describe the pairwise interactions. The non-linear effects, however, can be in- cluded into DLVO potential by explicitly accounting for the counterion condensation. This can be achieved by re- placing the bare colloidal charge in Eq. ([1]) by the renor- malized effective charge Z -^ Z^jj. The charge renor- malization accounts for the strong non-linear particle- counterion correlations near the colloidal surfaces. Besides the DLVO pair potential, the effective colloidal interactions in the OCM formalism also have the, so- called, volume terms which depend on colloidal density, but not on colloidal coordinates^. The volume terms were argued to play important role for the thermodynam- ics of charged colloidal suspension a^^i^^'^^ . For structural properties of the OCM, however, these terms do not play any role, since they do not depend on colloidal coordi- nates. This point must be considered with special care when one wants to study colloidal thermodynamics us- ing the 0CMi^i2£i^. In fact, there are some approaches that describe the effective interaction by simply defin- ing a pair potential which reproduces the correct colloid- colloid correlations in using the OCM. Clearly, such ap- proaches must loose some thermodynamic informations contained in the volume terms. The question of whether the effective potential models based purely on pair interactions arc sufficient to study the thermodynamics of a fully multi-component system is still under discussio n^^'^^ . In the case of charged colloidal systems, the problem is even more subtle, since such sys- tems must obey additional constraints i. e. global electro- neutrality and the well-known Stillinger-Lovett moment conditional^. As a consequence, many theoretical tools originally designed for unconstrained systems have to be reformulated before they can be applied to charged systems2L2^. The aim of this work is to address some thermody- namic inconsistencies which arise when different routes are used to calculate the thermodynamic functions of charged colloidal suspensions. To this end, we will use the Renormalized Jellium Model (RJM), from which both the renormalized charge and the osmotic pressure can be easily calculate d^^i"^" . From the renormalized charge, the effective pair potential — and hence the cohoid-coUoid pair correlation functions — can be obtained using the OCM Ornstein-Zernike (OZ) equationSl with an appro- priate closure. Knowing the correlations, it is possi- ble to calculate the osmotic compressibility using the Kirkwood-BufF (KB) fluctuation theorj«22. In this work we will compare the osmotic compressibilities of the RJM calculated using both the contact theorem and the KB fluctuation relations. The paper is organized as follows. In section II we will briefly review the theoretical methods used for the thermodynamic investigations — the RJM, the Donnan Equilibrium, and the Kirkwood-BufF relations. In sec- tion III, we will briefly discuss the simulation techniques employed in this study. The results will be presented in section IV, and conclusions, discussion, and suggestions for the future investigations will be given in section V. II. THEORETICAL BACKGROUND A. The Renormalized Jellium Model The RJM is a model that allows one to calculate the effective charge of colloidal particles and the thermody- namic properties of colloidal suspensions based on the mean-field Poisson-Boltzmann-like (PB) equation. RJM is known to be very accurate for salt-free colloidal sus- pensions with monovalent counterions. In contrast to the traditional Cell Model (CM), where a lattice-like struc- ture is assumed for colloidal particles, in the RJM the colloidal correlations are modeled by a uniform neutral- izing background. The major conceptual advantage of the RJM over the CM is that the pair potential Eq. ([T]) is exact within the RJM formalism, while for CM there is no pairwise interaction between the colloidal particles22. Thus, the effective charges calculated using CM have no clear connection with the DLVO potential. Recently, the RJM was successfully extended to incorporate inter- colloidal correlation9^2i2i, the multivalent counterions^^. and colloidal polydispersity^^. In the RJM, one colloidal particle of charge —Z^areq and radius a is fixed at the origin of the coordinate sys- tem. The distribution of free (uncondensed) ions around this particle is assumed to follow the Boltzmann distri- bution, p±{r) — p±e^^'^'^^^\ where p± are the counte- rion and colon mean densities, and ^(r) is the mean electrostatic potential. The remaining colloidal parti- cles, along with their condensed counterions, are taken to provide a uniform neutralizing background of charge density —Zej/qp- The reduced mean electrostatic poten- tial 0(r) — f3qil){r) then satisfies the Poisson-Boltzmann- Jellium (PBJ) equation: V2(/,(r) = -47rAB (p+e-^^'^^ - p^e^^"-^ - Z^ffp) . (2) This equation can be numerically solved with the bound- ary conditions oo) -^ and d<j){r Zhare^E dr !'■=" The first condition defines the zero of the electrostatic potential in the bulk of suspension, while the second one determines the electric field at the colloidal surface using the Gauss law. Far from the central colloidal particle — the region where the electrostatic potential is weak — the PBJ equation can be linearized, resulting in the following long- distance behavior: Z/pffABS e Hr) (1 + Ka) r (3) where k = ^47rAs(/9+ +p_) = ^/4^^XB(2p^+Z^ p) is the effective screening length, and where we have used the global charge neutrality condition pj^— p^ — P^eff = 0. Note that in the far-field, the bare charge Zbare is replaced by the renormalized charge Z^ff, reflecting the nonlinear correlations at the colloidal surface. For a given salt and colloidal concentrations, p_ and p, respectively, the effective charge is calculated by match- ing the numerical solution of Eq. ([2]) with the linearized potential Eq. ([3]), in the far- field. Since within the RJM the background charge arises from the smeared-out charge of colloidal particles and their condensed coun- terions. the self-consistency requires that the effective colloidal charge must be the same as the charge of the miiform neutralizing background. This procedure can be easily implemented numericallj*^. Suppose that we know Zeff, then from Eq. ^ we will also know the potential and the electric field in the far-field region. We can then integrate the PBJ equation using a standard Rounge- Kutta algorithm to obtain the electrostatic potential all the way up to the colloidal surface. The corresponding bare colloidal charge Z^are is obtained using the Gauss low at the colloidal surface. In reality, of course, one wants to calculate the effective charge for a given bare charge. This can be done by varying Z^ff until the de- sired Zfjare IS found. lu practicc, this can be easily imple- mented numerically by incorporating a Newton-Raphson root-finding subroutine in the PBJ solver. The osmotic pressure within the RJM is given by /3P = /?+ + p_, (4) where p± are the bulk concentrations of free colons and counterions. In spite of its apparent simplicity, this ideal- gas-like equation of state requires a knowledge of nii- croion concentrations in the far-field which, in turn, de- pend on the charge rcnormalization and osmotic equilib- rium with the salt reservoir. We should also note that unlike for CM, for which the contact value theorem is an exact statemenli^i^ai^-ii, Eq. (g]) of the RJM is only valid in the mean-field approximation. We will later ar- gue that the failure to properly account for ionic corre- lations leads to thermodynamics inconsistencies in the RJM. B. The Donnan Equilibrium In this work we will consider a colloidal suspension in contact with a salt reservoir. The system is separated from the reservoir by a semi-permeable membrane which allows for a free flux of microions. The ionic concen- tration inside the suspension will then be determined by the osmotic equilibrium with the salt reservoir. Contrary to uncharged systems, for which the osmotic equilibrium simply results in a solvent flow from a solute poor to a so- lute reach region, the osmotic equilibrium in charged sys- tems is also constrained by the overall charge neutrality of the system. Physically, this is refiected in the appear- ance of a potential difference across the semi-permeable membrane which controls the overall build up of charge in the systeroiiii^. This potential difference is known as the Donnan potential^. From a theoretical point of view, it can also be thought of as a Lagrange multiplier used to enforce the charge neutrality of the systemMii^. In equilibrium, the ionic electrochemical potentials in- side the system must be equal to the ones in the salt reservoir. Neglecting the electrostatic correlations be- tween the microions, the ionic concentrations in the bulk and reservoir are related by p± = psC^'^" , where ps is the salt concentration in the reservoir, and 0d is the adi- mensional Donnan potential. Using the charge neutrality condition for free ions, /9+ — p_ — pZ^jf = 0, the Donnan potential can be eliminated to yield the bulk concentra- tions of free (uncondensed) microions: P± = i (^^{pZefff + {2psf ± ZeffP^ . (5) This expression can be used, together with the equa- tion of state Eq. ^, to write the osmotic pressure (311 l3U = l3P~2p,=p+ ^{pZefff + {2psY - 2ps. (6) where we have also added the colloidal ideal gas con- tribution /3Pc = p. It is important to stress that the above expression for the osmotic pressure completely ig- nore the microion correlations. This can be justified as long as the concentration of colons in the bulk is very low. The colloid-counterion correlations are taken into account through the charge renormalization. Using Eq. ^, two important limits can be verified. For high salt concentrations — Zeffp/2ps ^ 1, salt- dominated regime — there is no significant variation in the ionic concentrations across the membrane and the osmotic pressure (j6]) is small. On the other hand, in the limit Zef fp/2ps ^1 — the eounterion-dominated regime — there is a significant variation in the microion concen- tration between the bulk suspension and the reservoir and the osmotic pressure is large^i. The inverse osmotic compressibility Xt ~ p I — — j follows directly from Eq. ©: PXt' p'zl ff VipzIffV+WJ^ dlog{r]) (7) where t] = 47ra^p/3 is the colloidal volume fraction. The derivative on the right-hand-side of this expression can be neglected, since in the RJM the effective charge depends only weakly on the colloidal volume fractio n^^i^" . C. The Kirkwood-Buff relation Once the nonlinear colloid-ion correlations are properly taken into account through the charge renormalization, the DLVO pair potential Ec^. (1) can be used to investi- gate the structural properties of the suspension. This can be done by solving the OCM Ornstein-Zernike equation: h{r) = c(r) + p / h{r')ci\r - r'|)dr', (8) where /i(r) and c(r) are the total and the direct cor- relation functions, respectively. This equation has to be supplemented by an appropriate closure relation between h(r) and c(r)^. Once the structural properties are known, the ther- modynamic informations can then be obtained using the Kirkwood-Buff (KB) fluctuation theory^^. KB theory al- lows us to express the thermodynamic functions, such as the osmotic coefficients and the compressibilities, as integrals over the pair correlation functions. Originally formulated for unconstrained mixtures, KB theory re- quires some extra care when extended to systems in which the number densities of different components are not independen t ^^i^^i^^ . This is precisely the case for the charged systems, for which long-range Coulomb in- teraction requires an overall charge neutrality. In ad- dition to this, there are also other constraints known as the Stillinger-Lovett moment conditions, that restrict the fluctuations of different components'^ of a charged sys- tem. A naive application of the original KB theory to charged systems leads to undetermined results^iS. One way of avoiding these difficulties is to study the KB inte- grals for arbitrary k vectors in the Fourier spaco^^, tak- ing the limit fc — > at the end of the calculations. The extended KB theory then relates the osmotic compress- ibility with the Fourier transform of the total correlation function /i(k). XT = 1 + P / hir)dr = 1 + ph{0), (9) Using OZ equation, this expression can be inverted to yield (^)..--<°'- (10) KB theory shows that the knowledge of colloidal pair correlation function is sufficient for calculating the equa- tion of state of the colloidal suspension. Curiously, Eqs. ^ and (jlO[) rely only on the pair correlations which are well modeled using only the effective pair potential, Eq. ([T]). This suggests that the zero-order volume terms, which depend on colloidal concentration^, are not very important for the thermodynamics. III. MONTE CARLO SIMULATIONS To explore the validity of the RJM model, we perform Monte Carlo simulations to obtain the "exact" pairwise interaction potential. The simulations are performed for several flxed distances R between two spherical colloidal particles of charge —ZtareQ, which are restricted to move along the main diagonal of a box of side length L ~ 180A. Colloid particle 1 is located at x,y,z = —R/2^/3, and colloidal particle 2 at x,y,z ~ i?/2\/3. In order to keep the electro-neutrality, 2Zi,are microions of charge q are also present in the simulation box. If salt is added to the system, then L^ ps microions of charge q and L^ ps mieroions of eharge —q are included inside the box. The total number of microions in the system is then N = 2L^ps + 2Zbare- The radii of all the ions are set to 2 A. The usual Coulomb potential is considered between all the charged species. The total energy used in the MC simulations is: „ N-l N 2 N „ AT „ ^B -1 11 ''*i 1 ^ii 1 ''2i Z— 1 J— 2+1 -^ 2—1 2 — 1 where Zi is the charge valence of the ion i (+1 or —1), rij is the distance between two ions i and j, rn and r2i are the distances between the ion i and the colloidal particles 1 and 2, respectively. Since we consider periodic boundary conditions, the Ewald summation technique is employed^. The equilibration is achieved after 2.5 x 10'^ simulation steps per particle; every 100 movements per particle an uncorrelated state is saved. The mean force is calculated using 1 x lO"' uncorrelated configurations. The average electrostatic force on a colloidal particle (positive force corresponds to repulsion), along the diag- onal direction is / N UR)^{Y. ZbareZi f COS 9u , COS 6*22 ^bare where Fe{R) ~ -^Fe{R), On and 621 are the angles be- tween the diagonal and the line connecting the particle i to the colloid 1 and the colloid 2, respectively. These distances are measured from the diagonal in the counter- clockwise, for particle 1, and in the clockwise, for particle 2 direction, respectively. The Ewald technique is used to calculate the electrostatic forces. Besides the aver- age electrostatic force, there is also an entropic depletion force which must be taken into account. To do this we use the method of Wu et al.— , which consists of a small displacement of the colloidal particles along the diago- nal (while the microions remain in a fixed configuration) in order to count the resulting overlaps between the col- loidal particle and the microions. This entropic force can be expressed as (ivf) - (nI) + (7V|) - (nI) where Fd{R) ~ ■^Fd{R), -/Vf is the number of overlaps of colloidal particle 1 with the microions (both anions and cation), after a small displacement Ai? (w lA) in the di- rection of the colloidal particle 2 (superscript c stands for closer) and Nl , is the number of overlaps after a displace- ment AR in the opposite direction (superscript f stands for farther). Similarly iV^ and iVg, are the number of overlaps of colloidal particle 2 with the microions after a displacement AR in the direction of the colloidal particle 1 and in the opposite direction respectively. The effec- tive pair potentials can then be calculated by integrating the mean force, -As jj^^^^ dR' [Fe{R') + Fd{R')] , where Rmax is the reference distance at which the interaction between the two colloidal particles is negligible. IV. RESULTS We are now in a position to compare the thermody- namic predictions from Eqs. ([7]) and Q. To this end, the OZ equation is numerically solved using the hipernetted- chain (HNC) closure: c(r) = hir) - log(ft.(r) -I- 1) - /3u(r). (14) This closure is known to be very accurate for Yukawa- like pair potentials'"^. For a given reservoir salt con- centration ps and volume fraction 77, the pair potential is given by H]), with the effective charge calculated using the RJM. In order to test the accuracy of the effective pair po- tential predicted by the RJM, in Fig. [1] we compare it with the results of the Monte Carlo simulations. As can be seen, the DLVO pair potential with the bare colloidal charge considerably overestimate the effective colloid-colloid interaction. On the other hand, the pair- potential predicted by the RJM agrees well with the MC simulations. Near the colloidal surface, however, a small FIG. 1: Effective pair potentials calculated using the MC sim- ulations (squares), bare DLVO pair potential (dashed line) and DLVO with RJM effective parameters (solid line) for Zbare = 20, a = lOA and Xb = 7.2A. 0.5- -(a) ' ' -RJM - Simulation - p^ = 24.9mlVl ireV^ = 21.6 - ri = 0.0084 2 4 6 8 10 1.5 0.5- -(b) -RJIVI - Simulation - p^ = 249mM 4reV^ = 21.6 - Ti = 0.0084 1 2 3 4 5 r/o r/o FIG. 2: Colloid-colloid pair correlation functions obtained us- ing the MC simulations (Ref.47) and the RJM-OZ approach, for a) ps = 24.9 mM and b) ps = 249 mM. In both cases, the bare charge is ZtareXB/a = 21.6, and the volume fraction is ri = 0.0084. deviation from the Yukawa functional form is evident. These non-linear screening effects are a consequence of electrostatic correlations between the counterions near the colloidal surface. In Fig. [21 the colloid-colloid pair correlation func- tion g{r) calculated using the RJM and the HNC in- tegral equation, is compared with the results of the MC simulations^ in the high salt concentration regime. Again, wc sec a good agreement between the theory and the simulations. Figs. [1] and [2] show that the effective charges calcu- lated using the RJM arc able to correctly predict both the pair interactions and the structural properties of col- loidal suspensions containing added electrolyte. We next check if this good agreement also extends to the ther- modynamic functions. Unfortunately, very quickly we run into difficulties. We find that for the intermediate salt concentrations, the osmotic compressibility calcu- lated using the KB fluctuation relation Eq. ([9]) strongly deviates from the one calculated using the RJM equation of state (JEOS), Eq. (|6]). The discrepancy between the two routes can be clearly seen in Fig. [3j which shows the osmotic compressibility Xosm as a function of the reser- voir salt concentration p^, for colloidal particles of bare charge Z = 1000 and various volume fractions. Although both routes agrees in the low-salt and high-salt regimes, there are strong deviations at intermediate salt concen- trations. Furthermore, as the colloidal concentration in- creases, the discrepancy between the two thermodynamic routes becomes stronger. At low volume fractions and high salt concentration, both routes approach the cor- rect ideal gas limit Xosm ~ 1, when strong screening makes the system to behave as a dilute suspension of hard spheres. The question that arises then is: Which thermody- namic route is more reliable? Unfortunately the answer is not very clear. Due to the difficulty of performing large scale simulations on suspensions containing elec- trolyte, there is very little data available to us to answer this question. Furthermore, there is also a scarcity of the experimental data dealing with osmotic properties of charged colloidal suspensions. In Fig.|4l we compare both the osmotic pressure calculated using the JEOS and the KB fluctuation theory with the experimental measure- ments of Rasa et al.— . Neither one of the thermody- namic routes seems to be able to accurately describe this experimental data. Most likely this is a consequence of the strong electrostatic correlations between the ions re- sulting from the use of a low dielectric solvent by Rasa et al. Nevertheless, the fluctuation route seems to give results in a closer agreement with the experimental data 10" 10 D ImMl (b) -JEOS -KB - 10 10' (c) -JEOS -KB - .-i/ '/^ ' 10 10' 10" 10 10 10 10" 10" p [mM] 0.5 1 kTp[Nm"-] 0.5 1 1.5 kJp[Nm"'] 10 10" 10 10" 10 p,[mM] FIG. 4: Comparison between the osmotic pressure calculated using the JEOS, (solid line) and using the explicit integration of Eq. ^ (dashed line) with the experimental results reported in Ref. 48. The reservoir salt concentration is ps = 8/iM, while the Bjerrum length is As = 2.38 nm, the colloidal radius is a = 21.9 nm, and colloidal charges are: a) Z = 34 and b) Z = 40. FIG. 3: Reduced osmotic compressibility x = px//^ eis a func- tion of the reservoir salt concentration ps for a colloidal par- ticles of radius a — 30Aand bare charge Z = 1000. The colloidal volume fractions are: &) r\ = 10~*, b) 77 = 10~*, c) 77 = 10""^ and d) 77 = 10~^. We see a dramatic discrep- ancy between the predictions of the JEOS (solid lines) and the Kirkwood-Buff fluctuation theory (dashed lines), especially at intermediate salt concentrations and high volume fractions. than the JEOS. This suggest that for the RJM the fluc- tuation route might be more reliable for calculating the thermodynamic functions. We will now explore the pos- sible causes of the discrepancy between the two thermo- dynamic routes. A. Colloid-colloid correlations One possibility is that the discrepancy observed in Fig. [3] is due to the way that colloidal correlations enter into the theory in the two thermodynamic route a^^'^° . In- deed, while the colloid-colloid correlations are neglected in the JEOS, they contribute to the osmotic pressure cal- culated using the KB formalism Eq. ([5|), since the HNC equatioD^ used to obtain the correlation function takes into account colloidal hard-cores. The colloid-colloid re- pulsion is particularly important for large volume frac- tions and high-salt concentration {Zpjlps ^1), when ionic contribution to osmotic pressure is small. To asses the relevance of these correlations, we can add to the JEOS the excess colloidal virial pressure, 27rp2 /3P" / 9iry^dr+2TTp^V g{r)^r^dr, (15) where u{r) is the effective pair potential in the OCM de- scription. The first term on the right hand side of this equation represents the standard excess virial pressure for the one-component system, while the second term ac- counts for the density dependent effective pair potential. This term is essential to reproduce the correct Debye- Hiickel limiting law in the infinite dilution limit-''. Sub- stitution of Eq. dl]) into Eq. (flS)) produces the following expression for the excess pressure, 2^p2 PP^ -np g(r)f3u(r){Kr + l)r dr g{r)(3u{r) nr — 2{Kaf 1 Ka (16) r dr. where u{r) is the effective colloidal pair potential, Eq. [T] For all the parameters studied here, however, we find that I pea; I ^ Pj^iu ^nd the cffcct of coUoidal correlations is too small to account for the strong discrepancy observed in Fig. 13 B. Ion-ion correlations As the salt concentration increases, the mean distance between the cations and anions becomes smaller, leading to strong inter-ionic correlations. Such correlations are completely ignored by the mean-field Poisson-Boltzmann equation, which is the basis of the RJM. Indeed, in the absence colons (and for monovalent counterions), the RJM model was found to provide an excellent ac- count of both thermodynamic and structural properties of charged suspensions22ii2£iSi24. This good accuracy of the model is the result of large characteristic distance be- tween the counterions inside a salt-free suspension. On the other hand, presence of salt leads to strong cation- anion correlations neglected in the RJM. In order to explore the influence of inter-ionic correla- tions on the osmotic pressure in a colloidal suspension, we modify the JEOS by adding the correlational Debye- Hiickcl contributioE^ii^, /3P" (17) Figure [S] shows the osmotic compressibilities resulting from addition of Eq. [T7]to the JEOS, Eq. [H As can be seen from this figure, incorporation of ionic correlations even at this leading-order level, significantly improves the agreement between the two thermodynamic routes, es- pecially at large colloidal volume fractions. This simple calculation suggests that the thermodynamic consistency of the RJM can be restored by incorporating the inter- ionic correlations into the RJM. Unfortunately, at the moment, it is not clear how the inter- ionic correlations can be included into the RJM in a fully self-consistent fashion. This will be the subject of future research. (a) 1 -JEOS - JEOS + Ion-Ion --Fluctuations a ,,--'' qO.5 0.5- 0.4- a ?<0.3- ca 0.2- (b) ■JEOS JEOS + Ion-Ion ■ Fluctuations 250 500 ),|mM] 750 1000 250 500 ),[inM] 750 1000 FIG. 5: Comparison between the osmotic compressibilities calculated using the JEOS Eq. [6] (solid lines), using the JEOS with explicit ionic correlations Eqs. ^ and {TTJ, and using the the KB fluctuation theory, Eq. (|9}. The radius of colloidal particles is a = lOA, the bare colloidal charge \s Z = 1000, and the Bjerrum length is A5 = 7.2A. The volume fractions are: a) 77 = 0.01 and b) 77 = 0.05. V. SUMMARY AND CONCLUSIONS We have reported an inconsistency arising when dif- ferent routes are employed to calculate the thermody- namic functions in the the RJM. The discrepancies are particularly strong at moderate salt concentrations. On the other hand, comparing the predictions of the RJM with the MC simulations, we see that the model accu- rately accounts for the effective pair interactions and the colloidal correlation functions, even for suspensions con- taining electrolyte. Thermodynamic inconsistency between different routes is not particular to the RJM and is found for many other system^. Even for a Debye-Hiickel elec- trolyte, the osmotic compressibility calculated via the PM virial equation is quite different from the predictions of the fluctuation theory^. In these cases, MC simula- tions are particularly helpful to choose the more accurate route to thermodynamics^. Unfortunately, simulations of charged colloidal suspensions at even moderate salt concentrations are still too computationally demanding while the experimental data is still very scarce. The experimental and the simulational data available to us seems to indicate that KB fluctuation relations 10 provide a more reliable route to thermodynamics of the RJM. The KB route seems to partially account for the inter-ionic correlations which are completely neglected by the JEOS. These correlations are negligible in the absence of colons, they however become relevant when salt concentration increases and the characteristic distance between the cations and the anions becomes smalli. In Section IV (B), we showed that even a simple incorporation of the DH contribution to the osmotic pressure already brings the JEOS and the fluctuation results into a closer agreement. A fully self-consistent incorporation of ionic correlations into the RJM requires, however, development of a new methodology closer in spirit to the density functional theory. The work in this direction is now in progress. VI. ACKNOWLEDGMENTS This work was partially supported by the CNPq, FAPERGS, INCT-FCx, and by the US-AFOSR under the grant FA9550-09-1-0283. ^ Y. Levin, Rep. Prog. Phys. 65, 1577 (2002). ^ R. Messina, J. Phys.: Condens. Matter 21, 113102 (2009). ^ L. Belloni, J. Phys.: Condens. Matter 12, R549 (2000). * K. Mahdi and M. O. de la Cruz, Macromolecules 33, 7649 (2000). ^ A. Diehl, M. C. Barbosa, and Y. Levin, Europhys. Lett. 53, 86 (2001). ® Y. Levin, E. Trizac, and L. Bocquet, J. Phys.: Cond. Mat. 15, S3523 (2003). '' A. R. Denton, Phys. Rev. E 2007, 051401 (76). ^ B. Lu and A. R. Denton, Phys. Rev. E 75, 061403 (2007). ^ J.-P. Hansen and H. Lowen, Annu. Rev. Phys. Chem. 51, 209 (2000). ^° G. Vernizzi, G. I. Guerrero-Garcia, and M. O. de la Cruz, Phys. Rev. E 84, 016707 (2011). ^^ M. Tamashiro, Y. Levin, and M. Barbosa, Eur. Phys. J. B 1, 337 (1998). ^^ M. Deserno and H. H. von Griinberg, Phys. Rev. E 66, 011401 (2002). " L. Belloni, Colloids and Surfaces A 140, 227 (1998). ^'' P. Gonzalez-Mozuleos and M. O. de la Cruz, Physica A 387, 5362 (2008). 1^ C. N. Likos, Physics Reports 348, 247 (2001). ^ A. Denton, in Nano structured Soft Matter: Experiment, Theory, Simulation and Perspectives (Springer, Dordrecht, 2007). " A. R. Denton, Phys. Rev. E 67, 011804 (2003). ^^ F. H. Stillinger, H. Sakai, and S. Torquato, J. Chem. Phys. 117, 288 (2002). ^^ A. A. Louis, J. Phys.: Condens. Matter 14, 9187 (2002). ^" M. Dijkstra, R. van Roij, and R. Evans, J. Chem. Phys. 113, 4799 (2000). ^^ E. Trizac et ai, Phys. Rev. E 75, 011401 (2007). ^^ B. V. Derjaguin and L. Landau, Acta Physicochim. URSS 14, 633 (1941). ^^ E. J. Verwey and J. T. G Overbeek, Theory of the Stability of Lyophobic Colloids (Elsevier, Amsterdam, 1948). 2" B. Zoetekouw and P. R. E. R. van Roij, Phys. Rev. E 73, 021403 (2006). ^^ D. Y. C. Chan, P. Linse, and S. N. Petris, Langmuir 17, 4202 (2001). ^■^ F. Stillinger and R. Lovett, J. Chem. Phys. 49, 1991 (1968). ■^^ H. L. Friedman and P. S. Ramanathan, J. Phys. Chem. 74, 3756 (1970). 2* P. G Kusalik and G N. Patey, J. Chem. Phys. 86, 5110 (1987). ^^ E. Trizac and Y. Levin, Phys. Rev. E 2004, 031403 (69). ^'^ S. Pianegonda, E. Trizac, and Y. Levin, J. Chem. Phys. 126, 014702 (2007). ^^ J.-P Hansen and I.R. McDonald, Theory os Simple Liquids (Academic Press, New York, 1976). ^^ J. G Kirkwood and F. P. BuflF, J. Chem. Phys. 19, 774 (1951). ^^ R. Castaneda-Priego, L. F. Rojas-Ochoa, V. Lobaskin, and J. C. Mixteco-Sanchez, Phys. Rev. E 74, 051408 (2006). 11 ^* T. E. CoUa, Y. Levin, and E. Trizac, J. Cliem. Phys. 131, 074115 (2009). ^^ T. E. CoUa and Y. Levin, J. Chem. Phys. 133, 234105 (2010). ^^ J. M. Falcon-Gonzalez and R. Castaiieda-Priego, Phys. Rev. E 83, 041401 (2011). ^^ J. M. Falcon-Gonzalez and R. C. Priego, J. Ghem. Phys. 133, 216101 (2010). ^* H. Wennerstrom, B. Jonsson, and P. Linse, J. Ghem. Phys. 76, 4665 (1982). ^® M. Deserno and G. Holm, in "Electrostatic Effects in Soft Matter and Biophysics", eds. G. Holm, P. Kekicheff, and R. Podgornik, NATO Science Series II - Mathematics, Physics and Ghemistry 46, (2001). ^° S. L. Garnie and D. Y. G. Ghan, Ghem. Phys. Lett. 77, 437 (1981). "1 R. A. Marcus, J. Ghem. Phys. 23, 1057 (1955). ^'^ F. G. Donnan, Ghem. Rev. 1, 73 (1924). *^ R. Behera, J. Ghem. Phys. 108, 3373 (1997). ■*■* M. P. Allen and D. J. Tildesley, Computer Simulations of Liquids (Oxford University Press, Oxford, 1987). "■■^ J. Z. Wu, D. Bratko, H. W. Blanch, and J. M. Prausnitz, J. Ghem. Phys. Ill, 7084 (1999). ■"^ P. Attard, Avd. Ghem. Phys. 92, 1 (1996). "■^ V. Lobaskin and K. Qamhieh, J. Phys. Ghem. B 107, 8022 (2003). "^ M. Rasa et al, J. Phys.: Gondens. Matter 17, 2293 (2005). ^^ L. Belloni, J. Ghem. Phys. 123, 204705 (2005). ^" J. Dobnikar, R. Gastafieda-Priego, H. von Griinberg, and E. Trizac, New Journal of Physics 8, 277 (2006). ^'^ Y. Levin and M. Fisher, Physica A 225, 164 (1996). ^^ J. G. Rasaiah and H. L. Friedman, J. Ghem. Phys. 48, 2742 (1968).