Nonlinear Electrostatics. The Poisson-Boltzmann Equation
NNonlinear Electrostatics. The Poisson-Boltzmann Equation
C. G. Gray* and P. J. Stiles * Department of Physics, University of Guelph, Guelph, ON N1G2W1, Canada ([email protected])
Department of Molecular Sciences, Macquarie University, NSW 2109, Australia ([email protected])
The description of a conducting medium in thermal equilibrium, such as an electrolyte solution or a plasma, involves nonlinear electrostatics, a subject rarely discussed in the standard electricity and magnetism textbooks. We consider in detail the case of the electrostatic double layer formed by an electrolyte solution near a uniformly charged wall, and we use mean-field or Poisson-Boltzmann (PB) theory to calculate the mean electrostatic potential and the mean ion concentrations, as functions of distance from the wall. PB theory is developed from the Gibbs variational principle for thermal equilibrium of minimizing the system free energy. We clarify the key issue of which free energy (Helmholtz, Gibbs, grand, …) should be used in the Gibbs principle; this turns out to depend not only on the specified conditions in the bulk electrolyte solution (e.g., fixed volume or fixed pressure), but also on the specified surface conditions, such as fixed surface charge or fixed surface potential. Despite its nonlinearity the PB equation for the mean electrostatic potential can be solved analytically for planar or wall geometry, and we present analytic solutions for both a full electrolyte, and for an ionic solution which contains only counterions, i.e. ions of sign opposite to that of the wall charge. This latter case has some novel features. We also use the free energy to discuss the inter-wall forces which arise when the two parallel charged walls are sufficiently close to permit their double layers to overlap. We consider situations where the two walls carry equal charges, and where they carry equal and opposite charges.
INTRODUCTION
Most electricity and magnetism textbooks, and pedagogical journal articles, discuss only linear electrostatics. That is, the charge density of the medium at a point x is either assumed to be fixed, or to be perturbed linearly by the electric field ( ) E x . The subject of nonlinear dielectric media has a large literature which is mostly confined to specialist journals and monographs . Another important case involving nonlinear electrostatics is a conducting medium in thermal equilibrium such as an electrolyte solution or a plasma. For example, a solution of sodium chloride dissolved in water produces hydrated ions Na and Cl . In the absence of an applied field, the solution is spatially uniform and overall neutral but locally the Cl ions tend to cluster around the Na ions, and vice-versa, resulting in a rapid screening out of the electric field around any given ion. The screening occurs essentially over a characteristic distance D , called the Debye length, where BD 2 k Te c . (1) Here r 0 is the permittivity of water (we use SI units), r the dimensionless dielectric constant ( r is about 80 for water at room temperature), the permittivity of free space, e is the magnitude of the electron charge, B k is Boltzmann’s constant, T the temperature, and c the bulk concentration or number density (in ions per m ) of the Na ions (or the Cl ions). Thus, each ion is, on average, surrounded by a spherically symmetric screening cloud of radius of order D , which is about 1 nm at room temperature for a bulk ion concentration of 100 mM. As we discuss later, we have here assumed the so-called primitive or implicit solvent model with the solvent (assumed to be water in this example) taken to be a dielectric continuum with permittivity , and we have also assumed point ions. A key assumption is that mean field theory applies, i.e., we have assumed that the electrostatic field at a position x from a central ion is the Coulomb field of the central ion together with the thermal average field due to all the other ions. Any ion near the central one is assumed to sense this total mean field. The ion-ion correlations in a screening cloud due to Coulombic interactions are neglected. Eq (1) was derived by Debye and Hückel based on the assumptions just stated. One writes down two equations: first the rigorous Poisson equation relating the total mean potential ( ) x at a position x from a central ion to the total mean charge density ( ) x due to the central ion (here called “fixed” for reasons clarified in the next paragraph), and to the other (here called “mobile”) ions, ( ) ( ) / x x , (2) with total mean charge density f m ( ) ( ) ( ) x x x , (3) where f ( ) x is the charge density of the fixed ion (a delta function, e.g., ( ) e x for a central positive ion (cation)) and m ( ) x is the local mean charge density of the mobile ions. By definition, we have m ( ) ( ) ( ) e c e c x x x , (4a) where ( ) c x are the local mean concentrations of the mobile ions, and, assuming Boltzmann relations in the approximate form ( ) exp ( ) c c e x x we thus have, m ( ) exp ( ) exp ( ) e c e e c e x x x , (4b) where B k T and c is the bulk concentration of the electrolyte at large distances from the central ion where ( ) x is assumed to approach zero. From (2) we see that depends on , and from (4b) depends on , so that we have two equations that must be solved self-consistently for these two quantities. Eqs (2-4) can be combined into a single nonlinear partial differential equation for ( ) x , called the Poisson-Boltzmann (PB) equation: ( ) ( / ) exp exp (1/ ) ( ) e c e e x x x x . (5) For weak electric fields one can linearize (5) in ( ) x , thereby deriving the linearized Poisson-Boltzmann (LPB) equation, ( ) ( ) (1 / ) ( ) x x x , (6) where D , as first done by Debye and Hückel and thus also called the DH equation. Because of the spherical symmetry of simple ions, the various functions of x above are in fact functions of just the radial coordinate r x , and (5) and (6) reduce to ordinary differential equations with the radial part of the Laplacian, / / (2 / ) / r d dr r d dr r d dr , replacing . We keep the general forms, however, for reasons that will become clear in the next paragraph. For f ( ) ( ) e x x the appropriate solution of (6) is easily verified to be ( ) / 4 exp( ) r e r r , variously called the DH, screened Coulomb or Yukawa potential, and which shows explicitly that D is the screening length, at least for weak fields. It is also the Green function for the operator occurring in equation (6). Interestingly, the full nonlinear PB equation (5) was derived before Debye and Hückel by Gouy and Chapman who considered the problem of a diffuse double layer formed by an electrolyte solution in the presence of a charged wall, e.g., an electrode or a charged biomembrane. The only formal change is that in this case the fixed charge density f ( ) x is now that due to the charged wall, taken in the simplest case to be a wall with uniform surface charge density , and x now represents an arbitrary point in the fluid. In this system the screening cloud due to the mobile ions forms a diffuse layer with net charge equal and opposite to that of the wall charge, thus producing the double layer, of infinite extent but decaying essentially exponentially as a function of distance from the wall with effective width of the order of D for not too large . (We assume here an infinite system with one charged wall. Later we also consider finite systems between two charged walls.) We will focus on (5) in the present paper for this double layer. In this case again because of system symmetry ( ) x and ( ) x reduce to functions of just x (the coordinate orthogonal to the charged wall), the full Laplacian can be replaced by / d d x , and (5) reduces to an ordinary nonlinear differential equation. As we shall see, despite its nonlinearity it can be solved analytically for this Gouy-Chapman (wall) geometry. For other geometries, e.g., charged cylindrical or spherical surfaces, numerical methods are needed for the full nonlinear PB equation even for the standard simple geometries . Such nonplanar geometries have been used for modelling double layers around cylindrical electrodes, colloids, and charged macromolecules such as proteins and DNA. As mentioned, standard textbook and pedagogical journal discussions of the PB (and LPB) equations are uncommon. But because of the importance of electrostatic screening in many systems, there are discussions in a host of monographs for specialized fields including biophysics
9, 23-34 , surface science , chemical physics , polymer physics , plasma physics , solid state physics , condensed matter physics , many-body theory , thermodynamics , statistical mechanics , liquid state theory , electrolyte solutions , electrochemistry , soft matter , physical chemistry , biophysical chemistry , biochemistry , medical physics , physiology , molecular biology , colloids , applied mathematics , materials science and technology . There is also a large literature on theories going beyond primitive model PB by including ion-ion correlations, finite ionic size and molecular structure of the solvent (assumed above to be water), nonuniformities of the wall charges, additional intermolecular and wall forces. We do not discuss these improvements here, nor do we discuss the quantum generalization of classical PB theory, but we indicate in the references some quantum discussions . Thus, for electrons in bulk metals and at metallic surfaces, at not too high temperatures the relevant mean field generalization is the Thomas-Fermi (TF) theory of screening, where the screening length TF is essentially (1) with B k T replaced by the Fermi energy F E . We focus on the double layer problem and carry out two main tasks. Firstly, we give a more systematic and physically rigorous derivation of (5) based on the Gibbs variational principle for thermal equilibrium , i.e., minimizing the system free energy. This has been done before , but we simplify and clarify earlier derivations by carefully identifying which free energy (e.g., Helmholtz, Gibbs, grand, …) is to be minimized. It turns out this depends not only on the assumed conditions of the bulk solution (e.g., fixed ion numbers, volume, and temperature), but also on the surface or boundary conditions chosen for the system, e.g., fixed wall charge or fixed wall potential. Secondly, we use free energy to derive the force on one of the walls when two opposing parallel walls of the container, or two fully immersed parallel surfaces, each has a double layer and the walls or surfaces are sufficiently close to permit overlap of their respective double layers. We consider in detail two walls of identical charge and then discuss more briefly two walls of opposite charge . We also briefly discuss the use of free energy to correct a recent PB calculation which erroneously predicted that the double layer collapses to finite width as the surface charge on a wall is raised beyond a threshold value. Clarke and Stiles were led astray in part by using a free energy inappropriate for their assumed boundary condition. The paper also includes the analytical solution to the PB equation for a single charged wall, both for a full electrolyte solution and for a solution containing only counterions; the latter case has some novel features. In the case of a single charged wall we also use PB theory to derive simple analytical expressions for observables such as the equilibrium values of various free energies and the differential capacitance. For two charged walls the appendices give convenient algorithms for determining the quantities necessary to calculate the interwall-force; the algorithms provide a simpler route to the required inputs to the force calculations than the standard procedure of fully solving the PB equation numerically. Numerical and graphical examples are given to illustrate some of the results for the one- and two-wall problems.
II. DERIVATION OF THE PB EQUATION FROM THE HELMHOLTZ FREE ENERGY F We consider a 1:1 univalent electrolyte solution with N cations and N anions, in thermal equilibrium at a temperature T in a cubic container of volume V L . The left wall at x is assumed to have a fixed uniform positive charge with surface density . This surface density may arise from various sources, e.g. from preferential adsorption of cations from the solution, or from the positive plate of a battery. A diffuse double layer forms at the x wall. In the simplest manifestation of the primitive model all ions in solution are represented as point charges and the solvent is replaced by a continuous dielectric fluid of uniform electric permittivity . The Helmholtz free energy F of this non-uniform system can be written as a volume integral of a free energy density ( ) f x , i.e., ( ) V F f dV x . (7) The free energy F U T S contains energetic and entropic components, where U is the internal energy and S the entropy. We approximate both components and write el id ( ) ( ) ( ) f f f x x x , where ( ) ( ) ( ) f x x x (8) is the electrostatic potential energy in the mean-field approximation in which V dV xx x x (9) is the mean Coulomb potential at x due to the mean charge density of the mobile electrolyte charges and the fixed wall charges. The ideal term ( ) ( ) ln ( ) 1 ( ) ln ( ) 1 f c c c c x x x x x (10) contains the thermally averaged kinetic energy and the ideal-gas Sackur-Tetrode entropic terms (which contain no entropic correlation) for the solution cations and anions. In (10) the thermal de Broglie wavelength B / 2 h m k T where h is the Planck constant and all ions are assumed to have equal mass m . We now introduce a Helmholtz free energy [ , ] F c c as a functional of the cation and anion density fields ( ) c x and ( ) c x (in ions per m ): other fields can be expressed in terms of them. In (10) id ( ) f x is already expressed in terms of c and c . To express el ( ) f x in terms of these ionic concentrations we combine (8) and (9) and leave implicit that the charge density is expressed in terms of the concentrations by (3) and (4a). Our functional [ , ] F c c is thus ( ) ( )[ , ] ( ) ln ( ) 1 ( ) ln ( ) 18 | |
V V V
F c c dV dV c c c c dV x x x x x xx x (11) By the Gibbs variational principle , the free energy functional, here expressed as a density or concentration functional, takes its equilibrium value for the values of the ionic densities that minimize it, subject to the constraints of fixed numbers ( ) V N c dV x of cations and ( ) V N c dV x of anions in the solution. We use the standard method of Lagrange multipliers to perform this constrained minimization. Thus, we extremize a functional [ , ] [ , ] V J c c F c c c c dV with no constraints where and are constant Lagrange multipliers and ( ) x is expressed in terms of the ionic densities by (3) and (4a). Varying c and c we thus find for [ , ] J c c : ( ) ( ) ( ) ( )8 | |( ) ln ( ) ( ) ln ( ) . (12) V VV
J dV dVc c c c dV x x x xx xx x x x
The two terms in the double integral are equal by symmetry, and in the first term we eliminate ( ) x by re-introducing ( ) x using (9) to give ( ) ( ) ( ) ln ( ) ( ) ln ( ) . V J dV c c c c x x x x x x (13) From (3) and (4a), we have ( ) ( ) ( ) e c c x x x . As we require J for arbitrary variations ( ) c x and ( ) c x in both ( ) c x and ( ) c x , these two conditions yield two Euler-Lagrange equations ( ) ln ( ) , ( ) ln ( ) e c e c x x x x , (14a,b) that we recognize as the constant electrochemical potential conditions for the two ionic species in the solution. Thus the and are the ionic electrochemical potentials usually denoted by and , respectively. We thereby see the physical significance of : J J F N N is the grand free energy or grand potential (denoted by F in Section IV.2.2 and Appendix B). We solve (14a,b) for ( ) c x to obtain ( ) exp ( ) c e x x . As ( ) x is short-ranged, we note that when x , ( ) 0 x so that ( ) c x takes the its limiting bulk-solution value exp which we denote by c . Similarly, exp c , so that we can eliminate and the 's in (14a) and (14b) to obtain the standard Boltzmann distributions for the cation and anion concentrations in a 1:1 electrolyte 0 ( ) exp ( ) c c e x x , (15a,b) in agreement with (4b), as we shall see next that c c . For a positive uniformly charged wall, global electroneutrality requires S e N N dS , (16) where N and N are the numbers of cations and anions in solution and S is the x surface with the fixed charge. Divide (16) by V L and note that the right side is O(1 / ) L compared to the left, and thus negligible in the thermodynamic limit ( , N , V / N V c fixed). Thus, in this limit we have c c c . As all ionic interactions are assumed to be Coulombic, the Poisson equation ( ) ( ) ( ) x x x is valid, so with (4a) and (15a,b) it becomes the Poisson-Boltzmann equation: ( ) exp ( ) exp ( ) ( ) e c e c e x x x x , (17) in agreement with (5) in the thermodynamic limit where c c c . In applying the Gibbs variational principle above, we have assumed the laws of electrostatics (Poisson, Coulomb) and have used the variational principle to derive the “Boltzmann” part of PB theory. We can also derive the “Poisson” part from the variational principle by introducing an extended Helmholtz free energy functional [ , , ] F c c depending on three variables and minimizing with respect to all three. Thus, we use the identities ( ) ( ) ( ) ( ) ( ) f x x x x x and rewrite F as
21 3 3
1[ , , ] ( ) ( ) ( )2 ( ) ln ( ) 1 ( ) ln ( ) 1 , (18)
V VV
F c c dV dVc c c c dV x x xx x x x
1 where depends on c as before. We extremize this F with the constraints of fixed N , or [ , , ] [ , , ] V J c c F c c c c dV with no constraints. Variations with respect to c and c yield the Boltzmann relations ( ) exp ( ) c c e x x as before and the variation with respect to now yields the Poisson equation ( ) ( ) / x x . In this section we have demonstrated that the Helmholtz free energy is the relevant free energy to use in the Gibbs variational principle when the bulk concentrations require fixed numbers of ions, and fixed volume and temperature, and the boundary conditions require fixed surface charge. In later sections we show that not only a change of bulk conditions (to, for example, fixed ionic chemical potentials, or fixed pressure) necessitates the use of a Legendre transformed free energy in the variational principle, but so does a change of boundary condition from fixed surface charge to fixed surface potential. III. ANALYTICAL SOLUTION OF THE PB EQUATION FOR WALL GEOMETRY
We again consider a univalent 1:1 electrolyte adjacent to a uniform positively charged wall of infinite extent and simplify the PB equation (5) to the form of a one-dimensional or ordinary differential equation (with x and a prime denoting / d dx ) ( ) (2 / ) sinh x e c e x . (19) We assume the boundary surface at x has uniform charge density . On multiplying (19) throughout by ( ) x we obtain
2( ) cosh ( )2 d c dx e xdx dx , (20) which is readily integrated to yield the first integral 2 ( ) 2 cosh ( ) / 2 cosh2 x c k T e x const c k T e , (21) where the last form follows from taking x and using the boundary condition x (0) (0) / E , with (0) . On noting that the electric field vanishes at infinity so that ( ) 0 and that our reference potential requires ( ) 0 , we see that an equivalent expression for the constant of integration is B const c k T . Using this value for const and integrating / d x d from x to x , we find the inverse form of the solution of (21), x , to be / x d , with
11 22 B c k T e where we have taken the negative square root here to ensure (0) 0 for . For a general : z z electrolyte this integral form would require, when generalized, elliptic integrals , or the equivalent direct form of the solution, ( ) x , involving Jacobi elliptic functions . In the case of a 1:1 electrolyte we get the relatively simple inverse form exp exp / 2 1 / exp / 2 1 x e e , corresponding to the direct form (which can be verified to satisfy the PB equation (19) or the first integral (21)) B k T xx e x , (22a) in which tanh( / 4) e and the electrolyte screening constant D is the reciprocal of the Debye screening length (1). The ionic concentrations, that follow from the Boltzmann equations ( ) exp ( ) c x c e x , are exp( )( ) exp( ) xc x c x . (22b) For weak fields (i.e., small or small ), (22a) reduces to the standard exponential expression ( ) exp x x which also follows directly as the relevant solution of the one 3 dimensional version of the DH equation (6) in the form (with x ) ( ) ( ) x x . In this limit the ionic concentrations reduce to ( ) 1 exp c x c e x . In the limit of large x with arbitrary, the limiting values of and c are B ( ) 4 / exp x k T e x and ( ) 1 4 exp c x c x . In Fig.1 we show two plots of the PB and LPB/DH curves for ( ) x . The Debye screening length is D nm in both panels. The surface potentials are
40 mV (corresponding to a surface charge density -2 obtained from (23) below) in panel A, and
200 mV (corresponding to -2 ) in panel B. Assuming we introduce a second characteristic length GC B k T e , the Gouy-Chapman length which characterizes the strength of the surface charge. (The length GC will reappear in section IV.2.6.) Thus, in panel A with GC we have GC D ,whereas in panel B with GC we have GC D . The DH curve decays over a distance D in both panels. In panel A, with small or large GC , the decay length of the PB curve is only slightly less than D . By contrast, in panel B with large or small GC , the decay length of the PB curve is significantly smaller than D and lies between GC and D . In this latter case, the stronger surface charge density pulls the counterions in more closely, causing the potential to be screened more effectively than predicted by linear DH theory, where the screening length D is independent of . In the limit where and approach infinity with D fixed, we have and GC , and the PB decay length approaches zero together with GC . 4 Figure 1. Plots of ( ) x vs x for the PB and DH theories. In both panels we have T , r ,
100 mM c and D . In panel A we assume
40 mV (corresponding to ) and in panel B
200 mV (corresponding to ). On equating our two equivalent expressions for const from the first integral we immediately obtain the Grahame equation c k T e which simplifies to c k T e , (23) 5 an important analytic result using PB theory to predict how the surface charge density is related to the surface potential . An experimentally related result known as the Lippmann equation enables us to determine the surface charge density of an electrified metallic droplet, such as liquid mercury, from the isothermal derivative of the surface tension with respect to the surface potential . Following Verwey and Overbeek , we note that the change in Helmholtz free energy F accompanying isothermal increases in the surface area A and the surface charge q of an electrified liquid metal is given by dF dA dq . Upon performing a Legendre transform from F to the new free energy F (which also plays a key role in section IV.2.4), F F q , (24) we find that dF dA q d . As this differential is exact we immediately obtain the Lippmann equation TA T qA . (25) It predicts that the variation of the surface tension with respect to the surface potential passes through a maximum value when the surface is uncharged, i.e., when there is no electric double layer. The Lippmann equation therefore predicts that an electric double layer reduces surface tension at the metal/electrolyte interface. One of the most studied observables for electric double layers is the differential capacitance defined by / / A T A T C , (26) where the second form in (26) follows from (25). An analytical expression for d C in PB theory therefore follows from the Grahame expression (23) as 6 C e c e . (27) The limiting value of (27) for small is
2d D
C e c . Comparing this value with the expression / for the capacitance per unit plate area of a standard parallel plate capacitor, where is the plate separation, we see that the effective ‘plate separation’ for the small-field double layer capacitor is D , a result we might have anticipated physically since D is the effective width of the double layer for small surface potential. For larger applied fields, from the parabola-like shape of cosh( / 2) e we see that d C is predicted to show a minimum at . This parabolic behavior is in fact observed near at low electrolyte concentrations c but deviations are observed at large values of and c . Explaining such deviations requires extensions of PB theory and is an area of current research (see Kornyshev ). IV. FREE ENERGY AND FORCES BETWEEN TWO CHARGED PLATES
Derivation of the force between two parallel charged plates separated by an initially homogeneous electrolyte solution can be based on a hydrostatic equilibrium argument and we give this in Appendix A. These two plates can take an assigned separation d by applying external constraint forces to them. In this paper we consider forces that result from the PB primitive model alone. When the two plates are immersed in a large reservoir of electrolyte so that each plate experiences forces from both the internal and external electrolyte (our scenario B) the inter-plate force can be deduced from PB theory most simply by using the grand free energy as indicated in subsection IV.2.2 and Appendix B. If, on the other hand, the electrolyte is confined to the region between the two plates and contains fixed numbers of 7 cations and anions - our scenario A - we can derive the relevant inter-plate forces for this two-plate problem straightforwardly from the Helmholtz free energy of the system as demonstrated next.
1. Scenario A. Electrolyte solution between plates of fixed charge. Helmholtz free energy and inter-plate force.
We initially consider two parallel square plates of area A separated by a fixed distance 2 d , each plate carrying the same constant surface charge density , with the electrolyte solution confined to the region between the plates. The remaining containing walls are assumed uncharged. We choose the x -direction perpendicular to the plates, with one plate defining the x plane and the other the x d plane. We shall later consider the plate in the x plane to have a uniformly positive surface charge density and the plate in the x d plane to have a uniformly negative surface charge density . We show that in the latter case the inter-plate force is attractive at small plate separations but is purely repulsive in the former case when the two plates have equal charges. We shall also consider systems of varying size d , with fixed T and . To maintain constant solvent density as d increases we add extra solvent. For these two-plate problems our derivation in Section II of the Boltzmann equation for the case of the thermodynamic limit requires slight modification for our new system with finite dimension in the x direction. For the infinite system considered previously we chose x as the reference point for which we assumed ( ) 0 and for our current finite system we now choose x d as the reference point for which we assume that ( ) 0 d . From (14a,b) we saw that ( ) exp ( ) c e x x so that we have 8 d 3 exp c where d ( ) c c d . We can again eliminate and from these relations to get d ( ) exp ( ) c x c e x , (28a,b ) relevant to cations and anions in the two-plate scenario A problem. The PB equation (5) or (17) generalizes, from an argument analogous to that in Section II, to the two-plate form d dm ( ) ( ) exp ( ) exp ( ) x x e c e x c e x , x d . (29) We will consider the boundary conditions to be fixed surface charges, i.e., (0 ) / , (2 ) / d where , the charge density on both plates, is assumed to be positive. Later we indicate the required changes in the force calculation when fixed surface-potential boundary conditions are assumed. In our scenario A the temperature T , the system volume V , and the total numbers of cations N and anions N are fixed and the electrolyte is restricted to the region between the two plates. All fields (e.g., ( ), ( ), ( ), . c x x x etc ) depend only on x provided that the plate areas are so large that edge effects can be neglected. The Helmholtz free energy F can again be used to define the total free energy density ( ) f x through / ( ) d F A f x dx , (30) where, as before, el id ( ) ( ) ( ) f x f x f x . The ideal solution term, defined by (10), can be recast, using Boltzmann relations (28a,b) and the definitions m f ( ) ( ) ( ) ( ) ( ) e c x c x x x x , and then added to el ( ) f x to give the total free energy density in the form d 3 d 3el f B B ( ) ( ) ( ) ( ) ln 1 ( ) ln 1 ( ) f x f x x x k T c c x k T c c x . (31) 9 We shall now show that minus the derivative of / F A with respect to d gives the same expression as the force on the plate at x d as can be obtained by a hydrostatic equilibrium argument (see eq. (A5) of Appendix A). The free energy density (31) is symmetric about the mid-plane x d so that
20 0 / ( ) 2 ( ) d d
F A f x dx f x dx . Thus, the force per unit area on the plate at x d , f ( / ) / (2 ) F A d , can be obtained from
2d 0 f / ( ) d d f x dx which, from (31), becomes explicitly d dB B2d el f d d0 0 0 0d dB Bel f d d0 0 0 0 f ( ) ( ) ( ) ( ) ( )( )( ) ( ) ( ) ( ) , (32) d d d dd d d d k T c k T cf x dx x x dx c x dx c x dxd c d c dk T c k T cxf x dx x dx c x dx c x dxd d c d c d where we have used the facts that f ( ) x and the integral ( ) ( ) d c x c x dx are independent of d . The latter result, re-expressed as / ( ) ( ) 0 d d c x c x dx together with the Boltzmann equations (28a,b) and the Leibniz rule, gives d d d dB B md d0 0 0 ( )( ) ( ) ( ) d d dB k T c k T c xc x dx c x dx k T c c x dxc d c d d . (33) Equation (32) therefore takes the form d d2d el B0 0 ( )f ( ) ( ) d d xf x dx k T c c x dxd d . (34) From (A5) based on a hydrostatic equilibrium argument we expect f to be equal to the second term in (34), so that the other terms should cancel, as we now show. Using / / / d d d and el f we find d d2d 0 0 ( )f ( ) ( ) d del B xf x dx k T c c x dxd d . (35) 0 The identity ( ) ( )2 f x x in conjunction with the Leibniz rule gives ( ) ( )f ( ) ( ) ( )2 d dB x xd x dx k T c c x dxd d . (36) Using Poisson’s equation for ( ) x and integrating the last integral on the right-hand side by parts we get ( ) ( ) ( )( ) ( ) ( ) dd d x x xx dx x x dxd d d , (37) in which both boundary terms ( ) / d d and (0) / d are zero because the boundary electric fields are independent of the inter-plate spacing d . Thus, (36) in conjunction with the symmetry condition x ( ) ( ) 0 E d d , simplifies to d d2d f B k T c c , (38) and agrees with the expression (A5) derived from a hydrostatic equilibrium argument. This inter-plate force (38) is always repulsive and depends only on the mid-plane ionic concentrations. Fig.2 The inter-plate force f vs plate separation d for scenario A with T , r , , and
16 2 / 8 10 m
N A . From electroneutrality
16 2 / / 2 / 3 10 m
N A N A e . 1 In Fig.2 we show an example of the inter-plate force as a function of the plate separation d . In Appendix C we give algorithms to determine the quantities d c and d c used to compute this force from (38). The force is purely repulsive. It can be shown to behave as d at large separations and thus has no well-defined range. When the two plates have surface charge densities equal in magnitude and of opposite sign, we have ( ) 0 d and d d d c c c . The sign of the inter-plate force is not now obvious, but as we shall discuss, the inter-plate force d 22 f 2 / 2 ( ) d B k T c E d is repulsive when the two plates are far apart and attractive when the two plates are close together. Finally, it is worth recording the first integral of the PB equation (29) for scenario A, which can be written as / 2 ( ) ( ) ( ) . x k T c x c x const (39) Interestingly, (39) is equivalent to the hydrostatic equilibrium relation (A4) and thus provides an alternative derivation of that relation. Again consulting Appendix A and choosing
0, 2 , x d and d in (39), we find that the constant of integration in (39) takes values f , f and d dB k T c c , respectively, when the charges on the plates are identical. These equivalent values for const also give the two-plate Grahame relation / (2 ) exp 1 exp 1 k T c e c e , (40) a direct generalization of the single-plate Grahame relation (23) (see also the version in the line above (23)) and to which it reduces in the limit as d , , N N , with / (2 )
N Ad and / (2 ) N Ad fixed at c . This is not an explicit relation between and as d c depend on ; an explicit algorithm to determine d c is given in Appendix C. This Grahame equation also differs from those of other scenarios, such as two plates totally 2 immersed in the electrolyte fluid (discussed in Appendix B), two plates with opposite potentials (see Appendix C), etc.
2. Other Scenarios
Here we consider briefly several other interesting scenarios involving electric double layers and their interactions.
Scenario B
In this scenario the two parallel plates are fully immersed in the electrolyte fluid, in contrast to Scenario A discussed above where the plates form two opposing boundary walls of the fluid. This was the case first discussed in the literature, and we give a detailed description in Appendix B for two identically charged plates. For reasons discussed there we restrict ourselves, in this scenario, to thick non-conducting plates. The original application was to colloid stability; in colloid-colloid interactions, there are long-range attractive dispersion forces, but the short-range repulsive double-layer forces prevent colloid precipitation. Fig.3 The inter-plate force f vs plate separation d for scenario B with T , r , , and
10 mM. c
3 We show in Fig.3 an example of the inter-plate force f calculated from (B11) as a function of plate separation d . This force is seen to be purely repulsive, as in Fig.2 for scenario A. Here, however, in contrast to scenario A, the force falls off essentially as D exp 2 / d at large d with a well-defined range D which is about 3 nm. In Appendix C we give an algorithm for the quantity d needed to compute this force from (B11). The case of plates with unequal surface charge density is mentioned below in subsection 2.3 and the case of plates with equal and opposite surface potentials is discussed in subsection 2.4. Grand Free Energy
When the electrolyte between the plates is in equilibrium with a large reservoir of electrolyte, it is often simpler to work with the grand free energy F (also called the grand potential) rather than the Helmholtz free energy F that we employed in section IV.1 for scenario A. The former is related to the latter by a Legendre transformation F F N N , (41) and the control variables for the transformed thermodynamic potential F are the electrochemical potentials rather than the ion numbers N relevant for F . The additional control variables are V , T and for both F and F . For fixed electrochemical potentials the ion numbers fluctuate, so that strictly speaking the N in (41) are average values, but the relative fluctuations are usually negligible for a large system. In Appendix B we employ this grand free energy to analyze scenario B. From F one can derive the PB equation and the inter-plate force by steps paralleling those taken above to derive these quantities from F . In 4 subsection 2.5 below we define and derive an expression for the equilibrium value of the renormalized grand free energy F for a single plate. Plates with Unequal Charge
In general, the left and right plates may have unequal charge,
L R . The general case is discussed by Ohshima for scenario B. For scenario A with oppositely charged plates and fixed surface charge densities, L R , the inter-plate force is given by (A5). We find the numerical calculations simpler if we employ the fixed surface potential boundary conditions considered in the next subsection; the same expression (A5) for the inter-plate force holds for fixed equal and opposite surface potentials.
Fixed Surface Potential Boundary Conditions
In our scenarios A and B we assumed fixed surface charge boundary conditions (fixed ). In some problems, however, it is more convenient or more appropriate to assume fixed surface potential boundary conditions (fixed ). A simple example involves the two plates connected to the two terminals of a battery. The original investigators of scenario B (see Appendix B for references) employed a renormalized free energy (see following subsection) and fixed surface potential boundary conditions. The explicit boundary conditions assumed by these original investigators were fixed on both plates; with a battery supplying the external potentials the boundary conditions become fixed on one plate and fixed on the other. The relevant free energy F for a fixed potential boundary condition at a single plate, together with fixed ( , , N V T ), is the Legendre transformation of F defined by (24), or equivalently 5 / / F A F A . (42) Equation (42) has been derived by detailed chemical arguments , but here we give a brief alternative heuristic physical argument. Compare the Legendre transformations (41) and (42). In (41) we transform from fixed bulk ionic numbers N for F to fixed bulk electrochemical potentials for F , with N then free to fluctuate. In (42) we transform from fixed surface charge density (fixed ) for F to fixed surface potential for F , with the surface charge q A then free to fluctuate. The role of the intensive electrostatic potential in the surface Legendre transformation (42) is analogous to the role of the intensive electrochemical potentials in the bulk Legendre transformation (41) of the grand free energy F . Thus, F is a surface grand free energy analogous to the bulk grand free energy F and in both cases the potentials ( and ) are the natural control variables for grand free energies. (A more formal argument is given in the line below (24).) To derive the PB equation and the inter-plate force from a free energy, F is the appropriate one to use if the bulk conditions are fixed numbers of ions N , volume and temperature, and the boundary conditions are fixed surface potentials. If the bulk conditions are fixed electrochemical potentials together with fixed ( , ) V T and the boundary conditions are fixed surface potentials, the appropriate free energy for the derivation of the PB equation and the inter-plate force is the combined Legendre transform of (41) and (42),
F F q F N N q . An analytic expression for the equilibrium value of the renormalized (see next subsection) version F of F for the single plate problem, in terms of its natural variable is seen from (53) to be
0B D / 8 cosh / 2 1
F A c k T e . 6 We turn now to the inter-plate force for a system with fixed electrochemical potentials and fixed surface-potential boundary conditions. From Section III and Appendices A and B, Figure 4. The inter-plate force f as a function of the inter-plate separation d assuming fixed surface potentials
10 mV on the left plate and
10 mV on the right plate. Panel A refers to scenario A (with
16 2 / / / 2.409 10 m
N A N A N A ) and panel B to scenario B (with
10 mM c ). the expressions used to calculate the inter-plate forces in Fig. 4 are f ( ) / 2 ( ) p d d in scenario A, and f / 2 ( ) d in B, where d d B ( ) p d c c k T is the mid-plane pressure, and as d by symmetry, d d d c c c . The reason why ( ) p d is absent in B, which leads to a purely attractive force, is that in scenario B (unlike A) the fluid on the outside of the fully immersed plate at x d exerts a negative pressure force B c k T on 7 the plate (expected physically, and shown formally in (B10)), and this cancels the ( ) p d term. In scenario B we have d d c c c as d d / exp c c e with d by symmetry, so that B ( ) 2 p d c k T . Appendix C gives the algorithms used to calculate the inputs to the force calculations, i.e., d c and ( ) d for Fig. 4A, and ( ) d for Fig. 4B. The inter-plate force is seen to have both attractive and repulsive regions in A, but is purely attractive in B. Just as pure repulsion is expected physically in Figs 2 and 3 (like charges repel), the pure attraction seen in Fig. 4B is not unexpected (unlike charges attract). The repulsive region at large d in Fig.4A deserves comment. A rough argument is that for large d , ( ) p d is of order B N Ad k T and ( ) d is of order d . Thus, for large values of d , f ( ) / 2 ( ) p d d is dominated by the repulsive d pressure term. Forces between surfaces of unlike polarity (not necessarily exactly opposite) have been studied by several groups, both experimental and theoretical . With one exception, outlined below, the scenario considered by the theoretical workers is a combination of our scenarios A and B; the mobile ions cannot access the outer surfaces of the plates (as in our scenario A) but in the region between the plates they are exchangeable with ions of a reservoir (as in our scenario B). Our results are thus not directly comparable with those of these other workers, but qualitative similarities occur. When the other workers restrict themselves to primitive model PB theory they find purely attractive forces between plates of unlike polarity, as we do in scenario B. In agreement with experiment, they find repulsion at large plate separations if they extend the theory beyond primitive model PB, or perform simulations, particularly for highly charged surfaces and multivalent mobile ions. Effects due to ion-ion Coulomb correlations, ionic finite size, and nonuniform surface charge, among others, are found to play important roles in generating the repulsive forces. The one 8 exception is the work of McCormack et al who qualitatively discuss the force for surfaces of equal and opposite potential for our scenario B in PB theory and find, as we do, a purely attractive force. Renormalized Free Energy
The quantities , F F , F and F are absolute free energies, as defined in statistical mechanics. It is possible to shift the zero-point of these quantities. If we add a constant to any free energy, this does not affect its variation, or its derivative with respect to d , so the PB equation and inter-plate forces are unaffected. Many workers use such renormalized free energies, sometimes implicitly. A common renormalization is to use the excess free energy over that of the uniform system in the absence of an external electric field (no surface charges), e.g., F F F for the renormalized Helmholtz free energy. With this definition F can be written as el id ( ) ( ) V F f f dV x x , where el ( ) f x is given by any of the forms used earlier, e.g., (8). Restricting ourselves to the equilibrium value of the free energy, and to a single plate, we first find a convenient expression for the equilibrium value of id ( ) f x , the renormalized ideal part of the Helmholtz free energy density. We start with the definition (10) and subtract the corresponding uniform-system value to find ( ) ln 1 ( ) ( ) 2( ) ln ( ( ) / ( ) ln ( ( ) / . (43) f k T c c c ck T c c c k T c c c x x xx x x x We use expression (10) for id ( ) f x to write the uniform-system free energy density f c k T c , and for completeness, record the total free energy F f V c k T c V , (44) 9 of the uniform system. F is seen to be extensive (proportional to V ), as expected. As F is a surface effect, we anticipate and find (see (50)), F to be proportional to the plate area A . The equilibrium Boltzmann relations ( ) / exp ( ) c c e x x , together with the definitions m f ( ) ( ) ( ) ( ) ( ) ec ec x x x x x and el ( ) ( ) 2 ( ) f x x x therefore give ( ) 2 ( ) ( ) ( ) ln 1 ( ) ( ) 2 f f k T c c c c x x x x x x . (45) When we add el ( ) f x to (45) we get the total renormalized free-energy density ( ) f x , ( ) ( ) ( ) ( ) ln 1 ( ) ( ) 2 f f k T c c c c x x x x x x . (46) It is simpler from here on to work with the grand free energy F (related to the Helmholtz free energy F by (41)), and its density ( ) f x , or rather the renormalized values F and ( ) ( ) ( ) ( ) f f c c c c x x x x . From the large distance behaviors of (14a,b) we see that ln k T c , so that we find el f B ( ) ( ) ( ) ( ) ( ) ( ) 2 f f k T c c c x x x x x x . (47) Note that the ln c terms have now been eliminated . Using the first integral (21) of the PB equation with B const c k T (see text below (21)) we can write el ( ) f x as el B B ( ) 2 cosh ( ) 1 ( ) ( ) 2 f c k T e k T c c c x x x x , (48) with the second form in (48) following from the Boltzmann relations for ( ) / c c x . From (47) and (48) we thus have B f ( ) 4 cosh ( ) 1 ( ) ( ) f c k T e x x x x . (49) The corresponding value of ( ) V F f dV x is therefore 0
0B 0 / 4 cosh ( ) 1
F A c k T e x dx , (50) where we have adopted the usual geometry with direction x perpendicular to the surface at x (assumed positively charged with surface density and area A ), and we have evaluated explicitly the surface term. We change the integration variable in (50) from x to using / dx d with given in the text below (21), to get
0B 0 / 2 cosh 1
F A c k T e d . (51) Another change of variable from to y e gives
01 12 2 / 2 / cosh 1 y F A c k T k T c e y dy , (52) where y e . The integral cosh 1 y y dy has the value y so that the equilibrium value of the renormalized grand free energy is / 8 cosh / 2 1 F A c k T e . (53) In (53) F is expressed in terms of both and ; to express it entirely in terms of its natural variable we would use the Grahame relation (23) between and . From (53) we can immediately write down expressions for the complete grand free energy F , and for other renormalized free energies such as / F A and / F A . As noted in the preceding subsection, the latter quantity, expressed as a function of its natural variable is just the first term in (53), i.e., (53) without the surface term . This is the free energy derived by Verwey and Overbeek but they fail to make the point that this free energy is / , F A not / . F A Corresponding derivations of various equilibrium free energies for two plates can also be carried out but the results are much more complicated . Despite these complications 1 for the free energies themselves, simple expressions for free-energy derivatives with respect to the inter-plate distance, which give the inter-plate forces, can be derived, as discussed with examples in section III and Appendix B.
Counterions Only
In this scenario there are only counterions in the solution, just enough to balance the charge on one or more plates. This case presents some new features, e.g., the electric field and counterion concentration fall off much more slowly at large distances from a plate than for a full electrolyte solution and there is no Grahame relation. Also, the PB equation can be solved analytically for a variety of both planar and non-planar geometries. The literature contains both theoretical and experimental studies. The PB equation for this scenario has just one of the two exponential terms in (5), e.g., the one with positive exponent when the counterions are negative, corresponding to a positively charged plate. In the mathematical literature this single exponential equation is known as the Liouville equation, after the mathematician Liouville who, in 1853, posed and solved it as a purely mathematical problem in two dimensions . It also arises in other physical problems . One of these problems concerns the potential in a space charge of electrons near a surface from which these electrons have been thermally emitted. We restrict our brief discussion to the single plate problem with surface charge density on the plate at x ; Andelman and Israelachvili have also considered the corresponding two-plate problem (for scenario B) and the associated inter-plate force. For x the PB equation takes the one-dimensional Liouville form ( ) / exp ( ) x e c e x , (54) where (0) c c , with ( ) c x the anion concentration. The boundary conditions are (0) / and ( ) 0 . As ( ) x behaves as ln( ) x const for large x we cannot use 2 the previous normalization ( ) 0 . We have therefore chosen (0) 0 as a convenient reference potential for (54). Multiplying (54) by ( ) x we find its first integral to be
21 B2 ( ) ( ) x k T c x const , where ( ) exp ( ) c x c e x . As both ( ) x and ( ) c x vanish at x , we see that const , so that we have
21 B2 ( ) ( ) x k T c x . (55) The solution of (54) or (55) for ( ) x is easily verified to be B GC ( ) 2 / ln 1 / x k T e x , (56) where
GC B k T e is the Gouy-Chapman length introduced in section III (see discussion of Fig. 1). Note that (0) 0 as stated above. The electric field in this case, x B GC ( ) ( ) 2 /
E x x k T e x , has range GC and falls off at large x as x . This contrasts with the case of the full electrolyte solution where (see Section III) the range depends on D as well as GC (it is essentially D for not too large ), and the fall-off is exponential at large x . Note that because , there is no Grahame relation between surface potential and surface charge density analogous to (23) for a full electrolyte. The corresponding anion concentration ( ) exp ( ) c x c e x is found, using (56), to be
20 GC ( ) / 1 / c x c x , (57) where c is found from the first integral (55) and the boundary condition (0) / to be given by / (2 ) 2 / ( ) c k T k T e . The range of ( ) c x is here controlled by GC and the fall-off at large x is x , in contrast to the case of the full electrolyte where the 3 range depends on both D and GC (but is approximately D for small ) and the fall-off at large x is exponential. The equilibrium value of the Helmholtz free energy per unit plate area can be written in the form (see (31)) / ( ) ( ) ( ) ln 1 ( ) F A f x x x k T c c x dx . (58) Since (0) 0 , the surface term f ( ) ( ) x x does not contribute to (58). The electrostatic term ( ) ( ) f x x can be expressed in terms of ( ) c x using (55) and the integral ( ) c x dx takes the value / e due to global electroneutrality. An analytical expression for the free energy F in terms of its natural variable then follows as / / ln 2 F A k T e c , (59) with c determined above. Note that F if . Due to electroneutrality there are then no ions left in the solution, in contrast to the case of the full electrolyte where equal numbers of cations and anions remain in solution, and F F as . Equation (44) for F shows that for a full electrolyte F remains non-zero as . V. CONCLUDING REMARKS
According to the fundamental variational principle of Gibbs for thermal systems, a state of thermodynamic equilibrium minimizes the system free energy. We have shown how this principle can be applied to an electrolyte solution in contact with charged walls to obtain the nonlinear PB equation by minimizing the free energy of the system. Our discussion emphasizes that for quasi one-dimensional geometry with a uniformly charged planar wall, the nonlinear PB equation is no more complicated to solve analytically than the traditional linear DH equation considered in a few electromagnetism texts and numerous other more 4 specialized texts. We have given the simple analytical solutions of the PB equation for both a full electrolyte such as potassium chloride in water, and for an electrolyte containing only counterions; the latter system shows some interesting novel features. Analytical expressions for the equilibrium value of the free energy are also derived. For two charged planar walls the solutions of the PB equation, and the expressions for the equilibrium values of the free energy, are much more complicated and are not derived explicitly. Despite these complications a simple expression for the derivative of the free energy with respect to the inter-wall separation, which gives the inter-wall force, can be derived. This expression for the force is shown to agree with that obtained from a hydrostatic equilibrium argument and is used to calculate the force for some examples as discussed below. The nonlinear primitive PB model provides a simple theoretical description of the spatially inhomogeneous electric double layer that forms at the boundary of any electrolyte solution and the charged surface of an electrode, a biological membrane, a colloidal particle, or a charged macromolecule such as a protein or DNA. The thermodynamically preferred structure of a double layer can be predicted using the criterion that this structure minimizes the system free energy, but, as we have emphasized, care must be taken to ensure that the free energy that is minimized is consistent with which bulk and surface properties of the system have been chosen as the fixed control parameters. It is not obvious from many cases discussed in the literature whether this free energy is that of Gibbs, Helmholtz, grand or some other. Thus, for example, it is not clear from the seminal treatise of Verwey and Overbeek whether their free energy of an infinite plate at constant surface potential immersed in an infinite reservoir of electrolyte refers to Helmholtz or to a grand free energy. Overbeek has since confirmed the former interpretation. Our analysis, however, supports the latter interpretation. A recent attempt by Clarke and Stiles to employ a free energy functional valid for fixed surface-charge boundary conditions mistakenly used a fixed surface-potential 5 free energy functional instead. This error led them to predict a ‘finite electric double layer’ at the interface between a highly charged metallic plate and a 1:1 electrolyte solution, and to imply a discontinuous force per unit area at the charged interface between their ‘finite electric double layer’ and the bulk electrolyte. When these points are corrected, only the classical diffuse Gouy-Chapman double layer of infinite spatial extent survives. From the perspective of the primitive model PB theory, as mentioned above forces between two parallel electrified plates separated by an electrolyte solution can be calculated from the hydrostatic equilibrium expression (A5) with a repulsive kinetic pressure component and an electrostatic component that is attractive if the plates have equal and opposite charges. Several examples are given. In calculations of forces between two parallel charged plates, care must also be taken to distinguish between cases where the electrolyte is confined to the region between the two plates (our scenario A) and situations where the two plates are immersed in an electrolyte reservoir so that the electrolyte is in contact with both the interior and exterior surfaces of each plate (our scenario B). As shown in our explicit examples, there are quantitative (see Figs. 2 and 3) and even qualitative (see Panels A and B of Figure 4) differences between these two scenarios. Figure 4A (for scenario A) indicates inter-plate attraction between plates with equal and opposite surface potentials for small plate separations and inter-plate repulsion for large plate separations, whereas Figure 4B (for scenario B) exhibits only attractive inter-plate forces. In conclusion, we think that PB theory deserves serious consideration for inclusion in electromagnetism courses and standard textbooks. Besides being an interesting and widely applicable aspect of nonlinear electrostatics, a subject often not mentioned, it can also serve as an excellent illustration of the power and usefulness of the fundamental Gibbs variational principle in thermodynamics and statistical mechanics. 6
APPENDIX A. HYDROSTATIC EQUILIBRIUM ARGUMENT FOR INTERPLATE FORCE
In general the hydrostatic equilibrium condition is B ( ) ( ) p x x f where ( ) p x is the pressure and B ( ) x f the body force density (per unit volume) at x due to any long-range forces acting on the fluid, e.g., gravitational or electric. For our system the body force is electric and is given by B m ( ) ( ) ( ) x x E x f where ( ) E x is the mean electric field and m ( ) x the mean charge density due to the mobile ions. Because of the simple wall geometry, with the charged walls perpendicular to the x axis, the equilibrium condition reduces to the one-dimensional form m ( ) ( ) ( ) p x x x , (A1) where B ( ) ( ) ( ) p x k T c x c x is the ideal or kinetic pressure due to cations and anions in the mean-field primitive model of the electrolyte. In the bulk fluid the one-dimensional Poisson equation is m ( ) ( ) x x . (A2) Eliminating m ( ) x between (A1) and (A2) gives ( ) ( ) ( ) ( )2 dp x x x xd x . (A3) Integration of (A3) gives the integrated form of the hydrostatic equilibrium condition ( ) ( ) .2 p x x const (A4) Since the left-hand side is a constant for all x in the range
0, 2 d , we can evaluate it at x d . When the two plates bear identical charges ( ) 0 d so that ( ) const p d , which is positive. The force per unit area f on the plate at x d is the sum of a repulsive pressure 7 and an attractive electrostatic contribution f (2 ) (2 )2 p d d . (A5) This is clearly positive since (A5) is equal to ( ) p d from (A4). When the plates have opposite polarity, then ( ) 0 d , and we have f ( ) / 2 ( ) p d d . In this case the sign of the net force on the plate at x d is not obvious; we show in section IV.2.4 and Appendix C that in scenario A the net force can be attractive or repulsive, depending on the value of d , whereas in scenario B the net force is purely attractive. As d d B ( ) p d c c k T , we see that determination of the inter-plate force requires d c and d c when the plates are equally charged, and requires d d c c and ( ) d for oppositely charged plates. Specific algorithms for determining these quantities are discussed in Appendix C, and examples are given in sections IV.2.1 and IV.2.4. APPENDIX B. SCENARIO B. PLATES OF FIXED CHARGE IMMERSED IN ELECTROLYTE SOLUTION. GRAND FREE ENERGY AND INTERPLATE FORCE
Here we consider the problem, relevant to the analysis of the stability of colloidal suspensions, where each of the finite parallel plates is immersed in a large reservoir of 1:1 electrolyte, initially having a spatially uniform concentration c . Rather than assume fixed numbers N of mobile ions as previously, we now consider the ionic electrochemical potentials and to be fixed. The electrolyte between the two plates is assumed to be in equilibrium with the electrolyte in the reservoir and we see from the large distance behaviors 8 of (14a,b) that ln k T c . The volume V and temperature T are specified as before. With specified thermodynamic parameters ( , , , T V ) the thermodynamic potential appropriate for use in the Gibbs variational principle and in the force calculation, is the grand free energy
F F N N , where here N denote the mean ionic numbers. We again neglect edge effects as the two plates are assumed to be large. To simplify this problem, we assume insulating plates with fixed inside surface charge densities and double layers only on the two inside surfaces. These plates are also subject to external constraint forces in the x -direction that maintain the plate separation at d . If the plates also carry charge on their outer surfaces, and thus have double layers there too, the inter-plate force will be unaffected. This is because the net force on the outside of a plate is unaffected by the presence of an outside double layer, as is easily seen from the hydrostatic equilibrium of any outer fluid region with one boundary on the outer surface of the plate, and the opposite planar boundary far from the plate where the concentration of each ionic species is c . We also assume large plate thicknesses so that the electric field within each of the closely spaced dielectric plates tends to zero and the effects of the outer electrolyte solution on the inner one, can be neglected . We consider circular plates, an inessential assumption but one which simplifies the description of the geometry. Consider two coaxial parallel circular plates of radius a in the yz planes x and x d . These define a cylinder of length d in the x direction inside a cubic container of volume V L , with both a and L large, but with a L so that edge effects can be neglected and d L . We regard the total grand free energy int ext F F F of the system to be the sum of an internal component int F corresponding to the electrolyte fluid inside the cylinder and an external component ext F corresponding to the fluid outside. Thus, 9 / ( ) d F A f x dx , (B1) and ext F can be taken as / ( ) ( ) LL d
F A f x dx f x dx , (B2) where the y and z parts of the original volume integrations, leading to the purely x -integrals (B1) and (B2), have been restricted to the region a , with area A a , where y z is the cylindrical radial coordinate. When edge effects are neglected these two equations give an excellent approximation to F when a . In the large system limit there is a negligible contribution to the inter-plate forces from external regions with a . We want to calculate the net force per unit area f between the two plates in equilibrium with the electrolyte. The relevant PB equation again takes the form (19), but here is chosen in the reservoir. The volume density int ( ) f x of the grand free energy per unit area int / F A is related to the density int ( ) f x of the Helmholtz free energy by int int ( ) ( ) ( ) ( ) f x f x c x c x . The internal Helmholtz free energy density is int intel id ( ) ( ) f x f x , where int 21el 2 ( ) ( ) ( ) ( / 2) ( ) f x x x x and intid ( ) f x is defined by (10). Ionic concentrations ( ) c x of the mobile ions are related to the reservoir concentration c by the Boltzmann expressions ( ) / exp ( ) , ( ) / exp ( ) c x c e x c x c e x . (B3a,b) For identical charges on the two plates int ( ) f x is symmetric about the mid-plane . x d It follows that the grand free energy inside the cylinder is related to the grand free energy density by / ( ) 2 ( ) d d
F A f x dx f x dx . 0 From int int ( ) ( ) ( ) ( ) f x f x c x c x , the relation ln k T c , and the Boltzmann distributions (B3a,b) we find that the internal grand free energy density can be written as int el Bf el B ( ) ( )( ) ( ) ( ) ( ) ( ) ln ( ) ln( ) ( ) ( ) ( ) ( ) . (B4) c x c xf x f x k T c x c x c x c xc cx x f x k T c x c x The inter-plate force due to the internal electrolyte can now be calculated from (B4) as int2d int int0 f / 2 / / ( ) d d F A d f x dx . (B5) We note that ( ) ( / 2) ( ) f x x , f ( ) ( ) ( 2 ) x x x d where the delta functions are those of Dirac, (2 ) d by symmetry, and that since ln k T c the condition of fixed corresponds to fixed c . We thus find from (B3a,b) and the Leibniz rule that (B5) becomes cosh ( )( )f ( / 2) ( ) ( ) 2 . d e xxd x k T c dx k T c cd d d (B6) From the hydrostatic equilibrium argument of Appendix A, we expect the result for int2d f to be just the final term in (B6), so that the other terms should cancel, which we now show. By symmetry, the mid-plane electric field ( ) 0 , d so ( ) ( )f ( ) 2 sinh ( ) . (B7) d x xx e c e x dx k T c cd d d On integrating the first integral in (B7) by parts we obtain 1 ( ) ( )f ( ) ( ) 2 sinh ( ) . (B8) d d x xx x e c e x dx k T c cd d d As, by symmetry the electric field ( ) 0 d and the surface charge density (0) , the two leading terms on the right-hand side of (B8) cancel. Also, from the PB equation (19) the integral in (B8) vanishes, so that (B8) simplifies to int d d d2d B B f 2 cosh k T c c k T c e . (B9) In calculating ext / / d F A using (B2), note that only the second term with x d will contribute, and when we use the fact that ext B ( ) 2 f x k T c is independent of x , we see that ext2d B B f / (2 ) 2 ( / 2 2 ) 2 d k T c L d k T c , (B10) i.e., the expected pure external pressure force pointing left. Thus, the net force per unit area on the right plate, int ext2d 2d 2d f f f , when the surface charges are fixed is given by d d d2d B B f 2 2 cosh 1 k T c c c k T c e . (B11) Identical expressions, describing net repulsion between the plates, were established by Langmuir , Derjaguin and Landau and Verwey and Overbeek for the scenario where the plate potentials are both fixed at . We briefly discuss this scenario in Section IV.2.4. As the plate separation becomes very large d d c c c and we see from (B3a,b) and (B11) that d and f 0 . 2 The first integral of the PB equation (29) for scenario B has the same general form (39) as for scenario A and we find analogous values int int d d0 2d B f f k T c c for the constant of integration. From these values we find the two-plate Grahame equation
11 22 c k T e e (B12) for scenario B, another generalization of (23) for a single plate, and to which it reduces in the limit . d Unfortunately, (B12) is not a self-contained relation between and , as d depends on (or ). Appendix C discusses a method for determining d . APPENDIX C. COMPUTATION OF INTER-PLATE FORCES
In this Appendix we describe computational methods used to produce our figures for inter-plate forces as a function of the plate separation d . Depending on the plate polarities the inter-plate force can be assessed. Expressions describing the inter-plate forces for various conditions on the electrolyte composition and electric boundary conditions at the plates have already been derived for two identical plates and plates of opposite polarity. The inter-plate force can be expressed in terms of mid-plane ionic concentrations d c , the mid-plane electric field x ( ) ( ) E d d , or the corresponding electric potential d . As observed by Derjaguin and Landau the first integral of the PB equation replaces the second-order nonlinear PB equation with appropriate boundary conditions by a simpler first-order differential equation. For the two-plate problem with identical plates the relevant Debye parameter / e c c k T , like the inter-plate force itself depends on the sum of the mid-plane ionic concentrations, so we can solve the first integral together with relevant physical constraints explicitly for these ionic concentrations. We consider scenarios where total ion 3 numbers in the electrolyte solution are fixed (Figs. 2 and 4A) and where the solution ionic chemical potentials (or the reservoir electrolyte concentration) have been fixed (Figs. 3 and 4B). Whereas Figs. 2 and 3 refer to fixed surface-charge densities at the plates, Figs. 4A and 4B refer to fixed surface potentials at these plates. Apart from standard values for the fundamental physical constants we assume that the temperature T and that for water at o
25 C , as solvent, the relative electric permittivity (or dielectric constant) r is . Figure 2. Scenario A for plates with identical surface charge densities
Input parameters include for the fixed surface charge densities and
16 -2 / 8 10 m
N A for the fixed number of anions in solution per square meter of each plate. The three unknown variables of interest for a stipulated value of the plate separation d are taken to be d c and the surface potential of either plate. As shown by (38) the mid-plane concentrations alone determine the repulsive inter-plate force. From the first integral (C1) of the relevant PB equation (29), and its boundary condition (0) / , we obtain the algebraic equation (40) that provides the first of three equations determining our unknown variables. On integrating / dx d between x and x d with provided by the first integral of (39) with d dB const k T c c , viz., B k T d d exp 1 exp 1 c e c e , (C1) with the negative sign ensuring that the electric field at x is positive when , we find
120 12 d dc e c e . (C2) 4 This is the second of our three equations. A third independent equation can be found from electroneutrality: d0 0 d d N A c x dx Ac e x dx N A e . Upon changing the integration variable from x to , we see that this equation takes the form
120 12 d0 d d exp 2 / (2 ) /exp 1 exp 1 c e d N A ec e c e . (C3) Using the FindRoot function of Wolfram’s
Mathematica we have solved the algebraic equation (40) together with the two integral relations (C2) and (C3) to find d c and . The inter-plate force d d2d B f k T c c can then be plotted as a function of the plate separation d for the assumed values of and / N A . Figure 3. Scenario B for plates with identical surface charge densities
As shown in (B11) the inter-plate force d2d B f 2 cosh 1 k T c e . Here, the input parameters are the surface charge density of each plate and the reservoir concentration
10 mM c . The two unknown variables of interest for a given plate separation d are taken to be the mid-plane potential d and the surface potential of each plate. They can be determined by solving the algebraic Grahame equation (B12) relevant to our scenario B together with the integral relation
120 1d 2 d d c de e , (C4) 5 derived from d0 / d d , after noting that because x ( ) ( ) 0 E d d , by symmetry, the first integral of the relevant PB equation (19) can be expressed in the form
12 12 d c e e . (C5) The solution to the two simultaneous equations (B12) and (C4) yields the mid-plane potential d and hence the inter-plate force for any required plate separation d . Figure 4A. Scenario A for plates with equal and opposite surface potentials
For this case the relevant inter-plate force is given just below equation (A5) as d 22d B f 2 / 2 ( ) k T c d . (C6) By symmetry, d , so that d d d c c c , and hence the ideal (kinetic) pressure becomes dB ( ) 2 p d k T c . We fix the surface potentials of the two plates at the values
10 mV and
10 mV , and also fix the total number of cations N in solution at
16 2 / 2.41 10 m
N A . We choose our three unknown variables to be the surface charge density on the left-hand plate, the mid-plane concentration d c of cations or anions and the gradient ( ) d of the mid-plane electrostatic potential. For a stipulated value of d we initially solve two integral relations for and d c , and then use the algebraic two-plate version of the Grahame equation / 2 ( / 2) ( ) 2 cosh( ) 1 d k Tc e , (C7) for equal and opposite plate polarities, to obtain the second term ( / 2) ( ) d in (C6). The first of the integral relations takes the form / d d , where / (4 / ) cosh( cosh k T c e e , (C8) 6 so that
10 2 d 0 2B ( / ) (4 / ) cosh( cosh dd k T c e e . (C9) A second integral relation in and d c comes from the relation
20 0 / ( ) ( ) ( ) d d
N A c x dx c x c x dx , (C10) which can be rewritten as d 0 / (2 ) cosh ( ) d N A c e x dx , and rewritten again, with as the integration variable, to give d 10 2 d 0 2B cosh/ 2 ( / ) (4 / ) cosh( cosh e dN A c k T c e e . (C11) We solve (C11) and (C9) for the two unknowns and d c , and then, using ( / 2) ( ) d from (C7), we obtain the force per unit area (C6) on the right-hand plate. Figure 4B. Scenario B for plates with equal and opposite surface potentials
As in the previous case d , so that d d d c c c . Here, however, the internal kinetic pressure d B B ( ) 2 2 p d c k T c k T on the plate at x d due to ions between the two plates is exactly cancelled by the external pressure due to ions with x d . Thus, in contrast to scenario A, above, the net force per unit area on the plate at x d takes the purely attractive form f / 2 ( ) d , (C12) directly proportional to the square of the mid-plane electric field. The first integral of the relevant PB equation (19) can be written as ( / 2) / 2 ( ) 2 cosh 1 2 cosh 1 f d c k T e c k T e . (C13) 7 To relate ( ) d to d we substitute the negative square root of (C13) into / d d to obtain dd c k T e ddc k T e For the fixed value
10 mV of the surface potential at the left-hand plate and any stipulated value of the plate separation d , the integral relation (C14) can be solved numerically for the force per unit area, f , on the right-hand plate. ACKNOWLEDMENT
CGG thanks NSERC (Canada) for financial support.
REFERENCES C.J.F. Böttcher, Theory of Electric Polarization, 2nd ed., Vol.1, Dielectrics in Static Fields (Elsevier, Amsterdam, 1973), Ch.7, p.289. 2.
B.K.P. Scaife, Principles of Dielectrics (Oxford University Press, Oxford, 1989), Ch.9, p.243. 3.
For brief coverage see P. Brito, C. Grosse and C. Halloy, Electrostatic problems in nonlinear media, Am. J. Phys. , 1014-1018 (1986). At very large fields, dielectric breakdown in an insulator like rock salt (NaCl) generates free charges (electrons or/and ions) which can lead to a double layer at the surface. This and other means of producing free charges and surface double layers (e.g., by heating, illumination, 8 charge injection) have been studied for numerous insulator and semiconductor systems. An early reference is A. F. Joffe, The Physics of Crystals (McGraw-Hill, New York, 1928). E. Garmire, Resource Letter NO-1: Nonlinear Optics, Am. J. Phys. , 245-255 (2011). The dynamic nonlinear electric susceptibilities arising in nonlinear optics are direct generalizations of the corresponding nonlinear electrostatic susceptibilities. 4. P. Debye and E. Hückel, Zur theorie der electrolyte I, gefrierpunktserniedrigung und vewandte erscheinungen, Phys. Zeit. , 185-206 (1923). 5. M.G. Gouy, Sur la constitution de la charge électrique à la surface d'un électrolyte, J. Phys. Radium , 457-468 (1910). 6. D.L. Chapman, A contribution to the theory of electrocapillarity, Phil. Mag. , 475- 481 (1913). 7. G. Lamm, The Poisson-Boltzmann equation, Rev. Comp. Chem. , 147-365 (2003). 8. W. Cai, Computational Methods for Electromagnetic Phenomena. Electrostatics in Solvation, Scattering and Electron Transport (Cambridge University Press, Cambridge, 2013), Ch.2 (PB and DH theory) and Ch.3 (numerical methods for solving the PB equation). 9.
H. Ohshima, Biophysical Chemistry of Biointerfaces (Wiley, Hoboken, 2010). PB and DH for planar, spherical and cylindrical double layers. 9 10.
A. Zangwill, Modern Electrodynamics (Cambridge University Press, Cambridge, 2013), p.150 (simplified treatments of classical (DH) and quantum (TF) screening of an inserted point charge in a plasma) and p.262 (PB for double layer at charged planar surface with medium containing only counterions). 11.
A. Das, Lectures on Electromagnetism, 2nd ed. (Hindustan Book Agency, New Delhi, 2013), p.376 (PB and DH for a classical plasma). 12.
W.T. Scott, The Physics of Electricity and Magnetism, 2nd ed. (Wiley, New York, 1966), p.261 (PB and DH for screening planar charged surface of a bulk electrolyte solution and a plasma). 13.
R.P. Feynman, R.B. Leighton and M. Sands, The Feynman Lectures on Physics, Vol.2 (Addison-Wesley, Reading, 1963), p.7-8 (PB and DH for planar double layer). 14.
A.M. Portis, Electromagnetic Fields: Sources and Media (Wiley, New York, 1978), p.164 (simplified treatments of classical (semiconductors) and quantum (metals) screening of electrons in solids) and p.169 (DH screening of a charged sphere inserted into a plasma). 15.
N. Meyer-Vernet, Aspects of Debye shielding, Am. J. Phys. , 249-257 (1993). 16. M. Uehara, K.K. Sakane, H.S. Maciel and W.I. Urruchi, Physics and biology: bio-plasma physics, Am. J. Phys. , 450-455 (2000). 0 17. D.C. Kelly, Plasma equation of state, Am. J. Phys. , 827-828 (1963). 18. O. Theimer and M.M. Theimer, Computer simulation of statistical plasma mechanics (an example of undergraduate research), Am. J. Phys. , 661-666 (1978). A.D. Boozer, Simulating a one-dimensional plasma, Am. J. Phys. , 580-584 (2010). 20. D. Marin and F. Mendicuti, Double layer I. Qualitative detection of adsorption of ions and neutral molecules on the mercury electrode and excess charge on the electrode, J. Chem. Ed. , 649-651 (1988), and Double layer II. Quantitative study of adsorption of ions on the mercury electrode, J. Chem. Ed. , 712-715 (1988). 21. J. O'M. Bockris, Teaching the double layer, J. Chem. Ed. , 265-268 (1983). 22. R. de Levie, Gouy, Debye-Hückel, and Fick: Understanding differential equations without solving them, J. Chem. Ed. , 129-132 (1999). 23. D. Andelman, Electrostatic properties of membranes; the Poisson-Boltzmann theory. Ch. 12, pp.603-642, of Handbook of Biological Physics, Volume 1, eds. R. Lipowski and E. Sackmann (Elsevier, Amsterdam, 1995). 24.
J. Claycomb and J.Q.P. Tran, Introductory Biophysics (Jones and Bartlett, Boston, 2011), p.127 (DH). 1 25.
K.A. Dill and S. Bromberg, Molecular Driving Forces, Statistical Thermodynamics in Chemistry and Biology (Garland, New York, 2003), p.434 (PB and DH for planar and spherical surfaces). 26.
D.T. Edmonds, Electricity and Magnetism in Biological Systems (Oxford University Press, Oxford, 2001), p.186 (PB and DH for double layer). 27.
G.B. Benedek and F.M.H. Villars, Physics, with Illustrative Examples from Medicine and Biology, Vol.3, Electricity and Magnetism (Addison-Wesley, Philippines, 1979), pp.3-25 (PB and DH for planar and spherical double layers). 28.
M. Daune, Molecular Biophysics (Oxford University Press, Oxford, 1999), p.323 (DH for electrolyte solutions and p.340 (PB for double layers of polyelectrolyte spheres and cylinders). 29.
R. Nossal and H. Lecar, Molecular and Cell Biophysics (Addison-Wesley, Redwood, 1991), p.20 (PB for double layer) and p.204 (DH for electrolyte solutions). 30.
R. Blossey, Computational Biology, a Statistical Mechanics Perspective (Chapman and Hall /CRC, Boca Raton, 2006), p.158 (PB and DH for planar, cylindrical and arbitrary (protein) surfaces). 31.
M.B. Jackson, Molecular and Cellular Biophysics (Cambridge University Press, Cambridge, 2006), p.277 (PB equation and the Debye length). 2 32.
P. Nelson, Biological Physics (Freeman, New York, 2004), p.267 (PB for double layer at charged planar surface with solution containing only counterions). 33.
R. Phillips, J. Kondev and J. Theriot, Physical Biology of the Cell, 2nd ed. (Garland, New York, 2009), p.346 (PB for planar surface) and p.349 (DH for spherical surface). 34.
T.A. Waigh, Applied Biophysics, a Molecular Approach for Physical Scientists (Wiley, Chichester, 2007), p.32 and p.205 (PB and DH for planar and spherical double layers). 35.
A.W. Adamson, Physical Chemistry of Surfaces, 5th ed. (Wiley, New York, 1990), p.203. 36.
J.J. Bikerman, Physical Surfaces (Academic Press, New York, 1970), p.370 (PB for planar double layer). 37.
H-J Butt, K. Graf and M. Kappl, Physics and Chemistry of Interfaces, 3rd ed. (Wiley-VCH, Weinheim, 2013), p.58. 38.
J. Israelachvili, Intermolecular and Surface Forces, with Applications to Colloid and Biological Systems, 2nd ed. (Academic Press, London, 1991), p.215 (PB for double layer at a charged planar surface with only counterions in the solution). The counter-ions only case was earlier discussed by J. Frenkel, Kinetic Theory of Liquids, (Oxford University Press, Oxford 1946), p.357. 3 39.
D.J. Shaw, Introduction to Colloid and Surface Chemistry, 3rd ed. (Butterworths, London, 1980), p.150 (PB for double layer). 40.
M.C. Petty, Langmuir-Blodgett Films, an Introduction (Cambridge University Press, Cambridge, 1996), p.33 (PB for planar double layer). 41.
K.S. Birdi, Introduction to Electrical Interface Phenomena (CRC, Boca Raton, 2010), p.19 (PB for planar double layer). 42.
A. Sanfield, Introduction to the Thermodynamics of Charged and Polarized Layers (Wiley-Interscience, London, 1968), p.103 (PB for planar double layer). 43.
J.G. Kirkwood and I. Oppenheim, Chemical Thermodynamics (McGraw Hill, New York, 1961), p.193 (DH for electrolyte solutions). 44.
S.R. Morrison, The Chemical Physics of Surfaces (Plenum, New York, 1977), p.35 (double layer at surfaces of solids), and p.263 (double layer at semiconductor/electrolyte solution interface). 45.
M. Muthukumar, Polymer Translocation (CRC Press, Boca Raton, 2011), p.48 (PB and DH theories). 46.
N.G. van Kampen and B.U. Felderhof, Theoretical Methods in Plasma Physics (North Holland, Amsterdam, 1967), p.82. 4 47.
S. Ichimaru, Statistical Plasma Physics, Vol.II, Condensed Plasmas (Addison-Wesley, Reading, 1984), p.17 (quantum plasmas), p.18 (classical plasmas), p.116 (electron liquids). 48.
A.G. Sitenko, Electromagnetic Fluctuations in Plasma (Academic Press, New York, 1967), p.55. 49.
J. Delcroix, Plasma Physics (Wiley, New York, 1965), p.188 (PB and DH for classical plasmas). 50.
W. Ebeling, W-D Kraeft and D. Kremp, Theory of Bound States and Ionization Equilibrium in Plasmas and Solids (Akademie-Verlag, Berlin, 1976), p.18 (classical PB and DH) and p.19 (quantum TF). 51.
N.W. Ashcroft and N.D. Mermin, Solid State Physics (Holt Rinehart and Winston, New York, 1976), p.340 (TF theory for metals). 52.
C. Kittel, Introduction to Solid State Physics, 7th ed. (Wiley, New York, 1996), p.280 (TF screening of a positive charge placed in a sea of electrons). 53.
P.M. Chaikin and T.C. Lubensky, Principles of Condensed Matter Physics (Cambridge University Press, Cambridge, 1995), p.204. 54.
R. Brout and P. Carruthers, Lectures on the Many-Electron Problem (Interscience, New York, 1963), p.21 (classical PB and DH for electron gas) and p.99 (quantum TF). 5 55.
G.D. Mahan, Many-Particle Physics, 2nd ed. (Plenum, New York, 1990), p.428 (quantum TF theory). 56.
R.D. Mattuck, A Guide to Feynman Diagrams in the Many-Body problem, 2nd ed. (McGraw Hill, New York, 1976), p.97 and p.191. 57.
A. Eschrig, The Fundamentals of Density Functional theory, 2nd ed. (Eagle Publishers, Leipzig, 2003), p.69 (TF theory). 58.
G.F. Giuliani and G. Vignale, Quantum Theory of the Electron Liquid (Cambridge University Press, Cambridge, 2003), p.198 (TF theory). 59.
G.N. Lewis and M. Randall, Thermodynamics, 2nd ed. (McGraw Hill, New York, 1961) p.335 (DH theory for electrolyte solutions). 60.
S. Glasstone, Thermodynamics for Chemists (Van Nostrand, New York 1947, reprinted by Krieger 1972), p.407 (DH theory for electrolyte solutions). 61.
L. Blum and D. Henderson, Statistical mechanics of electrolytes at interfaces, in D. Henderson, editor, Fundamentals of Inhomogeneous Fluids (Marcel Dekker, New York, 1992), pp.239-276. 62.
L.D. Landau and E.M. Lifshitz, Statistical Physics (Addison-Wesley, Reading, 1958), p.231 (classical PB and DH) and p.232 (quantum TF). 6 63.
T.L. Hill, Introduction to Statistical Thermodynamics (Addison Wesley, Reading, 1960), p.321 (PB and DH for electrolyte solutions). 64.
R.H. Fowler and E.A. Guggenheim, Statistical Thermodynamics (Cambridge University Press, Cambridge, 1939), p.390 (PB and DH for electrolyte solutions). 65.
D.A. McQuarrie, Statistical Mechanics (Harper and Row, New York, 1976), p.326 (PB and DH for electrolyte solutions). 66.
N. Davidson, Statistical Mechanics (McGraw Hill, New York, 1962, reprinted by Dover 2003), p.485 (PB and DH for electrolyte solutions). 67.
P. Attard, Thermodynamics and Statistical Mechanics (Academic Press, London, 2002), p.318 (PB and DH for electrolyte solutions) and p.322 (PB for double layer). 68.
H.T. Davis, Statistical Mechanics of Phases, Interfaces and Thin Films (VCH, New York, 1996), p.580. 69.
J. Goodisman, Statistical Mechanics for Chemists (Wiley, New York, 1997), p.249 (DH for electrolyte solutions). 70.
J-P Hansen and I.R. McDonald, Theory of Simple Liquids, with Applications to Soft Matter, 4th ed. (Elsevier, Amsterdam, 2013), p.133 (DH for electrolyte solutions), 7 p.413 (DH for plasmas and other classical ionic fluids), p.435 (PB for double layer), p.448 (TF for electron fluids). 71.
J-L Barrat and J-P Hansen, Basic Concepts for Simple and Complex Liquids (Cambridge University Press, Cambridge, 2003), p.194 (PB for double layer). 72.
W.R. Fawcett, Liquids, Solutions and Interfaces (Oxford University Press, New York, 2004), p.121 (DH for electrolytes), p.542 (PB for planar double layer). 73.
N.H. March and M.P. Tosi, Coulomb Liquids (Academic Press, London, 1984), p.18 (classical DH for bulk ionic fluids), p. 42 (quantum TF for electron fluids), p. 262 (classical PB for double layer). 74.
J.M.G. Barthel, H. Krienke and W. Kunz, Physical Chemistry of Electrolyte Solutions (Springer, Darmstadt, 1998), p.28. 75.
R.A. Robinson and R.H. Stokes, Electrolyte Solutions, 2nd ed. (Butterworths, London, 1959), p.73. 76.
S.A. Rice and M. Nagasawa, Polyelectrolyte Solutions, a Theoretical Introduction (Academic Press, New York,1961), p.99 (PB and DH theory). 77.
F. Oosawa, Polyelectrolytes (Marcel Dekker, New York,1971), p.28 (PB theory). 78.
H.C. Friedman, Ionic Solution Theory (Interscience, New York, 1962). PB and beyond for bulk ionic solutions. 8 79.
J. Goodisman, Electrochemistry: Theoretical Foundations (Wiley, New York, 1987), p.86 (classical PB for double layer), p.186 (quantum TF). 80.
A.J.Bard and L.R.Faulkner, Electrochemical Methods: Fundamentals and Application, 2nd. ed. (Wiley, New York, 2000), (PB and DH for electrolyte solutions). 81.
E. Gileadi, Physical Electrochemistry. Fundamentals, Techniques and Applications (Wiley-VCH, Weinheim, 2010), p.116 (PB for double layer). 82.
W. Schmickler, Interfacial Electrochemistry, 2nd ed. (Oxford University Press, New York, 2010), p.39. 83.
J.O’M Bockris and A.K.N. Reddy, Modern Electrochemistry, Vols. 1 and 2 (Plenum, New York, 1970), p.180 (PB and DH for electrolyte solutions) and p.722 (PB for double layer). 84.
J.O’M Bockris, B.E. Conway and E. Yeager, Comprehensive Treatise of Electrochemistry, Volume 1: The Double Layer (Plenum, New York, 1980). 85.
M.J. Sparnaay, The Double Layer (Pergamon Press, Oxford, 1972). 86.
L.S. Hirst, Fundamentals of Soft Matter Science (CRC Press, Boca Raton,2013), p.137. 87.
M. Doi, Soft Matter Physics (Oxford University Press, Oxford, 2016), p.204 (PB for double layer). 9 88.
N. Laanait, Ion Correlations at Electrified Soft Matter Interfaces (Springer, Heidelberg, 2013), p.5 (PB theory). 89.
D. Dean, J. Dobnikar, A. Naji and R. Podgornik, editors, Electrostatics of Soft and Disordered Matter (Pan Stanford Publishing, Singapore, 2014). 90.
C. Holm, P. Kekcheff and R. Podgornik, editors, Electrostatic Effects in Soft Matter and Biophysics (Kluwer, Dordrecht, 2001). 91.
W.K.C. Poon and D. Andelman, editors, Soft Condensed Matter Physics in Molecular and Cell Biology (Taylor and Francis, New York, 2006), p.98 and p.193 (PB for planar, spherical and cylindrical double layers). 92.
R.S. Berry, S.A. Rice and J. Ross, Physical Chemistry, 2nd ed. (Oxford University Press, New York, 2000), p.772 (PB and DH for electrolyte solutions) and p.790 (PB for double layer). 93.
D.A. McQuarrie and J.D. Simon, Physical Chemistry, a Molecular Approach (University Science Books, Sausalito, 1997), p.1044. 94.
P.W. Atkins, Physical Chemistry, 3rd ed. (Freeman, New York, 1986), p.237 (DH for bulk electrolytes). 0 95.
J.T. Edsall and J. Wyman, Biophysical Chemistry (Academic Press, New York, 1958), p.286 (DH for bulk electrolyte solutions). 96.
P.R. Bergethon, The Physical Basis of Biochemistry (Springer, New York, 1998), p.321 (DH for electrolyte solutions) and p.419 (PB for double layer). 97.
R.K. Hobbie, Intermediate Physics for Medicine and Biology, 2nd ed.(Wiley, New York, 1988), p.237. (PB for solutions and interfaces). 98.
L.E. Bayliss, Principles of General Physiology, 5th ed.,Vol.1, The Physico-Chemical Background (Longmans, London,1959), p.130 (DH for bulk electrolyte solutions). 99.
R. Hober, Physical Chemistry of Cells and Tissues (J. and A. Churchill, London, 1947), p.87 (DH for bulk electrolyte solutions). 100. K. Sneppen and G. Zocchi, Physics in Molecular Biology (Cambridge University Press, Cambridge, 2005), p.49 (DH double layer for spherical surface). 101. E.J.W. Verwey and J.Th.G. Overbeek, Theory of the Stability of Lyophobic Colloids (Elsevier, New York, 1948, reprinted by Dover 1999). This is the classic monograph on double layers in electrolyte solutions. PB theory starts on p.22. 102. D.F. Evans and H. Wennerstrom, The Colloidal Domain, where Physics, Chemistry and Biology Meet, 2nd ed. (Wiley-VCH, New York, 1999), p.131 (PB for double layer). 1 103. R.J. Hunter, Foundations of Colloid Science, 2nd ed. (Oxford University Press, Oxford, 2001), p.317, p.365, p.369 (PB for planar, spherical and cylindrical surfaces). 104. J. Lyklema, Fundamentals of Interface and Colloid Science, Vol.1 and Vol.2 (Academic Press, London, 1991 and 1995). 105. W.B. Russel, D.A. Saville and W.R. Schowalter, Colloidal Dispersions (Cambridge University Press, Cambridge, 1989), p.99 (PB and DH for planar and spherical double layers). 106. I. Rubinstein, Electro-diffusion of Ions (Society for Industrial and Applied Mathematics, Philadelphia, 1990), p.23 (PB for double layers). 107. B. Li, Minimization of the electrostatic free energy and the Poisson-Boltzmann equation for molecular solvation with implicit solvent, SIAM J. Math. Anal. 40, 2536-2566 (2009). 108. J-N Chazalviel, Coulomb Screening by Mobile Charges. Applications to Materials Science, Chemistry and Biology (Birkhaűser, Boston, 1999), p.14 (classical PB) and p.25 (quantum TF for electrons in metal). 109. For applications of double layers to (a) semiconductors, see Chavazalviel , J.P. McKelvey, Solid-State and Semiconductor Physics (Harper and Row, New York, 1966) p.489, R.B. Adler, 2 A.C. Smith and R.L. Longini, Introduction to Semiconductor Physics (Wiley, New York, 1964) p.149, J.L. Moll, Physics of Semiconductors (McGraw Hill, New York, 1964) p.121, S. Selberherr, Analysis and Simulation of Semiconductor Devices (Springer, Wien, 1984), p.28, F.F.Y. Wang, Introduction to Solid State Electronics, 2nd ed. (North Holland, Amsterdam, 1989), p. 226, I. Ipatova and V. Mitin, Introduction to Solid State Electronics (Addison-Wesley, Reading, 1996), p.281, J.O’M Bockris, A.K.N. Reddy and M. Gamboa-Aldeco, Modern Electrochemistry, Volume 2A, Fundamentals of Electrodics, 2 nd ed. (Kluwer/Plenum, New York, 2000), p. 984, Y.V. Pleskov, Electric double layers on semiconductor electrodes, chapter 6, pp 291-328, in Bockris et al. , H. Gerischher, Semiconductor electrochemistry, in Physical Chemistry, an Advanced Treatise, Vol. 9A/Electrochemistry, ed. H. Eyring (Academic Press, New York, 1970), p. 463, (b) so-called supercapacitors, see B.E. Conway, Electrochemical Supercapacitors, Scientific Fundamentals and Technological Applications (Springer, New York, 1999), (c) glass electrode pH meters, see D.J. Graham, B. Jaselskis and C.E. Moore, Development of the glass electrode and the response, J. Chem. Ed. , 345-351 (2013), and K.L. Chung, pH glass electrode and its mechanism, in J.T. Stock and M.V. Orna, editors, Electrochemistry Past and Present, ACS Symposium Series 390 (ACS, Washington, 1989), pp.286-302, (d) micro-and nano-fluidics, see H-C Chang and L.Y. Leo, Electrokinetically Driven Microfluidics and Nanofluidics (Cambridge University Press, Cambridge, 2010) and (e) reviews of various applications (e.g., batteries, fuel cells, nano- and flexible energy storage devices, electro-actuators) involving double layers of so-called room temperature ionic liquids (electrolytes without solvent), see A.A. Kornyshev, Double-layer in ionic liquids: paradigm change?, J. Phys. Chem. B , 5545-5557 (2007), and M.V 3 Fedorov and A.A. Kornyshev, Ionic liquids at electrified interfaces, Chem. Rev., , 2978-3036 (2014). 110. P. Grochowski and J. Trylska, Continuum molecular electrostatics, salt effects and counterion binding - a review of the Poisson-Boltzmann model and its modifications, Biopolymers , 93-113 (2008). 111. Y. Levin, Electrostatic correlations: from plasma to biology, Rep. Prog. Phys. , 1577-1632 (2002). 112. T. Colla, L. Nunes Lopes and A.P. dos Santos, Ionic size effects on the Poisson-Boltzmann theory, J. Chem. Phys. , 014104-1-8 (2017). 113. J.W. Gibbs, The Scientific Papers of J.W. Gibbs, Volume 1, Thermodynamics (Longmans, Green and Company, London, 1906; reprinted by Dover, 1961), p. 90. The original classic paper, On the Equilibrium of Heterogeneous Substances, was published in 1876 in the Transactions of the Connecticut Academy of Arts and Sciences. The paper’s length (300 pages) and typesetting requirements necessitated a fundraising campaign. 114. M. Bailyn, A Survey of Thermodynamics (AIP Press, New York, 1994), p.233. 115. S. Levine, On the interaction of two colloidal particles, using the complete Debye-Hückel equation, J. Chem. Phys. , 831-848 (1939), and The free energy of the double layer of a colloidal particle, Proc. Phys. Soc. A ,781-790 (1951). 4 116. R. A. Marcus, Calculation of thermodynamic properties of polyelectrolytes, J. Chem. Phys. , 1057-1068 (1955). 117. K.A. Sharp and B. Honig, Calculating total electrostatic energies with the nonlinear Poisson-Boltzmann equation, J. Phys. Chem. , 7684-7692 (1990) 118. M.K. Gilson, M.E. Davis, B.A. Luty and J.A. McCammon, Computation of electrostatic forces on solvated molecules using the Poisson-Boltzmann equation, J. Phys. Chem. , 3591-3600 (1993). 119. A.D. McGillvray and J.D. Swift, Variational principle for the Poisson-Boltzmann equation, J. Phys. Chem. , 3575-3580 (1968). 120. D.C. Chan and D.J. Mitchell, The free energy of an electrical double layer, J. Coll. Interface Sci. , 193-197 (1983). 121. T.J. Slukin, Applications of the density-functional theory of charged fluids, J. Chem. Soc. Faraday Trans. 2 , 575-586 (1981). 122. F. Fogolari and J.M. Briggs, On the variational approach to Poisson-Boltzmann free energies, Chem. Phys. Lett. , 135-139 (1997). 123. V. Kralj-Iglic and A. Iglic, A simple statistical mechanical approach to the free energy of the electric double layer including the excluded volume effect, J.Phys.II (France ) 6 , 477-491 (1996). 5 124.
E.S. Reiner and C.J. Radke, Variational approach to the electrostatic free energy in charged colloidal suspensions: general theory for open systems, J. Chem. Soc. Faraday Trans. , 3901-3912 (1990). 125. A.C. Maggs, A minimizing principle for the Poisson-Boltzmann equation, Eur. Phys. Lett. , 16012-1-5 (2012). 126. J.Th.G. Overbeek, The role of energy and entropy in the electrical double layer, Colloids and Surfaces , 61-75 (1990). 127. B. Derjaguin and L. Landau, Theory of the stability of strongly charged lyophobic sols and of the adhesion of strongly charged particles in solutions of electrolytes, Acta Physicochimica USSR , 633-662 (1941). 128. D. McCormack, S.L. Carnie and D.Y. Chan, Calculations of electric double-layer force and interaction free energy between dissimilar surfaces, J. Colloid Interface Sci. ,177-196 (1995). 129. M. Hatlo and L Lue, A field theory for ions near charged surfaces valid from weak to strong couplings, Soft Matter , 125-133 (2009). 130. A.V. Parsegian and D. Gingell, On the electrostatic interaction across a salt solution between two bodies bearing unequal charges, Biophys. J. , 1192-1204 (1972). 6 131. K. Besteman, M.A.G. Zevenbergen and S.G. Lemay. Charge inversion by multivalent ions: dependence on dielectric constant and surface charge density, Phys. Rev. E , 61501-1-9 (2005). 132. YS. Jho, F.L.H. Brown, MW. Kim and P.A. Pincus, Repulsion between oppositely charged planar macroions, PLOS One , 69436 1-7 (2013). 133. M. Trulsson, B. Jönsson, T. Åkesson, J. Forsman, and C. Labbez, Repulsion between oppositely charged macromolecules or particles, Langmuir , 11562-11569 (2007). 134. B.M.N. Clarke and P.J. Stiles, Finite electric boundary-layer solutions of a generalized Poisson-Boltzmann equation, Proc. Roy. Soc. A , 2015.0024 -1-16 (2015) 135. J.D. Jackson, Classical Electrodynamics, 3rd ed. (Wiley, New York, 1999), p.41. 136. The correct electrostatic energies of ions with charges e have potentials of mean force ( ) w x in place of the mean field energies ( ) e x . The potential of mean force (PMF) discussed in these references is that relevant to bulk electrolyte solutions, but a PMF is similarly the correct effective potential for the double layer. 137. See, e.g., D.A. McQuarrie , p.86. 138. B.W. Ninham and V.A. Parsegian, Electrostatic potential between surfaces bearing ionizable groups in ionic equilibrium with physiologic saline solution, J. Theoret. Biol. , 405-428 (1971). An earlier derivation of the solution to the PB equation in terms of 7 Jacobi elliptic functions is given by G. Jaffé, Theorie der leitfähigkeit polarisierbarer medien I, Ann. der Phys. , 217-248 (1933). 139. D.C. Grahame, The electrical double layer and the theory of electrocapillarity, Chem. Rev. , 441-501 (1947). 140. G. Lippmann, Relations entre phénomènes électriques et capillaires, Ann. Chim. Phys. (Paris) , 494-549 (1875). See also Gibbs , p.366. 141. Overbeek and Andelman do not have the ln c term in (46). They do not seem to distinguish between the renormalized Helmholtz free energy (with density (46)) and the renormalized grand free energy (with density (47)); we find it is the latter which has no ln c term. 142. Liouville first considered this equation in two dimensions (in connection with the curvature problem of two-dimensional surfaces in three dimensions) and found its general solution. Subsequently the equation was solved in one , three , four and more spatial dimensions. Planar, axial and spherical symmetries have been discussed. log / / 2 0 u v a , J. de Math. Pures et Appl. , 71-72 (1853). 8 144. H.T. Davis. Introduction to Nonlinear Differential and Integral Equations (Dover, New York, 1962), p.20, 377. 145. H.Bateman, Partial Differential Equations of Mathematical Physics (Cambridge University Press, Cambridge, 1964), p.166. 146. W.F. Ames, Nonlinear Partial Differential Equations in Engineering (Academic Press, New York, 1965), p.180. 147. J. Ockendon, S. Howison, A. Lacey and A. Movchan, Applied Partial Differential Equations, revised edition (Oxford University Press, Oxford, 2003), p.215, 221, 284. 148. O.W. Richardson, The Emission of Electricity from Hot Bodies, second ed. (Longmans, Green and Co., New York, 1921), p.50. 149. P.L. Chambré, On the solution of the Poisson-Boltzmann equation with applications to the theory of thermal explosions, J. Chem. Phys. , 1795-1797 (1952). 150. G. Leibbrandt, Nonlinear superpositions for Liouville’s equation in three spatial dimensions, Lett. Math. Phys. , 317-321 (1980); G. Leibbrandt, S-s Wang and N. Zamani, Bäcklund generated solutions of Liouville’s equation and their graphical representation in three spatial dimensions, J. Math. Phys. , 1566-1572 (1982). 9 151. D.T. Stoyanov, On the classical solutions of the Liouville equation in a four-dimensional Euclidean space, Lett. Math. Phys. , 93-96 (1986). 152. Y. Matsuno, Exact solutions for the nonlinear Klein-Gordon and Liouville equations in four-dimensional Euclidean space, J. Math. Phys. , 2317-2322 (1987). 153. To see that the electrostatic force on the x d plate is / 2 (2 ) / 2 (2 ) d E d in magnitude and points to the left (opposite to the pressure force), assume (for simplicity) that is positive and recall that the electric force/unit area on the right hand uniformly charged plate is in the direction of the surface field (2 ) d E , here pointing to the left, and with magnitude (2 ) / 2 E d . Since (2 ) E d , the result follows immediately. Thus, we may regard the total force f per unit area on the right-hand plate as the sum of a (repulsive) kinetic pressure force (2 ) p d (pointing right) and an attractive electrostatic force / 2 (2 ) E d (pointing left). The force f on the plate at x has the same magnitude as f but acts in the opposite direction. Both forces are equal to ( ) p d in magnitude when the two plates carry identical charge. 154. Ohshima discusses insulating plates of finite thickness. Conducting plates, of any thickness, are more complicated and, as far as we are aware have not been discussed in the literature. This is because the charge on conducting plate surfaces must distribute itself to make the conductor an equipotential region, on the surfaces and throughout the volume. The freely mobile surface charge is coupled to both the inner and outer electrolyte solutions, which are dissimilar for closely spaced plates, so that the inner and outer electrolyte solutions are coupled to each other. 0 155. See, e.g., Zangwill , p71. 156. I. Langmuir, The role of attractive and repulsive forces in the formation of tactoids, thixotropic gels, protein crystals and coacervates, J. Chem. Phys.6