147. -., “Concentrated Polymer 5olutions: Part 111. Nor- 156. Zwanzig, R.

Zwanzig, R., “Theoretical Basis for the Rouse-Zimm

mal Stresses in Simple Shear Flow,” ibid., 13, 955 Model in Polymer Solution Dynamics,” J . Chem. Phys.,
.
( 1967b )
157.
60,2717 (1974).
- ., J. Kiefer, and G . H. Weiss, “On the Validity of
148. -., “Correlation of Stress Data in Viscoelastic Poly-
mer Solutions,” ibid., 14,360 ( 1968). the Kirkwood-Riseman Theory,” Proc. Nat. Acad. Sci.,
149. Yamakawa, H., ‘Transport Properties of Polymer Chains 60,381 (1968).
in Dilute Solution: Hydrodynamic Interaction,” J . Chem.
Phys., 53,436 ( 1970). THE AUTHOR
150. Yamakawa, H., Modern Theory of Dilute Polymer Soh-
tions, Harper and Row, New York ( 1971). Mike Williams, born and bred in Wisconsin, took the con-
151. Zimm, B. H. “Dynamics of Polymer Molecules in Dilute venient but fortunate educational route through Madison. The
Solution: Viscoelasticity, Flow Birefringence and Dielec- B.S. led to a summer job at Los Alamos which stimulated his
tric Loss,” J. Chem. Phys., 24, 256 ( 1956). interest in rheology, and he returned to Madison to study con-
152. - ., “Theory of the Non-Newtonian Viscosity of High tinuum modeling with Bob Bird. After getting his Ph.D. in
Polymer Solutions,” Ann. N . Y . Acad. Sci., 89, 670 ( 1961). 1964, Mike spent a year at the University of Oregon with
153. -., and R. W. Kilb, “Dynamics of Branched Poly- Marshall Fixman and focused on molecular rheology. Since
mer Molecules in Dilute Solution,” J. Polymer Sci., 37, arriving at Berkeley, he has tied to synthesize these two re-
19 (1959). search interests and has also dabbled in drag reduction, block
154. Zimm, B. H., G. M. Roe, and L. F. Epstein, “Solution of a copolymer thermodynamics, and flow-induced blood damage.
Characteristic Value Problem from the Theory of Chain
Molecules,” J . Chem. Phys., 24, 279 ( 1956). His extracurricular activities include a ridiculous degree of
155. Zimmerman, R. D., and M. C. Williams, “Evaluation of participation in baseball and tennis and a fannish interest in
Internal Viscosity Models,” Trans. SOC. Rheol., 17, 23 all sports, but most of his outside time is engaged in career
(1973). guidance work with high school students.

JOURNAL REVIEW

Porous-Electrode Theory with

Battery Applications
JOHN NEWMAN
Lawrence Berkeley Laboratory and
Department of Chemical Engineering
Developments in the theory of flooded porous electrodes are reviewed University of California
with regard to simulation of primary and secondary batteries, adsorption Berkeley, California 94720
of ions and double-layer charging, and flow-through electrochemical re-
actors.
and
WILLIAM TIEDEMANN
Corporate Applied Research Group
Globe-Union, Ins., Mihvaukee, Wisconsin

Porous electrodes have numerous industrial applications the system, or the reactants may be dissolved and forced
primarily because they promote intimate contact of the through a porous electrode.
electrode material with the solution and possibly a gaseous 6. The compactness of porous electrodes can reduce the
phase. Specific factors are as follows: ohmic potential drop by reducing the distance through
1. The intrinsic rate of the heterogeneous electrochemi- which current must flow. This has obvious advantages in
cal reaction may be slow. A porous electrode can compen- reducing the losses in batteries and fuel cells. It may also
sate for this by providing a large interfacial area per unit permit operation without side reactions by providing po-
volume (for example, lo4 cm-1). tential control for the desired process.
2. Double-layer adsorption constitutes the basis for If porous electrodes were trivially different from plane
novel separation processes involving cycling of the elec- electrodes, there would be no motivation for their separate
trode potential. Just as in conventional fluid-solid adsorp- study. But here inherent complications arise because of
tion, a high specific interfacial area is desirable. the intimate contact of electrode and solution-the ohmic
3. Important reactants may be stored in the solution in potential drop and the mass transfer occur both in series
close proximity to the electrode surface by means of porous and in parallel with the electrode processes, with no way
electrodes. This permits sustained high-rate discharge of to separate them. One needs to develop an intuitive feeling
the lead-acid cell. for how and why the electrode processes occur nonuni-
4. A dilute contaminant can be removed effectively with formly through the depth of the electrode. Finally, we
a flow-through porous electrode. The proximity of the flow- seek methods for designing a porous electrode for a par-
ing stream to the electrode surface is again important. ticular application in such a way as to maximize the effi-
5. Similar arguments apply to nonconducting reactants ciency.
of low solubility. Then another solid phase (as in batteries) Electrodes with two fluid phases are excluded from con-
or a gas phase (as in fuel cells) may be incorporated into sideration here. Gas electrodes for fuel cells are reviewed

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 25

by Chizmadzhev et al. (1971) and by Bockris and Srini- Nonnenmacher ( 1960). Refinements (Posey and Moro-
vasan (1969) among others. The present article is also zumi, 1966; Johnson and Newman, 1971, for example)
more concerned with macroscopic porous-electrode theory permit the treatment of double-layer charging and make
than with the detailed morphology and reaction mecha- clearer the applicability to flow-through electrodes and
nisms of particular chemical systems (see, for example, electrodes with time-dependent structural parameters.
Burbank et al., 1971; Milner and Thomas, 1967) or with Average Quantities
the growth of crystals or layers of insoluble products (see
Hoar, 1959; Young, 1961; Vermilyea, 1963). In this macroscopic treatment, we disregard the actual
geometric detail of the pores. Thus, we can define a poten-
With these exclusions, a broad subject area remains. tial in the solid, conducting matrix material and another
Battery electrodes illustrate an industrially important ap- potential @z in the pore-filling electrolyte. These quantities,
plication of porous-electrode theory. Flow-through porous and others to be defined shortly, are assumed to be con-
electrodes have some promise for recovery and removal of tinuous functions of time and space coordinates. In effect,
electropositive metals (Ag, Au, Cu, Hg, Pb) from dilute the electrode is treated as the superposition of two con-
solutions, for electro-organic synthesis, and for oxidation tinua, one representing the solution and the other repre-
of unwanted organic pollutants and surfactants. Transient senting the matrix. I n the model, both are present at any
double-layer charging and adsorption are of interest in the point in space.
determination of the internal area of porous electrodes, in Dunning ( 1971) has included an extensive discussion
some separation processes, and in the interpretation of im- of average quantities and the derivation of certain equa-
pedance measurements on porous electrodes. tions in a manner designed to clarify the nature of the
Fundamentals of electrochemical thermodynamics, kinet- averages used.
ics, and transport processes are not explained at length. We Averaging is to be performed in a volume element within
build here on earlier expositions of these subjects (see, for the electrode. The porosity is the void volume fraction E
example, Newman, 1967, 1973). within the element, and this is taken here to be filled with
The earlier reviews of Grens and Tobias (1964), Posey electrolytic solution. The element also contains representa-
(1964), and de Levie ( 1967) are relevant and have been tive volumes of the several solid phases which may be
consulted particularly in regard to the early Russian litera- present. Let ci be the solution-phase concentration of spe-
ture. See also Gurevich et al. (1974). cies i, averaged over the pores. The superficial concentra-
tion, averaged over the volume of both matrix and pores,
MACROSCOPIC DESCRIPTION OF POROUS ELECTRODES is thus cci. For flow-through electrodes, ci is the preferred
average concentration because it is continuous as the
Porous electrodes consist of porous matrices of a single stream leaves the electrode. Furthermore, ci, rather than
reactive electronic conductor or mixtures of solids which E C ~ is
, likely to be used to correlate the composition depen-
include essentially nonconducting, reactive materials in dence of diffusion coefficients, activity coefficients, and
addition to electronic conductors. An electrolytic solution the conductivity of the solution phase.
penetrates the void spaces of the porous matrix. At a given The specific interfacial area a is the surface area of the
cime, there may be a large range of reaction rates within pore walls per unit volume of the total electrode. Let !in
the pores. The distribution of these rates will depend on be the pore-wall flux of species i averaged over this same
physical structure, conductivity of the matrix and of the interfacial area. The pore-wall flux to be averaged is the
electrolyte, and on parameters characterizing the electrode normal component of the flux of species i at the pore wall,
processes themselves. relative to the velocity of the pore wall, and in the direc-
In order to perform a theoretical analysis of such a com- tion pointing into the solution. The pore wall may be
plex problem, it is necessary to establish a model which moving slightly because of a dissolution process. Thus, ujin
accounts for the essential features of an actual electrode represents the rate of transfer of the species from the solid
without going into exact geometric detail. Furthermore, phases to the pore solution (per unit volume of the total
the model should be described by parameters which can electrode).
be obtained by suitably simple physical measurements. Next, let Ni be the average flux of species i in the pore
For example, a porous material of arbitrary, random struc- solution when averaged over the cross-sectional area of the
ture can be characterized by its porosity, average surface electrode. Thus, for a plane surface, of normal unit vector
area per unit volume, volume-average resistivity, etc. Simi- n, cutting the porous solid, n . Ni represents the amount of
larly, one can use a volume-average resistivity to describe species i crossing this plane in the solution phase, but re-
the electrolytic phase in the voids. A suitable model would ferred to the projected area of the whole plane rather than
involve averages of various variables over a region of the to the area of an individual phase.
electrode small with respect to the overall dimensions but The superficial current density i2 in the pore phase is
large compared to the pore structure. In such a model, due to the movement of charged solutes:
rates of reactions and double-layer charging in the pores
will have to be defined in terms of transferred current per
unit volume.
i2 = F C Z ~ N ~
i
A number of early and recent models represent the Similarly, the current density il in the matrix phase is de-
structure with straight pores, perpendicular to the external fined to refer to the superficial area and not to the area of
face of the electrode, and a one-dimensional approximation an individual phase.
is introduced and justified on the basis of the small diame-
ter of the pore compared to its length. As de Levie (1967) Material Balance for Solutes
has pointed out, the mathematical equations to treat are Within a pore, in the absence of homogeneous chemical
essentially identical with those of the macroscopic model reactions, a differential material balance can be written for
although a parameter such as the diffusion coefficient has a species i. This equation can be integrated over the vol-
a different interpretation. We shall prefer to speak in terms ume of the pores in an element of the electrode, and sur-
of the macroscopic model. The model set forth by New- face integrals can be introduced by means of the diver-
man and Tobias (1962) has elements which can be found gence theorem. Careful use of the definitions of average
in the earlier work of Coleman (1946), Daniel’-Bek quantities yields (Dunning, 1971) the material balance for
(1948), Ksenzhek and Stender (1956a, b ) , and Euler and species i:

Page 26 January, 1975 AlChE Journal (Vol. 21, No. 1)

 if the electrode is operating in a steady state or a pseudo
steady state where double-layer charging can be ignored.
Equation ( 2 ) becomes
This result applies to the solvent as well as the solutes.
Three different averages are represented in Equation
( 2 ) . ci is an average over the volume of the solution in
the pores. jin is an average over the interfacial area be-
tween the matrix and the pore solution, and N. is an aver- The polarization equation is necessary to express the de-
age over a cross section through the electrode, cutting pendence of the local rate of reaction on the various con-
matrix and pore. Here, it should be borne in mind that centrations and on the potential jump at the matrix-solution
the averages involve a volume which is large compared to interface. Electrode kinetics do not follow fundamental
the pore structure and small compared to the regions over laws which can be written down as reliably as the law of
which considerable macroscopic variations occur. conservation of matter. The situation is the same here as in
Equation ( 2 ) states that the concentration can change chemical kinetics or heterogeneous catalysis. The polariza-
at a point within the porous electrode because the species tion equation is the principal point where the macroscopic
moves away from the point (divergence of the flux Ni) or theory of porous electrodes will be subject to further re-
because the species is involved in electrode processes (fara- finement, as one tries to account not only for the mech-
daic electrochemical reactions or double-layer charging) anism of the charge-transfer process but also for the mor-
or simple dissolution of a solid material. This latter term phology of the electrode, the formation of covering layers
uii,, resembles the term which would describe the bulk or of crystallites of sparingly soluble species, and the trans-
production of a species by homogeneous chemical reac- port from such sparingly soluble phases to the site of the
tions. In the macroscopic model, the transfer or creation charge-transfer process.
from the matrix phases appears to occur throughout the It is common to begin with a polarization equation of
bulk of the electrode because of the averaging process. the form
Electroneutrality and Conservation of Charge
A volume element within the porous electrode will be, in
essence, electrically neutral because it requires a large elec- (9)
tric force to create an appreciable separation of charge over
where io is the exchange current density and 7, is the local
an appreciable distance. We shall also take each phase sep-
value of the surface overpotential. The surface overpoten-
arately to be electrically neutral. For the solution phase,
tial is the potential difference between the electrode and a
this takes the form
(perhaps imaginary) reference electrode of the same kind

c i
zici = 0 (3)
as the working electrode and positioned adjacent to it, just
beyond the double layer. The potential difference @I - @pz
is equal to T ~ plus
, an additive term which depends on the
local solution composition.
Our assumption here means that the interfacial region Consider a redox reaction like the oxidation of ferrous
which comprises the electric double layer (where de- ions to ferric ions, thereby leaving the electrode structure
partures from electroneutrality are significant) constitutes unchanged. One frequently assumes that the composition
only a small volume compared to any of the phases or dependence of the exchange current density and the com-
the electrode itself. This will not be true for finely pqrous
position dependence of the equilibrium potential cancel
media and very dilute solutions, where the diffuse layer
in such a way that the polarization equation can be written
may be more than 100 angstroms thick. Furthermore, we (see, for example, Newman, 1973, p. 388)
make no attempt here to treat electrokinetic effects like
electro-osmosis and the streaming potential.
It is a consequence of the assumption of electroneutrality
that the divergence of the total current density is zero. For
1
the macroscopic model, this is expressed as
V * il + V . iz = 0 (4)
Charge which leaves the matrix phases must enter the pore where io is a constant representing the exchange current
solution. In fact, combination of Equations ( 1), ( 2 ) , and density at the composition cI0, cZ0,which might conveni-
(3) gives ently be taken to be the initial concentrations of the re-
actants or the concentrations prevailing external to the
electrode.
In Equation ( l o ) , the reaction is first order with respect
to the reactant and product, at a given electrode potential.
where in is the average transfer current density (from the This would be obscured in Equation (9) because of the
matrix phase to the solution phase). V . i2 is the transfer shift of the equilibrium potential with concentration, and
current per unit volume of the electrode (A/cm3) and has io consequently has a fractional-power dependence on con-
the direction of an anodic current. centrations.
Electrode Processes For a single electrode reaction, a material balance on the
For a single electrode reaction, represented as solid phases shows how the porosity changes with the ex-
tent of reaction at each location within the electrode:
-ac= - A o V . i z
at
Faraday’s law is expressed as where

ajin= - - &mi= --V Si *iz (7)

nF nF

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 27

M i being the molecular weight and pt the density of the cioglu (1973) and Appel (1974) have concerned them-
solid phase (taken here to be a pure substance). selves with this problem. Parsons (1970) has summarized
In addition to faradaic reactions, electrode processes earlier efforts and the relevant experimental data.
can involve charging of the electric double layer at the The differential double-layer capacity
interface between the pore solution and the conducting
phases of the matrix. Let us, for the moment, exclude C=(g) =-F x,z i ( ” av.
) cj (17)
faradaic reactions and consider an ideally polarizable cj
electrod-ne which passes no steady current when held
is relatively easy to measure, and data are available over
at a constant potential (within a certain range of allowed
a wider range of concentrations.
potentials).
In the absence of transient concentration variations, and
For dependent variables we may choose q, the surface
charge density on the electrode side of the double layer, with a constant specific interfacial area, Equation (15)
reduces to
and ri, the surface excess or surface concentration of a
solute species i. Here, we mean averages over the surface
V .iz = uf (U’,
ci)
au’
+ aC -
of the pores. These surface quantities can be taken to
depend on the solution-phase composition, as represented
at
by ci and on the electrode potential @I - @2, where @2 an equation which can be used to analyze double-layer
is measured with a given reference electrode. The rela- charging in porous electrodes.
tionship holds Analysis of electrode processes is the basis for evaluating
the term ajin in the material balance Equation (2). The
general case involving double-layer charging and all the
differential coefficients in Equation ( 16) becomes too
since the interface as a whole is electrically neutral and complicated to pursue profitably at this point.
the charge on the solution side of the double layer is com- Transport Processes
prised of the contributions of adsorbed solute species. For In the matrix phase, the movement of electrons is gov-
a solid surface it is only changes in q and ri which can be erned by Ohm’s law
determined easily and which are involved in impedance
measurements and changes in solution composition. For a il = - o V @ ~ (19)
porous material of large specific area, one can, however,
grind up the electrode and determine the magnitude of ri where u is the effective conductivity of the matrix. This
by chemical analysis. quantity will be affected by the volume fraction of the
In the absence of electrokinetic effects, that is, ignoring conducting phase or phases, the inherent conductivity of
translation of the solution side of the double layer in a each conducting solid phase, and the manner in which
direction parallel to the surface, one can express the ma- granules of conducting phases are connected together.
terial balance for a solute species at the interface as (New- In a dilute electrolytic solution within the pores, the flux
man, 1973, p. 206) of mobile solutes can be attributed to diffusion, dispersion,
migration, and convection:
ami mi
-
at
= a iin.faradaic - a /in = - -Lfaradaic - a ji.
nF
(14)
where the subscript faradaic refers to charge or mass which Since Ni is the superficial flux based on the area of both
is actually transported through the interface or is involved matrix and pore, one can think of N ~ / Eas the flux in the
in a charge transfer reaction. It is assumed in the last form solution phase. Similarly, V / E will be roughly the velocity
that only a single electrode reaction is involved. For the in the solution phase if we take v to be a superficial bulk
ideally polarizable electrode, these faradaic terms are taken fluid velocity, for example, the volumetric flow rate enter-
to be zero. For a single electrode reaction, the left side of ing the electrode divided by its superficial area. The
Equation (9)O r (10) represents a &,farad&. molar-average velocity v* would thus be given by
Addition of Equation (14) according to Equation (5)
gives CTV* = 2 Ni
i

The current transferred from the matrix to the solution is

involved either in double-layer charging or in faradaic is the total solution concentration. (The solvent is to be
electrode reactions. included in both these sums.)
The differential of the surface concentration can be ex- The ionic diffusion coefficient and mobility of a free
pressed as solution require a correction for the tortuosity of the pores

dri = (g) Y
dU’+
k#O
(3)
ack U’.
dck
j#O,
cj
k
(16)
in order to yield Di and ui. A porosity factor has already
been taken out; that is, cDi might logically be regarded to
be the effective diffusion coefficient of the species in the
pore solution in the same way that u is the effective con-
where U’ = - @2. This formidable equation indicates ductivity of the matrix. Estimation of such physical proper-
all the differential coefficients which must be known for an ties is considered again in a later section.
exact treatment of transient processes involving double- D, represents the effect of axial dispersion-the attenua-
layer charging. These differential coefficients have been tion of concentration gradients as a fluid flows through a
determined carefully for ideally polarizable mercury sur- porous medium. Plug flow does not prevail in the pores;
faces in relatively simple electrolytic soutions. Their estima- fluid near the wall moves more slowly than fluid toward
tion becomes inherently more difficult for solid surfaces, the center of the void space. The compensation for this
particularly in the presence of faradaic reactions. Nigan- convective nonideality appears as a diffusion phenomenon.

Page 28 January, 1975 AlChE Journal (Vol. 21, No. 1)

However, the dispersion coefficient is not a fundamental In other words, the transference numbers of the ions are
transport property-it depends on the fluid mixing and assumed to be inversely proportional to their partial molar
vanishes in the absence ot convective fluid motion. A cor- volumes. However, this choice is based solely on conveni-
relation is given in the section on flow-through porous ence (Newman and Chapman, 1973), and no physical sig-
electrodes. Dispersion should properly be represented by nificance should be attached to it.
a tensor coefficient since dispersion is different in direc- The cation flux can now be expressed as
tions parallel to and perpendicular to the bulk flow veloc-
- c ( D + D ~ ) V C++ -iz
t+O
ity v. N + = C + .V (30)
The current density in the solution phase according to Z+F
Equation (1) now becomes with a similar expression for the anion flux, and the solvent
iz = - KV@Z- aF Ci
Di Vci (23)
flux is
+
No = co'V - c ( D Da) V C O (31)
where Here D is the effective diffusion coefficient of the electro-
lyte.
K = cF2 ZFqCi (24) Substitution of these fluxes into the material balance 2
i
and rearrangement gives two fundamental equations for
is the effective conductivity of the solution phase. The porous electrodes involving a binary electrolyte:
second term on the right in Equation (23) represents the V,t-O
- vet+o
diffusion potential. Convection and dispersion have made
no contribution in this equation since the solution is elec-
-de+ v - v . = a [ V o j o n + -
at V+
j+n -i--n+ V-
]
trically neutral. If the ionic diffusion coefficients are all
equal, the diff usion-potential term will disappear as well.
There are two limiting cases which are likely to receive
treatment. For the reaction of a minor component from
a solution with excess supporting electrolyte, migration is and
neglected. Equations (20) and (2) become
arci
-
at
+V [VCJ= uji,, +V * [(Di + Da) eVci]
(25)
For assessing the potential variation within the solution,
one is tempted to neglect the diffusion potential in Equa-
tion (23) and take the conductivity K to be constant (for
uniform porosity). However, on the basis of a small ratio Equation (32) is an overall conservation equation, ex-
of reactant to supporting electrolyte, one is occasionally pressing perhaps conservation of volume. I t shows how the
justified in neglecting the potential variation altogether. velocity V. changes as a result of an electrode reaction and
The binary electrolyte, the other limiting case, is simple the changing composition of the solution.
because only two ionic species are present. The electro- Equation (33) is a material balance for the electrolyte.
neutrality condition ( 3 ) then allows the electrolyte con- The corresponding equation in the dilute-solution approxi-
centration to be defined as mation does not reflect the dilution which would occur if
the solvent were to be produced in an electrode reaction.
c = c + / v + = c-/u- (26) The equations become quite similar if one is willing to
neglect the concentration dependence of both the partial
Space will not be devoted to dilute-solution theory for this
situation since it is mainly of interest in comparison with
molar volume ve
of the electrolyte and the cation trans-
ference number t+O with respect to the solvent velocity.
the more complete treatment in the next subsection. Instead of Equation (23), we may wish to express
Concentrated Binary Electrolyte Ohm's law in the form (see Newman, 1973, Equation
81-8)
Many battery systems involve solutions of a single elec-
trolyte. Consequently, it is desirable to develop the theory
with all possible exactness, particularly since the equations - -- - VQ., -
i2
K
s+
-+---
v+n
t+O
z+ v + nco ) (34)
to use are not hopelessly complicated and thermodynamic
and transport data are frequently available for binary solu- where @2 is the potential in the pore solution measured
tions. Newman et al. (1965) developed the transport equa- with a reference electrode having the stoichiometric coeffi-
tions in terms of the molar-average velocity. Newman cients si and number n of electrons transferred, and pe is
(1967) states the results with the mass-average velocity the chemical potential of the electrolyte.
and the solvent velocity, while Newman and Chapman
Surnrnory
(1973) showed how the volume-average velocity can be
used. The structure of equations describing the porous elec-
The superficial volume-average velocity .V is defined as trode does have order. A certain level of complication is
necessary in order to treat the physical processes which
V. = vo No + v+N + + v- N- (27) are involved. Equation (19) covers transport in the matrix,
while Equations ( l ) , ( 2 ) , ( 3 ) , and (20) relate to trans-
whereV+ and 7-satisfy port, conservation, and electroneutrality in the solution
- - - phase. However, the bulk appearance of a species is now
v+ v+ + u- v- = ve (28) related to electrode processes occurring throughout the
This is not sufficient to determine the partial molar vol- volume. Thus, Equations ( 4 ) and ( 9 ) , representing bound-
umes of the ions separately. Consequently, we let ary conditions between the phases, must apply here
- - - - throughout the volume.
V+ V + = t-OV, or t + O V+ V + = t-Ov- V - (29) Complications and uncertainties are largely related to

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 29

electrode processes. Changes in the structure and proper- large values of either 8 or 9,the ohmic effect dominates,
ties of the electrode need to be related to the local extent and the reaction distribution is nonuniform. The ratio a,/a,
of discharge or the integral of the local transfer current seems to be unavoidable, but its role is more difficult to
V - ip. discern. For small values of K / U , the reaction occurs prefer-
entially near the electrode-solution boundary at the ex-
NONUNIFORM REACTION RATES
pense of the regon near the backing plate.
Many special cases are treated in the literature. The The description of a porous electrode in the absence of
simplest assume that the solution phase is uniform in com- concentration variations is similar to that of the secondary
position-either because the current has just been switched current distribution at a disk electrode (Newman, 1973,
on and the concentrations have not had time to change or section 117). It is customary to introduce one of two ap-
because there is forced convection through the electrode so proximations to the polarization Equation ( 9 ) . For very
as to maintain the composition uniform. Double-layer low overpotentials, this equation can be linearized to read
charging is also to be ignored. Still open to the investi-
gator’s choice are the relative electric conductivities of
the two phases, the thickness of the electrode, and the
form of the polarization equation expressing the electro-
chemical kinetics. On the other hand, at very high overpotentials, one or the
Consider a porous electrode in the form of a slab of other of the terms on the right in Equation ( 9 ) can be
thickness L. The electrode is in contact with an equipoten- neglected. The first term is neglected for a cathode; the
tial metal surface on one side at x = L and in contact with last term for an anode. This is referred to as the Tafel up-
an electrolytic solution on the other at x = 0. In such prob- proximation.
lems it is convenient to specify the superficial current For systems with a finite conductivity of each phase, the
density I flowing through the electrode rather than the solution of these equations with the linear electrode kinetics
potential difference across it. Positive values of I will cor- has been given by Euler and Nonnenmacher (1960) and
respond to anodic currents. The structure of the electrode restated by Newman and Tobias (1962). These last work-
will be taken to be uniform. ers also obtained the solution for Tafel polarization, while
The four governing differential equations take on a one- Micka (1965) was able to treat the complete Equation
dimensional form. Equation (4) expresses conservation of ( 9 ) in the special case where a, = aC.Daniel’-Bek (1948)
charge, Equation (23) reduces to Ohm’s law for the pore had earlier presented the results for linear and Tafel kinet-
solution, and Equation (19) is the same law for the matrix. ics in the case where the current collector is positioned
The polarization Equation (9), with ?lS = @I - @2, de- at the electrode-solution boundary instead of being at x =
scribes the transfer of charge from the matrix to the solu- L.
tion. Sufficient boundary conditions include Many papers deal with matrix conductivities much
greater than the solution conductivity, so that the matrix
i 2 = -I, il =0, and @p= 0 at x = O (35) potential @I is uniform. It then becomes possible to treat
very thick electrodes, where the analysis is similar to that
i 2 = 0 at x = L (36) for the diffuse part of the electric double layer (Newman,
These say that at the electrode-solution interface (x = 0) 1973, section 52), as pointed out by Frumkin (1949).
the current is carried entirely by the pore electrolyte while Frumkin gave solutions for linear kinetics [Equation (39)]
at the metal backing, the current is carried entirely by the and also for the full Equation (9) for the special case
matrix. As an arbitrary reference of potential, we choose where aa = aC.He spoke, however, in terms of corrosion
@ p = 0 at x = 0. Somewhere within the electrode, between processes in the interior of a tube. Ksenzhek and Stender
x = 0 and x = L, the current is transferred from the solu- (1956b) gave the solution in the context of porous elec-
tion to the matrix, and the rate of reaction is proportional trodes. For semi-infinite electrodes, Posey ( 1964) extended
to di2/dx. this work to the cases a, = 2ac and 3ac (as well as the re-
The reaction distribution is generally nonuniform within ciprocal cases where aC equals 2aa or 3aa). For semi-in-
the electrode. To minimize the ohmic potential drop, the finite electrodes, the gross current-potential behavior can
current should be divided between the solution and the be obtained immediately for any ratio of aa to ac, by anal-
matrix in proportion to their effective conductivities. How- ogy with the theory for the diffuse part of the electric
ever, this requires high reaction rates near x = 0 and double layer. Then
x = L. Slow electrode reaction kinetics force the reaction
to be more uniformly distributed in order to reduce the
transfer current density. These competing effects of ohmic
12 = - ac exp
aaacF
[
2ai 0K R T
(F)
a,FV

potential drop and slow reaction kinetics determine the

resulting distribution.
For the problem as formulated above, four dimensionless
ratios govern the current distribution. These can be stated
as a dimensionless current density where V = - a2 at x = 0, the solution side of the
electrode.
(37) For finite electrodes and a high matrix conductivity,
RT Ksenzhek (1962b) and Winsel (1962) give solutions for
aa = aC in Equation ( 9 ) . Posey (1964) was able to in-
a dimensionless exchange current for the electrode clude the case a, = 2ac (or the converse, ac = 2 4 . He
also gives approximate solutions for all values of aa/ac.
(38) One feature of the results is the nonuniformity of the
RT
reaction rate within the electrode. Figure 1 shows the dis-
the ratio a,/ac of the transfer coefficients in the polariza- tribution of reaction rate for Tafel kinetics and for = K .
(T

tion Equation (9), and the ratio K / U of the effective con- As in the classical secondary-current-distribution problem,
ductivities of the solution and matrix phases. the value of aio is unimportant as long as the backward re-
The first two, 6 and P, are ratios of the competing effects action can truly be neglected. In the Tafel case, the cur-
of ohmic potential drop and slow electrode kinetics. For rent distribution depends only on the parameter 8 and on

Page 30 January, 1975 AlChE Journal (Vol. 21, No. 1)

 3 (For Tafel kinetics, with a given value of I , changes in a&
merely add a constant to a1 - @2 without affecting the
Ordinate current or potential distributions in any other way.) The
slope of the curves gives the current density flowing in
I 09594 1.084 either the matrix or the pore solution, according to Ohm’s
I0 0 693 I943
100 0 1693 12 6 9
law. Consequently, the second derivative of either 91 or
@2 is related to the local reaction rate or the rate at which
current is transferred from one phase to the other. The
difference @1 - aPzalso gives this reaction rate through
the polarization equation.
The total potential loss in the electrode, a combination
of kinetic or surface overpotential and ohmic potential
drop, is given by the difference between the potential 9 1
in the matrix at the current collector and the potential
in the pore solution at the pore mouth [where @pz was actu-
ally taken to be zero in Equation (35)1. For linear polar-
ization, this loss can be expressed as

- 0
0 0.2 0.4 0.6 0.8 1.0
y=x/L (43)
Fig. 1. Reduced current distributions for Tafel The potential loss for Tafel polarization is plotted in Fig-
polarization with u = K. ure 3 under the condition of Equation (42). This corre-
sponds to the so-called “Tafel plot” and has a slope of
2.303 for low values of 6. As 6 increases, the potential drop
the ratio K / U . (For cathodic polarization in the Tafel range, due to resistance becomes important. For a high conduc-
replace aa by - a c in Equation (37) in order to use these tivity of one phase, a double Tafel slope results, as pointed
results.) For a small value of 8, the reaction is uniform; but out by Ksenzhek and Stender (1956b) and by Winsel
for large values of 8, the reaction takes place mainly at the (1962), among others. Thus, the lower curve in Figure 3
electrode interfaces. The ratio K / U serves to shift the reac- attains a slope of 4.605 for large 6. If both conductivities
tion from one face to the other so that the reaction is some- are finite, the ohmic contribution to the potential loss will
what more uniform as K approaches u at constant 6. eventually dominate. Thus, the upper curve in Figure 3
Curves for the linear-polarization equation would have becomes linear in 6, not linear in the logarithm of 6. The
the same general appearance as Figure 1. In analogy with behavior of both curves for large 6 illustrates the general
the classical problem of the secondary current distribu- rule that the ohmic potential drop becomes more important
tion, the reduced distribution depends only on the param- at large currents.
eter v and on the ratio K / U and is independent of the The Tafel approximation becomes poor at low currents.
magnitude of the current. The distribution becomes non- One cannot extrapolate Figure 3 since p@1( L ) should ap-
uniform for large values of V, that is, for large values of proach zero as 6 approaches zero. The backward reaction
the exchange current density, the specific interfacial area,
or the electrode thickness or for small values of the con-
ductivities. The ratio K / U still shifts the reaction from one
face to the other.
The distance to which the reaction can penetrate the
electrode determines how thick an electrode can be effec-
tively utilized. This penetration depth is characterized by
the length

Frumkin (1949) pointed out the analogy of the penetra-

tion depth to the Debye length in diffuse-double-layer
theory. The concept of the penetration depth has also
been emphasized by Ksenzhek and Stender (1956b) and
by Winsel ( 1962). Electrodes much thinner than the pene-
tration depth behave like plane electrodes with an en-
hanced surface area. Electrodes much thicker are not fully
utilized. At high current levels, L/6 is a length which may
be more characteristic of the penetration of the reaction.
For diffusive transport of reactants or consumption of re-
actants stored in the matrix or the pore solution, other fac-
tors will enter into the optimization of the electrode thick-
ness.
Potential distributions for Tafel kinetics are illustrated
in Figure 2 for
Fig. 2. Potential distributions for Tafel polari-
zation with u = K. Here 4, = aQF/RT (or
RT --acF/RT for cathodic currents).

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 31

 30 c - exp
a,FV
]
( - F) (44)

and the current-potential relation is given by

where

LOG 6
Fig. 3. Potential of the metal backing plate as it depends on 6 .
Here B = &/RT (or -acF/RT for cathodic currents).

Mass transfer limitations external to the electrode should

should be accounted for, and eventually the linear ap- also be included. We have taken c,O to be the concentration
proximation becomes applicable at very low values of 6. at the face of the electrode, adjacent to the solution.
Equation (40) gives the current-potential relation for a The penetration depth can be taken to be the reciprocal
semi-infinite electrode with a high conductivity of one of kl (compare Ksenzhek, 1962a), and for large anodic
phase, and Equation (43) is the result without these re- polarization this becomes equal to L / y l . Equations (45)
strictions but instead restricted to the linear approximation. and (46) also give a double Tafel slope at high polariza-
tion. This phenomenon, due to a mass-transfer effect, has
been pointed out by Ksenzhek (1962a), by Gurevich and
MASS TRANSFER Bagotskii (1963b), and by Austin and Lerner (1964) and
has been contrasted with the double Tafel slope, men-
Steady Moss Transfer
tioned below Equation (43), due to an ohmic effect,
A solution-phase reactant is depleted during the opera- Austin (1964) has examined the relative importance of
tion of a porous electrode, and diffusion of this species these two effects on the basis of the ratio aanF2eDlclo/
from a reservoir at the face of the electrode represents a s ~ R T K 8/yl.
loss, in addition to the ohmic potential drop and surface Gurevich and Bagotskii have developed complicated
overpotential considered in the previous section. analytic solutions for the first special case. These become
Two special problems might be defined in this area. The so complicated that limiting cases (discussed above) must
first involves a redox reaction obeying Equation ( 10). The be examined or machine computation becomes convenient.
reactant and product species, whose stoichiometric coeffi- Bomben (1963) presents computer programs for this case
cients are taken to be +1 and -1, move by molecular and also the situation where the reverse reaction term in
diffusion alone since convection is assumed to be absent Equation (10) can be neglected. Grens and Tobias
and migration is negligible because an excess of support- (1965), using a computer program which includes mass
ing electrolyte is presumed to be present. This excess of transfer of each species, including the supporting electro-
supporting electrolyte is also used to justify the approxima- lyte, have made comparisons with analytic solutions con-
tion of a constant solution-phase conductivity K and neglect cerning variations of composition and conductivity and ap-
of the diffusion potential' so that Equation (23) reduces proximations to the polarization equation. In this manner,
to Ohm's law for the solution phase. The second special Grens (1970) has examined assumptions of the one-dimen-
case involves a binary electrolyte, where the solution-phase sional approximation (see also de Levie, 1967), of in-
conductivity can be taken to be proportional to the con- finitely thick electrodes, of uniform concentration or con-
centration. * ductivity, and of negligible potential gradient in the pores.
The first special case was treated by Newman and Tobias Today, one could also test (or relax) the assumptions of
[1962, with neglect of the reverse reaction in Equation invariant electrode matrix, constant diffusion coefficients
( l o ) ] , by Ksenzhek (1962a, also for an infinitely thick and mobilities, and absence of fluid flow, even that due
electrode and with negligible ohmic potential drop in either to the reaction itself. Austin (1969) has reasserted the
phase), and by Gurevich and Bagotskii (1963a, b, sum- value of analytic solutions for limiting situations.
marized and more readily accessible in 1964). For each Deposition from a binary electrolyte has been treated
species, two new parameters are introduced-a diffusion by Newman and Tobias (1962) and by hlicka (1966).
coefficient and a bulk or characteristic concentration cia. Bomben (1963) gives a computer program suitable for the
These can be combined into the dimensionless group yi = Tafel case.
siIL/nF E Dicio. Posey and Misra (1966) and later Alkire (1973) have
With negligible ohmic potential drop in either phase, treated, by computer simulation, bipolar porous electrodes,
the reaction rate distribution is given by that is, electrodes with no external electrical connection
but with induced polarization because a current passes
through them. The apparent resistance of the porous elec-
trode is different from that of an equal volume of solution
0 In the steady state, the diffusion potential can be included, without
a separate term, by using a modified conductivity:
not only because the porosity is less than one but also
because the current has the possibility of reacting cathodi-
cally on one side of the electrode, flowing through the
matrix, and emerginq by anodic reaction at the other side
of the electrode. With a bipolar tubular gold electrode,
Posey and Misra were able to determine the transfer co-
efficient of the ferricyanide-ferrocyanide redox reaction.

AlChE Journal Wol. 21, No. 1)

Page 32 January, 1975
Transient Man Transfer nonconducting crystallites of sparingly soluble fuel to con-
To follow the course of the discharge of a porous elec- ducting particles which are the site of the electrochemical
trode, beginning with a uniform solution composition and reaction. They obtained an analytic solution exhibiting a
following the development toward a steady state, requires limiting current due to internal mass-transfer limitations,
a consideration of the time derivative in Equation (8). and they discussed acceptable limits of the solubility of
Stein (1959, see also Newman, 1962) treated the change the sparingly soluble reactant. Self discharge can impose
of sulfuric acid concentration in the pores of a lead acid an upper limit, and adequate power density, a lower h i t .
battery. The performance during a very high rate discharge Later, Dunning et al. (1971) analyzed discharge and cy-
is limited by a severe depletion of acid directly at the cling, with the internal mass-transfer coefficient dependent
mouth of the pores of the positive plate. To study this on the local state of charge. The next work (Dunning et
effect, Stein was able to avo,d consideration of structural al., 1973) included discharge and cycling of silver-silver
changes, and he assumed that the electrode reaction itself chloride and cadmium-cadmium hydroxide electrodes in
occurred reversibly. binary electrolytes of NaCl and KOH, respectively. The
Grens and Tobias (1964) undertook a computer analysis mechanism of discharge included, in series, the kinetics of
including migration and diffusion of each solute species dissolution or precipitation of the sparingly soluble fuel, the
and a polarization equation like Equation (10). The cad- diffusion to the reaction site, and the electrochemical ki-
mium anode in KOH and the reduction of ferricyanide netics of the reaction-with regard for the changes in the
in NaOH received detailed treatment. areas available for these processes. Also accounted for were
A characteristic time for diffusion processes is L2/Di. porosity changes according to Equations (11) and (12),
For an electrode thickness of 0.1 cm and an effective fluid flow according to Equation (32) (or its equivalent),
diffusion coefficient of cmZ/s, this leads to diffusion variations in conductivity and activity coefficient in the use
times of the order of 1000 s. However, for thick electrodes, of Equation (34), and electrolyte transport according to
the presence of a reservoir of unreacted species in the depth Equation (33) (or its equivalent). However, the concen-
has little effect on the dominant processes occurring near tration derivative in this equation was neglected (pseudo
the solution side of the electrode. The penetration depth steady state).
[see, for example, Equation (46)] can then be used Alkire and Place (1972) have analyzed the consumption
rather than L in forming the characteristic time for diffu- of a solid reactant as did Winsel (1962) but with treat-
sion. Thus, 90% of the electrode potential change can ment of the solution composition according to the pseudo
occur in a time as short as 30 s. (The characteristic time steady state method. The calculations apply specifically to
for charging the double-layer capacity is usually much copper dissolution from an inert porous substrate, for ex-
less. The ratio of the double-layer-charging time to the ample, stainless steel.
diffusion time is D&/K. This ratio is greater than one only The lead dioxide electrode has been modeled recently
for very high interfacial areas and low conductivities.) by both Simonsson (1973b, c) and by Micka and Rouzar
( 1973). The former analysis appears to be comprehensive;
the latter follows Stein (1959) in assuming reversible elec-
BATTERY SIMULATION trode kinetics. This leads to the conclusion that the poten-
Porous electrodes used in primary and secondary bat- tial in the pore electrolyte is a unique function of the con-
teries invariably involve solid reactants and products, and centration, and this obscures the formulation of boundary
the matrix is changed during discharge. Consequently, conditions (Newman, 1962). In the two papers, the figures
no steady state operation is strictly possible. Such systems of concentration versus position within the electrode, with
are complex, and their simulation on the computer,is an time as a parameter, provide an interesting contrast.
active area of research. Simonsson assumed a constant concentration at the pore
mouth; Micka and Rou:ar show a concentration which
Structural Changes eventually goes to zero at the pore mouth.
Winsel (1962) allowed for the consumption of a solid Gidaspow and Baker (1973) emphasize porosity changes
reactant by setting the transfer current equal to zero at and blocking of the pores as a mechanism of battery fail-
any point in the electrode where the charge passed was ure.
equal to that which this fuel could supply. Because of the One now discerns a number of factors which can affect
nonuniform current distribution, the solid fuel is exhausted the extent of utilization of a battery as a function of rate
first at the side of the electrode adjacent to the solution. of discharge:
Cutting off the reaction in this region forces a higher 1. Higher current densities yield higher overpotentials,
average reaction rate in the remainder of the porous struc- and thus a given cutoff potential is reached sooner. How-
ture (for a constant-current discharge), yielding a higher ever, this must be considered in conjunction with a factor
electrode overpotential. The ohmic potential drop in the which leads to a time dependence of the potential.
solution in the depleted part of the electrode also con- 2. Acid depletion at the pore mouth (Stein, 1959)
tributes significantly to the overpotential. Winsel could hastens the end of discharge of the lead acid battery, par-
thus describe battery discharge without accounting for the ticularly at high current densities.
composition variation in the solution or changes in the
matrix porosity. 3. The pores may become constricted or even plugged
Alkire et al. (1969b) treated the change in pore size and with solid reaction products. A nonuniform reaction dis-
solution composition and the attendant fluid flow for dis- tribution will accentuate this problem at the mouth of the
solution of porous copper in sulfuric acid solution. The pores.
treatment of a straight-pore system is particularly appro- 4. Utilization of the solid fuel and covering of the reac-
priate here since they were able to wind copper wire care- tion surface with reaction products. Notice that both Dun-
fully and sinter the coil to produce a uniform pore struc- ning et al. (1973) and Simonsson (1973b) assumed that
ture (Alkire et al., 1969a). In contrast to the work of the local extent of utilization was constrained to be incom-
Grens and Tobias (1964), the time derivative of solution plete, and Simonsson allowed this limit to depend on the
concentration was neglected (resulting in a so-called discharge current density.
“pseudo steady state analysis”). 5. Rates of mass transfer between crystallites and the
Dunning and Bennion (1969) developed a model for reaction surface may become more limiting as the dis-
battery electrodes with a description of mass transfer from charge exhausts the front part of the electrode. This could

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 33

account for changes in the apparent limit of utilization lead dioxide electrode. Such electrodes or cells can be
with current density. cycled with current (or potential) as an arbitrary function
of time.
Optimization of Electrode Thickness
Advice on how thick to build a p o r y electrode DOUBLE-LAYER CHARGING AND ADSORPTION
(Ksenzhek, 1962b; Gurevich and Bagotskii, 1969) does
not go significantly beyond pointing out that polarization A measurement of the double-layer capacity of a porous
decreases little for increasing thickness beyond the pene- electrode most directly reflects the active surface area co-
tration depth [see Equations (41) and (46)], because of herently connected electrically and therefore accessible for
the poor utilization of the added material. We should ex- electrochemical reactions. Changes in the nature of the
pect, however, to find cost, weight, and volume optimiza- solid phase composition, surface crystal structure, adsorbed
tion for given duty cycles and cut-off potentials. materials, solution composition, temperature, and electrode
potential also influence the measured value. Such results
Machine Computation should be useful for the characterization of battery elec-
The complexity of the factors which we wish to include trodes and may be especially valuable since the electrode
in the analysis of the behavior of porous electrodes leads is not destroyed, and indeed need not be removed from
many of us to use a high-speed digital computer. the cell in which it is being studied.
For battery electrodes involving a binary electrolyte, let There have been many contributions to the theoretical
us suggest the following computational scheme (Dunning, understanding of double-layer charging. Ksenzhek and
1971). Equations (32) and (33) simplify considerably Stender (1956a) were the first to treat the subject, and
with assumptions of constant transference numbers and further refinements have included development of a macro-
partial molar volumes. Equations ( 7 ) , ( l l ) , and (32) can homogeneous model (Ksenzhek and Stender, 1956b; New-
then be combined and integrated to relate the velocity to man and Tobias, 1962) and treatment of coupled faradaic
the current density reactions (Ksenzhek, 1963; Austin and Gagnon, 1973),
changes in solution composition (Ksenzhek, 1964; John-
son and Newman, 1971), effects of external surface area
(Ksenzhek, 1964; Posey and Morozumi, 1966) finite con-
ductivity u of the electrode matrix (Posey and Morozumi;
(47) Johnson and Newman), effect of ohmic potential drop ex-
with the boundary condition that both v. and i2 vanish at ternal to the porous electrode (Posey and Morozumi; John-
a plane of symmetry or the backing plate. son and Newman; Tiedemann and Newman, 1974), and
For a highly conducting matrix (u >> K ) , one can use imposed alternating current (Winsel, 1962; de Levie,
four principal unknowns-the electrolyte concentration c, 1963, 1967; Darby, 1966a, b ) .
-
the current density i2, the transfer current i = V i2, and Experimental verification of these models has been per-
the solution potential ~2 (or, equivalently, the local elec- formed by Stender and Ksenzhek (1959), Bird et a].
trode overpotential). The velocity is eliminated by means (1969), Johnson and Newman (1971), Gagnon (1973,
of Equation (47), and the porosity is related to the trans- 1974), and Tiedemann and Newman (1974). The merits
fer current density and the time step by Equation ( 11). of each experimental technique are often specific to the
We have then programmed four equations for simul- system currently under investigation. Of the effects men-
taneous solution as coupled, ordinary, nonlinear, differen- tioned previously, the most difficult to allow for is that as-
tial equations at each time step, with boundary conditions sociated with faradaic reactions. Little detailed informa-
at two values of x (Newman, 1968, 1973-appendix C ) . tion concerning the electrode kinetics of most porous elec-
Straightforward to program are the relation i = V * i2 trodes is available. Also, the relative importance of faradaic
and Equation (34) representing Ohm's law. The activity reactions will change with the specific electrode composi-
coefficients (relating pe to c) and the conductivity can be tion and age (or cycling history). It is therefore desirable
taken to de end on the concentration. The porosity de- to employ experimental techniques which minimize the
P
pendence o K should be accounted for, and the equations
must be linearized properly to assure convergence.
effects of faradaic reactions in the measurement of the
double-layer capacity. We shall briefly mention three ex-
We have programmed Equation (33) symmetrically be- perimental techniques and refer the reader to de Levie
tween the old time step and the present one, in order to (1967) and Posey and Morozumi (1966) for more com-
attain stability. Simplifications were made for constant plete analysis.
transference numbers and partial molar volumes, and D, Bird et al. (1969) measured the double-layer capacity
was set equal to zero. Both the porosity and concentration of battery electrodes by applying a potential step and inte-
dependence of D were included. The pseudo steady state grating the resulting passage of current. A critical condi-
approximation results when the time derivative of concen- tion in this procedure is the absence of faradaic currents
tration is omitted from this equation. Simonsson ( 1 9 7 3 ~ ) in the region of the applied potential step. Because of the
finds that this approximation is valid for low discharge distributed capacitance within the electrode, it is not possi-
rates but can lead to appreciable error for rapid discharge. ble merely to subtract the faradaic portion of the charging
Finally, the polarization equation relates the transfer curve.
current i to the concentration and overpotential. One can Gagnon (1973, 1974) has employed a triangular voltage
include here complexing and mass transfer of a sparingly sweep method. While he has shown the effects of dis-
soluble reactant and electrochemical kinetics on a surface tributed capacitance, faradaic reactions, and electrode
area which changes with the local state of charge. thickness, best results are obtained when experiments are
The boundary conditions can be made to include a Em- performed at reduced temperatures (-51"C), which mini-
ited reservoir of solution adjacent to the electrode and mize faradaic reactions.
perhaps even a remote reservoir, connected to the adjacent Johnson and Newman (1971) have shown that the cur-
reservoir by a mass-transfer coefficient. This will allow rent response of a porous electrode to a step change in
simulation of the recovery of the electrode potential with the potential yields, under certain circumstances, a nearly
time as the electrolyte concentration is restored. A com- constant value of i d the product of the current density
plete lead-acid cell can be simulated by juxtaposition (in and the square root of time (also indicated by Ksenzhek
the computer) of a porous lead electrode and a porous and Stender, 1956a, Posey and Morozumi, 1966). A value

Page 34 January, 1975 AlChE Journal (Vol. 21, No. 1)

 relatively long contact times for treatment of very dilute
solutions. A fluidized-bed electrode (Fleischmann et al.,
1971) can replace the stationary porous matrix.
In modeling of the system for design purposes, there are
a number of factors which might be included in the analy-
sis. Many of these are already relevant for porous elec-
trodes without flow, for example:
1. Should a straight-pore model or a homogeneous
macroscopic model be used?
2. Should transport processes be described by dilute or
concentrated electrolyte theory? How are conductivities
and other transport properties to be estimated for porous
media? Should ohmic losses in the solid matrix be in-
cluded?
3. What are the kinetics of the reactions and the effects
of side reactions?
4. Is the system isothermal?
5. Is operation steady or transient?
New factors are also introduced specifically for flow-
through electrodes. These include:
1. Is the flow uniform through the electrode, or is there
3
channeling?
2. What is the direction of flowing solution with respect
Fig. 4. Comparison of experimental and theoretical results for to the counter electrode? The counter electrode can be
potentiostatic double-layer charging of porous PbOz electrodes. upstream or downstream or it can be beside the working
Area = 241 cmz, L = 0.095 cm, temperature = 28°C. a freshly electrode. Two, rather than one, dimensional consideration
prepared PbO2 electrode, AV = 2.51 mV, 7. = 1.33, aC = 23.33 of the distributions of concentration, current, and potential
f/cm3; 0 cycled PbO2 electrode, AV = 1.52 mV, 7. = 0.768, OC may be necessary.
= 26 f/cm3. (1 is a ratio of the external resistance, ohm - cm2, 3. How fast can reactants be transported from the flow-
to L/K.)Solid curves are theoretical. ing solution to the solid surface? Chemical-engineering cor-
relations of mass-transfer coefficients in packed beds (Bird
et al., 1960, pp. 411 and 679; Wilson and Geankoplis,
1966) are being extended to electrochemical situations by
of the double-layer capacity can be inferred from this con- Yip (1973), Gracon (1974), and Appel (1974).
stant value of z dz?However, most porous electrodes have 4. Do the flow atterns promote significant axial mixing?
nonzero resistances which prevent the attainment of a con-
stant value of i\','tat short times, and the capacity begins
B
This dispersion e ect is mathematically similar to diffusion
and can be characterized by a dispersion coefficient D,.
to saturate at long times. Tiedemann and Newman (1974) Sherwood et al. ( 1975) have recently correlated available
used the same technique to study the PbO2 and Pb elec- data on dispersion in packed beds and express their results
trodes in H2S04 solution. However, a plot of i d F v s . d? in Figure 5 by plotting a PCclet number proportional to
yielded a maximum in i f i , and this value was used to +
u / a ( D D,) against a Reynolds number proportional to
obtain a value for the double-layer capacity. Figure 4 d a u , with the Schmidt number' v / D as a parameter. (For
shows the comparison between theory and experiment for spheres, the effective particle diameter dp on the graph is
the charging curve of PbOz in solution. Emphasis related to the specific interfacial area by G$ = 6 ( 1 - c ) /
on the maximum of this charging curve minimizes the a , ) Dispersion in the cross-flow, radial direction is also
ohmic effects at short times as well as the faradaic effects included. At low flow rates, axial mixing becomes less im-
at long times. As faradaic reactions become more impor-
tant, a positive displacement of the experimental curve
from the theoretical curve is observed at long times. How- 10001 I I I / I I I/ / I
ever, the shape of the curve near the maximum will remain
relatively unchanged.
I00
Johnson and Newman also demonstrated that specific
adsorption of ions on a porous high-surface-area carbon
electrode could be used to bring about a separation or
concentration of soluble species. By alternating the applied a"
potential at specified time intervals, a NaCl solution could L
a loI \
n
be desalted during the adsorption cycle and concentrated
during the desorption cycle. This technique constitutes a f
z
novel means of separation.

FLOW-THROUGH ELECTROCHEMICAL REACTORS

Flow-through porous electrodes possess inherent ad- @A"
vantages over nonporous electrodes with flowing solutions Re =
or porous electrodes without flow. This has been well Fig. 5. Correlation of dispersion coefficients in packed beds, E =D
demonstrated by Kalnoki-Kis and Brodd (1973), who show + Da. The asymptotes for molecular diffusion are based on e =
increases in current by factors of lo2 to lo4 at a given 0.4 and a tortuosity factor of 1.4. is the superficial velocity v,
overpotential. High specific interfacial area allows the at- and for spheres d , = 6(1 - €)/a. (Sherwood e t a/., 1975).
tainment of relatively high volumetric rates of reaction, and
the flowing solution eliminates or reduces mass-transport 0 The Schmidt number being based on the true molecular diftusion

problems. Thick electrodes and low flow rates produce coefficient without any tortuosity factor.

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 35

portant than molecular diffusion, but the Reynolds number concentration at the walls of the pores is close to zero. This
at which this occurs decreases as the Schmidt number be- also brings about considerable simplification in the design
comes larger. calculations, since it decouples the potential distribution
5. What is the approach to limiting current and the from the mass transfer rates and current distribution.
degree of conversion?
Quontitative Design Principles
The extent to which any or all of the above are incorpo-
rated into the analysis is dependent on the specific system 1. We want to avoid side reactions. These could range
and given design and optimization requirements. Equations from unwanted evolution of hydrogen or oxygen to com-
( 8 ) , (lo), (19), (20), (23), and (33) facilitate the for- peting reactions or subsequent reactions in electro-organic
mulation of the problem, but to effect a solution may re- synthesis. Consequently, the potential variation in the solu-
quire various assumptions. tion phase should not exceed a certain limit, one or two
Austin et al. (1965) and Gurevich and Bagotski? tenths of a volt in moderately sensitive cases. For high de-
(1963c, 1964) were the first to treat, in detail, the polar- grees of completion, the design equation is
ization behavior of flow-through porous electrodes. As-
sumptions employed were a homogeneous porous electrode
in steady operation with uniform flow, a one-dimensional
treatment, first-order electrode kinetics with no side reac- which can be regarded as the product of a superficial cur-
tions, negligible ohmic losses in the solid matrix and a con- rent density (nE'oco) with the reciprocal of the effective
stant solution conductivity, negligible dispersion, and neg- conductivity ( U K )and an effective depth of penetration
ligible mass-transfer resistance. Solution was passed of the reaction into the electrode ( v / a k , ) . The mass
through from the back side, away from the counter elec- transfer coefficient k, itself depends on the velocity so
trode. that /4@Irnax is proportional to the velocity to about the
Johnson and Newman (1971) examined the adsorption 1.5 power.
and desorption of ions in a flow-through porous electrode In practice, we would use this equation to determine the
as a means of desalting water. Their analysis included maximum permissible velocity or throughput which can
ohmic losses in the carbon matrix and transient operation. be achieved while avoiding side reactions and yet operat-
Schmidt and Gidaspow (1973) have presented a similar ing near limiting current. Note how the difficulty of the
treatment. problem (or separation process) is governed by the ratio
Bennion and Newman (1972) emphasized the role of C ~ / Kof the reactant concentration to the conductivity and
the mass transfer coefficient in their concentration of dilute by the specific interfacial area u of the electrode. (After
solutions containing copper. Alkire and Ng (1974) ana- allowance for the dependence of k, on a, Ih@lrnax is pro-
lyzed the two-dimensional current distribution resulting portional to a to about the -1.5 power.)
from placement of the counter electrode parallel to the 2. The thickness of the electrode is now determined by
direction of flow within the working electrode. Other oper- the required degree of reaction. This is given by
ational aspects of flow-through systems have been consid-
ered by Gurevich et al. (1968a, b ) , Budeka et al. (1968),
Sohm (1962), Guillou and Buvet (1963), Sioda and
Kemula ( 1972), and Wroblowa ( 1973).
L=-
V

ak,
co
In -
CbL
- d/sSC'/6

U1.5
In -
co
cbL

Experimental investigations of single straight pore and

micromesh electrodes have been conducted by Blaedel
et al. (1963), Blaedel and Strohl (1964), Blaedel and
Klatt ( 1966), Klatt and Blaedel ( 1967, 1968), and Blaedel Costs of course depend on how co, c b L , K , and a affect
and Boyer (1971, 1973). These studies are helpful in de- 0 and L. Comparisons are given in Table 1 for the lead
termining the correspondence between experiment and and copper systems. As the feed concentration is increased
theory (since tortuosity and surface effects are small), as by a factor of 415, the cost per unit volume goes up by
shown by Sioda ( 1974b). Further experimental investiga- a factor of 90 while the cost per unit mass goes down by
tions by Sioda (1968a, b, 1970) and Wroblowa and Saun- a factor of 4.6. The pumping power makes a dramatic in-
ders (1973) have helped to develop mechanistic criteria
to treat these systems. Alkire and Mirarefi (1973) have TABLE1. OPERATINGCONDITIONS,
DESIGNRESULTS,
AND
analyzed the current distribution in a tubular electrode be- COSTSFOR REMOVALOF LEADAND COPPERIONS
FROM
low the limiting current. GIVENSOLUTIONS
Carlson and Estep (1972) have studied removal of
metal ions from aqueous waste streams, Beck and Giannini Pb system Cn system
(1972) have considered effluent treatment, Roe (1964)
has looked at electrochemical fractionation, and copper re- Feed concentration, muliter 1.45 667
moval has been effected by Bennion and Newman ( 1972), Product concentration, mg/liter 0.05 1
Wenger et al. ( 1973), and Posey ( 1973). Solution conductivity, mho/cm 0.8 0.17
Design and optimization is specific to the given system Allowable potential variation, V 0.1 0.2
Specific electrode area, cm-1 25 25
and have been discussed by Bennion and Newman (1972) Superficial velocity, cm/s 0.64 0.0036
and by Gurevich et al. (1968b). To illustrate the design Electrode thickness, cm 60 5
principles outlined by Bennion and Newman, we examine Time to plug, days 377 5
the removal of Cu + + and Pb + + .
Pb system Cu system
Qualitative Design Principles d/kgal $/lb. d/kgal $/lb.
1. We want to promote intimate contact between the
electrode and the solution in order to carry out reactions Electrode capital cost 0.2222 0.1836 26.5 0.0529
to a high degree of completion. This calls for small flow Operating labor 0.0218 0.0180 4.2 0.0083
channels and hence for a flow-through porous electrode. Electrical energy 0.0021 0.0018 3.3 0.0066
2. Carrying reactions to a high degree of completion Pumping energy 0.132 0.109 0.014 0.00003
- - _ _ -
also calls for operation at the limiting current so that the Total 0.38 0.31 34 0.068

Page 36 January, 1975 AlChE Journal (Vol. 21, No. 1)

crease in importance for dilute solutions. ( 1958). However, Simonsson ( 1973a) showed that this
A variety of processes of potential interest for applica- discrepancy was the result of their wash-dry procedure.
tion of the concept of flow-through, porous-electrode elec- Haebler et al. ( 1970) and Szpak et al. (1966) removed
trochemical reactors is as follows: mass transport limitations by flowing the solution through
1. Removal of electropositive cations by electrodeposi- the electrode. Such effects had been examined already by
tion. Liebenow ( 1897). Recently, flow-through electrodes have
emerged as a potentially useful industrial process. Bennion
a. Cu 600 pprn to < 1ppm. and Newman ( 1972) thus concentrated copper solutions.
b. Pb 1.45 to 0.05 ppm from HzS04 solutions. Carlson and Estep (1972) used flow-through porous elec-
c. Hg 10 to 0.01 ppm from solution 280 g/l NaCl. trodes to remove trace metal impurities from waste streams.
d. Ag processing of waste photographic emulsions. Austin et al. (1965) and Wroblowa and Saunders (1973)
e. Au from waste plating solutions. have examined the reactions of organic and inorganic mate-
rials. Sioda (1970, 1971, 1972a, b, c, 1974a) has examined
2. Fe + + to Fe + + . Acid mine drainage waters.
+

3. Oxidation of organic pollutants and of CN- (24 to flow-through electrodes with an emphasis on electroanalyti-
0.1 pprn). cal applications.
4. Oxidation of organic surfactants. Reduction of foam- Only a small number of experimental porous electrode
ing in Solvay process for manufacture of NaZC03. studies have yielded kinetic information (Posey and Misra,
1966; Austin and Lerner, 1964), the potential-time dis-
5. Electro-organic synthesis where the need for careful
control of potential may have ruined attempts at scale up. charge behavior (Simonsson, 1973b, c ) , or the electro-
Many of these-both oxidations and reductions-are re- chemically active surface area (Euler, 1961). Information
lated to water pollution abatement. Others may have in- in these areas is needed to refine the models which are be-
dependent economic incentives. ing developed.
Some of these processes raise questions about new as-
pects of the system design such as
ESTIMATION OF PHYSICAL PROPERTIES
1. How to design a counter electrode to carry out a gas-
evolution reaction. The porosity of an electrode plays a major role in deter-
2. How to separate anode and cathode processes. mining its polarization behavior. The void volume con-
3. How to regenerate the electrodes in the case of lead tained within the electrode provides not only intimate con-
deposition. tact between solution and solid phases, as well as a reser-
4. How to carry out both reduction and oxidation on a voir of soluble species available for consumption, but also
waste stream. the necessary space for storage of solid products of reac-
5. How to balance oxidation needed in one process with tion. The effective solution and solid phase conductivities
reduction needed in a separate process. and the solution diffusion coefficient are found to be func-
tions of the porosity. The correspondence between theory
and experiment is strongly dependent on the values se-
EXPERIMENTAL INVESTIGATIONS lected for the effective conductivities and diffusion coeffi-
cients as well as their functional dependence on porosity.
Current distribution within an electrode has been deter- Therefore, it is important that these parameters be either
mined by measuring the distribution of discharged mate- measured experimentally or predicted from appropriate
rial through the interior (see, for example, Brodd, 1966; equations.
Bro and Kang, 1971; Nagy and Bockris, 1972; Alkire, Meredith and Tobias ( 1962) have reviewed extensively
1968; Euler and Horn, 1965; Bode and Euler, 1966a, b; conduction in heterogeneous media. Equations were de-
Bode et al., 1966). This determination involves a wash-dry rived which predict the effective conductivity of solutions
operation followed by electrode sectioning, and this pro- in multiphase dispersions as a function of the volume frac-
cedure can result in a net reduction and a redistribution of tion occupied by the solution. The utility of these equations
material within the electrode. Sectioning the electrode also lies in their ability to account for changes in the effective
results in averaging the material in question over a non- solution conductivity of porous electrodes uhdergoing struc-
zero thickness and may result in an inaccurate picture of tflral changes during testing. Their results %uggest that the
the real current distribution. effective conductivity can be found by multiplying the
Current distribution can also be inferred by measuring bulk conductivity times the porosity raised to the three-
the potential distribution with reference electrodes located halves power. Gagnon (1973) showed that this relation-
in the electrode (Daniel’-Bek, 1948; Brodd, 1966) or by ood agreement with experimental results. Scho-
using a sectioned electrode and directly measuring the cur-
rent in each element (Coleman, 1951). However, sectioned
ship infi
field is
and akshinamurti (1948) have shown that the same
porosity factor can be used for the diffusion coefficient.
electrodes are usually inferior to sectioning because of the (The effective diffusion coefficient considered here is cDi
small distance involved, an exception being the MnOz elec- in terms of the diffusion coefficienyt introduced in Equa-
trode (Coleman, 1951). Studies on simulated single pores tion (20) .) Ksenzhek, Kalinovskii, and Baskin (1964)
(Will and Hess, 1973) have helped to examine micro- measured the conductivity of 1.ON sodium hydroxide solu-
structure problems. Gagnon and Austin (1971) formed a tion in porous nickel and carbon electrodes. Their results
composite Ag-AgzO electrode from five thin sections and suggest that the effective conductivity of the contained
determined the current distribution by disassembling and solution is proportional to 2 rather than as suggested
completing the discharge in each section. Panov et al. by Meredith and Tobias.
(1969) used a composite electrode of 15 sections with Direct measurement of the effective conductivity of a
provision for varying both solid and solution phase resist- solution in an inert porous material can be readily accom-
ance. The results are in good agreement with the calcula- plished and offers a check on the correlations mentioned
tions of Daniel’-Bek (1966) and of Newman and Tobias above. However, since porous electrodes have an electroni-
(1962). cally conducting solid phase, the passage of alternating or
Most investigations have tended to support the predic- direct current through the electrode (with the electrode
tions of porous-electrode theory. Haebler et al. (1970) acting as an inert separator) will be affected by the amount
reported results at odds with the predictions of Stein of current which passes through the solid phase by way of

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 37

double-layer charging or faradaic reactions. Posey and f represents current density i,, in the polarization
Misra (1966) have analyzed bipolar porous electrodes to equation, A/cm2
show how to correct the measured conductivity for the F F'araday s constant, 96487 C/equiv
presence of faradaic ieactions. Experimental studies by i0 exchange current density, A/cm2
Romanova and Selitskii (1970) and by Simonsson (1973a) il superficial current density in the matrix, A/cm2
have shown the importance of gas bubbles on the measure- 12 superficial current density in pore phase, A/cm2
ment. Oldham and Topol (1967, 1969) have developed in transfer current per unit of interfacial area, A/cm2
equations for potentiostatic and galvanostatic diffusion I superficial current density to an electrode, A/cm2
studies on porous media and have obtained experimental i V . i2, transfer current per unit volume of elec-
verification (Topol and Oldham, 1969) of their predic- trode, A/cm3
tions. jin pore-wall flux of species i, mole/cm2-s
The value of the true electrochemically active interfacial kl reciprocal penetration depth for diffusion, see
area is difficult to determine. A BET measurement reflects Equation (46), cm-I
all the area present, including matrix, binders, and non- km coeflicient of mass transfer between flowing solu-
conducting fuel. Tiedemann and Newman (1974) and tion and electrode surface, cm/s
Gagnon ( 1973) have developed experimental techniques, L thickness of porous electrode, cm
involving double-layer charging, to assess the electrochem- Mi symbol for the chemical formula of species i
ically active area. Katan and Bauman (1974) have shown Mi molecular weight of species i, g/mole
how various geometric variables of a porous electrode can n number of electrons transferred in electrode reac-
be used to estimate the specific interfacial area. tion
n normal unit vector
CONCLUSIONS Ni superficial flux of species i, mole/cm2-s
4 surface charge density on solid side of double
The theory of the operation and behavior of porous elec- layer, C/cm2
trodes has been developed to the point where almost any R universal gas constant, 8.3143 J/mole-deg
system can be treated. In particular, equations were devel- Si stoichiometric coefficient of species i in electrode
oped to provide the bases for examining the behavior of reaction
specific systems such as primary and secondary batteries, sc Schmidt number, u / D
adsorption and double-layer charging, and flow-through t time, s
electrochemical reactors. ti0 transference number of species i with respect to
Experiments have verified the predictions of nonuniform the solvent velocity
current distribution in several types of porous electrodes, T absolute temperature, OK
but experimental information on the transient behavior of ui mobility of species i, cm2-mole/J-s
porous electrodes is scarce. The polarization expression ap- U' @l - @2, v
plicable to a given system is subject to refinement to ac- V superficial fluid velocity, cm/s
count for the morphology of the electrode, the formation .V superficial volume average velocity, cm/s
of covering layers or of crystallites of sparingly soluble V* molar average velocity, cm/s
species, and the transport from such sparingly soluble V @ l ( L ) - @2(0),volt
phases to the site of the charge-transfer process. partial molar volume of species i, cm3/mole
Areas which remain to be examined in greater detail are: X distance through porous electrode, cm
( 1 ) complete analysis of an entire electrochemical cell; ( 2 ) zi valence or charge number of species i
prediction of the entire polarization curve for a given
battery; ( 3 ) dynamic behavior of flow-through electro- Greek Letters
chemical reactors; (4) complete optimization of a given
electrochemical system. aa = transfer coefficient in anodic direction
a, = transfer coefficient in cathodic direction
~9 = eaF/RTor -acr,F/RT
ACKNOWLEDGMENT yi = dimensionless diffusion parameter
This work was supported by the United States Atomic En- ri = surface concentration of species i, mole/cm2
ergy Commission and Globe-Union, Inc. 6 = d'mensionless current density defined in Equation
(37)
E = porosity or void volume fraction
NOTATION qs = surface overpotential, V
a = specific interfacial area, cm-1 K = effective conductivity of solution, mho/cm
= constant defined by Equation (12) K' = modified solution conductivity, mho/cm
A,,
c = concentration of binary electrolyte, mole/cm3 pe = chemical potential of electrolyte, J/mole
= concentration of species i per unit volume of solu- Y = kinematic viscosity of flowing solution, cm2/s
ci Y = square root of a dimensionless exchange current;
tion, mole/cm3
= reference or bulk concentration, mole/cm3 see Equation (38)
ciO = numbers of moles of cations and anions produced
co = solvent concentration, mole/cm3 Y + , Y-

co = feed concentration to flow-through electrode, by the dissociation of a mole of electrolyte

mole/cm3 pi = density of a solid phase of species i, g/cm3
cbL = exit concentration for flow-through electrode, u = effective conductivity of the solid matrix, mho/cm
mole/cm3 @l = electric potential in the matrix, V
cT = total solution concentration, mole/cm3 @2 = electric potential in the solution, V
C = differential double-layer capacity, f/cm2
dp = effective particle diameter of bed packing, cm Subscripts
D = diffusion coefficient of electrolyte, cm2/s 0 = solvent
Da = dispersion coefficient, cm2/s e = electrolyte
Di = diffusion coefficient of species i, cm2/s + = cation
E = D i- D ,cm2/s - = anion
Page 38 January, 1975 AlChE Journal (Vol. 21, No. 1)
LITERATURE CITED Battery Electrodes,” Electrochim. Acta, 11,1107 $1966).
Budeka, Yu. F., I. G. Gurevich, and V. S. Ba otskii, “0 vliyanii
Alkire, Richard C., “A Theoretical Study of Bipolar Porous parametrov vkhoda na rabotu zhidkostnyfh poristykh elek-
Electrodes,” J. Electrochem. SOC.,120,900 ( 1973). trodov v rezhime prinuditel’noi podachi v nikh reagenta,”
-., “Reaction Distribution in a Dissolving Porous Anode,” Elektrokhimiya, 4,874 ( 1968).
Dissertation, Univ. California, Berkeley ( 1968). Burbank, Jeanne, Albert C. Simon, and Eugene Willihnganz,
-., Edward A. Grens 11, and Charles W. Tobias, “Prepara- “The Lead-Acid Cell,” Charles W. Tobias (ed.), Aduan.
tion of Uniform Pore Structures for Porous Electrode Electrochem. Ebctrochem. Eng., 8, 157 ( 1971).
Studies,” J. Electrochem. SOC., 116, 809 ( 1969). Carlson, G. A., and E. E. Estep, “Porous Cathode Cell for
., “A Theory for Porous Electrodes Undergoing Metals Removal from Aqueous Solution,” J. Electrochem.
Structural Change by Anodic Dissolution,” ibid., 116, 1328 SOC.,119,114C (1972).
(1969). Chizmadzhev, Yu. A,, V. S. Markin, M. R. Tarasevich, and
Alkire, Richard, and Ali Asghar Mirarefi, ”The Current Dis- Yu. G. Chirkov, Makrokinetika protsessov poristykh
tribution Within Tubular Electrodes under Laminar Flow,” sredakh ( Toplivnye ekmenty ), Izdatel’stvo “Nauka,” Mos-
ibid., 120, 1507 (1973). cow ( 1971).
Alkire, Richard, and Patrick K. Ng, “Two-Dimensional Current Coleman, J. J., “Dry Cell Dynamics: the Bobbin,” Trans. Ebc-
Distribution Withill a Packed-Bed Electrochemical Flow Re- trochem. SOC.,90,545 ( 1946 ).
bctor,” ibid., 121,95 ( 1974 ). ., “Distribution of Current in Porous Electrodes,” I.
Alkire, Richard, and Barry Place, “Transient Behavior During Electrochem. Soc., 98,26 ( 1951 ).
Reactant Depletion in Porous Electrodes,” ibid., 119, 1687 Daniel’-Bek, V. S., “K voprosu o polyarizatsii poristykh elektro-
(1972). dov. I. 0 paspredelenii toka i potentsiala vnutri elektroda,”
Appel, Peter, “Electrochemical Systems-Impedance of a Ro- Zh. Fiz. Khim., 22,697 ( 1948).
tating Disk and Mass Transfer in Packed Beds,” Ph.D. -., “K voprosu o polyariza$~ii poristykh elektrodov. IV.
thesis, LBL-2280, Univ. Calif., Berkeley, in prep. ( 1974). Vliyanie soprotivleniya tverdoi fazy na raspredelenie PO-
Austin, L. G., “Tafel Slopes for Flooded Diffusion Electrodes,” tentsiala i toka v elektrode,” Elektrokhimiya, 2, 672 (1966).
Trans. Faraday SOC.,60, 1319 ( 1964). Darby, R., “Faradaic Impedance of Polarized Porous Elec-
-., “Note on ‘The Influence of Electrode Reaction Kinet- trodes I. First Order Kinetics,” J. Electrochem. SOC., 113, 392
ics on the Polarization of Flooded Porous Electrodes,’ by (1966).
E. A. Grens I1 and C. W. Tobias,” Electrochim. Acta, 14, ., “Faradaic Impedance of Polarized Porous Electrodes
639 (1969). 11. Nonlinear Kinetics,” ibid., 113,496 (1966).
., and E. G. Gagnon, “The Triangular Voltage Sweep de Levie, Robert, “On Porous Electrodes in Electrolyte Soh-
Method for Determining Double-Layer Capacity of Porous tions,” Electrochim. Acta, 8, 751 ( 1963).
Electrodes, Part I. Theory,” J. Electrochem. SOC.,120, 251 -., “Electrochemical Response of Porous and Rough
(1973). Electrodes,” Paul Delahay ( ed. ), Adoan. Electrocha. Elec-
Austin, L. G., and H. Lerner, “The Mode of Operation of trochem. Eng., 6,329 ( 1967).
Porous Diffusion Electrodes-I. Simple Redox Systems,” Dunning, John Stephen, Analysis of Porous Electrodes with
Electrochim. Acta, 9, 1469 ( 1964). Sparingly Soluble Reactants, dissertation, Univ. of California,
Austin, L. G., Pierre Palasi, and Richard R. Klimpel, “Polariza- Los Angela ( 1971 ) .
tion at Porous Flow-Through Electrodes,” Fuel Cell Systems, ., and D. N. Bennion, “A Model for Secondary Battery
pp. 35-60, Adv. Chem. Series, No. 47, Am. Chem. SOC., Electrodes,” Proc. Advances in Buttery Technology Symp.,
Washington, D.C. (1965). Southern Calif.-Neveda Section of the Electrochem. SOC.,
Beck, E. C., and A. P. Giannini, “Basic Electr‘c Cell in Ef- Vol. 5, 135 (1969).
fluent Treatment,” J. Electrochem. SOC., 119, 114C (1972). ., and John Newman, “Analysis of Porous Elec-
Bennion, Douglas N., and John Newman, “Electrochemical trodes with Sparingly Soluble Reactants,” J. Electrochem.
removal of copper ions from very dilute solutions,” J. Appl. Soc., 118, 1251 (1971).
Electrochem., 2, 113 (1972). ., “Analysis of Porous Electrodes with Sparingly
Bird, Judith, Josk Giner, and Michael Turchan, “Determination Soluble Reactants 11. Variable Solution Properties, Convec-
of Surface Area of Battery Electrodes by Potentiostatic Dou- tion, and Complexing,” ibid., 120, 906 ( 1973).
ble Layer Capacity Measurement,” U.S. Army Materiel Com- Euler, J., “Porendimensionen und Oberflachenkapazitat von
mand, Harry Diamond Lab., Washington, D.C. ( 1969 ). Braunstein,” Electrochim. Acta, 4, 27 ( 1961 ) ,
Bird, R. Byron, Warren E. Stewart, and Edwin N. Lightfoot, ., and L. Horn, “Stromverteilun in Akkumulatorenplat-
Transport Phenomena, Wiley, New York ( 1960). ten aus rohrformigen Elementen,” ibi%.,10, 1057 ( 1965).
Blaedel, W. J., and S. L. Boyer, “Submicromolar Concentration Euler, J., and W. Nonnenmacher, “Stromverteilung in porosen
Measurements with Tubular Electrodes,” Anal. Chem., 43, Elektroden,” ibid., 2, 268 ( 1960).
1538 (1971). Fleischmann, M., J. W. Oldfield, and L. Tennakoon, “Fluidized
“Electrochemical Characteristics of the Gold bed electrodes Part IV. E1ectrodepositi:n of co per in a
Micromesh Electrode,” ibid., 45,258 ( 1973).
Blaedel, W. J., and L. N. Klatt, “Reversible Charge Transfer
P
fluidized bed of copper-coated spheres, J. A p p . Eledro-
chem.,‘l,103 ( 1971).
at the Tubular Platinum Electrode,” ibid., 38, 879 ( 1966). Frumkin, A. N., “0 raspredelenii korrozionnogo protsessa PO
Blaedel, W. J., C. L. Olson, and L. R. Sharma, “The Tubular dline trubki,” Zh. Fiz. Khim., 23, 1477 ( 1949 ) .
Platinum Electrode,” ibid., 35, 2100 ( 1963). Gagnon, E. G., “The Triangular Voltage Sweep Method for
Blaedel, W. J., and J. H. Strohl, “Continuous Quantitative Determining Double-Layer Capacity of Porous Electrodes 11.
Electrolysis,” ibid., 36, 1245 ( 1964). Porous Silver in Potassium Hydroxide,” J. Electrochem. SOC.,
Bockris, J. OM., and S . Srinivasan, Fuel Celk: Their Electro- 120,1052 (1973).
chemistry, McGraw-Hill, New York ( 1969). ., “The Triangular Voltage Sweep Method for Determin-
Bode, H., and J. Euler, “Autoradiographische Untersuchung der ing Double-Layer Capacity of Porous Electrodes 111. Porous
Stromverteilung auf Platten von Bleiakkumulatoren-I,” Nickel in Potassium Hydroxide,” ibid., 121,512 ( 1974).
Electrochim. Acta, 11, 1211 (1966). ., and L. G. Austin, “Low-Temperature Cathode Per-
., “Autoradiographische Untersuchung der Strom- formance of Porous Ag/AgO Electrodes,” ibid., 118, 497
verteilung auf Platten von Bleiakkumulatoren-11,” ibid., 11, (1971).
1221 ( 1966). Gidaspow, Dimitri, and Bernard S. Baker, “A Model for Dis-
., E. Rieder, and H. Schmitt, “Autoradiographische charge of Storage Batteries,” ibid., 120, 1005 (1973).
Untersuchung der Stromverteilung auf Platten von Bleiak- Gracon, Brian Eugene, “Experimental Studies of Flow-through
kumulatoren-111,” ibid., 11, 1231 ( 1966). Porous Electrodes,” M.S. thesis, Univ. of Illinois, Urbana
Bomben, John L., Current Distribution in a Porous Electrode. (1974).
M.S. thesis, Univ. of California, Berkeley ( 1963). Grens, E. A., 11, “On the Assumptions Underlying Theoretical
Bro, P., and H. Y. Kang, “Discharge Profiles in a Porous Cad- Models for Flooded Porous Electrodes,” Electrochim. Acts,
mium Electrode,” J. Electrochem. SOC., 118, 519 ( 1971). 15, 1047 ( 1970).
Brodd, R. J., “Experimental Current Distribution in Porous ., and Charles W. Tobias, “Analysis of the Dynamic

AlChE Journal Wol. 21, No. 1) Jonuary, 1975 Page 39

 Behavior of Flooded Porous Electrodes,” Ber. Buncienges. American Chemical Society, Washington, D.C. ( 1965).
Phys. Chem., 68,236 ( 1964 ) . Micka, K., “Zur Theorie der porosen Elektroden V. Fall eines
., “The Influence of Electrode Reaction Kinetics binaren Elektrolyten,” Collect. Czech. Chem. Commun., 31,
on the Polarization of Flooded Porous Electrodes,” Electro- 3653 ( 1966).
chim. A&, 10,761 ( 1965). ., and Ivo RouEar, “‘Theory of Porous Electrodes-XI.
Guillou, M., and R. Buvet, “Repartition du potentiel dans les The Positive Plate of the Lead-Acid Battery,” Electrochem.
electrodes de puissance a circulation &electrolyte porteur- Acta, 18,629 ( 1973).
11. Theorie generale et etude d’un fritte multitubulaire,” Milner, Paul C., and Upton B. Thomas, “The Nickel-Cadium
ibid., 8,489 ( 1963). Cell,” Charles W. Tobias (ed.), Adcan. Electrochem. Ekc-
Gurevich, I. G., and V. S. Bagotskic “Rabota poristykh elektro- trochem. Eng., 5, 1 ( 1967).
dov v diffuzionnom rezhime podachi reagentov,” Inzh. Fiz. Nagy, Z., and J. O’M. Bockris, “On the Electrochemistry
Zh., 6(2 ) , 60 (1963). of Porous Zinc Electrodes in Alkaline Solutions,” J. Electro-
., “Effektivnost’ ispol’zovaniya poristykh elek- chem. SOC.,119,1129 (1972).
trodov, rabotayushchikh v diffuzionnom rezhime podachi Newman, John, Transient Phenomena in the Discharge of
reagenta,” ibid., 6 ( 3 ) , 6 9 ( 1963 ) . Porous Electrodes, Internal tech. report, NOLC project, Dept
., “Rabota zhidkostnvkh poristykh elektrodov v of Chem. Eng., Univ. California, Berkeley ( 1962).
rezhime s vynuzhdennoi podachei reagenta,” ibid., 6( 5 ) , 75 ., ‘Transport Processes in Electrolytic Solutions,”
(1963). Charles W. Tobias (ed. ), Aduan. Electrochem. Electrochem.
., “Porous Electrodes with Liquid Reactants un- Eng., 5,87 (1967).
der Steady-state Operating Conditions,” Electrochim. Acta, -., “Numerical Solution of Coupled, Ordinary Differ-
9, 1151 (1964). ential Equations,” Ind. Eng. Chem. Fundamentals, 7, 514
., “K voprosu optimizatskii eometrii ( tolshchiny (1968).
zhidkosthykh poristykh elektrodov,” Ele%trokhimiya, 5, 1297 Newman, John S., Electrochemical Systems, Prentice-Hall,
(1969). Englewood Cliffs, N.J. (1973).
., and Yu. F. Budekp .“K optimizatsii para- Newman, John, Douglas Bennion, and Charles W. Tobias,
metrov vkhoda elektrokhimicheskoi sistemy s zhidkostnymi “Mass Transfer in Concentrated Binary Electrolytes,” Ber.
poristymi elektrodami, rabotaynshchimi* v rezhime prinu- Bunsenges. Phys. Chem., 69, 608-612 (1965). [for correc-
ditel’no: podachi reagenta pri obratimoi reaktsii,” ibid., 4, tions, see ibid., 70, 493 ( 1966).I
321 ( 1968). Newman, John, and Thomas W. Chapman, “Restricted Diffu-
Gurevich, I. G., Yu. F. Budeka, and V. S. Bagotskic “Zavisi- sion in Binary Solutions,” AIChE J., 19,343 ( 1973).
most’ ,raboty zhidkostnykh oristykh elekbodov s prinudi- Newman, John S., and Charles W. Tobias, “Theoretical Anal-
tel’noi podachei reagenta ot ioeffitsienta ego ispol’zovaniya,” ysis of Current Distribution in Porous Electrodes,” J . Elec-
ibid., 4, 1251 ( 1968). trochem. SOC., 109,1183 (1962).
Gurevich, I. C., Yu. M. Vol’fkovich, and V. S. Bagotskii, Zhid- Niqancioglu, Kemal Mustafa, “Current Distribution and Mass
kostnye Poristye Elektrody, Nauka i Tekhnika, Minsk ( 1974 ) . Transfer in Rotating Electrode Systems,” Dissertation, ( LBL-
Haebler, H., H. Panesar, and E. Voss, “Elektrolytstromung und 1880), Univ. California, Berkeley ( 1973).
Masseausnutzung von Porosen Bleidioxidelektroden,” Elec- Oldham, Keith B., and Leo E. Topol, “Electrochemical Inves-
trochim. Acta, 15,1421 (1970). ti ation of Porous Media I. Theory of Potentiostatic Meth-
Hoar,T. P., “The Anodic Behaviour of Metals,” Mod. Aspects ofs,’’ J. Phys. Chem., 71,3007 ( 1967).
Electrochem., 2,262 ( 1959). ., “Electrochemical Investigation of Porous Media
Johnson, A. M., and John Newman, “Desalting by Means of 111. Theory of Galvanostatic Methods,” ibid., 73, 1462
Porous Carbon Electrodes,” J . Electrochem. SOC., 118, 510 (1969).
( 1971 ). Panov, V. A., V. S. Borovkov, and P. D. Lukovtsev, “Eksperi-
Kalnoki-Kis, T., and R. J. Brodd, “Characterization of Porous mental’noe izuchenie raspredeleniya toka v poristom elek-
Flow Through Electrodes,” ibid., 120, 109C ( 1973). trode,” Elektrokhimiya, 5, 180 ( 1969).
Katan, T., and H. F. Bauman, “Relating Geometrical Variables Parsons, Roger, “Faradaic and Nonfaradaic Processes,” Adcan.
of Porous Electrodes,” ibid., 121, 124C ( 1974). Electrochem. Electrochem. Eng., 7, 177 ( 1970).
Klatt, L. N., and W. J. Blaedel, “Quasi-Reversible and Irre- Posey, F. A., “Methods for the Calculation of Polarization in
versible Charge Transfer at the Tubular Electrode,” A d . Porous Electrodes,” J. Electrochem. SOC., 111, 1173 ( 1964).
Chem., 39, 1065 ( 1967). Posey, F. A,, Electrolytic Demonstration Unit for Copper Re-
“‘Catalytic Reactions at Tubular Electrodes,” moval From Distillation Plant Blowdown, Oak Ridge Na-
ibid., 40,Sli’ (1968). tional Lab., Oak Ridge, Tenn. ( 1973).
Ksenzhek, 0. S., “DiffuzionnG rekim paboty poristykh elektro- Posey, F. A,, and S. S. Misra, “Induced Polarization of Porous
dov,” Zh. Fiz. Khim., 36, 243 ( 1962). and Tubular Electrodes,” J. Electrochem. SOC., 113, 608
., “Polyarizatsiya tonkikh poristykh elektrodov,” ibid., ( 1966).
36,633 ( 1962). Posey, F. A., and T. Morozumi, “Theory of Potentiostatic and
-., ‘”Perekhodnye protsessy pri zaryazhenii poristykh Galvanostatic Charging of the Double Layer in Porous Elec-
elektrodov,” ibid., 37, 2007 ( 1963 ) . trodes,” ibid., 113, 176 (1966).
-., “Perekhodnye protsessy pri zaryazhenii poristykh Roe, D. K., “Electrochemical Fractionation: Potentiostatic
elektrodov 11. Vliyanie gneshnei poverkhnosti elektroda; Chromatography and Elution Voltammetry,” Anal. Chem.,
vliyanie kontsentratsionnoi polyarizatsii,” ibid., 38, 1846 36,2371 (1964).
(1964). Romanova, I. L., and I. A. Selitski?, “Izuchenie poristogo
-., E. A. Kalinovski: and E. L. Baskin, “Provodimost’ elektroda svintsovogo akkumulyafora metodom izmereniya
elektrolita v poristykh nikelevykh elektrodakh,” Zh. Prik. effektivnogo soprotivleniya zhidkoi fazy,” Elektrokhimiya, 6,
Khim., 37,1045 (1964). 1776 ( 1970).
Ksenzhek, 0. S., and V. V. Stender, “Opredelenie udel’noy Schmidt, T., and D. Gidaspow, “Electrosorption with Porous
Electrodes,” Abstract No. 250, Chicago meeting, J. Electro-
poverkhnosti poristykh elektrodov metodami izmereniya em- chsm. SOC., 120, llOC ( 1973).
kosti,” Dokl. Akad. Nauk S S S R , 106,487 (1956). Schofield, R. K., and C. Dakshinamurti, “Ionic Diffusion and
“Raspredelenie toka v poristom elektrode,” Electrical Conductivity in Sands and Clays,” Discuss. Fara-
Dokl. Akad.’kauk S S S R , 107,280 ( 1956). day SOC., no. 3,56 ( 1948).
Liebenow, “Wber die Berechnung der Kapazitat eines Sherwood, Thomas K., Robert L. Pigford, and Charles R.
Bleiakkumulators bei variabeler Stromstarke,” Z . Elektro- Wilke, Mass Transfer, McGraw-Hill, New York (1975).
chem.,?, 58 (particularly p. 63) ( 1897 ). Simonsson, Daniel, “Current Distribution in the Porous Lead
Meredith, Robert E., and Charles W. Tobias, “Conduction in Dioxide Electrode,” J . Electrochem. SOC., 120, 151 ( 1973).
Heterogeneous Systems,” Charles W. Tobias (ed. ), Adcan. ., “A mathematical model for the porous lead dioxide
Electrochem. Electrochem. Eng., 2, 15 ( 1962 ). electrode,” J. Appl. Electrochem., 3, 261 ( 1973).
Micka, Karel, “Theory of Polarization of Porous Electrodes,” Simonsson, Daniel, “The Discharge Process of the Porous Lead
Fuel Cell Systems, pp. 73-82, Adv. Chem. Series, No. 47, Dioxide Electrode,” Dissertation, Inst. kemisk teknologi,

Page 40 January, 1975 AlChE Journal (Vol. 21, No. 1)

 Kungl. Tekniska Hogskolan, Stockholm ( 1973). J . Phys. Chem., 73,1455 ( 1969).
Sioda, R. E., “Electrolysis with Flowing Solution,” Electrochim. Vermilyea, D. A.,‘ “Anodic Films,” Aduan. Electrochem. Elec-
Acfa, 13, 375 (1968). trochem. Eng., 3,211 (1963).
,, ‘”Electrolysis with a Flowing Solution on Graphite Wenger, R. S., D. N. Bennion, and J. Newman, ‘Wectrochemi-
Packing,” ibid., 13, 1559 ( 1968). cal Concentratin and Purifying from Dilute Copper Solu-
-., “Electrolysis with Flowing Solution on Porous and tions,” J. EZectroc%em. Soc,, 120, 109C ( 1973)
Wire Electrodes,” ibid., 15,783 ( 1970). Will, F. G., and H. J. Hess, “Morphology and Capacity of a
., “‘Distributionof Potential in a Porous Electrode under Cadmium Electrode. Studies on a Simulated Pore,” ibid., 1.
Conditions of Flow Electrolysis,” ibid., 16, 1569 ( 1971). Wilson, E. J., and C. J. Geankoplis, “Li uid Mass Transfer at
., “Current-potential Dependence in the Flow Electrol- Very Low Reynolds Numbers in Paczed Beds,” Ind. Eng.
ysis on a Porous Electrode,” J. EZectroanal. Chem. Interfacial Chem. Fundamentah, 5,9 ( 1966 ).
Electrochem., 34,399 ( 1972). Winsel, A., “Beitrage zur Kenntnis der Stromverteilung in
-., “Flow Electrolysis on a Porous Electrode Composed of porosen Elektroden,” Z . Elektrochem., 66, 287 ( 1962).
Parallel Grids,” ibid., 34,411 ( 1972). Wroblowa, Halina S., “Flow-through Electrodes I. Dia
-., “Limiting Current in Flow Electrolysis on Porous Mechanistic Criteria,” J . Electrwnal. Chem. Interfacia?:;$
Electrode,” Electrochem. Acta, 17, 1939 ( 1972). trochem., 42, 321 (1973).
-., “Certain Preparative Aspects of the Flow Electrolysis -., and A. Saunders, “Flow-through Electrodes 11. The
on Porous Electrodes,” ibid., 19, 57 ( 1974). I3-/I- Redox Couple,” ibid., 42, 329 ( 1973).
., “Flow Electrolysis on Grid Electrodes,” Anal. Chem., Yip, Harry Hung-Kwan, “Mass Transfer Coefficient in Packed
46,964 ( 1974). Beds at Low Reynolds Numbers,” M.S. thesis, Univ. Cali-
., and W. Kemula, “Application of Flow Electrolysis on fornia, Berkeley ( 1973).
Porous Electrodes for the Electroreduction of Potassium Yonng, L., Anodic Oxide Films, Academic Press, London
p-Nitrobenzoate,” Electrochim. Acta, 17, 1171 ( 1972). (1961).
Sohn, J. C., “Les electrodes A circulation de liquide,” ibid., 7 ,
629 ( 1962).
Stein, Werner, “Die Vorgange in den Poren von Masseplatten THE AUTHORS
bei der Entladung des Bleisammlers mit grossen Stromdich-
ten,” Die Naturwissenschuften, 45, 459 ( 1958). William Tiedemann received his Ph.D. in Electrochemical
Stein, Werner, “The Physical Processes in the Pores of Plates Engineering from the University of California at Los Angeles
During the Discharge of a Lead Storage Battery with Large in 1971. Since joining the Corporate Applied Research Group
Current Densities,” Ph.D. thesis, Rheinisch-Westfalischen of Globe-Union Inc. as a Senior Electrochemical Engineer, his
Techn. Hochschule Aachen ( 1959). research activities have been concerned with fundamental stud-
Stender, V. V., and 0. S. Ksenzhek, “Grafitirovaynnye anody ies of the lithium-fluorographite nonaqueous primary battery
pri elekrolize vodnykh rastvorov khloristykh solei,” Zh. Prikl. and porous electrode modeling of the lead-acid battery.
Khim., 32, 110 ( 1959). John Newman was trained in chemical engineering at North-
Szpak, S., J. D. Elwin, and T. Katan, “Current Distribution in western University (B.S., 1960) and the University of Califor-
Porous Electrodes Operating under Forced Flow,” Electro- nia, Berkeley (M.S., 1962; Ph.D., 1963). He is currently Pro-
chim. Acta, 11, 934 (1966). fessor of Chemical Engineering at the University of California
Tiedemann, William, and John Newman, “Douhle-layer Ca- at Berkeley and a Principal Investigator in the Inorganic Mate-
pacity Determination of Porous Electrodes,” J. Electrochem. rials Research Division at the Lawrence Berkeley Laboratory.
SOC., in press ( 1974 ). His work in electrolytic mass transfer and electrochemical sys-
Topol, Leo E., and Keith B. Oldham, “Electrochemical In- tems design earned him the Young Author’s Prize of the Elec-
vestigation of Porous Media 11. Potentiostatic Experiments,” trochemical Society for both 1966 and 1969.

OsciIIating Behavior on Distillation Trays4

This paper describes an extension of an earlier study on oscillations
which can develop on distillation trays. The oscillations take the form of
violent lateral movements of the gadliquid biphase, and they are generally
MICHAEL W. BIDDULPH
undesirable for good column performance. It has been found that lower
density gases need higher velocities to initiate oscillation, and vice versa fdr Department of Chemical Engineering
higher densities. The predictive method for the onset of oscillations proposed University of Nottinghom
previously appears to remain valid for the range of gas density covered, Nottingham, England
and for small hole size trays. The presence of oscillations has been shown
to increase weeping significantly,

SCOPE
The objective of this work was to extend the study by results were on the system air/water, and so the depen-
Biddulph and Stephens (1974) of the phenomenon of dence on gas density could not be confirmed. In this exten-
oscillations on sieve trays. The dimensionless group which sion of the work, the apparatus was modified to allow
was proposed earlier as a criterion for the onset of oscilla- various gas mixtures to be used. This was achieved by
tions was based on the work of Hinze (1965)and on dimen- converting the open system used previously into a sealed,
sional analysis. The group did appear to have consistent closed-circuit system. Mixtures of helium and air of vari-
critical values for full-wave oscillation and half-wave oui compositions were used to obtain densities lower than
oscillation for the results then available. However all these that of air. Mixtures of carbon dioxide and air were used

AlChE Journal (Vol. 21, No. 1) January, 1975 Page 41