Advances
in Colloid and Interface Science. 16 (1982) 281-320 ElsevierScientificPublishingCompany,Amsterdam-Printed in TheNetherlands
281
THE RESPONSE OF A COLLOIDAL SUSPENSION TO AN ALTERNATING ELECTRIC FIELD R-W. O'BRIEN Dent. of Theoretical and Applied Mechanics, School of Mathematics, The University of New South !Jales,P_ O_ Box 1, Kensington, New South Wales 2033, Australia CONTENTS ABSTRACT ___.______________________-____.-_-.____-______________..___. 281 I. INTRODUCTION ________~__..._...__._._......__......_._....__........_.282 II. 283 A. Outline cf the paper ~-_..~_~.~~~.~~~...~~~~~.~~~~~~~.~~~~..~.~~~~ III_ DEFINITION OF COMPLEX CONDUCTIVITY ___.____.________.__.________._._-_ 283 A. The measurement of K* ____________________________._______________ 285 IV. THE RELATIONSHIP BETWEEN COMPLEX CONDUCTIVITY AND PARTICLE DIPOLE STRENGTH IN A DILUTE SUSPENSION ___________.__________.__..___.___.___ 285 286 V. THE ELECTROKINETIC EDUATIONS ____..______._*___.________--____.-.___-_ VI_
VII. VIII IX_ X. XIXII_ XIII XIV.
xv. XVI.
THE PERTURBATION SCHEME _____.___________________________.____________ 288 A. The dioole coefficient of a particle with small dielectric constant _________________________________.______._._______._________ 293 B. The zero frequency limit ____.____________________________________ 295 THE CALCULATION OF THE DIPOLE COEFFICIENT TO O(r,)________________.___296 THE &) FDRNULk FOR THE DIFOLE COEFFICIENT __-____-_-_-___-_-_-_____299 COMPARISON OF THE APPGOXIMATE FORMULA !.IITH THE COMPUTED RESULTS ______ 302 THE EFFECT OF SURFACE DISSOCIATION REACTIONS ON ME DIPOLE STRENGTH _. 305 APPENDIX I - THE MEASUREMENT OF COMPLEX CONDUCTIVITY _______._____.___308 APPENDIX II - DERIVING THE FORMULAE (EQS. 5.24 AND 6.3) FOR 6. and uo. 370 APPENDIX III - A METHOD FOR EVALUATING THE FORMULA (ED. 7.7) for CE __ 312 APPENDIX IV.- THE DIPOLE COEFFICIENT OF A PARTICLE IN A SYMMETRIC ELECTROLYTE WITH IONS OF APPROXIMATELY UNIFORM DIFFUSIVITIES __________-______________________.______.____ 314 ACKNOWLEDGEMENT ___________-_____-_-_____________-_______-_____-___-_317 LIST OF SYFISOLS ___._____._..____._.______________~_____~_____~_-_~_~~317
I_ ABSTRACT This paper is concerned with the calculation of the complex conductivity K* of a susoension, a quantity which may be determined exoerimentally from the measurement of the alternating current which flows between a pair of electrodes in the suspension due to an alternating voltaqe difference. A semi-analytic formula is derived for the complex conductivity of a dilute suspension of spherical particles with small dielectric constant which is reasonably accurate for c-potentials
of less than 50 mV_ For such suspensions this formula represents a very economical alternative to the exact computer calculation of K* described by DeLacey and klhite (ref. 2). Although the formula for K* is derived for particles with fixed surface charge, it
is
shown
that
the
formula
can
also be applied to a more general
class of susoensions, in which the surface charge arises from the dissociation
of a single type of surface group_ II.
INTRODUCTION East of the theoretical studies of colloidal suspensions in electric fieids have been concerned with the interpretation of measurements of the electrophoretic mobility and electrical conductivity, quantities which are measured in a steady electric field_ Such experiments are, however, limited in the amount of information they can provide; i-e_ they only yield one number for a given suspension_ This number can be used to determine only one parameter, such as the c-potential of the particles_ Furthermore, it has been shown (ref. I) that electroohoretic mobility and static conductivity are insensitive to the particle charge determining mechanism, and hence such measurements cannot provide information about such things as the dissociation constants of surface groups. As these limitations do not apply to alternating field measurements performed over a range of frequencies, there is a possibility that such measurements may yield much more information than can be obtained from the steady field measurements which are currently in use.. In this gaper we present a theoretical study of the behavior of a dilute suspension of spherical particles with low C-potential in an alternating electric field. In particular, we shall be conct:-ned with the calculation of the complex conductivity of the suspension, a quantity which is related to the complex impedance of a conductance cell containing suspension_ For a dilute suspension, the interactions between the particles are negligible, and we can determine the complex conductivity from an analysis of the behavior of an isolated particle placed in an infinite electrolyte solution with a spatially uniform alternating electric field. In general both particle and ions will be sqt in motion by the field, with the particle and counterions moving in opposite directions. As a result the double layer distorts in an alternating fashion, giving rise to an electric field, which at large distances from the particle has the same form as that of an alternating electric dipole. It can be shown (ref_ 2) that the complex conductivity of a dilute suspension may be determined from the eiectrical dipole strength of the particles_ To determine the phase and amplitude of the dipole strength of a particle, we must solve the electrokinetic equations which describe the distribution of electrical potential, ion densities and fluid flow around the particle. At present there is only one analytic solution to this problem, namely Dukhin and Shilov's formula (ref3) for a particle with a thin double layer in an electrolyte with two types of ions.
283 Although there are no analytic formulae available for particles with larger double layers or more complicated electrolytes, it is now possible to determine the dioole strength of a particle with fixed surface charge, using OeLacey and uhite's (ref. 2) recently develoned computer program to solve the electrokinetic equations numerically. Although this computer solution may be obtained for any surface potential and electrolyte, the program is very expensive to run, and the numbers which appear in the output provide little physical insight into the problem_ Outline of the Paper Our aim in this paper is to derive an approximate formula for the dipole strength of a particle with low c-potential. In the following section we define the complex conductivity K* and we briefly describe the experimental determination of this quantity. In Section IV we give the relationship between K* and the dipole strength for a dilute suspension of spherical particles. The remainder of the paper is concerned with the calculation of the dipole strength of a spherical oarticle with low zeta potential, small dielectric constant and fixed surface charge. In Section V we set out the electrokinetic equations
A.
which must be solved in order to determine the dipole strength and in Section VI we describe the perturbation scheme for solving these equations for a particle with low zeta potential_ The first step in the perturbation scheme (involving the solution ofthe electrokinetic equations for an uncharged particle) is described in Section VI. In Section VII we then solve the next set of equations to obtain the O(c) correction to the dioole strength and in Section VIII we obtain a formula for
the
dipole
strength
which
is correct
to O(s*).
Although
the
formula
contains
a number of integrals which must be evaluated numerically, the integration is straightforward and can even be carried out on a programmable hand calculator_ In Section IX we compare values of the dipole strength obtained with this approximate formula with values obtained from DeLacey and !Jhite's(ref. 2) computer program; it is found that the O(,;*)formula gives reasonably accurate results up to 5 values of about 50 mV_ For c-potentials in this range the formula represents a very economical alternative to the exact computer solution of the problem_ This comparison also provides the first independent check of DeLacey and klhite'sprogram to be reparted in the literatureIn Section X we remove the fixed surface charge condition and allow for variable dissociation of an ionogenic surface group on the particle. It is found that to D(,;*)the dipole strength is unchanged by this modificationIII.
DEFIIJITIONOF COMPLEX CONDUCTIVITY Consider a colloidal suspension subjected to an alternating macroscopic electric field. In general, the local electric field and current density in the suspension fluctuate with position on a microscopic length scale (of the order of the particle
254 radius) about average values which themselves vary on a macroscopic length scale. it is these average values which are of interest here since these are the quantities which are accessible to measurement_ Ue denote the average values by angle brackets; for example cE, denotes the average electric f-reld,which may be defined mathematically as a volume
average,
viz.:
= E dV_ : _I-
(2-l) VHere c is the local electric field and V is a sample volume which contains a statistically significant number of particles but which is small in comparison with the macroscopic length scale. Other mean values are defined similarly. The local current density in the suspension is given by
3P _ where if is the current density due to the free charges and s 1s the polarization charge contribution, P being the dioole strength per unit volume. In general the charge density at any point in the suspension varies with time and thus the current density has a non-zero divergence_ However. by adding &g to the current density we obtain the divergence free vector f, given by (ref. 4, pp_ llO-117): i=i
-f
1 +=z
sD
-
(2.2)
where LJis the electrical displacement vector. Since it is customary to refer to &gas the displacement current, we shall refer to i_as the nett current density. It is assumed that the average electric field and nett current density vary sinusoidally with time; hence we may writer cE> = Eoeiwt , and = i,eiwt )
(2.3)
where as usual cF> and -=i_~ are piven by the real parts of the complex quantities on the right-hand side of these expressions and 2 TIWis the frequency. At 10~1field strengths, we expect that the quantities j. and E-Owill be linearly related. For a statistically isotropic suspension, this linear relation takes the simple form i -0 = K*Eo , where K* is defined as the complex conductivity of the suspension.
(2-4)
285 A.
The measurement of K* In the earlier works on this subject (ref. 2,4), the experimental determination of K* is only described for the case of z conductance cell with parallel plate electrodes
_ For completeness we note here that for other types of conductance the procedure for measuring K* 1s _ similar to the static conductivity procedure. If 2, denotes the complex impedance of the cell for a standard liquid of known comolex conductivity Kc, and 2 is the impedance for the suspension, then
cells
z Kf
=
K*
s
-?
_
2
(Z-5)
This result (which is derived in Appendix I) is only valid if the frequency large to restrict any inhomogeneities in the
of the applied field is sufficiently
suspension to thin layers around the electrodes_ IV_
THE RELATIONSHIP BETWEEN COMPLEX CONDUCTIVITY AND PARTICLE DIPOLE STRENGTH IN A DILUTE SUSPENSION In a dilute suspension, we may treat each of the particles as being alone in an infinite electrolyte with uniform ambient electric field EoeiL>t . In the following section it will be shown that the electrical potential at large distances from an isolated spherical particle has the form (-1
f
Co/r3)r
-
Foe itit,
(3.1)
where c is the position vector from the center of the sphere. As mentioned in the introduction, the disturbance field COT -7
r3
- (
!ioelLt
has the same form as the field due to an electrical dipole with dipole strength i:_;t proportional to COEOe _ We shall refer to the quantity Co as the dipole coefficient particle. of the The radius tained For simply
variables in Eq. 3.1 are assumed to be non-dimensional with the sphere a being the unit of length and F is the unit of ootential; thus ED is obby multiplying the dimensional field strength by E _ a dilute monodisperse susoension of spheres, the complex conductivity is given by (ref. 2)
K* = Kz (l+3+CO) =
(3.2)
where Q is the particle volume fraction and KE is the complex conductivity of the uniform electrolyte which lies beyond the equilibrium double layer. If the electrolye is dilute (an assumption which is implicitly made in the formulation of
286 electrokinetic equations in Section V) and if the frequency of the applied field is much smaller than the relaxation frequency of a water molecule, then the complex conductivity of the electrolyte is simply given by (3-3) where Km and E are the static conductivity and dielectric constant of the electrolyte V_
THE ELECTROKINETIC EQUATIONS In this section we set out the differential equations which govern the distribution of ions, electrical potential and velocity in the electrolyte. These equations must be solved in order to determine the particle dipole coefficient CQ. As
mentioned in the previous section, we will
obtained
by dividing
the
electrical
potentials
be using
non-dimensional
variables,
by
kT/e, dividing the position vector by the particle radius a and dividing the ion densities by twice the ionic strength of the electrolyte multiplied by Avogadro's constant. Finally, the fluid velocity will be multiplied by
where n is the viscosity of the e-lectrolyte. it is assumed that the electrical potential and ion densities are As usual, only slightly altered by the applied field_ Me use a r;prefix to denote these small perturbations; hence the perturbation in the-density of the type j ions at iiJt while i+eib:tand lipe a point r in the electrolyte is denoted by finj(r)elfst, denote the changes in electrical potential and fluid pressure. Since the fluid velocity is zero in the absence of an aoplied field, we will dispense with the
r2;z>t (,:a)’
Zj3nj =0 , -Lx j=I
T-lJ=o,
(4.1) (4.2)
and 0 ,Q, + Zvjdn_ili, , - ~-VI-I. 7'- (osnj + Z.n.?jb 3 J J
(4-4)
values of the potential and j= I,..-,N. Here 9’ and ny are the equilibrium ion density in the absence of an applied fieid. N is the number of ionic species
for
287
in the electrolyte,
Zj
and Dj are the valency and diffusivity of the type j ions
and 2 m. = E(kT) J 6ane2Dj is the usual non-dimensional
ionic drag coefficient.
Eq_ 4.1 is Poisson's equation, Eq. 4-2 is the incompressibility
constraint
and Eq. 4.3 is the force balance equation for the liquid; in deriving the latter equation, it is assumed that the inertial forces are negligible, a result which is valid for the frequency range of interest here (ref. 2). Thus the flow field is able to adjust instantaneously
to the variations in the body force term on
the right-hand side of Eq. 4.3. Eqs-
4.1
presence
of
to 4-4 the
are
the same as for
the
static
field
problem,
except
for
the
term
- 2 Sri7iJa J --D.j on the right-hand side of the ion conservation equation (Eq. 4.4).
As we shall
see, this extra term considerably complicates matters. In the static field problem, O'Brien and Glhite (ref. 1) were able to eliminate Poisson's equation tial not
functions possible
(Eq_ 4.1) from the problem by working in terms of ionic
related
to A;> and Znj_
In the alternating
to remove the electrostatic
field
problem,
potenit
is
equation (Eq. 4.1) and we will work here
in terms of the original variables 8:) and .t.nj_ For a coordinate system fixed to u_ and An. take the form the particle, the far field boundary conditions for .';:::, J
(4-5) J Here ME is the non-dimensional
electrophoretic
mobility of the particle defined
by:
(4.6) where
3 is
the
dimensional
mobility_
The boundary conditions to be applied on the particle surface depend on the nature of the particle charge mechanism.
To begin with, we will assume that the
particle has uniform surface charge density, unaffected by the applied field. In Section X we will allow for variable dissociation of an ionogenic surface group.
258
For a particle with fixed charge, the surface boundary conditions are
(4-7) and u __= 0 on the particle surface, where E is the particle dielectric constant and a$~ is P p the potential variation inside the particleTHE PERTURBATION SCHEME VI. As in the earlier studies of the conductivity (ref. 5, 6) and electrophoresis (ref_ 7) of particles with 10~~<-potential, we assume that each of the variables in Eqs_ 4-i to 4-4 can be expanded in power series in eG/kT_ Superscripts will be used to denote the order of the coefficients in these expansions; for example, the expansions for &nj and 5-3are zn_
J
=
&rigf (31
Bnjlt (2)'
6nS + ___, (5.1)
VJhilethe expansion for the particle dipole coefficient takes the form (5.2) where the quantities Ci, CA and Cg are obtained from the far-field forms of the potentials 6~>",~g1and cc', respectively. Our aim is to determine these quantities for a particle with a small dielectric constant. In order to avoid confusion with the "0" superscript in the equilibrium potential 0 0 and ion density ny, we will reolace e" at this point by the symbol Y and use the Boltzmann formula:
ny=n3
exp(-ZjY) ,
in place of ny, where ng is the uniform ion density at large distances from the particle_ On substituting the perturbation expansions for each of the variables in Eqs. 4-l to 4-4 and boundary conditions 4.5 and 4.7, and equating coefficients of like powers of equations and boundary conditions for each set of _ c.. _we obtain _ b-a’, @I’ and IA’. coefficients 6n:,
In
this
section
we will
determine
the
leading
terms
in
the
above
expansions;
these are the quantities associated with an uncharged particle in an alternating In the following section we look at the O(C) terms
field_
Section VIII we obtain the O(<*) term in the expansion
and then finally in
(Eq. 5.2) for the dipole
coefficient. The quantities 6o", Any, 6p” N 2 Z.&n! = 0. ,2&$0 + (i--a) t j=l J J
and u" satisfy the equations: (5.3)
v-u 0 =o,
(5.4) O=O _,
V2"O _-yap
(5.5)
2 6n.=O 0 itia and 7*&n! -F Z.n"Iv26~~0-J JJ Dj J
,
(5.6)
with boundary conditions:
QJOS-E “0 _
-0
-c,
0 s -MEEo
(5.7)
, asr-=.
andk!$_>$=O, 0
38n ---J f
(0 -= at* Zjnj
at-
and
u”
atr=
ar
(5.8)
0 =
= 0
I_
The
equations
by a zero
velocity
(5-&j
and
and
and boundary conditions for
(S-5)
pressure
field,
with
zero
electrophoretic
u_’ are mobility
satisfied (ME =o
Thus an uncharged particle and the surrounding solvent are not set in motion by an alternating field. The remaining equations involving 6,' and brig may be rewritten in the compact vector form V2V0_
-
(5-J)
witere
V
O=
(5-10;
I-
and 5 is the matrix defined by
(5-11)
Here Tj = Z$-I~ '
(5-12)
(5.13) In deriving Eq. 5.9, we have eliminated T2G@' from Eq. 5.6 by using Poisson's equation (Eq. 5.3)_ Eq. 5.9 is the vector analog of the familiar diffusion equation which describes (among other things) the temperature distribution around a body in a continuous medium, where the body contains a sinusoidally varying heat source. In this case the temperature field due to the body is found to alternate in sign with increasing distance from the body. with an exponentially decaying amplitudeAt small frequencies the decay length is large and so the disturbance in the temperature field due to the body penetrates a long way into the surrounding medium, while at high frequencies the disturbance is confined to a thin layer on the surface of the body. As we shall see, the disturbances in the ion densities obtained from the solution of Eq. 5.9 have a similar frequency dependence, although in this case there is not a single decay length but a number of decay lengths associated with the eigenvalues of the matrix A. = !-fe begin by assuming that the solution v0 to Eq. 5.9 can be expressed as a linear combination of the eigenvectors of the matrix 6, i.e. M v0 =
z j=O
u(j)8 J
'
(5.14)
where the u(j)'s are a set of linearly independent eigenvectors. On substituting the form (iq. 5.14) for y" in Eq. 5-9 and using the fact that the eigenvectors are linearly independent, we obtain the simple equations:
02*g -
(ra)2Xjfg = 0
(5.15)
the new unknobms fy- Here Xj is the eiqenvalue associated with the eigenvector ,(j) and the index j ranges from zero to M_ for
291 In order to determine the quantities fy we must first obtain the eigenvalues .'. =; these quantities are the roots of the characteristic equation: J
?.$I = 0 ,
IA-
where i is the (N+l)x (N+l) unit matrix. !Gth the aid of the formula (Eq. 5.11) for h, we can write the characteristic equation in the more explicit form:
*
(5.16)
Clearly one of the roots of the characteristic equation (lo' say) is zero- By solving the eigenvector equation:
with k = 0, we find that the eigenvector u (0) corresponding to the zero eigenvalue is simply
(5.17)
With the aid of this observation, we can write the outer boundary condition (Eq. 0 and nni in the simple vector form 5.7) for Sri
y” -
-(E.
(ci)as r --r 10 _ - rlu - -
From the expression (Eq. 5.14) for v(O), it follows that the outer boundary conditions for the fy's are simply * -E f0 -0-r 0
9
(S.lE?)
and fy s 0 for i = l,Z,...,M asr+-. From the symmetry and linearity of the problem, it follo~rsthat each of the fy functions has the form (ref. I, Section 6): f:(r) = gi(rjEo - F
,
where F= C/r_ Substituting the above forms for f:(r) in Eo. 5-15 and solvinq sub.jectto the outer boundary conditions (Eq. 5-181, VJe obtain
0 f,tr)
= -go - r+
0 CoEo-
r/r
3
, (3_13)
and
for j = 1,2,...M. The constants CC! are to be determiiled from the boundary condition (Eq. 5.8) on the particle surfase,and the square root in the exponent is taken to have a positive real part (to ensure Mith
the aid
of
a decaying
solution).
the expressions
{Eq.
5.19)
for
fp and the
and 5.17) for v_' and u_(O), we see that the perturbation
formuiae ;Eqs. 5.14
in the ion densities decay
exponentially with increasing distance from the particle, while the perturbation in potential at large distances from the particle is dominated by the fi component and has the form s::r mr -5
0
0 - r+COEO
-r/r zasr___
fhus the coefficient Co ' in the oirter form for -fi is the same as the dipole coefficient appearing in the outer form The exponential scale
of order' R(rra+T}-';
acteristic equation one
which
for
the potential.
term in the expression
approaches
(Eq. 5.19) for
fi
decays
on a length
as w approaches zero, it can be seen from the char-
(Eq. 5.16) that all the eigenvalues approach zero, except for unity.
At
high frequencies, on the other hand, it can be
shown that Xj is approximately equal to Y-, a quantity which increases in proportion to C.1. From Eq_ 5.14 it follows that the gerturbations
in ion density extend a
large distance into the electrolyte at low frequencies, while at high frequencies the perturbations which
are confined to a thin layer on the particle surface; a result,
(as we mentioned before) is similar to the solution for the scalar diffusion
equation. The constants Ci in Eq. 5-19 are obtained by rewriting the inner boundary conditions (Eq. 5-8) in terms of the new unknowns fi and substituting
the formulae
of Eq. 5.19; the result is a set of linear simultaneous equations for the unknowns Co.
In or.fer to set up these equations, it is necessary to obtain formulae for
tjhe eigenvalues
X. .
it
Unfortunately,
does
not
seem possible
to obtain
such
formulae for a geieral electrolyte, and in order to obtain any analytic results, it is necessary to place restrictions on the nature of the electrolyte or on the dielectric constant of the particle_
In this paper it will be assumed that the
particle dielectric constant is much smaller than that of the solvent, an assumption which is ‘As
valid for
many aqueous
suspensions_
usual, R denotes the real part oi the quantity_
293
The dipole coefficient of a particle with small dielectric constant For a particle with zero dielectric constant, the ion densities are unaltered by the application of the f-ield. The potential field a::~' has the same form as
A.
for a particle in a pure dielectric, viz.:' ,j,~
0
(5.20)
and
inside the particle; a result which may be verified by substitution in the governing equations and boundary conditions. Thus for an uncharged particle the dipole coefficient is simpiy
co=_l
(5.21)
2 -
0
For a particle with a non-zero dielectric constant the normal component of the electric field at the particle surface is non-zero and thus there i'sa tendency for the field to drive ions alternately into and away from the impermeable particle surface. The resulting disturbances in the ion densities then diffuse out into the electrolyte, leading to a redistribution of charge and consequently to an alteration in the particle dipole coefficient. To obtain a formula for the dipole coefficient for a particle with a small but 0 in - the boundary condition (Eq. 5.8) non-zero dielectric constant, we replace 61; by the formula (Eq. 5-20) for a particle w?th zero dielectric constant. Since this term is mu'ltipliedby E /E in the boundary condition, the errors in this approximation The inner boundary condition (Eq. 5.8) may now be written will be of order (cp/c)5 in the simple vector form 2V
?%
0
=?%,
-
2 E -O-r
Tl -1
at r = 1.
Assuming that the column vector on the right-hand side may be expressed as a linear combination of eigenvectors*, viz:
'It is possible to show that any vector can be written in this form if the electrolyte has only two ionic species, or if all the ions have uniform diffusivity. Although it seems likely that this result will hold for a general electrolyte, the proof is not available at present-
294
-1
I_! 11 = M 1 Sj”(j)
(5.22)
9
j=O
1
IN
and substituting the form (Eq. 5.14) for v" in the above boundary condition, we find
5r
-
--_Y_E
2 5
(5.23)
.r
J-0
at r = 1, for j = 0, l,...,N. Although it is not possible to determine all of the sj's for a general electrolyte it is shown in Appendix II that ;o=-(I+~*j,.jjjI
)
(5.24)
which is all tnat is required for the calculation of the particle dipole coefficient 0 in - the boundary condito O(C,/S). On substituting the formula (Eq. 5.19) for f. tion (Eq. 5.23) and using the above expression for so, vie find that the dipole coefficient is given by: (5.25)
where we have used the fact that N
_ Ikf'k _ 4_iiK_ 1Wf
(5.26)
From Eq. 5.25, it can be seen that for
the dipole strength is the same as for an uncharged sphere in a pure dielectric since (as mentioned earlier) the perturbations in ion density are small and are confined to a thin layer on the particle surface. At low frequencies the ions are arranged around the particle in such a way that the dipole coefficient is the same as for a sphere with zero dielectric constant; a result which is consistent with O'Brien and White's (ref- I) conclusions on the effect of particle dielectric constant in the static field problem.
295
6.
The
zero
Before particle, for
frequency
have
the
shown,
order
rived
5.19)
here lrnj
(Eqs.
for
for
4.1
the
an
uncharged
4.4)
at
for
As we mentioned approach (Eq.
zero
with
5.19)for
the
more
earlier
in
formula
the
past,
some
formula,
a charged (Eq.
3.2)
researchers
taking
This
we turn
0 Sn J-;
they
a few
this
section,
give
since are
than
(ref
CO to
procedure
be
has
Thus
with
the
same
of at
small
expressions formulae
been
far-field
as
forms
those
from
the
for
the
for
an
$5:;
uncharged
particle
surfaces.
except
frequencies
eigenvalues
de-
equations
eigenvalues
low
(Eqs. were
electrokinetic
lengths
all
the
these
the
the
the
Oebve
frequency.
to
al though
also
charge,
associated
of
of
field.
error,
particle
decreasing fj's
the
arbitrary
of
the
this
a static
Si~!~”and
a charged
distances
coefficient
5).
of
oarticle,
of
from
in
(ref.
source
dipole
application In
conductivity
quantities
a particle to
particle
for
the
the
the
problem.
a particle
incorrect
locate
of
discuss
field
static of
be
to
static
the
to
to
calculation
here
the
coefficient
and
and
to
however,
In
the
pause
calculated
dioole
5.14
to
we shall
conductivity
8)
limit
proceeding
the
take
one
expresssions
the approximate
form
in
the
region
(5.27)
where of
we nave
arbitrary
tial in
619 has this
on
fj
the
the
the
f.
the
3.3)
(Eq.
3.1) the
the
have
5.27
3.1).
for
surrounding
particle (Eqs..
for
in
static
this
region;
value
exponentially and
the
now considering
v_, it and
follows in
fact
as (Ai vanishes_ decaying
potential
terms
ceases
to
a particle
that the bI< th
in
the
have
the
poten-
coefficient increasing
equation
a dipole
(Eq. form_
region
(Eq.
from
we are
5.14)
formula
significant
form
charge
since
(Eq.
particle,
except
zero
“O”,
the
form
fj’s
(Eq.
superscript
approaches
dipole
expression
regions
dipole
become
in
the
the From
form
from for
Finally,
all
the
dipole
distance 5.19)
drooped charge.
the the
and
as
and
5.28)
decayed It
to is
zero,
.&LA
conductivity particle
a result can
(ref_
extends the
and
dipole
dipole
be quite
the
potential
2) ; in
out
to
which
general,
distances
coefficients
different.
once
coefficient
the of
again
takes
appears
in
region
order
associated
the
of non1 jxj -!z,;a
with
the
_
296
THE CALCULATION OF THE DIPOLE COEFFICIENT TO O(c) VII. To determine the O(c) terms in the perturbation expansions for 69, anj, U and Bp for a particle with zero dielectric constant, we must solve the equations
v2y1 _ (c-)24 - y1 = WY1 - V&z> Ow 1
v-g
(6.11
,
=o,
(6.2)
and .2IjI- r&p I3=y
(,,)2Y&!O,
(6.3)
L
where v
1 =
s+1
9
and VJ=
1 Zldlll
0
IlZl I222 (6-4)
1 ZN6nN . INZN
In keeping with the superscript convention introduced in the previous section, Y' is the coefficient=of e&kT in the perturbation series for the equilibrium potential Y, given by Yl
= exprKa(l-r);/rl
(6-5)
for a spherical particle_ In deriving Eqs. 6.1 to 6.3, we have used the fact that the order one terms any, CD and 6p" are zero for a particle with zero dielectric constant, as shown in the previous section. From the boundary conditions 4.5 and 4.7, we find that
9'~
-M:Eo
as r-t -
(6.6)
,
and
av’ --=
ar
0
-
and u1 _ = 0 _
(6.7)
on the particle surface. Eqs.
6.2
and 6.3
are
the
same as for
the
static
field case. Thus to O(5) the
particle moves with the same mobility as in the static instantaneously
to
variations
in the
applied
field.
field
problem,
responding
29-i
The remaining equation (Eq. 6.1) is simply an inhomogeneous form of the vector diffusion equation (Eq. 5.9) discussed in the previous section; the same general conclusions concerning the frequency dependence of the solution apply here, except that in this case, since there are sources'distributed throughout the double layer, the disturbances in the ion densities at high frequencies will no longer be confined to a thin layer on the particle surface. This source term arises from the action of the field vOg" on the equilibrium charge cloud. This action results in an alternating build-up and depletion of ions on opposite sides of the particle; for example, the positive ions in the double layer around a negatively charged particle will be swept past the particle, resulting in a build-up of these ions on the downstream side of the particle. At the same time, the field will move the negative co-ions in the opposite direction, pushing them into the downstream portion of the double layer. Thus there is a build-up of both types of ions on the downstream side of the double layer and a corresponding depletion on the upstream side; the sign of the dipole coefficient CA is determined by the competition between these ions. To calculate CA, we begin by assuming (as in the previous section) that the unknown vector v1 in Eq. 6.1 can be expanded in terms of the eigenvectors of e, viz. v
1
!A_
L
=
y(j),! .I -
.j=O Substituting this form in Eq. 6.1 and using the fact that the eigenvectors are independent, we get $ff
- (,a)*X_f! =
J
3j7Yl
- vsp,
JJ
(6.9)
for j = O,l,...M, where the aj's are the coefficients in the expansion M
y=
z u(% ’_
j=O-
for the vector fi 'or 0
w.
Similarly, the boundary conditions 6.6 and 6.7 qive
as r -rm
and 1 s=Oatr=l for j = O,l,..., M.
,
,
(6.10)
298
As we mentioned in the previous section, the quantities 8~ and &nj have the same far-field form for a charged particle as for an uncharged particle. Thus at distances of more than a few Debye lengths from the particle, the quantities fI have the same exponentially decaying form (Eq. 5.19) as fy for j non-zero (vritn -3 replaced by Cj). while the iar fieid form of fi is given by Cj J.
1
co-F
fcls Co
r2
asr--0:
_
(6-H)
To determine the quantity Ci, we need only solve the equation ,2fI = +I 0
- .&.O
(6.12)
for f:, subject to the homogeneous boundary conditions 6.10. is sho>;in that the quantity ZD is given by ei: 'f I-2-D. jE1 J 3 3 $;<= - ' 0 (ias+ 4yK-)
In Appendix II it
(6.13)
The solution fi to Eq. 6.12 may be written as an integral involving the righthand side of the equation, using the procedure described on pp_ 242-243 of Refo5. On substituting the form of Eq_ 5-20 for 3-3 ln this integral and computing the far-field form of fg, we find that the required dipole coefficient is given
by CA =
-uoL(ra)
(6.14)
,
where L(x) = * + I
+
eXE (x) 25
and E5 is an exponential integral (See Chapter 5
X2 of Ref. 10). On substituting the result (Eq. 6.14) and the formula (Eq. 5.21) for CA and CE in the perturbation expansion (Eq. 5.2) for the dipole coefficient. we find that
Co = -l/Z - uoL(ita)ec/kT f O[(e
(6.15)
where no is given by the formula of Eq. 6.13. In Section IX we shall be comparing the approximate formula for Co to O(c2) with computed values for a KC1 electrolyte; in this case the ions have approximately equal diffusivities and the the O(s) correction to Co in Eq_ 6.15 is very small because the build-up in negative ions on the downstream side of the particles is approximately cancelled by the corresponding build-up in the positive ions. For other types of electrolytes (such as acids), in which the ions have very different diffusivities, the O(c) correction to Co can be quite significant; a point which is discussed in Section VI of Ref. 5 for the static field case.
299
VIII.
THE O(,;*) FORWLA
FOR THE OlPOLE COEFFICiENT
The c* coefficient Cz 7n _ the perturbation expansion for the dipole coefficient CO is obtained from the equations for the quantities .?:>* and 6,:.
As before, we
write these equations in the vector form:
~~*,* - (ra) 24 - v 2 = -n
f VY l
1
.yl
I 0
Z26n1 1-l
!!
(7-l)
-2 1 ZN91 i: ! and we look
for
a solution
of the
form
11
.* -
=
-x
f?"(j) J-
J=
_
(7.2)
As in the previous sections, we will compute the dipole coefficient from the asymptotic formula
(7.3) for fz at larae r.
- Eq. 7-l by the expansion On replacing "2 In
(Eq. 7_2), and
expressing
the column vectors on the right-hand side of Eq. 7-l in terms of the eiaenvectors ,li), we obtain the following equation for f;: IN N
N
Z;,?
-I-
- (6nlVY'j + ,/Y'j
j=l where we have used the result of Eq. A.7 in obtaining the u (0) component of the vectors on the right-hand side of Eq. 7.1, together with the fact that 6~3' satisfies Laplace's equation.
In a similar manner, we find that the boundary condition for
fg on the particle surface is
(7.5)
at r = 1. It is possible to represent the required quantity I$ as an integral involving the terms on the right-hand side of Eq. 7.4 by using the relation 0 -_ ,y* 3n
f0 -M:$O 2 an
(7.6)
61>'GdV, V
which follows from Green's identity (ref. 9). Here G is the quantity on the right-hand side of Eq. 7.4; A denotes the surface of the particle together with a large spherical surface of radius R (centered on the origin); and V is the volume lying between these two surfaces_ If the radius R is sufficiently large, we may 0 aoproximate 6:> and fi in the integral in Eq. 7.6 over the outer surface by their asymptotic forms (Eqs. 5.7 and 7.3)_ On evaluating the resulting integral and substituting the expression from the right-hand side of Eq_ 7-4 for G, we obtain (after an application of the divergence theorem)
c; =
N 1 EG2 I .Z?D. f WI) 335~ Eg(1WE+4iiKm) - ' j=l V {
c
i-t
IjZjDj J(oY'. 06+')6$
j=l
2 - v(S:b")2dV
dV f ;
.Y2 - r(~.')~ dV I
V
I
V
N
+x
Z$DjI
j=l
(,YI - ~6:2')&nidV -f
'jmjDj
j=I
V
(uyI _
>
(7.7)
V
where E. = IEOL Unfortunately, it is only possible to evaluate two of the integrals in this 0 formula analytically. !*lith the aid of the formulae of Eqs.5.20 and 6.5 for 6$ and YI , we find
f V
-
v(~v”)2d,,
=
-8a E2 o 3
$ exp(2ra)E6(2
(7.8)
On comparing the form of Eq- 7-7 in the limit of zero frequency with a formula for Co obtained from Eq_ 5.35 of O'Brien's static conductivity paper (ref. 5), we obtain
301
J
(OYl - gl),+'dV = -4nE;N(~a)
,
(7.9)
V
where the function N(za) is defined by the formula (Eq. 5.34) in Ref. 5. The remaining integrals in the expression numerically.
(Eq. 7.7) for CE must be evaluated
Of these, the easiest to evaluate is the integral involving Y2.
We let
vY2
-
N
o(t;l>')'dV= -4;iEg ~n~Z~F(~ra).
(7.10)
j=I Values of F(t:a) (obtained by the Romberg Integration technique (Ref. 11, Section 8.3) are set out.? in Table 1 and the function is plotted in Figure 1.
Unfor-
tunately, the remaining integrals involving &ni and 61:;~ depend on so many parameters that it is impossible to provide an adquate representation a single table.
of these functions in
The best that can be done is to reduce these integrals to simpler
TAeLE 1 Computed values of the function F(,:a) which appears in the formula F(ra)
0.0219 0.0249 0.0306 0.0314 0.0422 0.0528 0.0712 0.110 O-124 0.141 0.164 0.196
of Eq. 7-10.
-ra
F(r:a)
1.1 I.0 0.9 0.8 0.7 0.6 O-5 0.4 0.3 0.2 0.1 0.01
0.216 0.242 0.274 0.315 0.369 0.445 0.554 0.726 l-028 1.673 3.778 46.95
Note: For large ka values, F = l/d ::a - 7/lr(ka)'; while at small -:a values, F = I/(2
'ldith the aid of this table. it is possible to extend O'Brien's (ref_ 5) formula for the static conductivity to the case of an unsymmetrical electrolyte; the only modification required is the addition of the term
to O'Brien's expression
(Eq_ 5-35‘1 for conductivity-
Fig. 1. The function F(s:a),related to the integrals (Eq. 7.10), which appears the formula (Eq. 7.7) for the 52 coefficient of the particle dipole strength.
in
terms which can be easily evaluated numerically. To this end, it is shown in Apoendix III that the remaining integrals in Eq. 7.7 can be calculated once the integrals ‘.,e-XE1(2b:a) - eXE1(2rra+x) ,I 0
x
3
exp(-x/q/+:a)dx
have been computed for each distinct eiqenvalue Ai. With the aid of rational approximation for El (qiven on op. 231 of Ref. lo), it is possible to evaluate the above integral on a programmable calculator. The details of the numerical procedure for evaluating the formula of Eq. 7.7 for Cg are described in Appendix III. This procedure for calculating Cg to O(c2) requires much less computer time tnan the exact numerical computation described by Delacey and White (see ref. 2), but the results are, of course, limited to low zeta potentials. In the following section we will compare our results With DeLacey and White's computed values to obtain some idea of the range of validity of our formula for dipole strength. IX.
COMPARISOh OF THE APPROXIMATE FORMULA HITH THE COMPUTED RESULTS In this section we compare values for the dipole coefficient obtained from the
O($) formula with values obtained by the DeLacey and White (ref. 2) computer program for the case of KC1 electrolyte with a rcavalue of 10. For an electrolyte such as KC1 (in which the ions have approximately equal diffusivities), it is shown in Appendix IV that the formula (Eq. 7.7) for the c2
303
term, Ci in the expansion for the dipole coefficient reduces to the much simpler form (Eq. A-30). On combining this formula with the O(c) expression (Eq. 6.15) for CO, we get N E’:*b er, L(ka) CD=-+ IjZjDj c (iGc+4nK"T - kT D j=l (8.11
1
J=I IjZ5 [G(ra, $)
* i(-&
+ e2"aE6 (*,:a)+ m~(,:a)]]}
where 6 is a weighted mean diffusivity. given by N
7
Z;IjDj.
ij = Li
2 x 'k'k k=l and m _
,(kT)* 6:ne26
L and il are given is As mentioned earlier, formulae for the quantities The only term in Eq. 8.1 which requires numerical evaluation is G(-a, L&defined by
!?ef.
5.
,
(8.3)
where g(r) is the decaying solution to the equation , satisfying the boundary condition dg/dr = 0 at r = 1. A method for simplifying _:_ and evaluating integrals in the form of Eq. 8.3 is described in Appendix III'. In the derivation of the formula of Eq. 8.1, it is assumed that the particle dielectric constant c is zero. DeLacey and White use the more realistic value P we can expect discrepancies of order cp/c between the DeLacey and White values of Ci and the values obtained from the approximate formula (Eq. 8.1). At low zeta potentials and high ka values, the O(cp/c) error in the leading term for CO may be comparable with O(r,)and 0(5*) terms for Co. In order to of
2
for
E
and
thus
P’
.z.
-2 _ 'In this case kk = iw/D
304 2 make the G formula accurate in this region, of Eq.3.1 by the quantity:
we rep lace the -1/Z in the formula
"'P -l/2 -i-i3 4 (iLlei-4i;Km)' obtained from the formula of Eq. 5.25 for the dipole strength of an uncharged particle with small dielectric constant. With this modification, the formula of Eq. 8.1 provides a vet-yaccurate representation of DeLacey and White's data at c=25 mV,with errors of one or two percent in the imaginary part of Co and smaller errors in the real part. As expected, the errors in the approximate formula for Co increase with c_ In Figure 2 we show a comparison of the computed and approximate values
-0.30 0
-o-31-
F( [C,)
. x - 0.35 0
f 0.1
I
1
1 0. .5
w/&c2
ii. 5 Fig. 2. A comparison of the computed values of the real part (a) and the imaghnary part (b) of the particle dipole coefficient with -valuesobtained from the O(r ) formula for Ka=lO, with a KC1 electrolyte. The crosses represent the computed values.
305 for the real and imaginary parts of C0 for a zeta value of 50 mV over the frequency range for which computer results are available-
From Figure 2 it can be seen that
the errors in the imaginary part are of the order of ten percent, while the errors in the real part of CO are only a couple of percent.
The errors in the imaginary
part of CO may not be quite as serious as they sound, for since this quantity varies approximately like 5 2 , the formula for the imaginary part of CO could be used
to
compute
5 from measured
values
of
CO with
an error
of
only
five
percent.
The results displayed in Figure 2 are somewhat surprising; the real part of CO at first increases with frequency and over the entire range for which computed results are available, the computed value of this quantity is greater than the value obtained for a static field_
Furthermore, the imaginary part of CO changes sign
at low frequencies and thereafter increases monotonically
in magnitude as frequency
is increased. A more complete picture may be obtained from the curves in Figure 3, which cover a much greater frequency range.
The curves were obtained from the approximate
formula of Eq. 8.1 for CO, with ;.= 25 mV and a .ra value of unity-
As expected,
the imaginary part of CO decays to zero at large frequencies and the real part tends to the value appropriate to an uncharged sphere in a pure the
particle
ica value,
to those
charge the
has a much greater
curves
in Figure
have a similar
2,
except
that
effect
on the
dipole
dielectric.
strength
Although
at this
lower
form in the frequency range
there
is
no small
positive
maximum in the imag-
inary part of Co in this case. X.
THE EFFECT OF SURFACE DISSOCIATION
REACTIONS ON THE DIPOLE STRENGTH
The preceding analysis is based on the assumption that the surface charge density
is
unaffected
by the
application
of
the
external
field.
For those suspensions
in which the surface charge density is determined by the dissociation of ionizable surface groups, this assumption is unlikely to be true in general because the external field alters the concentration of counterions at the surface, leading to a change in the number of ionized groups. In this section, we consider a particle whose charge is determined by the dissociation of a single type of surface group-
Upon dissociating,
the
groups
yield ions of type j which go into solution; for example, if the surface reaction is SOH Z SO- + H+, where S denotes
surface
group,
the
hydrogen
ions
are the type j ions.
306
5=25mv I
b IO
Fig_ 3. The variation of the real part (a) and the imaginary part (b) of the particle dipole coefficient with frequency for a KC1 electrolyte with ::a=l. The curveswre obtained using the O(c2) formula of Eq. 8.1 for the dipole coefficient. The local charge density o is determined from the equation (ref. 12)
where Ks is the dissociation constant and Ns is the number of ionizable sites per The quantity n, in the expression here denotes the density unit area of surface. J of type j ions just beyond the surface. If the frequency of the applied field is such that the equilibrium equation (Eq. 9.1) is instantaneously satisfied, then the variation in charge density 60 due to the variation in local ion density 6nj wil 1 be given by
307
OO6” _ (9.2)
’
where the superscript "0" denotes the equilibrium value of the quantity, in the absence of the applied field. In Section VI we showed that the ion densities around an uncharged particle with
zero
dielectric
constant
are
unaffected
by
the
applied
field.
Thus
,zn. and
J
I-O
are both 0(&z)quantities and, therefore, the perturbation Lc;in charge density given by the formula of Eq. 9.2 is an O(.Q~)quantity. To O(r,)the problem of determining the ion densities and electrical potential is the same as for a particle of
fixed
charge.
The O(c*) problem is, however, different; in this case the boundary conditions associated with Eq. 7.1 for :7.>2 are no longer homogeneous. Rather surprisingly, the variation in the boundary conditions has no effect on the quantity Cg, and thus to 0(c2) the particle has the same dipole strength as a fixed charge particle. This result can in fact be proved for quite general boundary conditions. If finsdenotes the change in surface density of jth ions adsorbed onto the surface (assumed to be an O(s)' quantity),then the boundary conditions for the unknown vector _v2becomes 2 2V _-= at-
2A-v s --dY1 (>:a) dr = -
(9.3)
at
r= 1, where
form of Eq. 7.2, as a linear As usual, vie assume that v2 may be written in combination of the eigenvectors of the matrix A; the dipole coefficient = to determine .-I if to O(sL), we need only compute the leading coefficient f; in this expansion. the quantity vs has a similar expansion, that is, if M S v
= z j=O
zjp
,
(9.4)
.. 308 then the quantity e- y ’ component,for M
in the boundary condition (Eq. 9.3) will have no 5 (0)
since the eigenvalue ?.gis zero. Thtisif the eigenvector expansion (Eq. 9.4) for vs 7.5 - valid, the boundary condition for fz (obtained from Eq. 9.3) is the sameas for a fixed charge particle; and the dipole coefficient C2 (obtained from the 0 7. far-field form of fi) is unaltered by the change in boundary conditions. This result is rather disappointing because, as we mentioned at the outset, it was hoped that measurements of complex conductivity could provide information about surface dissociation reactions. kieshould emphasize, however, that the resuit derived here is not general; it only applies to suspensions of particles with low r:potential, and with only one type of SUrfaCe reaction. The result is not valid for amphoteric or zwitterionic latices in which the surface charge is determined by the competition between different surface reactions, for in this case the charge perturbation 6~ may not be an O(c,')quantity. Although complex conductivity measurements cannot be used in the study of surface reactions for the low r, suspensions considered here, these measurements can still provide more information about the suspension than can be conveniently obtained from electrophoresis and static conductivity measurements. By adjusting the parameters such as r potential, particle radius, ionic strengths and particle volume fraction so that the theoretical variation of dipole strength with frequency agrees with the measured variation, it should be possible to determine these quantities simultaneously. XI.
APPENOIX I - THE MEASUREMENT OF COMPLEX CONDUCTIVITY Consider the problem of determining the current flowing between a pair of electrodes in a suspension due to the application of an alternating potential difference. Since the net current density j is divergence free,
(A. 1) for any closed surface A, where n denotes the unit normal directed outwards from A. If A encloses one of the electrodes, then it must cut across the lead which brings the current to that electrode. In this case we may write the identity (Eq. A-1) in the form (A-2)
I(t)
where
that the
is
portion fact
by the If
the
current
which
cuts
the
integral
that free
the
charge radii
of
mean value
since
(Eq.
Eq. A.2
2.4)
for
where
of
the
this
cross-section
the
surface
A are
we may replace
A lies
the
surface
result, of
A,
minus
we have used
the
outside
in the
flowing
everywhere
j in the
fluctuations
a homogeneous
current
IO = K*
lead
is
dominated
the
much greater
intergral
are
(Eq-
cancelled
out
inhomogeneous
region
by K*$,eiWt,
using
integral suspension.
Thus in this the electrode becomes
onto
than
A-2)
by its the
in
inte-
surrounding the
case
the
constitutive the
formula
suspension
is
= -v+
I,,
given
of
range
fact
these
that
2.4)
potential
follows
interest
here,
and
suspension.
homogeneous
rj satisfies
portion
of
the
Laplace’s
equation:
the
follows
formula
the
to
the
(Eq.
E. has zero
result
is
curl
for
obtained
the constitutive A-4)
relation
we then
for
Eo,
are
thin,
the
by using obtain
Laplace’s
ratio
(Eq.
2.5)
electrodes take
Laplace's
the
boundary
with
difference
potential
in $9 between
if
and electric
difference
field
and are
we have two suspensions
from Eq. A.3 the
the
we may
and
if the
condition the
drop
(Eq.
A.5)
that
.; is
equation
in
to uniform
electrodes
is
equal
suspension
depend
difference.
potential
Hence,
suspension. it
case
applied
around
suspension,
and that
potential
Thus in this
proportional
that
substituting
expression
is small,
in the
electrode
applied
layers
layers
everywhere
on the
the
fact
the second
A.5).
over
each
from the while
divergence;
using
inhomogeneous
potential
be valid
results
of
-:i>
(Eq.
If the
the
electrical
has zero
for
equation
ities,
in the
(A-5)
frequency
the
F.
go in the
.
The first
only
field
field
(A-4)
0
the
the
the
,
.?> =
over
problem,
by
the macroscopic
(Eq.
we must determine field
where
the
(A- 3)
CO * ndA .
As in the static
to
if
for
the
Thus to calculate
to
In deriving
is
A'
and
= Ioeiwt,
I(t)
E.
electrode,
lead.
the microscopic
we may replace
relation of
the
j - fi over
scale,
Furthermore,
electrode,
the
of
curvature
length
gration.
across
to
contribution.
the microscopic
flowing
of
that
the
relating
the
ratio
complex complex
of
in the
independent of
different
the measured
conductivities. conductivities
This
of
the
properties
complex currents result
to measured
of
conductivwill is
be
equivalent
impedances.
310 APPENDIX II - DERIVING THE FORMULAE FOR EQS. 5.24 AND 6.13for so AND a0 XII. Both these expressions are special cases of the following general result. If the column vector
can be written as a linear combination of the eigenvectors of A, viz. M x= -1
yju(j)
(A-6)
j=O then the coefficient y. in this expansion is given by N 2 L-. 5 Djxj = xo+ i= y0 (itic +4:KY‘) .
(A-7)
The result (Eq. 5.24) is obtained by replacing r by the column vector on the left-hand side of Eq. 5.22, with the yj's replaced by sj_ Similarly, replacing _xby thevector y and yj by aj, we obtain the formula (Eq.6.i3) for uo. To prove the general result (Eq- A-7), we begin with the case of an electrolyte in which the ions all have different diffusivities. Clearly, "=-ji is not a root (Eq. of the characteristic equation (Eq.5.16) in this case. From the expression 5.11) for the components of the matrix A, we find that the eigenvectors u (k) cor= responding to eigenvalue xk are scalar multiples of the vector 1 'lXk -'I-xk ??'k '2-"k
(A-8)
INXk ‘N+
Thus to determine the coefficients yj in the expansion (Eq. A-6) for 5, we must solve the simultaneous equations
311
(A-9)
INIl -fN-X
. _ _ 1
where
M is the number of distinct non-zero eigenvalues. The required result (Eq. A-7) is obtained by premultiplying both sides of the above equation by the row vector
(A-10) and using the fact that 'k ----=o, - 'j for each eigen;alue hj, a result obtained by dividing the characteristic equation (Eq- 5.1') bY~~~(Yi- 'j). Only a slight modification is required to obtain the result for an electrolyte in which some of the ions have equal diffusivities.
Suppose, for instance. that
only ions 1 and N have equal diffusivities. Then one root of the characteristic
1_ Form the form (Eq. 5.11) of the equation is y,; we let this eigenvalue be r. A it follows that
matrix
L-11 is a suitable eigenvector. The only modification to the equations ( Eq. A-9) for the coefficients yj in this case is that the second column is replaced by the above column vector u (1)_ As before, the required result (Eq. A.7) is obtained by pre-multipl.yingboth sides of the equations for the yi's by the row vector (Eq. A-10) and using the fact that yl=-fN in this case.
XIII. APPENDIX III - A METHOD FOR EVALUATING THE FORMULA (Eq. 7.7) FOR C; If the electrolyte ions have approximately uniform diffusivities, the formula (Eq. 7.7) for Cg reduces to the simpler form (Eq. 8.1); the procedure for evaluating CE in this case is described in Appendix IV. In this section we describe a method for cdmputing CE for a general electrolyte. The terms in Eq. 7.7 ebich require numerical integration may be written as a single integral, viz.
*L
IjZjDj&
f
j=l Wth the aid of the eigenvector expansion (Eq. 6.8) for vl, we can replace 691 and &ni in this integral by linear combinations of the quantities fi, which satisfy the simple differential equations (Eq. 6.9). For an e7ectrolyte in which no two ionic species have the same diffusivity. the eigenvectors in the expansion (Eq. 6.8) have the form of Eq_ A-8 and the above integral may be rewritten an
0 VY1 - 065,
r.Z.D.fI +Ls f’ 3 J J 0 k&-I k
dV_
(A-11)
. To evaluate the coefficients of the ft's in the integrand, it is necessary to determine the eigenvalues of the matrix 4; thus the first step in the evaluation of c; 7s - the evaluation of the roots Xk of the characteristic equation (Eq. 5.16). The next step involves the computation of the integrals
/ V
VY
1 - v&$'f'dV k
for k=O.l,___ .M_ Wth
(A-12) the aid of the divergence theorem, it can be shown that
OY1 - o&~Of~dV = 4;iEgM(ka)f 2
.(Y1)' - v(&b")2dV
where M(ka) is given by the formula (Eq. 5.33) in Ref. 5, and the second term on the right-hand side of this expression can be eva?uated using the formu7a of Eq_ 7-8. together with the expression (Eq. 6.13) for the coefficient aoTo calculate the integral (Eq. A-12) for non-zero k, we begin by recall. ing that the quantities fi satisfy the differential equations 1 -V&J, 0 , 021 fk- (ka)21 Xkfk = (LEVY subject to the boundary conditions
313 1
afk= 0 ar
at
r
=
1
,
where the o k'~ are the (given
coefficients
in
tile
eigenvector
expansion
of
the
vector
w
in Eq. 6.4).
From the symmetry and linearity of the problem for fi, it follows that this function has the form f' k = gk (r)E -O-r^
(A-13)
-
On substituting this form for ft in Eq. 6.9, we obtain a differential equation Sy the usual Green's function technique, we write gk in terms of for g,(r). integrals involving .Y' - v&i; 0 _ Replacing fi in the integral (Eq_ A-12) by the form of Eq. A.13 and using the Green's function formula for gk, we get
where
and
L = (ka$)-'
,
_e21L 2 - 2L-l-k 'k =
(A-15) L-2
2[2+2Cl+C2]
1 -
(A. 16)
The integrals in the formula of Eq. A.14 can be easily evaluated numerically for each eigenvalue. However, for those who do not have ready access to a computer, we shall briefly describe how the integrals may be evaluated on a programmable hand calculator. Replacing 6.;~' and Y1 in the second integral on the right-hand side of Eq. A-14 by the expressions of Eqs.5.20 and 6.5, we can write this integral in terms of exponential integrals, viz. (A. 17)
where aG = -a3=ka/L, aI=-a4=ka+L and
-1 ,
a2 = -a5 = 1.
The first exponential integral El is tabulated for complex arguments on pp- 249 to 251 of Ref. 10; the higher exponential integrals in Eq. A.17 can be evaluated from El with the aid of the recurrence relation (Eq. 5-l-14) given in that reference_
314
On replacing 9,' and Y1 ln _ the remaining integral on the right-hand side of Eq. A.14 by the forms of Eqs. 5.20 and 6.5, we obtain
where ee-(s
= n,m
-
,-(ia+L-1)x dx dr
r ‘I
_Ir
r"
_
(A-19)
Xrn
Clearly En(2+:a) J
n,O =(raiL-I)
(A-20)
'
and
Jo,l = I, E 1 (.-;a+L-')e
-(qa-L-I) _ EI(2$:a)/(::a-L-l). I
The remaining integrals Jn m can be related to the Jn I 's by repeated use of the > > recurrence relation Jn,m+I =
En+m(2xa) m
(ka+L -1 ) J m n,m
(A-21)
-
Finally, the Jm I's can be computed from JI I by using the recurrence relation 3 , Jn+I,I = 11
- (?:a-L-')Jn,I+El("aCL-l)e-(~a-L-')
-
_
Thus the only integral that need be calculated numerically in the evaluation of the integral (Eq. 14) is JI I, which can be written in the form of a single , integral, namely -caXEI(2~a)-e“aXEI(2~aIx+-Il) JI,I = XIV-
e-x/Ldx
-
X
APPENDIX IV - THE DIPOLE COEFFICIENT OF A PARTICLE IN A SYMNETRIC ELECTROLYTE .WITH IONS OF APPROXIMATELY UNIFORM DIFFUSIVITIES
k!e let
Dj = D + "Df ,
(A-23)
where, as usual, Dj denotes the diffusivity of type j ions and fiis a weighted mean diffusivity. The quantity Dj is a constant and A is a small parameter. Gle seek an expression for the dipole coefficient Co which is correct to O(A)_
To evaluate the formula (Eq. 7.7) for the O(s') correction to the dipole coeffi1 1 cient, we require expressions for the quantities 1+2 and 13n.,obtained from the J solution of the vector equation (Eq. 6.1). For small A, we assume that the unknown . vector v? in _ this equation can be written in the form (A-24) On substituting this form in Eq- 6.1 and equating the O(1) and O(:.,) terms, we obtain the equations v
2
1 - (-a)2A - v 1 = ,yl _ "&)O, , VO =o -0
(A-25)
and 2A ,lzvl - (bra) ' = (ka)2e1- 2: -1 =o - v-1
=
(A-26)
where the vector y is defined in Eq. 6.4, and the matrices froand cl are defined by the formula
thus the components of /IOmay be obtained from the expression (Eq. 5.11) for e by replacing the -tjisby the uniform value * and r.D 0
0
_--
0
0;
0
__-
0
0
Di
e-e
0
0
0
By using the fact that
%
1
(A-27)
N II Ij'j = O ' j=l for a symmetric e?ectro?yte, it can be shown that !I is an eigenvector of the =. corresponding to eigenvalue 3 _ Thus the solution to the equation matrix A DK (Eq. A. 25) for y: is simply ' = g(r)(E -0 - p)w YO
'
where g(r) is the decaying solution to the differential equation
C
(~.28)
316
satisfying the boundary condition @X=0 dr at r = l_ From the formula of Eq. 6.4. it can be seen that the first component of 9 is zero. Thus from Eq_ A-28 it follows that the potential perturbation a-+ ' is zero for an electrolyte in which the ions have uniform diffusivity; the ion densities are altered by the field, but the ions are arranged in such a way that the charge density is everywhere zero. Thus the integral involving 6:) ' in the formula (Eq. 7.7) for CE is an O(a) quantity, and since it is multiplied by a term which is itself of O(n) for a symmetric electrolyte, we may neglect this term in the calculation of Cg to O(L)_ To calculate the terms involving 5ni in Eq. 7.7, we first rewrite these terms in the vector form iV
.yl .v,j;Ola-++-
+0(C2)1
dV ,
(A-29)
where 5 = (O,Z,D,,Z,,D,,--.,Z,ON), and _b= 6(0,z,,z,,._.2,) In formulating this expression we have used the approximate formula (Eq. A-24) for v_l_ Pre-multiplying both sides of Eq. A-26 by b_,and using the formula (Eq_ A-28) for yi, we find that the quantity _S- yi (which appears in the integral in Eq- A-29) satisfies the equation v
N
2
g(r)5 - r j=l
with the usual homogeneous boundary conditions. Thus if we define our mean diffusivity 6 by N
jj
=$ Zj20jEj/~Z~Ii j=l
,
i=l
the sum on the right-hand side of the equation for k - yi
I-y:=0
-
Thus the integral (Eq. A-29) reduces to OY
dV >
is zero and we find
31'7 where we have used the formula (Eq. A-28) for vl_ Substituting the form (Eq. :O 5.20) for &:3'and integrating with respect to r we obtain
2
E$f
IjZ5
j=l
J-g(r) I
$-
(I--$)r'dr _ r
Substituting this expression for the integrals involving c;niin Eq. 7.7 and using the formulae of Eqs. 7.8 and 7.9, we get
xv.
ACKNOWLEDGEMENT I would like to thank Pk. Emma DeLacey and Dr. Lee White for providing the computed complex conductivities used in Section IX. This work has been partially supported by a grant from the Australian-American Educational Foundation. XVI.
LIST OF SYMBOLS
a
particle radius
ai
coefficients defined after the formula of Eq. A.17
a
a row vector defined in Appendix
A
a
50
a square matrix, defined by Eq. 5.11
BOY!&
matrices defined in Appendix IV
b
a row vector defined in Appendix IV
cO C;
particle dipole coefficient, defined by Eq. 3.1
b 'j
IV
closed surface
coefficients in the asymptotic forms of Eq. 5.19 mean
ionic
diffusivity.
defined
by Eq_
8.2
diffusivity of type j ions
P
electric displacement vector
e
the electron charge
_E
electric field strength
CO
amplitude of the mean field strength (see Eq. 2.3)
EO
the magnitude of co
E,(x)
the nth exponential integral of x, defined
in
Section
VI
of Ref. 10.
31s fk J F(+:a)
coefficients
g(r)
a function defined after Eq_ 8.3
i
nett current density, defined by Eq. 2.2
ff
free charge current density
!*
amplitude of average nett current density, defined by Eq_ 2-3
in eigenvector expansion of vector yk
a function defined in Eq_ 7-10
the current flowing between the electrodes; see Appendix IO
the amplitude of I
‘j
a quantity related to ionic concentration,
J
an integral defined by Eq. A.19
n,m k-i
I
defined by Eq. 5.12
the absolute temperature, in energy units
K*
complex conductivity of suspension, defined by Eq_ 2-4
K" E 0, K
complex conductivity of electrolyte
KS
dissociation constant of surface group
static conductivity of electrolyte
a complex quantity defined by Eq. A-15 L(-:a)
a function defined after Eq. 6.14
iii
a mean ionic drag coefficient, defined after Eq_ 8.2
m. J M
non-dimensional
H(r:a)
a function defined in Eq. 5.33 of Ref. 5
ME
ionic drag coefficient, defined after Eq. 4.4
the number of linearly independent eigenvectors of-e, minus one
non-dimensional
electrophoretic
mobility, defined by Eq. 4.6
n. J no-' J
number density of type j ions
n
unit outward normal
N
the number of ionic species in the electrolyte
N(ka)
a function defined by Eq. 5.34 of Ref. 5
Ns
density of type j ions beyond the equilibrium double layer
number density of ionizable sites on the particle surface
P
fluid pressure
P
polarization vector
r
distance from particle center
319 r
position vector
r
unit vector in the direction of r
_U ,(j)
fluid velocity jth eigenvector of the matrix B
VU,? 2 column vectors defined in Sections VI, VII and VIII, respectively _ a!! W
a column vector, defined in Eq_ 6.4
Y
non-dimensional equilibrium potential
'j
valences of type j ions
Greek Symbols coefficients in the eigenvector expansion of the vector M coefficients in the expansion (Eq. 5.22) non-dimensional frequency, defined by Eq. 5.13 a prefix, denoting a small perturbation in the quantity concerned the deviation of the diffusivity of -Lype J_ ions from the mean value D dielectric constant of solvent particle dielectric constant electrical potential on the particle shear plane viscosity of the solvent Debye length of the bulk electrolyte jth eigenvalue of the matrix A electrophoretic mobility of the particle particle surface charge density particle volume fraction macroscopic e7ectrica7 potentia7 electrical potentials in the electrolyte and particle, respectively angular frequency of the applied field The angle brackets around a quantity denote a volume average of that quantity, as in Eq. Z-I_ Unless otherwise noted, the superscripts on a quantity refer to the coefficients in the perturbation expansion for that quantity, as in Eqs. 5-l and 5.2.
320
XVII. REFERENCES R-G:.O'Brien and L-R. White, J. Chem. Sot_ Faraday Trans. II, 74(1978)1607. E.H.B. OeLacey and L-R. blhite,J. Chem. Sot. Faraday Trans. II, to appear. V.N. Shilov and S.S. Dukhin, Kolloidny Zh., 32(1970)117S-S. Dukhin, Surface and Colloid Science, 3(1971)83. R.GJ. O'Brien, J. Colloid Interface Sci., 81(1981)234_ D.A_ Saville, J. Colloid Interface Sci., 72(1979)477. F_ Booth, Proc. Roy- Sot- London Ser. A, 203(1950)5. S-S. Dukhin and 8-V. Derjaguin, Surface and Colloid Science, 7(1974)49. 14-H.Protter and H-F. bleinberger,Maximum Principles in Oifferentiai Equations, Englewood Cliffs, N-J., Prentice-Hall, 1967. 10 M. Abromowitz and I-A_ Stegun, Handbook of Mathematical Functions, Dover, New York, 1972. II R-W_ Hornbeck, Numerical plethods,Quantum, New York, 1975. 12 R-0. James, J-A_ Davis and J-O_ Leckie, J. Colloid Interface Sci., 65(1978) 331.