Comparison of the Zeta Potential with the Diffuse Layer

Potential fromCharge Titration

Phi l Attard,* D. Antel mi , and I . Larson
I an Wark Research I nstitute, University of South Australia, Mawson Lakes, SA 5095 Australia
Received April 23, 1999. I n Final Form: October 21, 1999
The hypernetted chai n (HNC) theory i s used to compare charge ti trati on and zeta potenti al data for
several col l oi ds. I t i s found that the HNC di ffuse l ayer potenti al i s cl oser to the zeta potenti al than that
cal cul ated wi th the tradi ti onal Poi sson-Bol tzmann (PB)approxi mati on, and hence the amount of counteri on
bi ndi ng requi red to bri ng the two i nto conformi ty i s reduced. Further i mprovement i s obtai ned by usi ng
an effecti ve PB di ffuse l ayer potenti al , cal cul ated from both the numeri cal HNC theory and the anal yti c
extended PB approxi mati on. The rati onal e for usi ng the effecti ve PB potenti al i s that the conversi on from
the measured el ectrophoreti c mobi l i ty to the zeta potenti al i nvokes the PB approxi mati on. I n the case of
si l ver i odi de quanti tati ve agreement i s obtai ned wi thout any fi tti ng parameters, whi ch confi rms that no
counteri on bi ndi ng occurs. For the metal oxi des anal yzed (ti tani um di oxi de, al umi num oxi de, and si l i con
di oxi de) the two measurements can onl y be reconci l ed by postul ati ng counteri on bi ndi ng or surface charge
mobi l i ty.
Introduction
The el ectri c doubl e l ayer pl ays a major rol e i n di verse
areas such as adhesi on, sel f-assembl y, fi l trati on, wetti ng,
el ectroki neti cs, and i t i s perhaps the major determi nant
of col l oi dal i nteracti ons and col l oi d stabi l i ty. Consequentl y
the charge and potenti al of a range of surfaces and col l oi dal
parti cl es have been measured i n order to quanti fy the
el ectri c doubl e l ayer for vari ous systems. Surface charge
i s determi ned by charge ti trati on, whereas the so-cal l ed
zeta potenti al i s general l y determi ned by el ectroki neti c
techni ques (e.g., el ectrophoresi s, streami ng potenti al ,
el ectroosmosi s).
1-3
The di sti ncti on between the surface
potenti al and the zeta potenti al i s that the l atter appl i es
at the pl ane of shear, whi ch may be up to 2-3 water
di ameters away from the surface.
4-6
I n the basi cmodel of the el ectri cdoubl e l ayer, the col l oi d
parti cl e i s pl anar, spheri cal , or cyl i ndri cal i n shape, wi th
a smooth and uni forml y charged surface, and the sol vent
i s treated as a di el ectri c conti nuum. The doubl e l ayer i s
di vi ded i nto a compact part adjacent to the surface, and
a di ffuse part. I n the di ffuse part the i ons are mobi l e and
obey the l aws of stati sti cal mechani cs: the counteri ons
(i .e., those of opposi te charge to the surface) are attracted
toward the surface and the co-i ons are repel l ed. Thi s gi ves
ri se to an i on densi ty profi l e that decays to the bul k
concentrati on of the el ectrol yte far from the surface.
Si mi l arl y the el ectrostati c potenti al profi l e decays to zero
i n the bul k. The rate of decay or screeni ng l ength i s taken
to be the Debye l ength.
The compact part of the doubl e l ayer can be defi ned by
the i nner and outer Hel mhol tz pl anes. The i nner Hel m-
hol tz pl ane represents the pl ane of cl osest approach of
nonhydrated i ons to the surface, (0.1-0.2 nm from the
surface), where a physi cal (nonel ectrostati c) adsorpti on
may occur. The outer Hel mhol tz pl ane represents the
cl osest approach of the hydrated i ons, (0.6 nm from the
surface) The rel ati ve di el ectri c permi tti vi ty of the regi on
between the surface and the i nner Hel mhol tz pl ane,
(someti mes cal l ed the Stern l ayer), i s taken to be
r1
5,
and that between the two Hel mhol tz pl anes i s
r2
20,
whi ch are si gni fi cantl y l ess than that of bul k water (
r
)
78)due to surface-i nduced structure.
1-3
Apart from those
adsorbed at the i nner Hel mhol tz pl ane, i ons are excl uded
from both i nner regi ons of the doubl e l ayer, so that there
the potenti al consi sts of twol i near regi mes. The potenti al
at the outer Hel mhol tz pl ane i s cal l ed the di ffuse l ayer
potenti al . I nsofar as the shear pl ane l i es cl ose tothe outer
Hel mhol tz pl ane,
4-6
the di ffuse l ayer potenti al may be
i denti fi ed wi th the zeta potenti al .
There i s a degree of evi dence to show that thi s model
of the el ectri c doubl e l ayer, al though grossl y si mpl i fi ed i n
many respects (i t i gnores, for exampl e, the di screteness
of the surface charge and the mol ecul ar nature of water)
i s suffi ci ent toencompass the essenti al features of real i ty.
Conversel y, al though the compact regi on may appear
somewhat arti fi ci al , i t al so appears necessary for the ful l
descri pti on of the avai l abl e data.
1-3
Much of the evi dence
for the nature of the compact regi on comes from el ectro-
chemi cal measurements, (e.g., the capaci tance of the
mercury el ectrode). Addi ti onal evi dence i s obtai ned from
the compari son of charge ti trati on data wi th measured
zeta potenti al s, whi ch shows that the surface potenti al
cal cul ated from the former i s much greater than the zeta
potenti al . By i ncl udi ng the adsorbed counteri ons and the
capaci tance of the compact regi on the two measurements
may be made consi stent.
Of course some theory i s requi red tocompare the model
wi th measurement, and the earl i est and sti l l most
wi despread approach i s that of Gouy and Chapman,
7,8
(the above model i s often cal l ed the Gouy-Chapman-
Grahame-Stern or GCGS model ). Thi s theory i nvokes
the Poi sson-Bol tzmann (PB) approxi mati on, whi ch i s a
mean fi el d approach that equates the potenti al of mean
* Correspondi ng author.
(1) Hunter, R. J. ZetaPotential inColloidScience; Academi c: London,
1981.
(2) Hunter, R. J. Foundations of Colloid Science; Oxford Uni versi ty
Press: Oxford, 1986; Vol s. I and I I .
(3) Lykl ema, J. Fundamentals of I nterface and Colloid Science;
Academi c: London, 1991; Vol s. I and I I .
(4) Chan, D. Y. C.; Horn, R. G. J . Chem. Phys. 1985, 83, 5311.
(5) I srael achvi l i , J. N. J . Colloid I nterfaceSci. 1986, 110, 263.
(6) Horn, R. G.; Smi th, D. T.; Hal l er, W. Chem. Phys. Lett. 1989, 162,
404.
(7) Gouy, G. J . Phys. Radium1910, 9, 457.
(8) Chapman, D. L. Philos. Mag. 1913, 25, 475.
1542 Langmuir 2000, 16, 1542-1552
10.1021/l a990487t CCC: $19.00 2000 Ameri can Chemi cal Soci ety
Publ i shed on Web 01/08/2000
force of an i on i n the doubl e l ayer tothe mean el ectrostati c
potenti al . As such PB negl ects i oni c correl ati ons, whi ch
ari se from the el ectrostati c i nteracti ons between the i ons,
and from excl uded vol ume effects due toi on si ze. I n recent
years a vari ety of sophi sti cated stati sti cal mechani cal
techni ques have been appl i ed to i nhomogeneous fl ui ds.
These i ncl ude si mul ati ons, i ntegral equati ons, and densi ty
functi onal approaches; the resul ts for the el ectri c doubl e
l ayer have been revi ewed.
9,10
What has emerged from tests
agai nst these more exact approaches i s that for a gi ven
model the PBapproxi mati on can be quanti tati vel y i n error,
even for l ow el ectrol yte concentrati ons and moderate
surface charges.
9-14
I n vi ew of these tests i t i s questi onabl e whether the PB
approxi mati on real l y provi des a rel i abl e justi fi cati on for
any parti cul ar model of the i nner regi on of the doubl e
l ayer. I t coul d wel l be argued that the val ues of any fi tted
parameters are just effecti ve quanti ti es that bri ng the
two experi mental techni ques i nto conformi ty by compen-
sati ng for the shortcomi ngs of the PB approxi mati on. The
i ssue i s qui te i mportant because much of the quanti tati ve
model i ng of the surface chemi stry of col l oi d parti cl es (e.g.,
i on bi ndi ng coeffi ci ents)i s predi cated upon the val i di ty of
the GCGS and more el aborate model s of the doubl e l ayer.
The obvi ous questi on then i s whether the di screpancy
between the charge ti trati on and zeta potenti al measure-
ments can be resol ved, or at l east reduced, i f the di ffuse
l ayer potenti al i s cal cul ated usi ng a more sophi sti cated
theory. I n thi s paper we reanal yze l i terature data for a
sel ecti on of col l oi ds usi ng the hypernetted chai n (HNC)
i ntegral equati on, whi ch accuratel y i ncl udes the excl uded
vol ume and el ectrostati c correl ati on effects negl ected by
the PB approxi mati on. Appl yi ng Occams razor we use
the si mpl est model of the el ectri c doubl e l ayer, whi ch has
coi nci dent Hel mhol tz pl anes and no speci fi c (i .e., non-
el ectrostati c) i on adsorpti on. The rati onal e for thi s ap-
proach i s to see whether the experi mental data can be
reconci l ed by an accurate sol uti on of the si mpl est model
of the el ectri c doubl e l ayer. Whi l e undoubtedl y there are
many physi cal effects negl ected i n thi s model , the present
procedure wi l l al l ow an esti mate of the extent to whi ch
these effects are negl i gi bl e i n practi ce. I f one attri butes
any di screpancy that remai ns after the present anal ysi s
tospeci fi c counteri on adsorpti on, then the present resul ts
shoul d enabl e a more accurate quanti tati ve esti mate of
the bi ndi ng constants than was previ ousl y obtai ned wi th
the PB approxi mati on.
I. Electric Double Layer
A. The Basic Model. The most basi c model of the
el ectri c doubl e l ayer uti l i zes the so-cal l ed restri cted
pri mi ti ve model of the el ectrol yte. Here the sol vent i s taken
to be a di el ectri c conti nuum of rel ati ve permi tti vi ty
r
.
The i ons are model ed as hard spheres al l wi th di ameter
d, wi th a poi nt charge (qat thei r center. By el ectroneu-
tral i ty, the concentrati ons of the cati ons and the ani ons
are equal .
I n the present case we shal l consi der the pl anar doubl e
l ayer, so that the col l oi d parti cl e i s model ed as a smooth
pl anar wal l (semi -i nfi ni te hal f-space)wi th uni form surface
charge densi ty . The wal l i s made of the same di el ectri c
materi al as the sol vent (and so are the i ons)so that i mage
charges are negl ected. The di stance of cl osest approach
of the i ons tothe wal l i s d/2, (coi nci dent Hel mhol tz pl anes),
and the part of the doubl e l ayer between thi s pl ane and
the wal l i s cal l ed the Stern l ayer.
Due to the pl anar geometry the functi ons descri bi ng
the el ectri c doubl e l ayer vary i n the z-di recti on normal to
the wal l . These i ncl ude the mean el ectrostati c potenti al
(z) and the i on number densi ty profi l es F
(
(z). We choose
the coordi nate system such that the pl ane of cl osest
approach to the surface i s at z ) 0, so that the potenti al
drop across the doubl e l ayer i s (0). Thi s i s al so cal l ed the
di ffuse l ayer potenti al and i s often denoted
d
. I n the Stern
l ayer, -d/2 < z < 0, the potenti al i s gi ven by (z) ) (0)
- 4z/, where ) 4
0

r
,
0
bei ng the permi tti vi ty of
free space (SI uni ts). Our sol e concern here i s wi th the
potenti al i n the di ffuse part of the doubl e l ayer. (I t i s
possi bl e tomodel the Stern l ayer wi th a di el ectri cconstant
di fferent tothat of the bul k, but thi s onl y affects the surface
potenti al by modi fi cati on of the above formul a, not the
di ffuse l ayer potenti al i tsel f.)
B. Theories. 1. ThePoisson-Boltzmann Approxima-
tion. The Poi sson-Bol tzmann (PB) approxi mati on was
used by Gouy and Chapman
7,8
todescri be the above model
of the el ectri c doubl e l ayer. The formal l y exact Poi ssons
equati on rel ates the mean el ectrostati c potenti al to the
i on densi ty profi l es
Thi s i s combi ned wi th a Bol tzmann-l i ke expressi on for
the densi ty profi l es,
where k
B
) 1.38 10
-23
J K
-1
i s Bol tzmanns constant, T
i s the absol ute temperature, and F F
+
) F
-
i s the bul k
number densi ty (concentrati on) of the i ons.
I n thi s expressi on the potenti al of mean force of an i on
i s approxi mated by i ts charge ti mes the mean el ectrostati c
potenti al
What the approxi mati on negl ects i s the correl ati ons
between the i ons. For exampl e, due to thei r si ze and thei r
Coul ombi c i nteracti ons, each counteri on i n the doubl e
l ayer pushes asi de other counteri ons, so that the l ocal
potenti al that i t experi ences i s i n general greater than
the mean el ectrostati cpotenti al . That i s, correl ati ons cause
the counteri ons tobe more strongl y attracted tothe surface
than the Bol tzmann approxi mati on al l ows, and conse-
quentl y the surface charge i s actual l y screened more
qui ckl y, at l east i ni ti al l y, than i s predi cted by the mean
fi el d PB approxi mati on.
When the potenti al i s l ow, the exponenti al may be
l i neari zed. The l i near PB equati on i s
wi th sol uti on
(9) Bl um, L. Adv. Chem. Phys. 1990, 78, 171.
(10) Attard, P. Adv. Chem. Phys. 1996, 92, 1.
(11) Bal l one, P.; Pastore, G.; Tosi , M. P. J . Chem. Phys. 1984, 81,
3174.
(12) Kjel l ander, R.; Marcel ja, S. Chem. Phys. Lett. 1986, 127 402.
(13) Henderson, D.; Pl i schke, M. J . Phys. Chem. 1988, 92, 7177.
(14) Tang, Z.; Scri ven, L. E.; Davi s, H. T. J . Chem. Phys. 1992, 97,
9258.
d
2
(z)
dz
2
)
-4q

[F
+
(z) - F
-
(z)] (1)
F
(
(z) ) F e
-q(z)/k
B
T
(2)

(
(z) ) (q (z) (3)
d
2
(z)
dz
2
)
8q
2
F
k
B
T
(z) (4)
(z) ) (0) e
-z
(5)
Zeta Potential Observations fromChargeTitration Langmuir, Vol. 16, No. 4, 2000 1543
Here the Debye parameter
sets the l ength scal e of the doubl e l ayer, whi ch i s the
di stance over whi ch the surface charge i s screened. From
the el ectroneutral i ty of the doubl e l ayer, the rel ati onshi p
between surface charge densi ty and di ffuse l ayer potenti al
i s gi ven by
or,
The nonl i near PB equati on may al so be sol ved. I n thi s
case the el ectrostati c potenti al i s gi ven by
The di ffuse l ayer potenti al gi ven by
where here and throughout we use a di mensi onl ess surface
charge
As i n the l i near case, far from the surface the mean
el ectrostati c potenti al decays exponenti al l y to zero
The second equal i ty defi nes an effecti ve surface charge,
whi ch after some mani pul ati on of eq 10 may be shown to
be gi ven by
15
where agai n s
l PB
)4q
l PB
/k
B
T. The poi nt of thi s effecti ve
surface charge i s that i f thi s
l PB
were used i nstead of the
actual i n the linear PBapproxi mati on togi ve an effecti ve
di ffuse l ayer potenti al , eq 8, then i n the asymptoti cregi me
the effecti ve l i near PB potenti al , eq 5, woul d be i ndi s-
ti ngui shabl e from the actual potenti al predi cted by the
nonlinear PB approxi mati on wi th the actual surface
charge, eqs 9 and 10. I n other words, i f the nonl i near PB
approxi mati on were accurate and the mean el ectrostati c
potenti al were measured far from the surface, then usi ng
the l i near PB approxi mati on to fi t the resul ts woul d gi ve
thi s effecti ve surface charge, not the actual surface charge
of the system. Thi s poi nt hol ds more general l y and wi l l
be expl oi ted extensi vel y bel ow.
Al though our concern i n thi s paper i s wi th the el ec-
trostati c potenti al for an i sol ated doubl e l ayer, i t i s worth
menti oni ng that the val ue of the effecti ve surface charge
al so hol ds for other properti es of the doubl e l ayer. For
exampl e, the pressure between two charge wal l s of
separati on h i n the l i near PB approxi mati on i s gi ven by
That gi ven by the nonl i near PB approxi mati on i n the
asymptoti c regi me i s
where A 16[-1 + 1+s
2
/4]/s. I n other words, i f one
measured the force between twocol l oi ds i n the asymptoti c
regi me i n a system accuratel y descri bed by the nonl i near
PB approxi mati on, then the data woul d be wel l -fi tted by
the l i near PB approxi mati on wi th an effecti ve surface
charge. Si nce the val ue of thi s surface charge i s the same
as woul d have been obtai ned from a measurement of the
mean el ectrostati c potenti al of an i sol ated col l oi d, there
woul d be no contradi cti on between the two experi ments
and one woul d have no reason to bel i eve that thi s was not
the actual surface charge of the system.
2. FormallyExactTheory. The formal l y exact expressi on
for the wal l -i on potenti al of mean force i s
10,16
where, for the wal l -i on pai r, h i s the total correl ati on
functi on, i s the short-range part of the di rect correl ati on
functi on, and d i s the bri dge functi on. The fi rst two are
rel ated by the Ornstei n-Zerni ke equati on
10,16
where r
2
)(z -z)
2
+s
2
, and
(,R
(r)i s the short-range part
of the bul k i on-i on di rect correl ati on functi on. I f both the
wal l -i on bri dge functi on and the short-range part of the
bul k i on-i on di rect correl ati on functi on are negl ected,
the cl osure reduces to the nonl i near PB approxi mati on.
Formal l y exact asymptoti canal ysi s shows that the mean
el ectrostati c potenti al behaves as
10,16,17
Thi s has been del i beratel y cast i n l i near PB form. The
(15) Attard, P. J . Phys. Chem. 1995, 99, 14174.
(16) Attard, P. Phys. Rev. E 1993, 48, 3604.
(17) Enni s, J.; Kjel l ander, R.; Mi tchel l , D. J. J . Chem. Phys. 1995,
102, 975.

8q
2
F
k
B
T
(6)
) -q

dz[F
+
(z) - F
-
(z)]
)

dz
2q
2
F(0)
k
B
T
e
-z
)
2q
2
F
k
B
T
(0) (7)
(0) )
4

(8)
tanh
q(z)
4k
B
T
) tanh
[
q(0)
4k
B
T
]
e
-z
(9)
(0) )
2k
B
T
q
l n[s/2 + 1 + s
2
/4] (10)
s
4q
k
B
T
(11)
(z)
4k
B
T
q
tanh
[
q(0)
4k
B
T
]
e
-z
, z f

l PB
4
e
-z
(12)
s
l PB
)
-8
s
(1 - 1 + s
2
/4) (13)
P(h) )
8

2
e
-h
(14)
P(h) Fk
B
TA
2
e
-h
, h f
)
8

l PB
2
e
-h
(15)

(
(z) ) -k
B
T[h
(
(z) -
(
(z) + d
(
(z)] (16)
h
(
(z) -
(
(z) ) (q(z)/k
B
T +
2

R)(

ds s

dz
(,R
(r)h
R
(z) (17)
(z)
-4
l PB

e
-z
, z f (18)
1544 Langmuir, Vol. 16, No. 4, 2000 Attard et al.
effecti ve surface charge
l PB
may be formal l y expressed i n
terms of
(
16
and the factor v, the effecti ve i on charge q, and the actual
screeni ng l ength
-1
may al sobe expressed i n terms of the
correl ati on functi ons of the bul k el ectrol yte.
10,16,17
I n most cases of practi cal i nterest, to a very good
approxi mati on v ) 1, q ) q, and ) . For exampl e i n
monoval ent el ectrol yte the departure i s l ess than 3% at
0.1 M, and even smal l er for l ower concentrati ons.
15
Hence
the mai n concern i s to obtai n the effecti ve surface charge.
Thi s i s the charge that one woul d obtai n by fi tti ng the
l i near PB approxi mati on to measured potenti al data. As
di scussed i n the precedi ng secti on, thi s same effecti ve
surface charge appears i n the formal l y exact asymptote
for the pressure between two wal l s
and so force measurements and potenti al measurements
anal yzed wi th the l i near PBtheory wi l l be consi stent wi th
each other even though they donot yi el d the actual surface
charge of the system. Fi tti ng the nonl i near PB theory
i nstead wi l l yi el d the effecti ve surface charge
whi ch fol l ows from rearrangement of eq 13. (Here agai n,
s ) 4q/k
B
T).
3. TheHypernetted Chain Approximation. The hyper-
netted chai n (HNC)approxi mati on consi sts of the negl ect
of the bri dge functi on, d
(
(z) )0. Si nce the densi ty profi l e
may be wri tten
and si nce the mean el ectrostati c potenti al may be wri tten
one has twoequati ons tosol ve for twounknown functi ons.
(The bul k i on-i on di rect correl ati on functi on may be
assumed known; i n practi ce i t i s often most conveni ent to
obtai n i t by sol vi ng the HNC equati on for the bul k
el ectrol yte.)The HNCapproxi mati on has been extensi vel y
tested agai nst si mul ati ons and i s known to be accurate
for both bul k el ectrol ytes and for the el ectri c doubl e
l ayer.
9,10
There exi st systemati c methods for i mprovi ng i t
by i ncl udi ng bri dge di agrams of i ncreasi ng compl ex-
i ty,
11,15,18
but thi s wi l l not be done i n the present anal ysi s.
4. TheExtended Poisson-Boltzmann Approximation.
The HNC approxi mati on just di scussed i s cal l ed a si ngl et
theory, because i t yi el ds the si ngl e parti cl e densi ty profi l es.
An al ternati ve approach to the stati sti cal mechani cs of
the doubl e l ayer i s based upon the i nhomogeneous
Ornstei n-Zerni ke equati on
whi ch, when coupl ed wi th a cl osure and a profi l e equati on,
yi el ds the i nhomogeneous pai r correl ati on functi ons.
Numeri cal resul ts have been obtai ned for the el ectri c
doubl e l ayer usi ng the hypernetted chai n cl osure.
12,13,19
An anal yti c sol uti on to thi s equati on has been obtai ned
by Attard et al .
20,21
i n the case of the mean spheri cal cl osure
for zero-si zed i ons
To sol ve the i nhomogeneous Ornstei n-Zerni ke equati on
i t was necessary touse the nonl i near PBi on profi l es, rather
than sol vi ng for them sel f-consi stentl y. Thi s approxi ma-
ti on was justi fi ed by the vari ati onal formul ati on that was
used.
20,21
Thi s theory for the doubl e l ayer i s cal l ed the
extended Poisson-Boltzmann (EPB) approximation. When
appl i ed tothe bul k el ectrol yte i t i s equi val ent toi ncl udi ng
the el ectrostati c correl ati ons at the l evel of the mean
spheri cal model . Li ke that bul k theory the EPB ap-
proxi mati on i s expected to be an exact l i mi ti ng l aw at
l ower el ectrol yte concentrati ons. Tests of the approxi ma-
ti on confi rm i ts accuracy at l ower el ectrol yte coupl i ngs
and surface charge densi ti es.
15,20,21
The EPB approxi ma-
ti on i s useful for i sol ati ng the effects of correl ati ons due
to el ectrostati cs from those of i on si ze, si nce i t i ncl udes
the former but not the l atter, and i t gi ves the pri mary
correcti on to the PB approxi mati on at l ower coupl i ngs.
The EPB approxi mati on has been sol ved for the case
when the wal l s are made of di fferent di el ectri c materi al
tothe sol vent (i .e., i t i ncl udes the effects of i mage charges).
However here we shal l concentrate on the basi c model
where i mage charges are negl ected. I n thi s case the resul t
for the effecti ve surface charge requi red by the l i near PB
theory i s a si mpl e al gebrai c expressi on, namel y
15,21
where
the di mensi onl ess surface charge agai n i s s 4q/,
and I i s gi ven by
22
where z (1 + s
2
/4)
1/2
.
C. Effective Surface Charge. Fi gure 1 shows the
mean el ectrostati c potenti al i n the di ffuse part of the
doubl e l ayer for a 1 mM monoval ent el ectrol yte and a
surface charge of 4 C cm
-2
, (4 nm
2
per uni t charge). The
HNC resul t may be consi dered exact at thi s concentrati on.
The hydrated di ameter of cati ons such as sodi um or
potassi um may be taken to be d ) 0.4 nm, and that of
(18) Attard, P.; Mi kl avi c, S. J. J . Chem. Phys. 1993, 99, 6078.
(19) Kjel l ander, R.; Marcel ja, S. J . Chem. Phys. 1988, 88, 7138.
(20) Attard, P.; Mi tchel l , D. J.; Ni nham, B. W. J . Chem. Phys. 1988,
88, 4987.; 1990, 92, 3248.
(21) Attard, P.; Mi tchel l , D. J.; Ni nham, B. W. J . Chem. Phys. 1988,
89, 4358.
(22) Thi s i s eq 3.23 of ref 21. Equati on 3.20 from whi ch i t deri ves
shoul d have a factor of x mul ti pl yi ng - i n the denomi nator of the
i ntegrand. Thi s i s a typographi cal error onl y and none of the remai ni ng
resul ts i n the paper are affected.

l PB
)
1
2
[ + q{j
+
(i) - j
-
(i)}] (19)
P(h)
8
l PB
2

e
-h
, h f (20)
s
nl
) s
l PB
/(1 - s
l PB
2
/16) (21)
F
(
(z) ) F[1 + h
(
(z)] ) Fe
-
(
(z)/k
B
T
(22)
(z) )
-4qF

dz[h
+
(z) - h
-
(z)](z - z) (23)
h
R,
(r, s) ) c
R,
(r, s) +

dt F

(t
z
)h
R,
(r, t)c
,
(t, s)
(24)
c
R,
(r, s) ) -u
R,
Coul
(r, s)/k
B
T (25)

l PB
)
A
2s
[
1 +
q
2

4
{2I + l n 2}
]
1/2
(26)
A 16(-1 + 1 + s
2
/4)/s
I )
1
2
(
1 +
2z
2
- 3
(2z
2
- 1)
3)
l n 2 +
2 - 2z
3
+ z
2z(2z
2
- 1)
2
-
z
2
- 1
z
(
1 +
2z
2
+ 1
(2z
2
- 1)
3)
tan
-1

z - 1
z + 1
-
1
2
(
1 -
2z
2
- 3
(2z
2
- 1)
3)
l n(z + z
2
) (27)
Zeta Potential Observations fromChargeTitration Langmuir, Vol. 16, No. 4, 2000 1545
ani ons such as chl ori de or ni trate to be d ) 0.2 nm. The
present cal cul ati ons are carri ed out for a symmetri c
el ectrol yte, but one expects the cati ons to domi nate the
properti es of the doubl e l ayer for negati vel y charged
col l oi ds and the ani ons to domi nate for posi ti vel y charged
ones. The HNC gi ves a di ffuse l ayer potenti al of (0) )
146 mV when both i ons have a di ameter of 0.4 nm. Thi s
decreases by about 2% when both i ons have a di ameter
of 0.2 nm. Thi s i s typi cal of the di fference between the
two; the l arger di ameter wi l l be used for al l the cal cul ati ons
reported here.
The l i near and the nonl i near PB approxi mati on are
tested i n Fi gure 1. I t i s cl ear that i n general thi s
approxi mati on overesti mates the mean el ectrostati c po-
tenti al of the doubl e l ayer, wi th the l i near approxi mati on
bei ng the worse of the two. For the gi ven surface charge
(4 C cm
-2
) the l i near PB gi ves a surface potenti al of 556
mV, and the nonl i near PB gi ves 158 mV, whi ch are 280%
and 8% too hi gh, respecti vel y. Usi ng an effecti ve surface
charge densi ty of
l PB
) 0.61 C cm
-2
i n the l i near case,
and
nl
)1.95 C cm
-2
i n the nonl i near case, ensures that
the mean el ectrostati c potenti al predi cted by these two
approxi mati ons agrees wi th the HNC resul t far from the
surface. These are the surface charge densi ti es that woul d
be i nferred by anal yzi ng the experi mental data wi th the
PB approxi mati on i n the asymptoti c regi me. (I n Fi gure
1, the Debye l ength
-1
) 9.60 nm was used for the PB
cal cul ati ons, whereas the HNC screeni ng l ength i s
-1
)
9.61 nm; the di fference between two woul d not be
experi mental l y measurabl e.) These effecti ve surface
charges correspond to effecti ve di ffuse l ayer potenti al s of

l PB
(0))85 mVand
nl
(0))122 mV, whi ch underesti mate
the actual di ffuse l ayer potenti al of (0) ) 146 mV. The
concl usi on to be drawn from Fi gure 1 i s that usi ng an
effecti ve di ffuse l ayer potenti al makes the PB approxi ma-
ti on exact i n the asymptoti c regi me.
Fi gure 2A shows the actual and the effecti ve di ffuse
l ayer potenti al as a functi on of the actual surface charge
densi ty i n 1 mM monoval ent el ectrol yte. The vari ous
potenti al s are defi ned i n the fi gure capti on and al so i n
Tabl e 1. At l ow surface charge densi ti es the PB ap-
proxi mati on i s accurate, and the regi me of val i di ty of the
nonl i near PB approxi mati on i s l arger than that of the
l i near PB approxi mati on. As the surface charge densi ty
i s i ncreased, the rate of i ncrease of the actual di ffuse l ayer
potenti al gi ven by the HNC theory decreases. Thi s
fl atteni ng al sooccurs i n the nonl i near PBapproxi mati on,
but neverthel ess thi s approxi mati on i ncreasi ngl y over-
esti mates the actual di ffuse l ayer potenti al . The effecti ve
di ffuse l ayer potenti al s, whi ch woul d be obtai ned by fi tti ng
the PB approxi mati ons to the exact resul ts for the
el ectrostati c potenti al profi l e i n the asymptoti c regi me,
are markedl y l ess than the actual di ffuse l ayer potenti al
at hi gher surface charges. I n fact, the effecti ve potenti al
drop requi red by the PB approxi mati ons decreases wi th
i ncreasi ng surface charge densi ty at hi gh enough surface
charge densi ti es. I n the 1 mM el ectrol yte, at a surface
Figure1. The l ogari thm of the el ectrostati c potenti al profi l e
i n the di ffuse part of a doubl e l ayer wi th surface charge densi ty
) 4 C cm
2
i n 1 mM monoval ent el ectrol yte. The sol i d curve
i s the HNC predi cti on (d ) 0.4 nm), whi ch gi ves as the di ffuse
l ayer potenti al (0) ) 146 mV. I n (A), the dotted curve i s the
l i near PB predi cti on usi ng the actual surface charge, eqs 5 and
8, whi ch gi ves (0))556 mV. The dashed curve i s al so a l i near
PB predi cti on, eq 5, but wi th an effecti ve surface charge l PB
) 0.61 C cm
2
, eq 19, whi ch gi ves (0) ) 85 mV, eq 8. I n (B),
the nonl i near PBapproxi mati on, eq 9, i s used. The dotted curve
uses the actual surface charge and gi ves (0) )158 mV, eq 10.
The dashed curve uses an effecti ve surface charge nl ) 1.95
C cm
2
, eqs 19 and 21, whi ch gi ves nl (0) ) 122 mV, eq 10.
Figure2. The actual and effecti ve di ffuse l ayer potenti al as
a functi on of the actual surface charge i n monoval ent el ectrol yte.
The sol i d curve wi th tri angl es i s the HNC actual potenti al drop,
the dotted curves are the PB actual potenti al drop, (eqs 8 and
10), the sol i d curves are the PB effecti ve potenti al drop usi ng
the HNC, (eqs 8 and 19, and eqs 10, 19, and 21), and the dashed
curves are the PB effecti ve potenti al drop usi ng the EPB, (eqs
8 and 25, and eqs 10, 25, and 21). I n each case the nonl i near
PB approxi mati on l i es cl oser to the HNC actual potenti al drop
than does the correspondi ng l i near PB approxi mati on. Part A
i s for 1 mM, and part B i s for 10 mM.
1546 Langmuir, Vol. 16, No. 4, 2000 Attard et al.
charge densi ty of ) 4 C cm
-2
, (4 nm
2
per uni t surface
charge), the actual di ffuse l ayer potenti al i s (0) )146.1
mV, whereas the effecti ve di ffuse l ayer potenti al s are

l PB
(0) ) 85.17 mV and
nl
(0) ) 121.65 mV. For an
el ectrol yte wi th i ons of di ameter d ) 0.2 nm the resul ts
are 85.11 mVand 121.44 mV, respecti vel y. Soat al l surface
charge densi ti es Fi gure 2 confi rms the resul t establ i shed
i n Fi gure 1, that the effecti ve di ffuse l ayer potenti al i n
both the l i near and the nonl i near PB approxi mati ons
underesti mates the actual di ffuse l ayer potenti al .
Fi gure 2A al so compares the EPB approxi mati on for
the effecti ve di ffuse l ayer potenti al s wi th that gi ven by
the HNC theory. (Note that the EPBi s notbei ng compared
to the actual di ffuse l ayer potenti al .) I t may be seen that
the EPB i s qui te accurate for al l surface charge densi ti es
i n predi cti ng the effecti ve potenti al requi red by the l i near
PB approxi mati on, and up to a surface charge densi ty of
about )2 C cm
-2
for the nonl i near PB approxi mati on.
Even beyond thi s i t remai ns a useful approxi mati on up
toqui te hi gh surface charge densi ti es. Hence as an anal yti c
resul t i t provi des a useful compl ement to the effecti ve
potenti al deri ved from the HNC numeri cal approach. At
) 4 C cm
-2
, the EPB resul t for the effecti ve di ffuse
l ayer potenti al s are
l PB
(0))89.30 mVand
nl
(0))136.48
mV.
Fi gure 2B shows a si mi l ar compari son at the hi gher
monoval ent el ectrol yte concentrati on of 10 mM. I n thi s
case the decrease i n the HNC effecti ve di ffuse l ayer
potenti al wi th i ncreasi ng surface charge densi ty i s more
marked, parti cul arl y for the nonl i near case. The fact that
the HNC effecti ve potenti al i s nonmonotoni c means that
the charge cannot be unambi guousl y deduced from the
measured potenti al wi thout addi ti onal i nformati on. The
EPB remai ns quanti tati vel y accurate up to about 5 C
cm
-2
, and qual i tati vel y confi rms the HNC resul t that the
effecti ve di ffuse l ayer potenti al i s l ess than the actual
di ffuse l ayer potenti al for both the l i near and the nonl i near
PB approxi mati ons.
II. Analysis
A. Experimental Techniques. 1. ChargeTitration.
Charge ti trati on consi sts of addi ng a known amount of
the potenti al determi ni ng i ons to a sol uti on of known
col l oi d vol ume fracti on.
1-3
Measurement of the free
concentrati on of i ons yi el ds the amount of i ons adsorbed
to the col l oi d. Provi ded that one knows the surface area
of the col l oi ds, whi ch i n some cases can be probl emati c,
3,23
thi s gi ves a rel ati ve measure of the surface charge densi ty.
To obtai n the absol ute surface charge densi ty measure-
ments are carri ed out at several background concentra-
ti ons of i ndi fferent (i .e., nonbi ndi ng) el ectrol yte. The
uni que poi nt of i ntersecti on of the ti trati on curves cor-
responds to the poi nt of zero charge. For the si l ver i odi de
col l oi ds anal yzed bel ow the charge i s measured as a
functi on of the concentrati on of si l ver i ons i n a background
of KNO
3
or NaNO
3
. For the metal oxi des the pH determi nes
the potenti al and the typi cal i ndi fferent el ectrol yte i s KNO
3
or NaCl .
2. Electrophoresis. Col l oi d parti cl es i n a di l ute di spersi on
move i n response to an appl i ed el ectri c fi el d. Thi s
phenomenon i s known as el ectrophoresi s, and measure-
ment of the mobi l i ty (the vel oci ty di vi ded by the fi el d
strength) gi ves the zeta potenti al , whi ch i s bel i eved to be
rel ated to the di ffuse l ayer potenti al .
1-3
Consi der a si ngl e spheri cal col l oi d of radi us aand central
poi nt charge Q, whi ch i s equi val ent to a uni form surface
charge densi ty of ) Q/4a
2
. The force acti ng on i t due
to an external el ectri c fi el d of magni tude E i s F
E
) QE.
For a steady vel oci ty vthi s i s bal anced by the Stokes drag
F

)6av, where i s the shear vi scosi ty of water. Hence

the el ectrophoreti c mobi l i ty i s
Taki ng the zeta potenti al to be equal to the surface
potenti al , Coul ombs l aw gi ves (a) ) Q/a, (recal l
that ) 4
0

r
). Hence one sees that the mobi l i ty i s
proporti onal to the zeta potenti al
Thi s i s known as the Huckel -Onsager equati on.
The above negl ected the doubl e l ayer compl etel y. For
the case of a spheri cal charged col l oi d, the l i near PB
approxi mati on gi ves the surface potenti al as (a))Q/a(1
+ a), so that the above becomes
The Huckel -Onsager equati on i s therefore expected to
be val i d for l ow potenti al s, l ow el ectrol yte concentrati ons
( f 0), and smal l parti cl es (a f 0).
The opposi te to thi s l ast l i mi t i s a pl anar wal l , and we
now consi der a sl i t compri si ng two paral l el pl anar wal l s
of surface charge densi ty , separated by h, and of breadth
B . h, and l ength L . h. For conveni ence i t i s easi er to
i magi ne that the wal l s are fi xed and that the counteri ons
and water fl ow i n response to an appl i ed el ectri c fi el d,
(the phenomenon known as el ectroosmosi s), but i t ought
to be evi dent that i n the l i mi t hf thi s rel ati ve moti on
of fl ui d and wal l ought be the same as the el ectrophoreti c
moti on of a l arge parti cl e. Wi th the i on profi l es i n the z
di recti on normal to the surfaces, and the appl i ed fi el d
and vel oci ti es i n the x di recti on paral l el to the surfaces,
the water vel oci ty profi l e may be denoted v(z). The
tangenti al force per uni t vol ume acti ng on the water due
to shear i s
(23) Overbeek, J. Th. G. I n Colloid Science; Kruyt, H. R., Ed.;
El sevi er: Amsterdam, 1952; Vol . 1, p 115.
Table 1. The Diffuse Layer Potentials
symbol name equati ons
l PB l i near Poi sson-Bol tzmann 8
nl nonl i near Poi sson-Bol tzmann 10
HNC hypernetted chai n 23

l PB
HNC
effecti ve l i near Poi sson-Bol tzmann usi ng the hypernetted chai n 8,19

nl
HNC
effecti ve nonl i near Poi sson-Bol tzmann usi ng the hypernetted chai n 10,19,21

l PB
EPB
effecti ve l i near Poi sson-Bol tzmann usi ng the extended Poi sson-Bol tzmann 8,25

nl
EPB
effecti ve nonl i near Poi sson-Bol tzmann usi ng the extended Poi sson-Bol tzmann 10,21,25

E

v
E
)
Q
6a
(28)

E
)

6
(29)

E
)

6
(1 + a) (30)
F

2
v(z)
z
2
(31)
Zeta Potential Observations fromChargeTitration Langmuir, Vol. 16, No. 4, 2000 1547
The forces acti ng on the i ons compri se the el ectri c force
due to the appl i ed fi el d, and the force due to the Stokes
drag, whi ch ari ses from the moti on of the i ons rel ati ve to
the water
where F
(
(z) and v
(
(z) are the densi ty and vel oci ty profi l es
of the i ons. Note that thi s formul a i s PB-l i ke i n the sense
that onl y el ectrostati cs and drag contri bute tothe potenti al
of mean force; i on correl ati ons are negl ected. Further, i t
i s the bare appl i ed fi el d that enters here rather than the
mean fi el d; i on rearrangement and i nduced streami ng
potenti al s, whi ch woul d decrease the moti ve force, are
al so negl ected.
I n the steady state the force on the i on vani shes, whi ch
gi ves the Stokes drag i n terms of the appl i ed fi el d. The
drag force on the i on i s equal and opposi te to the force
exerted by the i ons on the water, so that the total force
on the water i s
Thi s al sovani shes i n the steady state, and usi ng Poi ssons
equati on gi ves
I ntegrati ng twi ce, one has
The coeffi ci ent of the l i near term vani shes (for a symmetri c
system). Thi s may be eval uated at the mi dpl ane (where
for h .
-1
the potenti al i s zero and the water vel oci ty
may be denoted v)and at the shear pl ane (where the water
vel oci ty vani shes, and where the potenti al i s cal l ed the
zeta potenti al ). Subtracti ng the two to el i mi nate c
0
one
obtai ns
Thi s i s known as the Hel mhol tz-Smol uchowski equati on.
The pl ane of shear i s expected to be cl ose to the outer
Hel mhol tz pl ane, and so one expects the zeta potenti al to
be approxi matel y equal to the di ffuse l ayer potenti al .
The zeta potenti al deduced from the mobi l i ty usi ng the
Hel mhol tz-Smol uchowski equati on i s two-thi rds that
deduced from the Huckel -Onsager equati on. There has
been a hi story of refi nements to the Huckel -Onsager
approxi mati on for spheri cal col l oi ds, begi nni ng wi th the
work of Henry
24
and i ncl udi ng the more recent and
rel ati vel y sophi sti cated works of Overbeek and Wi erse-
ma
25,26
OBri en and Whi te,
27
and Dukhi n.
28
(For di scussi on
of these and other approaches, see refs 1-3, 28.)I n general
these approaches express the el ectrophoreti c mobi l i ty as
where f(a) f 1, a f 0, whi ch i s the Huckel -Onsager
l i mi t, and f(a) f 3/2, a f , whi ch i s the Hel mhol tz-
Smol uchowski l i mi t. At i ntermedi ate val ues of a, the
val ue of f(a) di ffers between the di fferent approaches.
Whi l e the experi mental l y determi ned quanti ty i s the
mobi l i ty, i t i s the zeta potenti al that i s usual l y reported,
and the above shows that thi s depends on the parti cul ar
approxi mati on used to rel ate the two. I n turn thi s
determi nes the speci fi c compari son that i s to be made
wi th the charge ti trati on data. I f the rel ati onshi p between
the mobi l i ty and the zeta potenti al were exact, then one
shoul d compare the zeta potenti al wi th the doubl e l ayer
potenti al cal cul ated exactl y from the ti trated surface
charge. Normal l y however the rel ati onshi p i s based upon
the PB approxi mati on, and i t therefore fol l ows that the
appropri ate compari son i s wi th the doubl e l ayer potenti al
cal cul ated from the effecti ve surface charge. I n turn, the
effecti ve surface charge wi l l depend on whether the
expressi on for the el ectrophoreti c mobi l i ty has used the
l i near or the nonl i near PB approxi mati on.
Henry
24
used the l i near PB approxi mati on i n deri vi ng
hi s formul a for the zeta potenti al , so arguabl y i n thi s case
the l i near PB effecti ve doubl e l ayer potenti al i s the
appropri ate poi nt of compari son. The nonl i near PB
approxi mati on, (among other si mpl i fyi ng approxi mati ons),
was used i n the l ater anal yses,
25-28
and so i n these cases
the nonl i near PB effecti ve doubl e l ayer potenti al shoul d
arguabl y be used. The si tuati on i s not compl etel y cl ear
because of these addi ti onal approxi mati ons, and al so
because the contri buti on from the asymptoti c regi me to
the fi nal formul a i s not necessari l y domi nant.
B. Comparison. Thi s secti on presents measured and
predi cted zeta potenti al s for several col l oi ds. As far as
possi bl e the ti trati on and el ectrophoreti c data taken from
the l i terature emanates from the same l aboratory i n each
case. Thi s restri cti on was observed because i t i s known
that the surface properti es of col l oi dal parti cl es can be
sensi ti ve to the mode of preparati on and hi story of the
sampl e, and by ensuri ng that the two types of measure-
ments are performed on i denti cal sampl es one can have
some confi dence i n compari ng the two.
The zeta potenti al i n general was taken as gi ven by the
authors of the ori gi nal papers. The conversi on from the
measured mobi l i ty obvi ousl y depends on the parti cul ar
approxi mati on used by the ori gi nal authors. I n most cases
thi s was that of Wi ersema and Overbeek
25,26
or that of
OBri en and Whi te.
27
I n some cases the Hel mhol tz-
Smol uchowski formul a, eq 36, and the Henry formul a
24
were used. These experi mental zeta potenti al s are com-
pared wi th vari ous theoreti cal potenti al s as a functi on of
concentrati on of the potenti al determi ni ng i on. The surface
charge densi ty used i n the theoreti cal cal cul ati ons was
obtai ned from a smooth i nterpol ati on of the charge
ti trati on data. I n some cases thi s i nvol ved extrapol ati on
beyond the range of the measured data to enabl e com-
pari son wi th the ful l mobi l i ty data, and a degree of cauti on
shoul d be exerci sed i n these regi mes.
The theoreti cal potenti al s that are shown are the HNC
and PBdi ffuse l ayer potenti al s, and the PBeffecti ve di ffuse
l ayer potenti al s cal cul ated wi th the HNC theory. For the
(24) Henry, D. C. Proc. R. Soc. (London) Ser. A 1931, 133, 106.
(25) Wi ersema P. H.; Loeb, A. L.; Overbeek, J. Th. G. J . Colloid
I nterfaceSci. 1966, 22, 78.
(26) Overbeek, J. Th. G.; Wi ersema P. H. I n Electrophoresis; Bi er,
M., Ed.; Academi c: New York, 1967; Vol . 2, Ch. 1.
(27) OBri en, R. W.; Whi te, L. R. J . Chem. Soc., FaradayI I 1978, 74,
1607.
(28) Dukhi n, S. S.; Derjagui n, B. V. I n Surfaceand Colloid Science;
Mati jevi c, E., Ed.; Wi l ey: New York, 1974; Vol . 7, pp 52-72.
F
(
) (qF
(
(z)E - 6a
(
F
(
(z)[v
(
(z) - v(z)] (32)
F )

2
v(z)
z
2
+ 6{a
+
F
+
(z)[v
+
(z) - v(z)] +
a
-
F
-
(z)[v
-
(z) - v(z)]} )

2
v(z)
z
2
+ q[F
+
(z) - F
-
(z)]E (33)

2
v(z)
z
2
)
E
4

2
(z)
z
2
(34)
(z) )
4
E
v(z) + c
1
z + c
0
(35)

v
E
)

4
(36)
)

6
f(a) (37)
1548 Langmuir, Vol. 16, No. 4, 2000 Attard et al.
actual and effecti ve PB quanti ti es, both the l i near and
the nonl i near approxi mati ons were exami ned. As shown
by Fi gure 2, the effecti ve PB di ffuse l ayer potenti al
cal cul ated wi th the EPB approxi mati on i s qui te cl ose to
the HNC resul t for thi s quanti ty, except at hi gher surface
charges. The EPB curves woul d i n most cases overl ap
wi th the HNC effecti ve potenti al curves i n the fol l owi ng
fi gures, and for thi s reason the EPBresul ts are not shown.
1. Silver I odide. Si l ver i odi de (AgI )i s perhaps the col l oi d
that has been studi ed quanti tati vel y for the l ongest peri od
of ti me, and for thi s reason i s qui te wel l characteri zed. I t
has a poi nt of zero charge at pAg )5.44, (pI )10.56), and,
i n addi ti on to the el ectroki neti c measurements anal yzed
here, vari ous el ectrochemi cal measurements that uti l i ze
i ts conducti vi ty have been made.
29
Fi gure 3 shows the
zeta potenti al of AgI i n 1 mM KNO
3
. I n thi s case the zeta
potenti al data and the charge ti trati on data are from two
di fferent sources,
23,30
but as an i oni ccrystal i n equi l i bri um
wi th the el ectrol yte i n thi s case one expects l i ttl e vari ati on
between data from di fferent l aboratori es. The surface
charge densi ty obtai ned from the ti trati ons depends on
the total surface area of the col l oi d, and there i s some
uncertai nty i n the esti mati on of thi s quanti ty.
3,23
I t may be seen from Fi gure 3 that at l ow potenti al s al l
the theoreti cal curves coi nci de wi th each other and wi th
the measured zeta potenti al . At hi gher potenti al s the
l i near PBapproxi mati on greatl y overesti mates the di ffuse
l ayer potenti al , si nce the HNC resul t may be regarded as
the exact di ffuse l ayer potenti al for thi s parti cul ar model .
The experi mental zeta potenti al was obtai ned wi th the
Henry approxi mati on,
30
and thi s i s compared wi th the
OBri en-Whi te predi cti on at the extremi ti es,
1
where the
di screpancy between the two i s the greatest. For the case
of the Henry approxi mati on, whi ch i nvokes the l i near PB
approxi mati on, the zeta potenti al l i es cl ose to both the
HNC di ffuse l ayer potenti al , and al so the effecti ve l i near
PB di ffuse l ayer potenti al . I n vi ew of the experi mental
scatter i t i s di ffi cul t to say wi th ful l confi dence whi ch of
the two the measured zeta potenti al actual l y corresponds
to, but focusi ng on the two extreme poi nts, (pAg ) 3 and
pAg )8), suggests that i t i s the effecti ve l i near PB di ffuse
l ayer potenti al .
For the nonl i near PB approxi mati on, the effecti ve
potenti al l i es cl oser to the HNC di ffuse l ayer potenti al
than i n the l i near case. Usi ng the OBri en-Whi te formul a
shi fts the zeta potenti al i n the same di recti on, whi ch
rei nforces the i dea that the zeta potenti al gi ves the
effecti ve PB di ffuse l ayer potenti al .
Fi gure 4 shows the potenti al resul ts for AgI i n 12 mM
KNO
3
. (The charge used for the theoreti cal cal cul ati ons
comes from ti trati on data for 10 mM KNO
3
, as shown i n
the i nset.) The zeta potenti al was deri ved from the
measured mobi l i ty usi ng the Hel mhol tz-Smol uchowski
equati on.
23,31
Agai n at l ow potenti al s al l the theoreti cal
potenti al s coi nci de, (the measured mobi l i ty has a rel ati vel y
l arge error i n thi s regi me), and at thi s hi gher concentrati on
they di verge at hi gher potenti al s somewhat more markedl y
than i n Fi gure 3. At i ntermedi ate potenti al s the zeta
potenti al i s cl oser to the HNC di ffuse l ayer potenti al than
the effecti ve l i near PB potenti al , whi l e the converse hol ds
at hi gher potenti al s. The qual i tati ve saturati on of both
the zeta potenti al and the effecti ve potenti al s, whi ch does
not occur to nearl y the same extent for the HNC di ffuse
l ayer potenti al , agai n suggests that i t i s the effecti ve
potenti al that shoul d be compared to the zeta potenti al .
I t shoul d be stressed that no fi tti ng parameters have
been used here i n compari ng the theoreti cal and experi -
mental potenti al s. The quanti ty used i n the HNC and PB
cal cul ati ons i s the ti trated charge. I n the l i terature the
zeta potenti al i s usual l y compared wi th the di ffuse l ayer
potenti al cal cul ated from the CGCS model and the PB
approxi mati on usi ng a cal cul ated surface charge. The
l atter i s obtai ned from the Nernst expressi on, whi ch
i nvol ves several fi tti ng parameters. I n the case of AgI
where i t i s not necessary to i nvoke counteri on bi ndi ng,
onl y one parameter i s used, and thi s i s the capaci tance
of the Stern l ayer.
1-3
2. TitaniumDioxide. Ti tani um di oxi de, Ti O
2
, i s the fi rst
of the metal oxi des that we shal l anal yze. I n these cases
the surface charge ari ses from the protonati on and
deprotonati on of the metal hydroxyl group
I n general i ncreasi ng the pH deprotonates the surface
causi ng a negati ve surface charge, whereas l oweri ng the
(29) Lykl ema, J.; Overbeek, J. Th. G. J . Colloid I nterfaceSci. 1961,
16, 595.
(30) Osseo-Assare, K. A. M. Sc. Thesi s, Uni versi ty of Cal i forni a,
Berkel ey, 1972.; Osseo-Assare, K. A.; Fuerstenau, D. W.; Ottewi l l , R.
H. Am. Chem. Soc. SymposiumNo. 8, 1978, p 63. (31) Troel stra, S. A.; Kruyt, H. R. Kolloid Z. 1942, 101, 182.
Figure3. The zeta potenti al measured for AgI i n 1 mM KNO3
compared to the di ffuse l ayer potenti al cal cul ated from the
ti trated charge. The ci rcl es are measured data from Osseo-
Asara et al .
30
(sol i d ci rcl es Henry formul a; open ci rcl es OBri en-
Whi te formul a
1
). The sol i d curve i s the HNC, the dashed curves
are the nonl i near PB, and the dotted curves are the l i near PB.
I n the case of the PB both the actual and the effecti ve di ffuse
l ayer potenti al i s shown, whi ch have l arger and smal l er
magni tudes than the HNC, respecti vel y. The HNC theory i s
used to cal cul ate the effecti ve potenti al s. The i nset shows the
measured ti trated charge (ci rcl es, ref 23) and the smooth fi t
gi vi ng the surface charge used i n the theoreti cal cal cul ati ons.
Figure4. Si l ver i odi de i n 12 mM KNO3, (al l curves as i n Fi gure
3). The measured zeta potenti al i s from Troel stra and Kruyt
31
and the ti trated charge i s from Overbeek
23
at an el ectrol yte
concentrati on of 10 mM.
-M
+
-OH
2
H -M-OH H -M-O
-
(38)
Zeta Potential Observations fromChargeTitration Langmuir, Vol. 16, No. 4, 2000 1549
pH gi ves a posi ti ve surface charge. The poi nt of zerocharge
for these parti cul ar sampl es of Ti O
2
occurs at a pH of
5.8.
32,33
These surfaces carry a hi gher surface charge than
does AgI .
Fi gure 5 shows that the zeta potenti al l i es cl oser to the
effecti ve di ffuse l ayer potenti al than i t does to the actual
di ffuse l ayer potenti al . The l i near effecti ve potenti al i s i n
better agreement than i s the nonl i near effecti ve potenti al ,
even though the zeta potenti al was obtai ned usi ng the
formul a of Wi ersema and Overbeek,
25,26
whi ch i n part
uti l i zes the nonl i near PB approxi mati on.
I n Fi gure 6 the concentrati on of i ndi fferent el ectrol yte
has been i ncreased to 10 mM, whi ch agai n magni fi es the
di screpancy between the vari ous approaches. Strangel y
enough, at smal l potenti al s where one expects the theori es
to be most accurate the experi mental zeta potenti al
appears tohave a di fferent sl ope tothe theoreti cal curves.
I t i s agai n the case that the zeta potenti al l i es cl oser to
the l i near rather than the nonl i near PBeffecti ve potenti al .
3. Aluminum Oxide. Fi gure 7 shows zeta potenti al
measurements for -Al
2
O
3
from two l aboratori es,
34,35
and
one can see that apart from a smal l di screpancy i n the
poi nt of zero charge (8.1 vs 8.5), there i s good agreement
between the twomethods. The di screpancy for the ti trated
surface charge i s a l i ttl e greater, (not shown), and we
have chosen touse the data of Sprycha
34
for the theoreti cal
cal cul ati ons. Thi s does not greatl y affect the resul ts
because the cal cul ated potenti al i s not a sensi ti ve functi on
of the surface charge densi ty at hi gher surface charge
densi ti es.
For the rel ati vel y l ow surface charge densi ti es that
devel op i n thi s case, there i s l i ttl e di fference between the
cal cul ated potenti al s, parti cul arl y i n the case of the
nonl i near PB approxi mati on. The zeta potenti al , deri ved
from the measured mobi l i ty usi ng the formul a of Wi erse-
ma,
25
i s most cl osel y gi ven by the l i near PBeffecti ve di ffuse
l ayer potenti al .
At the hi gher concentrati on of background el ectrol yte,
(10 mM NaCl , Fi gure 8), the surface charge densi ty i s
hi gher, but the same concl usi on hol ds, namel y that the
zeta potenti al probabl y corresponds to the l i near PB
effecti ve di ffuse l ayer potenti al rather than ei ther the exact
di ffuse l ayer potenti al or the nonl i near PBeffecti ve di ffuse
l ayer potenti al . Overal l , the agreement i s l ess convi nci ng
for al umi num oxi de than i t was for ti tani um di oxi de or
si l ver i odi de.
4. Silicon Dioxide. The fi nal col l oi d that we have
anal yzed i s si l i con di oxi de (Fi gures 9 and 10). Charac-
teri zati on of Si O
2
has hi stori cal l y been i mportant, i n part
because i t i s sowi despread i n the terrestri al sphere. There
i s a great deal of evi dence that the surface properti es of
Si O
2
are sensi ti ve to the preparati on and cl eani ng
procedure,
1
and soi n thi s case i t was consi dered i mportant
to use mobi l i ty and ti trati on data from the same source,
even though thi s restri cted the amount of data that we
coul d anal yze. I n the case of 10 mM background el ectrol yte
(Fi gure 9)there i s a rel ati vel y hi gh surface charge densi ty
at hi gher pH and a correspondi ngl y l arge separati on
between the vari ous theori es. Of al l the theori es, the zeta
potenti al corresponds most cl osel y to the effecti ve di ffuse
l ayer potenti al i n the l i near PB approxi mati on.
(32) Wi ese, G. R. Ph.D. Thesi s, Uni versi ty of Mel bourne, 1973. Wi ese,
G. R.; Heal y, T. W. J . Colloid I nterfaceSci. 1975, 51, 427.
(33) Yates, D. E. Ph.D. Thesi s, Uni versi ty of Mel bourne, 1975. Yates,
D. E.; Heal y, T. W. J . Chem. Soc., Faraday I 1980, 76, 9.
(34) Sprycha, R. J . Colloid I nterfaceSci. 1989, 127, 1.
(35) Huang, C. P. Ph.D. Thesi s, Harvard Uni versi ty, 1971. Huang,
C. P.; Stumm, W. J . Colloid I nterfaceSci. 1973, 43, 409.
Figure5. Ti tani um di oxi de i n 1 mM KNO3, (al l curves as i n
Fi gure 3). The el ectrophoresi s data i s from Wi ese
32
and the
ti trati on data i s from Yates.
33
Figure6. Ti tani um di oxi de i n 10 mM KNO3, (al l curves as i n
Fi gure 3, and the experi mental data i s from the same sources
as i n Fi gure 5).
Figure 7. Al umi num oxi de i n 1 mM NaCl , (al l curves as i n
Fi gure 3). The sol i d ci rcl es represent data from Sprycha
34
and
the open ci rcl es are from Huang.
35
Figure8. Al umi num oxi de i n 10 mM NaCl , (al l curves as i n
Fi gure 3). The experi mental data i s from Sprycha.
34
1550 Langmuir, Vol. 16, No. 4, 2000 Attard et al.
Thi s concl usi on i s rei nforced by the data for 100 mM
background el ectrol yte, (Fi gure 10). Unfortunatel y there
are onl y twoexperi mental data poi nts for the zeta potenti al
i n thi s case, and the most that one can concl ude i s that
for thi s very hi gh surface charge densi ty, the zeta potenti al
has more or l ess saturated at 26 mV.
36
Of al l the theoreti cal
quanti ti es onl y the effecti ve di ffuse l ayer potenti al s show
a comparabl e saturati on. One concl udes that at l east
qual i tati vel y the zeta potenti al i s gi ven by the effecti ve
PB di ffuse l ayer potenti al s.
III. Discussion and Conclusion
At hi gh surface charge densi ti es, (hi gh concentrati ons
of the potenti al determi ni ng i on), the predi cted potenti al s
tend to saturate, as does the measured zeta potenti al . I n
Tabl e 2 the saturated val ue of the zeta potenti al i s
compared to the HNC di ffuse l ayer potenti al and to the
effecti ve l i near PB potenti al at the extremi ti es. I t can be
seen that the HNC di ffuse l ayer potenti al overesti mates
the zeta potenti al by a factor of 2-4. (From the fi gures,
the bare PBdi ffuse l ayer potenti al s are i n obvi ousl y worse
agreement wi th the zeta potenti al .) There i s fai rl y good
agreement between the HNC and the EPB val ues for the
effecti ve l i near PB potenti al , whi ch confi rms the val ue of
thi s anal yti c approxi mati on. The effecti ve potenti al i s
general l y wi thi n a factor of 2 of the zeta potenti al , and i n
the case of si l ver i odi de the agreement may be cal l ed
quanti tati ve. Two questi ons ari se: why i s the effecti ve
l i near PB potenti al the most appropri ate one to compare
to the zeta potenti al , and why i s the agreement i n the
case of si l ver i odi de so much better than for the metal
oxi des.
As menti oned i n the text, the conversi on from the
measured mobi l i ty to the zeta potenti al requi res some
theory, and al l theory to date has i nvoked the PB
approxi mati on. Tests of the PBapproxi mati on have shown
that i n general i t overesti mates the mean el ectrostati c
potenti al , but that usi ng an effecti ve surface charge,
(equi val entl y an effecti ve di ffuse l ayer potenti al ), secures
agreement i n the asymptoti c regi me (Fi gure 1). The
requi red effecti ve charge (potenti al ) i s a wel l -defi ned
functi on of the actual surface charge, (Fi gure 2), and so
one expects that the theori es for the zeta potenti al wi l l
yi el d the exact mobi l i ty provi ded that the effecti ve surface
charge i s used. To put i t another way, the zeta potenti al
i s the effecti ve potenti al that corresponds tothe measured
mobi l i ty, and i ts val ue depends on the approxi mati on that
i s i nvoked.
Most of the theori es for the zeta potenti al i nvoke the
nonl i near PB approxi mati on, and so i t i s surpri si ng that
i t i s the effecti ve l i near PBpotenti al that i s the one cl osest
to the measured zeta potenti al . Thi s coul d wel l be
coi nci dental because qui te general l y the l i near PBeffecti ve
potenti al al ways has a l ower magni tude than the nonl i near
PBeffecti ve potenti al . Hence the good agreement obtai ned
for the l i near PB coul d be compensati ng for the si mpl i fi ed
nature of the basi c model . Any physi cal effect negl ected
i n the model that tended to decrease the di ffuse l ayer
potenti al woul d parti al l y compensate for the errors i n the
l i near PB approxi mati on.
One such possi bi l i ty i s counteri on bi ndi ng i nsi de the
shear pl ane. I t has been proposed that a gel or porous
l ayer contai ni ng counteri ons exi sts on the surface of some
metal oxi des,
37
so that the ti tratabl e charge i s greater
than the net charge that contri butes to the zeta
potenti al .
38-40
There i s evi dence both for and agai nst such
a gel l ayer, and there are confl i cti ng vi ews on i ts
exi stence.
1-3
A si mpl er possi bi l i ty i s embodi ed i n the GCGS model ,
whi ch al l ows for counteri ons to adsorb at the i nner
Hel mhol tz pl ane.
41-43
Ori gi nal l y proposed on the basi s of
the PB approxi mati on, the expl anati on woul d obvi ousl y
work equal l y wel l , (i f not better due to the l ower amount
of bi ndi ng requi red), for the present HNC cal cul ati ons.
One coul d concl ude that a degree of such bi ndi ng must be
occurri ng for the metal oxi des, and i t i s thi s that causes
the magni tude of the zeta potenti al to be l ess than the
effecti ve di ffuse l ayer potenti al . Conversel y, i n the case
(36) James, R. O. Adv. Ceram. 1987, 21, 349.
(37) Lykl ema, J. Croatica. Chem. Acta 1971, 43, 249.
(38) Lykl ema, J. J . Electroanal. Chem. 1968, 18, 341.
(39) Wri ght, H. J. L.; Hunter, R. J. Aust. J . Chem. 1973, 26, 1191.
(40) Perram, J. W. J . Chem. Soc., Faraday I I 1973, 7, 993.
(41) Levi ne, S.; Smi th, A. L. Discuss. Faraday Soc. 1971, 52, 290.
(42) Perrram, J. W.; Hunter, R. J.; Wri ght, H. J. L. Aust. J . Chem.
1974, 27, 461.
(43) Yates, D. E.; Levi ne, S.; Heal y, T. W. J . Chem. Soc., FaradayI ,
1974, 70, 1807.
Figure 9. Si l i con di oxi de i n 10 mM monoval ent el ectrol yte,
(al l curves as i n Fi gure 3). The experi mental data i s from
James.
36
Figure10. Si l i con di oxi de i n 100 mM monoval ent el ectrol yte,
(al l curves as i n Fi gure 3; experi mental data as i n the precedi ng
fi gure).
Table 2. The Zeta Potential Compared to the Actual and
Effective Linear PB Diffuse Layer Potentials at the
Extremities Where They Appear to Saturate
col l oi d conc (mM) (mV) (0) (mV)
l PB
HNC
(0) (mV)
l PB
EPB
(0) (mV)
AgI 1 72 100 75 76
-68 -140 -85 -88
12 -80 -105 -72 -75
Ti O
2 1 -65 -160 -80 -92
10 -42 -125 -74 -79
Al 2O3 1 40 110 80 80
-40 -100 -75 -76
10 28 100 70 72
-25 -130 -74 -77
Si O
2 10 -60 -130 -70 -79
100 -26 -100 -57 -48
Zeta Potential Observations fromChargeTitration Langmuir, Vol. 16, No. 4, 2000 1551
of si l ver i odi de one coul d concl ude that nosuch counteri on
bi ndi ng occurs.
Thi s expl anati on i n terms of counteri on bi ndi ng i s
certai nl y the conventi onal one,
1-3
but there are si gni fi cant
probl ems wi th i t. One di ffi cul ty wi th the i on bi ndi ng
pi cture i s that an i ndi fferent el ectrol yte i s used i n the
experi ments, and by defi ni ti on these i ons cannot have a
speci fi c i nteracti on wi th the surface. I f they di d, then the
mobi l i ty or ti trati on curves for di fferent el ectrol yte
concentrati ons woul d not al l i ntersect at a uni que poi nt,
(the poi nt of zero charge for ti trati on, and the coi nci dent
i soel ectri c poi nt for el ectroki neti cs). Furthermore, the
vi scosi ty of water confi ned between mol ecul arl y smooth
surfaces shows i t to have the same val ue as i n bul k for
separati ons as smal l as 1 nm,
4-6
and the fact that one can
actual l y bri ng such surfaces i nto i nti mate mol ecul ar
contact i ndi cates that the water cannot be i mmobi l i zed.
I n other words, the evi dence i s agai nst both adsorpti on
and i mmobi l i zati on of counteri ons.
The present anal ysi s may suggest an al ternati ve
expl anati on for the l ower than expected zeta potenti al .
The fact that for si l ver i odi de the zeta potenti al i s equal
to the effecti ve potenti al , whereas for the metal oxi des i t
i s si gni fi cantl y l ess than the effecti ve potenti al , suggests
that i t i s the nature of the surface charge that i s i mportant.
Si l ver i odi de i s an i oni c crystal , and the surface charge
ari ses from an excess of the i oni cal l y bonded consti tuent
i ons. The surface charge for the metal oxi des ari ses from
the hydrol ysi s of the metal atom to M-OH, and the
consequent bi ndi ng or di ssol uti on of H
+
i ons. We suggest
that whereas for AgI the surface charges are i mmobi l i zed
by the strong i oni c bond, for the metal oxi des the surface
charges may have a degree of mobi l i ty al ong the surface.
The fact that chemi cal equi l i bri um i s establ i shed for the
surface groups shows that the charges cannot be com-
pl etel y fi xed, and nodoubt there i s a rel ati onshi p between
thei r mobi l i ty and thei r bi ndi ng constants. I t fol l ows that
an external l y appl i ed fi el d woul d cause the surface charges
to move rel ati ve to the sol i d, whi ch woul d reduce the
mobi l i ty of the col l oi d compared tothe fi xed surface charge
case. (Thi s putati ve surface charge mobi l i ty i s di sti nct
from the conducti vi ty of the sol i d, and al so from the so-
cal l ed surface conducti vi ty, whi ch i s i n fact the conducti vi ty
of the di ffuse part of the doubl e l ayer.
1-3
)
Obvi ousl y the present resul ts donot on thei r own resol ve
the di screpancy between the di ffuse l ayer potenti al
cal cul ated from the ti trated charge and the zeta potenti al
for the metal oxi des. I t i s al so possi bl e that the theory
rel ati ng the zeta potenti al s tothe measured mobi l i ty needs
further refi nement. Detai l ed model i ng of the compact
regi on of the doubl e l ayer, (charge and counteri on adsorp-
ti on, capaci tance, etc.), woul d be requi red for a more
defi ni ti ve pi cture, and the present HNCapproach provi des
a sound quanti tati ve basi s for such anal ysi s. The present
resul ts al ready i ndi cate that i n the case of si l ver i odi de
no counteri on bi ndi ng occurs, and i n the case of the metal
oxi des i f counteri on bi ndi ng does occur, then i t i s much
l ess than has been previ ousl y thought on the basi s of the
PB approxi mati on. I n the future we i ntend to model the
dependence of the surface charge densi ty upon the
concentrati on of the potenti al determi ni ng i ons, whi ch
we hope wi l l further cl ari fy the i ssue.
LA990487T
1552 Langmuir, Vol. 16, No. 4, 2000 Attard et al.