Published September, 1992
Proton Surface-Charge Density in Soils with Structural
and pH-Dependent Charge
Sharon J. Anderson* and Garrison Sposito
ABSTRACT
Nearly all soils contain both variable- and permanent-charge sur-
face sites. The objective of this research was to quantify the variable-
charge properties of three soils that contain both permanent structural
charge (primarily on external siloxane surfaces of mica and collapsed
vermiculite) and variable (pH-dependent) charge. When the three soils
were suspended in 0.02 or 0.05 mol kg~' NaCI solutions at eight equi-
librium pH values between 4 and 8, the net ion adsorption was ap-
proximately two times greater at pH 8 than at pH 4 and decreased
with increasing ionic strength. The point of zero net proton charge
(PZNPC), which was calculated from proton titration data, net NaCI
adsorption data, and previously measured values of the structural
charge density, ranged from 4 to 5. When the soils were suspended
in 0.05 mol kg ' CsCI solutions at equilibrium pH values between 4
and 9, the PZNPC was at least 1 pH unit lower than for NaCI sus-
pensions. Cesium adsorption was greater than Na adsorption at low
pH and was less pH dependent; the net proton surface-charge density
also was less pH dependent in CsCI suspensions. The three soils con-
tain <10 g kg-' organic C, yet the proton-titration, ion-adsorption,
and PZNPC data suggest that organic matter is extremely imortant
in the surface chemistry of the three soils.
M OST SOILS contain both variable-charge surface
functional groups (ionizable groups bound to
inorganic or organic adsorbents) and permanent-charge
surface functional groups (created by ionic substitu-
tion within the crystal structure of minerals). The gen-
eral charge-balance equation for particles bearing both
variable and structural charge can be expressed as the
sum of constituent charge densities (Sposito, 1984, p.
79; 1989):
<r0 + OH + <rls + obs + Ob = O [1]
where cr0 is the net permanent or structural charge
density, crH is the net proton surface-charge density
S.J. Anderson, Dep. of Crop and Soil Science, Michigan State
Univ., East Lansing, MI 48824-1325; and G. Sposito, Dep. of
Soil Science, Univ. of California, Berkeley, CA 94720. Contri-
bution from the Dep. of Soil and Environmental Science, Univ.
of California, Riverside. Received 2 Aug. 1991. "Corresponding
author.
Published in Soil Sci. Soc. Am. J. 56:1437-1443 (1992).
(usually representing pH-dependent or variable charge),
o-IS is the net surface charge brought by ions that are
bound in inner-sphere surface complexes, and eros is
the similarly defined density of outer-sphere surface-
complex charge. The dissociated or diffuse-layer sur-
face-charge density, <rD, is the net surface charge con-
tributed by adsorbed aqueous ions whose mobility is
comparable to that of ions in bulk solution.
In general, when charge densities are expressed on
a unit-mass basis, the sum of inner-sphere, outer-sphere,
and dissociated surface-charge densities is equal to the
net adsorbed-ion charge, (n+ n_), where n+ and
n_ are the adsorbed positive and negative charges,
respectively:
<*is + OQS + ob = (n+ - n.). [2]
Combination of Eq. [1] and [2] yields
[3]
Ob + OH = -(+ - "-),
which provides a basis for relating ion-adsorption
measurements to structural and variable surface-charge
densities. Equation [3] shows that the net adsorbed-
ion charge may be pH dependent in soils that contain
a mixture of permanent-charge clay minerals and var-
iable-charge adsorbents such as organic matter and
hydrous metal oxides.
Points of zero charge are pH values at which one
or more of the individual components of the surface-
charge density specified in Eq. [1] are equal to zero
(see Sposito [1984, 1989] for detailed discussion of
the relationships among the various points of zero
charge). The PZNPC, which is the pH value where
OH = O, depends on the concentrations of ionizable
surface functional groups and on the composition of
Abbreviations: PZNPC, point of zero net proton charge; PZSE,
point of zero salt effect; cr0, structural surface-charge density; <TH,
net proton surface-charge density; cr,s, inner-sphere complex sur-
face-charge density; cr^, outer-sphere complex surface-charge
density; o-D, diffuse-layer surface-charge density; o-H-titr, apparent
proton surface-charge density; o-H-tltr(PZNPC), value of <r Hlilr at
the PZNPC; AES, atomic emission spectrophotometry; CEC, ca-
tion-exchange capacity.
 1438 SOIL SCI. SOC. AM. J., VOL. 56, SEPTEMBER-OCTOBER 1992
the solution phase. The PZNPC is the most important
point of zero charge for soils that contain both per-
manent and variable charge because it is the only point
of zero charge in which the contribution of crH is con-
sidered separately from that of cr0. In spite of its im-
portance, the PZNPC has not been measured previously
for adsorbents comprising significant structural charge.
Many other investigators (e.g., Uehara and Gillman,
1980) have assumed that the PZNPC simply is equal
to the point of zero salt effect (PZSE, the pH value at
which the curves of apparent crH vs. pH for different
ionic strengths intersect [Parker et al., 1979]). Equal-
ity between the PZNPC and the PZSE, however, re-
quires the special conditions that the net adsorbed-ion
charge at the PZNPC be independent of ionic strength
and that CTO be equal to zero (Sposito, 1989), condi-
tions that seldom have been verified in published stud-
ies.
In order to measure either PZNPC or crH when both
variable-charge and permanent-charge adsorbents are
present, it is necessary first to measure the permanent
charge density accessible to adsorptive ions (effective
CTO) by an independent method. One such method is
the Cs adsorption method described by Anderson and
Sposito (1991). In that method, a Cs-saturated adsor-
bent is dried to promote the formation of inner-sphere
surface complexes, then is washed once with a dilute
solution of LiCl. Because the structural-charge sites
on the basal planes of 2:1 layer-type clay minerals
have a very low affinity for Li, Li preferentially dis-
places Cs from variable-charge sites and leaves Cs
adsorbed on structural-charge surface sites. Finally,
Cs adsorbed on structural-charge sites is extracted and
quantified by reacting the adsorbent several times with
NH4OAc solution. After the effective CTO is measured,
the PZNPC then may be found using charge balance:
the PZNPC is the pH value at which the net adsorption
of an index cation and anion (e.g., Na + and Cl~) is
equal in magnitude, but opposite in sign, to the ef-
fective CTO:
(M - L) = -Ob = PZNPC) [4]
where M and n L represent surface excesses of the
index cation and anion, respectively. The numerical
values of nM and n L may differ from n+ and _ in
Eq. [3], just as the effective a-0 may differ from the
total <r0. For example, on the basal-plane surfaces of
mica, n+ would include all interlayer K + ions, which
are not readily displaced by other cations, whereas
M, as defined here, would represent only the number
of moles of index cation adsorbed on structural-charge
and ionizable surface sites that were accessible to it
during an adsorption experiment. Equation [4] is true
if the structural charge accessible to the index cation
and anion is equal numerically to the structural charge
accessible to the cation used to determine effective
O-Q.
The main objective of this research is to show that
PZNPC and o-H of soils containing both permanent and
variable charge can be measured with the help of the
charge-balance constraints in Eq. [3] and [4], pro-
vided that effective cr0 has been measured previously.
The net proton surface charge and adsorbed-ion charge
of three representative agricultural soils were mea-
sured in NaCl and CsCl suspensions at pH 4 to 9.
Sodium chloride was chosen as an "indifferent" elec-
trolyte that does not form inner-sphere surface com-
plexes, whereas Cs+, which is used to measure effective
cr0 (Anderson and Sposito, 1991), forms inner-sphere
surface complexes with siloxane surface groups (Bar-
shad, 1948; Posner and Quirk, 1964; Edwards et al.,
1965).
MATERIALS AND METHODS
Ion Adsorption and Proton Titration
The < 106-fj.m fractions of the Fallbrook soil (fine-loamy,
mixed, thermic Typic Haploxeralf), Bonsall soil (fine,
montmorillonitic, thermic Natric Palexeralf), and Manead-
ero soil (collected in Mexico; tentatively classified as fine-
loamy, mixed, thermic Typic Xerochrept) were separated
by dry sieving. The primary criteria used to select these
soils were the presence of permanent-charge minerals and
very low carbonate content. The latter criterion was nec-
essary because reactions between H ions and carbonate
minerals would greatly complicate the interpretation of pro-
ton titration data. In addition, the three soil samples selected
had the following characteristics: (i) near-neutral pH of a
saturated soil paste, and (ii) low organic matter and Fe
oxide content, but with differences among the soils. Se-
lected properties of the soil separates are reported in Table
1; additional chemical and physical properties have been
presented by Anderson and Sposito (1991).
To saturate the soils with NaCl or CsCl in preparation
for the ion adsorption and proton titration experiments de-
scribed below, 0.02-kg samples of air-dried soil were sus-
pended in 0.2 kg of 0.5 mol kg-1, pH 6.5 MCI solutions
(M = Na or Cs) and shaken for 30 min. The suspensions
were centrifuged and the supernatant solutions were dis-
carded. The same shaking and centrifugation procedure was
used to react the soils twice with a 0.1 mol kg-1 MCI
solution and once with 1a solution of 0.05 mol kg-1 MCI.
After the 0.05 mol kg- MCI supernatant solution was dis-
carded, the soil pastes that remained in the centrifuge bot-
tles were mixed thoroughly.
The surface charge and ion adsorption characteristics of
subsamples of the NaCl- and CsCl-saturated soil pastes were
investigated using a discontinuous proton titration and ion
adsorption procedure: triplicate subsamples of MCl-satu-
rated soil paste containing 0.35 g oven-dry soil were added
to weighed, 40-mL polyallomer centrifuge tubes. The moist
pastes then were suspended in 0.035 kg of NaCl or CsCl
solution that contained HC1, NaOH, or CsOH to adjust the
equilibrium supernatant solution pH to a desired value be-
tween 4 and 9. All solutions were prepared on a mass, not
volume, basis, and the ionic strength of each solution was
exactly 0.02 or 0.05 mol kg-1. The soil suspensions were
equilibrated for 1 h in a shaking, temperature-controlled
Eberbach water bath at 25 1 C. The samples then were
centrifuged at 27 000 x g for 20 min. The supernatant
solutions were removed using a Pasteur pipette attached to
a 30-mL plastic syringe and were transferred to plastic vials.
The mass of entrained solution remaining in each centrifuge
tube was determined by weighing each tube and its contents
(soil plus entrained solution). The pH value of each super-
natant solution was measured immediately with a Ross
combination pH electrode (Orion Research, Boston, MA)
that was calibrated before each set of 12 pH measurements
in either pH 4 and 7 or pH 7 and 10 buffer solutions. The
supernatant solutions were saved for Cs or Na analysis.
Because KC1 often leaks from a pH electrode during mea-
surements, 10-mL solution aliquots were reserved for Cl
analysis prior to pH measurements.
 ANDERSON & SPOSITO: PROTON SURFACE-CHARGE DENSITY IN SOILS 1439

Table 1. Properties of the <0.106-jum soil fractions.

Clay
Soil Effective tr0t mineralogy! PH,P Clay Organic C FeCBD1I CaC03#
mmolc kg-1
Fallbrook -64.5 0.2 V,K 6.66 0.07 320 10 7.0 0.7 5.8 0.3 1.8 0.7
Bonsall -43 1 V,K 6.23 0.01 240 10 4.7 0.5 4.6 0.3 0.0
Maneadero -47.4 0.4 V,K,M 7.21 0.04 220 10 8.3 0.4 11.3 0.9 2.5 0.1
t Effective structural surface-charge density. Method of Anderson and Sposito (1991), based on Cs+ adsorption.
t V = repotassified vermiculite with 1.0-nm d spacing when Mg saturated; K = kaolinite; M = mica.
pH of saturated paste.
f Results for <2-mm fraction; Fe extractable by citrate-dithionite-bicarbonate method of Mehra and Jackson (1960).
# Results for <2-mm fraction; NH<OAc extraction at pH 5.5 (Holtzclaw et al., 1986).

During the discontinuous proton titrations, pure electro- n

lyte solutions (blanks) at each ionic strength and chosen
initial H + or OH" concentration were treated in exactly the
same manner as the soil suspensions. Other authors (Sigg
and Stumm, 1981; Schulthess and Sparks, 1986) have pro-
posed methods to correct for proton consumption by dis-
solved species released during weathering reactions. Back-
titration methods (Schulthess and Sparks, 1986), however,
require the arbitrary choice of a back-titration endpoint,
whereas filtration methods (Sigg and Stumm, 1981) raise
the question of what size of colloids are considered to be
part of the solid phase, and exactly what cutoff between
dissolved and colloidal material should be used. In our study,
no matrix corrections were made. The final pH of each
electrolyte blank solution was measured after the pH of the
corresponding supernatant solutions. Atmospheric CO2 was
not controlled in samples or in blanks. Adsorbed Na and
Cs were displaced from the soils by three (for Na) or four
(for Cs) extractions with 0.03 kg of 1 mol kg-1 NH4OAc
(Anderson and Sposito, 1991).
The total concentrations of Na and Cs in supernatant and
extractant solutions were measured using AES on a Perkin-
Elmer Model 5000 atomic absorption spectrophotometer
(Perkin-Elmer, Norwalk, CT) in an air-acetylene flame.
The relative precision of the AES measurements ranged
from 0.2 to 1.4%. Standards were prepared from Fisher
reference standards (Fisher Scientific, Springfield, NJ) in
appropriate background solutions. All standards and sample
dilutions1 were prepared on a mass basis and contained 5 g
La kg" to suppress ionization. Chloride was measured in
NaCl supernatant and blank solutions using an Aminco Cl
titrator (American Instrument Co., Silver Spring, MD).
Chloride adsorption was not measured during the Cs ad-
sorption experiments because Edwards et al. (1965) re-
ported that no net Cl adsorption occurs on Cs-saturated,
collapsed micaceous minerals, which were the only per-
manent-charge minerals in the Fallbrook, Bonsall, and Ma-
neadero soil samples (Table 1). Thus, it was assumed that
nc, was negligible for these CsCl-saturated soils.
C\ ~ '"ci.sample) [6]
where Msoln here refers to the mass, per unit soil mass, of
NaCl solution equilibrated with the soil. The difference
between the Cl concentrations 1 of blank and sample solu-
tions (0.2 to 0.5 mmol kg- solution) was similar to the
precision of the measurements (0.3 mmol kg-1 solution).
Consequently, the relative error in ncl was quite large. The
calculated values of ncl were not corrected for the small
amount of Cl entrained in the soil paste at the beginning of
an adsorption experiment because the difference between
corrected and uncorrected concentrations (0-0.06 mmol kg-1
solution) was nearly an order of magnitude smaller than the
uncertainty in the measurements. Chloride could not be
measured directly in the NH4OAc extracts (pH 7) because
the Cl titration method required low solution pH.
The apparent proton surface-charge density measured
during a titration experiment [crH-,i,r(pH,I)] is simply the
difference between the final H* or OH- concentration of
a sample and that of an identical electrolyte solution with
no added soil. To simplify the notation, crH>titr(pH,I) will
be abbreviated hereafter as a-Hftitr, unless a particular pH
value (e.g., the PZNPC) is specified. Values of crH titr at
each pH and ionic strength were calculated with the equa-
tion:
H.d* = Msoln[(10-PH'blank - 10-P
H blank
(10 P ' 10 -POH,sample\y (-]~\
where Mso,n has the same meaning as in Eq. [6]. Single-
ion activity coefficients, y+, were calculated with the Dav-
ies equation (Davies, 1962).
The net proton surface-charge density at a specified pH
value and ionic strength, o-H(pH,I), hereafter abbreviated
as o-H, differs from o-H,titr by the value of OH>titr at the PZNPC,
which is the pH where crH = 0:
= o-H,titr - <TH,titr(PZNPC). [8]

Data Analysis
The relative surface excess (,) of a cationic species i
was calculated with the equation:
= ,-,tot(NH4) (i = Na or Cs)
where n,:>tot(NH4) is the total number of moles of Na or Cs
extracted by NH4OAc from the soil per unit soil mass, Mentr
is the mass of entrained NaCl or CsCl solution per unit soil
mass, and mi is the molinity (mol kg-1 solution) of Na or
Cs in the NaCl or CsCl supernatant solutions.
The relative surface excess of Cl (ncl) was calculated by
measuring the Cl concentrations in blank NaCl solutions
and in NaCl supernatant solutions that had been equilibrated
with the soil samples:
The PZNPC and OH,,itr(PZNPC) were calculated using two
different procedures. In one method, crHttitr(PZNPC) was
calculated with the charge-balance constraint in Eq. [3] and
the definition of erH in Eq. [8]:
[5] = nM - n L o-H,titr <r [9]
The value of <TH-titr(PZNPC) calculated with Eq. [9] is the
average value across all pH for a given ionic strength and
adsorptive electrolyte. Then the PZNPC for each ionic
strength in NaCl and CsCl was found by fitting erH titr data
to an equation of the form
<7H,.itr = A + BX + CX2 [lOa]
where X is the equilibrium supernatant solution pH and A,
B, and C are adjustable parameters. A second-order poly-
 1440 SOIL SCI. SOC. AM. J., VOL. 56, SEPTEMBER-OCTOBER 1992

Table 2. Sodium and Cl adsorption (nNa and nc,), apparent proton surface-charge density ("H,mr)? and net proton surface-charge
density (crH) on the Fallbrook and Bonsall soil separates.
Ionic Equilibrium (UN. - He, +
strength PH N. *c, H,lllr OH.IH, + b)t <r
tnoi K mmolc kg-' soil
Fallbrook
0.05 4.20 O.Olt 67 9* 4 8t 48.1 0.3^ 46 12 4 121
4.59 0.02 69 2 -3 4 35.4 0.2 44 4 -9 4
5.53 0.04 82 1 -12 4 16.20 0.04 46 4 -28 4
5.84 0.04 86 5 -5 3 12.40 0.05 39 6 -32 6
6.52 0.07 93 7 -9 3 2.50 0.01 41 8 -42 8
7.04 0.04 96 3 -8 3 0.03 0.01 39 4 -44 4
7.30 0.08 101 1 -16 2 -0.08 0.01 53 2 -44 2
7.84 0.08 107 4 -8 4 -1.60 0.08 48 6 -46 6
0.02 4.30 0.04 73 1 -8 3 56.1 0.6 72 3 -6 3
4.76 0.03 84 1 NA# 37.4 0.2 NA -25 6
5.35 0.02 93 1 -10 1 25.0 0.1 63 1 -38 1
5.82 0.02 100 1 -8 4 17.02 0.06 61 4 -46 4
6.51 0.01 110 1 -14 1 7.50 0.01 67 1 -55 1
7.08 0.07 114 2 -4 1 0.24 0.01 54 2 -62 2
7.56 0.09 118 1 -4 4 -0.17 0.01 58 4 -62 4
7.91 0.09 121 2 -7 2 -0.97 0.01 63 3 -64 3
Bonsall
0.05 4.00 0.02 25 2 -9 2 43.9 0.5 34 8 20 8
4.37 0.01 32 3 -1 3 33.5 0.2 24 8 9 8
5.22 0.01 36 3 -11 2 15.88 0.06 19 8 -8 8
5.61 0.02 45 2 -10 6 12.32 0.06 24 9 -12 9
6.37 0.01 51 1 -10 7 2.50 0.01 20 10 -22 10
6.78 0.06 54 2 -6 5 0.01 0.01 17 9 -24 9
7.14 0.09 61 2 -8 3 -0.11 0.01 26 8 -24 8
7.7 0.1 63 1 -9 3 -1.78 0.01 27 8 -26 8
0.02 4.05 0.01 38 1 -3 2 51.3 0.1 49 7 77
4.47 0.01 48 3 NA# 35.71 0.05 NA# -9 7
5.05 0.01 58 2 -6.8 0.1 24.48 0.08 46 1 -20 7
5.55 0.01 62 1 -5 2 16.9 0.1 41 7 -27 7
6.20 0.02 74 2 -11 8 7.46 0.02 49 11 -37 11
6.81 0.02 77 2 -3 2 0.23 0.01 37 8 -44 8
7.40 0.03 80 1 -5.6 0.5 -0.19 0.01 43 7 -44 7
7.6 0.1 81 1 -8 3 -1.00 0.01 45 8 -45 8
t ob values (effective structural surface-charge density) taken from Table 1.
i Mean SD of triplicate measurements.
Uncertainty = (SD2Na + SD2CI + + SD2^)"2
11 Uncertainty in o-H is assumed equal to uncertainty in a-nfia at the point of zero net proton charge (PZNPC) or in BC, + rH + ), whichever is
larger.
# Data not available.

nomial providedz the best fit of the data across the pH range
investigated (R > 0.96), but the equations should not be
used to extrapolate to pH values much higher or lower than
those in the original data set. The resulting equation then
was solved for X under the condition crHititr = crH titr(PZNPC)
(i.e., crn = 0) to obtain the PZNPC. The PZNPC calculated
in this manner was dependent on both proton titration and
ion adsorption data, and on the calculated value of
<7H,titr(PZNPC).
In the other method used to calculate PZNPC and
o- H , titr (PZNPC), the PZNPC was calculated before
fH.titXPZNPC) and was independent of the proton titration
data. The PZNPC was found by fitting net ion adsorption
data (M nL) at each ionic strength to the equation
(nM - = a + + 8X2
calculated with Eq. [8], using the Eq. [9] values of
erH)tjtr(PZNPC). The value of crH thus obtained may be in-
terpreted as a property of variable-charge sites, provided
that crH>tjtr represents only those protons complexed by ion-
izable surface functional groups. Potentiometric titrations
should discriminate between protons that form inner-sphere
complexes with ionizable surface sites (crH) and any rela-
tively mobile protons that may neutralize permanent-charge
sites, because protons adsorbed on permanent-charge sur-
faces of H-saturated clays are extremely mobile and prob-
ably reside in the diffuse-ion swarm (Coleman and Harward,
1953; Harward and Coleman, 1954; Low, 1955). In any
case, the number of protons adsorbed on permanent-charge
sites is likely to be quite small except in extremely dilute
or acidic solutions (Gilbert and Laudelout, 1965; Foscolos
[lOb] and Barshad, 1969).

where X is the equilibrium supernatant solution pH and a,

P, and d are adjustable parameters. Equation [lOb] was
solved for X under the condition (nM L) = o-0. When
Eq. [lOb] was used to determine the PZNPC, o-H,titr(PZNPC)
was found using Eq. [lOa] to calculate the value of <rH titr
when X = PZNPC. Comparison of the PZNPC calculated
with Eq. [9] and [lOa] and that calculated with Eq.[10b]
provides a test of the titration and ion adsorption data, as
discussed below.
The value of <TH at each pH and ionic strength then was
RESULTS
The Nad adsorption and proton titration data for
ionic strengths of 0.02 and 0.05 mol kg-1 are pre-
sented in Table 2. The amount of Na adsorbed by
Fallbrook and Bonsall soil separates increased signif-
icantly with increasing pH, which indicates that ion-
izable surface functional groups are important in the
surface chemistry of these soils. Net Cl adsorption
 ANDERSON & SPOSITO: PROTON SURFACE-CHARGE DENSITY IN SOILS 1441

Table 3. Point of zero net proton charge (PZNPC) and apparent Table 4. Cesium adsorption (nCs), apparent proton charge
proton charge at the PZNPC [<rH ,,,r(PZNPC)] calculated density (<rH ,itr), and absolute proton surface-charge density
either with Eq. [9] and [lOa] or Eql [lOb] and [10a]. (<rH) data for the Fallbrook, Bonsall, and Maneadero soil
separates suspended in 0.05 mol kg-' CsCl solutions.
PZNPC o-H.itr(PZNPC)
Background Eq.[9] PZNPC o-H111P(PZNPC) Eq. [lOb] Equilibrium
electrolyte and [lOa] Eq. [lOb] ' Eq. [9] and [lOa] pH DCS H,titr "Cs + H,UI, + Obt WH
1
mol kg"1 mmolc kg soil-1
Fallbrook Fallbrook
0.05 NaCl 4.3 0.3t 4.2 03$ 446 39 611 3.93 0.01 92.9 0.6:|: 27.9 0.2t 56.3 0.5 -14 511
0.02 NaCl 4.1 0.2 3.7 0.3 63 5 67 6 4.39 0.01 94.6 0.3 21.3 o!2 51.4 0.5 -21 5
0.05 CsCl 2.9 0.3 NA# 42 5 NA 5.18 0.10 97.9 0.4 12.2 0.2 45.6 0.3 -30 5
Bonsall 5.84 0.02 101.9 0.4 5.04 0.08 42.5 0.3 -37 5
6.35 0.02 101.4 0.3 1.55 0.01 38.5 0.3 -40 5
0.05 NaCl 4.8 0.4 4.8 0.5 24 7 20 7 6.59 0.01 104.4 0.3 -0.17 0.01 39.8 0.3 -42 5
0.02 NaCl 4.2 0.3 4.1 0.3 44 7 42 7 6.89 0.01 106.7 0.4 -2.28 0.01 39.9 0.4 -44 5
0.05 CsCl 2.8 0.4 NA 45 6 NA 7.57 0.03 111 4 -5.12 0.03 41 4 -47 5
Maneadero 8.06 0.05 111.6 0.6 -15.6 0.1 31.4 0.5 -58 5
9.09 0.04 116.3 0.5 -18.1 0.1 33.7 0.6 -60 5
0.05 NaCltt 5.1 0.3 4.7 0.2 48 7 58 7
0.05 CsCl 4.2 0.4 NA 50 7 NA Bonsall
4.02 0.01 76 3 30.7 0.3 63 8 -14 8
t Nominal value worst-case error estimate based on uncertainty in 4.38 0.01 75.0 0.9 24.47 0.08 56 7 -21 7
o-H,lllr(PZNPC). 5.25 0.01 79.4 0.6 12.21 0.03 49 7 -33 7
t Nominal value worst-case error estimate based on the larger of 5.85 0.01 83.0 0.4 5.02 0.09 45 7 -40 7
SD^a, or SD*,. 6.28 0.01 83.6 0.3 1.53 0.01 42 7 -43 7
Mean SD 1of (<rc + <rH au + %. - a^); N1 = 24 for suspensions in 6.58 0.02 86.0 0.5 -0.17 0.01 43 7 -45 7
0.05 mol kg- NaCl; N = 21 for 0.02 mol kg- suspensions; N = 27 for 6.87 0.01 88.5 0.7 -2.30 0.02 43 7 -47 7
suspensions in CsCl. 7.45 0.02 90.5 0.5 -5.18 0.02 42 7 -50 7
11 Nominal value worst-case error estimate based on error in PZNPC. 7.92 0.03 93.7 0.4 -15.9 0.1 35 7 -61 7
# NA = data not available. Equation [lOb] gave PZNPC < 0 for Bonsall 8.71 0.06 97.8 0.9 -19.6 0.1 35 7 -65 7
and Fallbrook CsCl suspensions and < 1 for Maneadero CsCl suspensions;
OH,lltr(PZNPC) could not be calculated. Maneadero
tt Based on preliminary Na adsorption data from Anderson (1989). 4.09 0.01 66.5 0.4 49.2 0.3 68.4 0.4 1 7
4.46 0.02 70.0 0.8 45.1 0.3 68.0 0.9 -3 7
5.07 0.03 75.0 0.3 33.0 0.1 60.6 0.5 -15 7
5.67 0.01 80 3 21.1 0.1 54 3 -27 7
was negative and did not vary systematically with pH 6.37 0.02 82.5 0.6 14.2 0.1 49.3 0.5 -34 7
or ionic strength (P < 0.05). Net ion adsorption (nNa 6.55 0.02 87.5 0.5 4.64 0.01 44.7 0.5 -43 7
6.75 0.02 90 1 0.42 0.01 43 1 -48 7
nc)) increased with increasing pH, but this increase 7.58 0.03 92 4 -0.91 0.01 44 4 -49 7
was compensated by a decrease in o-H titr. Thus, as it 8.28 0.05 96.0 0.4 -6.73 0.01 41.9 0.4 -55 7
should be according to charge balance, the quantity 9.30 0.04 100.9 0.3 -14.6 0.1 38.8 0.2 -63 7
P
9.63 0.04 103.3 0.9 -20.9 0.2 35 1 -69 7
(Na - C1 + OH,titr + ^o) = <%,titr( ZNPC) W3S
independent of pH within the experimental precision t <r0 values (effective structural surface-charge density) taken from Table
1.
(Table 2; P < 0.05). Values of o-H,titr(PZNPC) and t Mean SD of triplicate measurements. 2
PZNPC, respectively, were calculated with Eq. [9] Uncertainty = (SE> + SD^,,,,, + SDJ,,)" .
and [lOa], using the effective o-0 values in Table 1. 1 Uncertainty in o-H is assumed equal to uncertainty in <TH litr(PZNPC,I),
These results were compared with values of PZNPC where PZNPC is the point of zero net proton charge.
and OH)titr(PZNPC) calculated with Eq. [lOb] and [lOa],
respectively (Table 3). The two calculation methods
gave identical PZNPC and OH>titr(PZNPC) values within When only those data for the pH range across which
experimental precision, indicating that the ion adsorp- (no, + a-Htitr + cr0) was pH independent (6<pH<8)
tion and proton titration data are mutually consistent. were included, OH>titr(PZNPC) and PZNPC were not
For the Bonsall and Fallbrook soils suspended in NaCl, significantly different (2 mmol kg-1 and 0.1 pH higher,
PZNPC was independent of ionic strength within the respectively) from the values reported in Table 3. The
precision of the measurements (Table 3). PZNPC values for the three soils suspended in 0.05
Cesium adsorption and proton titration data at an mol kg-1 CsCl ranged from 2.8 to 4.2 and were =1
ionic strength of 0.05 mol kg-1 are presented in Table pH unit lower than those for the soils suspended in
4. Cesium adsorption increased and o-Htitr decreased 0.05 mol kg-1 NaCl. This result can be interpreted
with increasing pH, but the decrease in erH titr was generally as a consequence of charge balance (Eq.
greater than the increase in Cs adsorption. According [3]). Since n^ > nNa at any pH value, both <rH and
to the charge-balance constraint of Eq. [3] and [9], PZNPC were lower in CsCl than NaCl suspensions.
(nCs + CTH titr + OQ) = OH titr(PZNPC) should be in-
dependent of pH if both proton adsorption in the dif- DISCUSSION
fuse-ion swarm and Cl adsorption are negligible. But Adsorption of Na and Cs on the Fallbrook, Bonsall,
(nCs + OH ,itr + cr0) decreased significantly (P < 0.01) and Maneadero soil separates (Tables 2 and 4) was
as the pH increased from 4 to 9 and was only inde- much more pH dependent than has been reported for
pendent of pH between pH 6.3 and 8. In spite of this, cation adsorption on inorganic hydroxylated surfaces
values of OHjti,r(PZNPC) were determined with Eq. [9] (Amphlett and MacDonald, 1956; Greenland, 1974;
by averaging (ncs + crH titr + cr0) across all pH (Table Ferris and Jepson, 1975). Cation adsorption on ion-
3). Then PZNPC was calculated with Eq. [lOa], and izable functional groups of soil organic matter may
Eq. [8] was used to calculate <TH (Table 4; Fig. Ib). have caused much of the increase in Na and Cs ad-
 1442 SOIL SCI. SOC. AM. J., VOL. 56, SEPTEMBER-OCTOBER 1992

U m' curred when the soils were suspended in NaCl at low

NaCl Fallbrook Bonsall
10 pH, then (nNa nc, + trH,titt + <TO), which was in-

s
"o
E -20
0
-10
CU

0.05 mol kg ']
O 0.02 mol kg "1 n
0)
-30
-40 o a
o 1 (U mtnj 0
-50
o
-60 ao ooo
-70
4 5 6 7 8 ' 9 '
10 CsCI
Fallbrook
, _ A D Bonsall
V 0 A A Maneadero
OJ
-10
M CD A
Q
-20
-30 fj A A ionizable sites existed when the soils were suspended
-40 fl ffm
o1 -50
-60
in
EQUILIBRIUM SUPERNATANT SOLUTION pH
Fig. 1. Net proton surface-charge density (trH) calculated with
Eq. [8], [9], and [10a]. Data points represent individual
measurements: (A) Fallbrook and Bonsall soils suspended
in 0.05 or 0.02 mol kg-1 NaCl; (B) Maneadero, Fallbrook,
and Bonsall soils suspended in 0.05 mol kg-1 CsCI.
sorption as the pH increased from 4 to 8. Even a small
quantity of organic matter can contribute relatively
large amounts of pH-dependent charge because 1the
CEC of soil organic matter may be >4 molc kg- at
pH 7 (Stevenson, 1982, p. 235; Perdue, 1985, p. 513).
The PZNPC values, which ranged from 4 to 5 for
the soils suspended in NaCl, indicate that organic mat-
ter and kaolinite are relatively more important than Al
and Fe oxides in the variable-charge properties of these
soils. The PZNPC of soil organic matter may be well
below 3, because the carboxyl groups on organic mat-
ter are more strongly acidic than simple carboxylic
acids (Sposito et al., 1982; Stevenson, 1982, Ch. 13;
Perdue and Lytle, 1983; Perdue, 1985, 1990; Gregor
and Powell, 1988). Points of zero net proton charge
for Fe and Al oxides suspended in monovalent elec-
trolyte solutions are >7 (Atkinson, et al.,1967; Parks,
1967; Breeuwsma and Lyklema, 1971; Schwarz et
al., 1984), whereas the PZNPC of kaolinite is prob-
ably between 4 and 5 (Ferris and Jepson, 1975; Car-
roll-Webb and Walther, 1988), although values as low
as 2.8 (Gallez et al., 1976) and as high as pH 7 (Rand
and Melton, 1975) have been reported. The PZNPC
for the Maneadero soil was greater than for the Bon-
sall and Fallbrook soils, perhaps because of the higher
Fe oxide concentration in the former soil.
It is unlikely that the pH-dependent cation adsorp-
tion was caused by competing proton consumption
reactions, such as mineral dissolution or adsorption
on siloxane surfaces, rather than proton complexation
by variable-charge surfaces. First, if significant min-
eral dissolution and consequent Al adsorption had oc-
dependent of pH in these experiments, should have
decreased with increasing pH. Second, illite and
montmorillonite
+
exhibit a strong preference for Cs+
over H (Sawhney, 1965), and the Na-H exchange
selectivity coefficient of montmorillonite is near unity
(Gilbert and Laudelout, 1965; Foscolos and Barshad,
1969). Thus, proton desorption from permanent-charge
sites cannot account for the nearly twofold increase in
Na as pH increased from 4 to 8, because protons
should occupy <0.5% of the permanent-charge sites
when the soils are suspended in 0.02 mol kg-1 NaCl
at pH 4.
The effect of ionic strength on Na adsorption and
OH (Table 2) indicates that fewer protons dissociated
from ionizable sites as the NaCl concentration in-
creased and, therefore, that fewer negatively charged
in the higher ionic strength. In contrast, the surface
charge of inorganic surface hydroxyl groups usually
increases with increasing ionic strength (Breeuwsma
^
A
and Lyklema, 1971; Madrid et al., 1984; Schindler et
10
al., 1987; Sprycha, 1989). Therefore, the decrease in
Na adsorption and in |<TH| with increasing ionic strength
may be another indicator of the importance of surface-
bound organic matter in the present experiments. Co-
agulation of humic substances and other polyelectro-
lytes occurs with increasing ionic strength (Tanford,
1961, Ch. 8; Morawetz, 1975, Ch. 7; Ghosh and
Schnitzer, 1980; Stevenson, 1982; Hayes et al., 1989,
Ch. 15). It is possible that, with decreasing ionic
strength and pH > PZNPC, conformational changes in
surface-bound organic matter expose additional ion-
izable functional groups that can complex more cat-
ions.
In CsCI suspensions, crH was more pH dependent
than was Cs adsorption. It is unlikely that competing
protonation reactions caused this result, because nei-
ther proton adsorption in the diffuse ion swarm nor
H + or OH" comsumption in mineral dissolution re-
actions should have been more important in CsCI than
NaCl suspensions, and there was no evidence for such
reactions in NaCl suspensions, as discussed above.
But it is possible that competing cation-exchange re-
actions may have complicated the CsCI adsorption and
titration data. Cesium adsorption was much greater
than Na adsorption at low pH and was less pH de-
pendent (Tables 2 and 4). Perhaps low pH facilitated
Cs-K (but not Na-K) exchange in collapsed interlay-
ers of 2:1 clay minerals. Or Cs adsorption at low pH
could have been balanced by polyvalent metal ion (not
proton) desorption if the "Cs-saturated soils" con-
tained some hydrolyzable metal cations that were not
displaced during Cs saturation at pH 7 but were dis-
placed by Cs at low pH (see Foscolos, 1968). Alter-
nately, at low pH, more Cs than Na may have been
adsorbed by variable-charge surface functional groups
such as soil organic matter. These explanations also
are consistent with the proton titration data: to reach
a given equilibrium supernatant solution pH<6, the
crHtitr values were lower (i.e., fewer protons con-1
sumed or more protons released) in 0.05 mol kg-
CsCI than in 0.05 mol kg-1 NaCl (Tables 2 and 4).
 ANDERSON & SPOSITO: PROTON SURFACE-CHARGE DENSITY IN SOILS 1443

In conclusion, the NaCl and CsCl adsorption data,

proton titration data, and measured PZNPC values in-
dicate that organic matter dominates, and kaolinite
and metal oxides contribute to, the variable-charge
behavior of the Fallbrook, Bonsall, and Maneadero
soils. When the net ion adsorption data and previously
measured values of effective tr0 were used to deter-
mine PZNPC values, PZNPC was significantly lower
in CsCl than in NaCl suspensions because more Cs*
than Na+ was adsorbed at any given pH. In NaCl
suspensions, the net ion adsorption decreased with in-
creasing ionic strength, possibly because of coagula-
tion of adsorbed organic matter.
ACKNOWLEDGMENTS
The authors wish to thank Ms. Adrienne Mackzum for
technical support. This work was partially supported by a
National Science Foundation Graduate Fellowship and a
University of California Dissertation-Year Fellowship for
S. Anderson.