Anal. Chem.
1996, 68, 3973-3978
Derivative Analysis of Potentiometric Titration
Data To Obtain Protonation Constants
Jian-Feng Chen, Yuan-Xian Xia, and Gregory R. Choppin*
Department of Chemistry, The Florida State University, Tallahassee, Florida 32306-3006
A methodology is described to calibrate glass electrodes
and to analyze potentiometric titration data to calculate
protonation constants. The analysis uses the variation of
dV/dpH with titrant addition in terms of two physical
parameters which involve the concentrations of H+, OH-,
and HmA. The data for titration of acetic acid and
8-hydroxyquinoline in 0.10-5.0 m NaCl media are
analyzed by this method to obtain the stoichiometric
protonation constants of the acids, the ionization con-
stants of water, and the parameters s and b in the pH
electrode calibration equation, pcH ) spHm + b, where
pcH ) -log [H+], and pHm is the pH meter reading.
The measurement of pH by conventional glass electrodes is a
continuing problem in the study of the chemistry of concentrated
electrolyte solutions.1-3 Although pH reference solutions can be
used as primary or secondary standards for pH electrode calibra-
tion for dilute solutions, their value for concentrated solutions is
limited by the significant differences in the activity coefficients of
the reference solutions and the test solutions. In addition, the
liquid junction potential for the common glass combination
electrodes causes further problems.4-8 The latter problem is
increased if solutions of salts other than potassium chloride are
used to avoid precipitation in the interface.
Various methods9-14 have been proposed to calibrate glass
electrodes to correct for the liquid junction potential. To avoid
concerns related to the activity coefficients in concentrated
solutions, often attention is devoted to the measurement of pcH
()-log [H+]). A simple and convenient approach in this case is
to measure several solutions of known concentrations of strong
acid or base, assuming complete deprotonation in the solutions.
However, there remains the difficulty of obtaining reproducible
meter readings (pHm) in the neutral range. In low (pH < 4-5)
and high (pH > 9-10) pH regions, linear correlations are
obtained, but often these linear calibration curves for the two
regions do not coincide. To extend the calibration over the entire
(1) Byrne, R. H.; Breland, J. A. Deap Sea Res. 1989, 36 (5), 803.
(2) Byrne, R. H.; Baldo, G. R.; Thompson, G.; Chen, C. T. A. Deap Sea Res.
1988, 35, 1405.
(3) Baldo, G. R.; Morris, M. J.; Byrne, R. H. Anal. Chem. 1985, 57, 2564.
(4) Kristensen, H. B.; Salomon, A.; Kokholm, G. Anal. Chem. 1991, 63, 885A.
(5) Knauss, K. G.; Wolery, T. J.; Jackson, K. J. Geochem. Cosmochim. Acta 1990,
55, 1519.
(6) Mesmer, R. E.; Holmes, H. F. J. Solution Chem. 1992, 21, 725.
(7) Bates, R. G. Determination of pH; Wiley: New York, 1973.
(8) May, P. M.; Williams, D. R. Talanta 1982, 29, 249.
(9) Hedwig, G. R.; Powell, H. K. J. Anal. Chem. 1971, 43, 1206.
(10) Avdeef, A.,; Bucher, J. Anal. Chem. 1978, 50, 2137.
(11) Irving, H. M.; Miles, M. G.; Petti, L. D. Anal. Chim. Acta 1967, 38, 475.
(12) Powell, H. K. J.; Curtis, N. F. J. Chem. Soc. B 1966, 1205.
(13) Yamazaki, H.; Sperline, R. P.; Freiser, H. Anal. Chem. 1992, 64, 2720.
(14) Slyke, D. D. J. Biol. Chem. 1922, 52, 525.
S0003-2700(96)00138-2 CCC: $12.00 © 1996 American Chemical Society
pH range, a common practice is to use buffer systems which
consist of a weak acid or mixture of several weak acids.9-12 An
advantage of the use of a buffer system is that the hydrogen ion
concentration is not affected by acid impurities, dissolved carbon
dioxide, etc. An alternative procedure involves the use of
spectroscopic indicator dyes13 for calibrating glass electrodes, in
which the ratio of the absorbence of the acidic form to the basic
form of an indicator is used to calculate pcH from known acid
deprotonation concentration constants.
Both of the latter methods are successful over the entire pH
range for the purpose of the calibration. However, their use in
potentiometric titration has limitations. First, the value of pcH
depends on knowledge of the acid deprotonation concentration
constants; second, it is difficult to ensure reproducibility of the
liquid junction potential in different experiments. The purpose
of this paper is to propose a new methodology to analyze
potentiometric titration data to obtain simultaneously information
on the pHm-pcH calibration as well as on the protonation
constants. Two physical quantities, Γ and Q, are defined to
characterize the pH (or pcH) change of a test solution upon
addition of a small amount of titrant solution. A plot of Q vs pHm
or pcH can be used to reflect the protonation (or deprotonation)
process. The results of analysis by this method are reported for
two acids in NaCl media over a range of ionic strength.
THEORY
The proton mass balance during titration of a weak acid (HmA)
can be written as
j ) m - [(V - V0)(COH° - CH°) + V0([OH-]0 - [H+]0) +
n
V([H+] - [OH-])]/[V0CHmA°] (1)
and
n
∑i[H A]i
(2)
CHmA
j)
V0 and V are respectively the initial volume and the total volume,
[H+]0 and [OH-]0 the initial concentrations of the proton and the
hydroxide ions in the test solution, CH° and COH° the concentra-
tions of the proton and the hydroxide ions in the titrant solution,
and CHmA° the initial concentration of the acid. The calibration
equation is defined as
pcH ) spHm + b (3)
Following the work of Slyke14 on buffer capacity of acid or base
Analytical Chemistry, Vol. 68, No. 22, November 15, 1996 3973
solutions, a general equation can be derived by combining eqs hydroxide and hydrochloric acid were 1.00 M standard solutions
1-3 and taking the first derivative with respect to pHm: (Fisher). All other reagents used were analytical grade.
Measurement of pcH. The meter readings, pHm, were
Γ ) (COH° - CH° - [OH-]0 + [H+]0) + obtained with an Accumet 950 pH/ion meter equipped with a
0.4343 dV Corning Semimicro combination glass electrode. The outer sleeve
(n j )CHmA
j0 - n (4)
sV dpHm of the reference cell of the electrode was filled with saturated NaCl
solution. The electrode was immersed in a saturated NaCl
solution and pretreated with 0.1 M HCl solution prior to use so
Γ is defined as that a stable and fast response could be assured. The glass
electrode was calibrated to the NBS scale using two standard pH
Γ ) Q + [H+] + [OH-] (5)
buffer solutions: 0.05 M potassium biphthalate buffer at pH 4.00
at 25 °C (Fisher) and 0.05 M potassium phosphate monobasic-
and Q as sodium hydroxide buffer at pH 7.00 at 25 °C (Fisher).
Potentiometric Titration. The potentiometric titration were
Q ) Ĥn
j CHmA (6)
conducted in a 50 mL jacketed cell controlled to 25.0 ( 0.1 °C
with an Isotemp constant temperature circulator system. The cell
was capped with a rubber stopper with a single hole (diameter of
Ĥ is an operator equal to d/d ln [H+]. As a “normalized” quantity,
1 mm) open to air. Nitrogen gas was bubbled through the cell
Γ is a measure of the test solution and is independent of the
solution to remove dissolved CO2. Titrant solutions (1-30 mM
titration conditions. It can be calculated directly from experimen-
NaOH in NaCl) were delivered to the cell via Teflon tubing by a
tal data if the concentration of the titrant is much higher than the
Schott Gerate automatic buret. A variable volume of titrant was
sum of [H+]0 and [OH-]0 in the test solution. If this condition is
added to obtain an approximately constant increment of pHm in
not met, Γ can be obtained by an iterative calculation. The
the cell. The relative error of the buret was less than (0.2% for
usefulness of the quantity Q derives from the fact that Q is a
titrant additions of larger than 1.0 mL and was estimated to be
function of the concentration and the protonation constants of the
less than (10% for the addition at microliter level. The pHm
ligand being titrated as well as of the proton concentration.
readings were recorded at a fixed time interval (1 min) after
The value Ĥnj can be shown to have the relation
addition of the titrant volume. In the buffer regions, a steady pH
A(ii) - A(ij) meter reading was reached within 30-40 s for the small-volume
Ĥn (7)
addition used. In less buffered regions (e.g., pcH 6-8), times
j)
A(11)
longer than 1 min are required in order to obtain a constant
reading. However, the data in these regions have no significance
and for the calculation of the quantities to be determined. The
m m uncertainties in both volume and pH meter reading in the less
A(kl) ) ∑∑klβ β [H i j
+ i+j
] (8) buffered regions produce some scatter in the plots of Q or Γ vs
i)0 j)0 pHm but have little effect on the peak position and, hence, on the
log K values.
Calculation. A spreadsheet program coded in the Quattro
where βi are the overall protonation constants (β0 ) 1 and βi )
Pro (Borland version 5.0) software was developed to process
∏j)1
i
Ki). If the interval between the two adjacent protonation
potentiometric titration data. The data fittings were performed
constants (∆ log K ) log βi - log βi-1) is more than 1.5 (the case
using the Optimizer function of the software. Values of dV/dpHm
for the acids investigated herein), this equation can be simplified
were calculated with eq 10, which was derived from the Lagrangian
to
polynomial:15

( ) ( )
m Ki[H+]
Ĥn
j) ∑
i)1 (K
2
(9)
dV
dpHm i
=
1 1
V +
pHm-1 - pHmi pHm-1 - pHm+1 i-1
-
+ [H ])+

( )
i
1 1
V +
pHmi - pHm+1 pHmi - pHm-1 i
+

Equation 9 indicates that Q is a quasi-parabolic function of pcH,
with each peak corresponding to a protonation and the peak
maximum value, Qmax ) 0.25CHmA for pcH ) log Ki.14 This
indicates the nature of the graphical pattern observed in analyzing
the potentiometric titration data by this approach.
EXPERIMENTAL SECTION
Reagents. 8-Hydroxyquinoline (HOxn, Aldrich, ACS grade),
sodium acetate (Aldrich, ACS grade), and 2-amino-2-(hydroxy-
methyl)-1,3-propanediol (TRIS, Fisher, reagent grade) were used
without purification. The stock solutions were prepared by
dissolving required amounts of the reagents in corresponding
NaCl solutions. To completely dissolve the HOxn, the solutions
were stirred for several days. The stock solutions of sodium
( )
1 1
V (10)
pHm+1 - pHmi pHm+1 - pHm-1 i+1
-
RESULTS AND DISCUSSION
The potentiometric titration data were processed to calculate
the electrode calibration parameter, b, the protonation constants,
log Ki, and the ionization products of water, pKw. This was
achieved by adjusting the three parameters to minimize the
deviation, σ ) (∑(Qi(exp) - Qi(cal))2/N)1/2 (N is the number of data
points) between the Q values derived with eqs 4-6 using
(15) Berden, R. L.; Fatres, J. D. Numerical Analysis, 5th ed.; PWS Publishing
Corp.: Boston, MA, 1993; p 159.

3974 Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
Table 1. Potentiometric Titration Data of 1.0 mM
Acetic Acida

vol pHm vol pHm vol pHm vol pHm

0.000 1.814 4.564 4.011 5.004 6.285 5.230 9.977
0.260 1.848 4.588 4.069 5.006 6.427 5.238 9.993
0.540 1.885 4.612 4.128 5.010 6.870 5.248 10.013
0.772 1.918 4.636 4.186 5.012 7.337 5.260 10.035
0.996 1.953 4.656 4.234 5.014 7.817 5.268 10.050
1.248 1.993 4.676 4.283 5.016 8.106 5.276 10.062
1.486 2.034 4.690 4.318 5.018 8.296 5.286 10.080
1.726 2.079 4.706 4.357 5.020 8.423 5.296 10.094
1.936 2.120 4.724 4.404 5.024 8.607 5.308 10.113
2.162 2.168 4.742 4.451 5.026 8.674 5.318 10.129
2.344 2.209 4.768 4.520 5.028 8.748 5.340 10.159
2.542 2.259 4.786 4.570 5.030 8.804 5.370 10.196
2.748 2.315 4.806 4.629 5.032 8.909 5.406 10.238
2.954 2.379 4.826 4.693 5.040 9.014 5.430 10.264
3.184 2.460 4.846 4.760 5.050 9.157 5.490 10.320
3.404 2.552 4.872 4.858 5.060 9.271 5.544 10.365
3.506 2.602 4.890 4.933 5.068 9.345 5.590 10.402
3.624 2.666 4.908 5.019 5.084 9.461 5.636 10.433
3.722 2.725 4.930 5.139 5.094 9.524 5.732 10.492
3.820 2.794 4.938 5.192 5.104 9.579 5.810 10.535
3.920 2.875 4.946 5.249 5.112 9.619 5.890 10.575
4.020 2.972 4.952 5.298 5.120 9.654 5.952 10.604
4.090 3.052 4.964 5.404 5.126 9.679 6.074 10.653
4.162 3.148 4.968 5.447 5.134 9.712 6.196 10.696
4.230 3.257 4.974 5.518 5.140 9.732 6.312 10.734
4.268 3.326 4.978 5.573 5.148 9.762 6.416 10.765
4.308 3.405 4.982 5.634 5.158 9.794 6.524 10.795
4.350 3.496 4.984 5.670 5.166 9.817 6.628 10.822 Figure 1. (a) Plot of Γ vs pHm (15 mL of 1.0 mM acetic acid in 1.0
4.390 3.589 4.988 5.746 5.174 9.842 6.734 10.846 m NaCl titrated with 30.0 mM NaOH). (b) Plot of Q vs pHm
4.424 3.669 4.990 5.785 5.182 9.864 6.850 10.872 corresponding to (a).
4.448 3.727 4.992 5.838 5.190 9.885 6.978 10.898
4.470 3.781 4.994 5.887 5.200 9.904 7.102 10.920
4.494 3.839 4.998 6.009 5.208 9.925 7.214 10.943
4.514 3.890 5.000 6.086 5.216 9.947 7.344 10.966 system are listed in Table 2, together with the protonation
4.538 3.948 5.002 6.174 5.222 9.958 7.610 11.005 constants of acetate and the ionization products of water. Figure
a 15.0 mL of 1.0 mM acetic acid (10.0 mM HCl as excess acid) was 2 presents two sets of the electrode calibration parameter b (see
titrated with 30.0 mM NaOH. T ) 25.0 °C, I ) 1.0 m (NaCl). Volume eq 3), respectively from acetic acid and 8-hydroxyquinoline
is given in milliliters. titrations. Despite a deviation of (0.04 pH unit, both sets of data
exhibit a consistent variation with background electrolyte con-
centration. Negative values of b were obtained for concentrations
experimental data and those calculated by eqs 7 and 8. Since below ∼0.05 m, while b f -∞ as the ionic strength approached
eqs 4-5 contain the quantities [H+] and [OH-], [H+]0 and [OH-]0, zero. This contrasts with the values of 0.088 reported by Hedwing
which require calculation from the parameters to be determined, and Powell9 and 0.2269 by Avdeef and Bucher10 at zero ionic
an iterative approach was used to reach a “self-consistent” result. strength for the combination glass electrodes filled with KCl
In systems where the ligand being titrated is sufficiently concen- solution. A semiempirical equation,
trated or its protonation occurs in the neutral region, the value of
Q is less dependent on the three parameters. The initial guess C 0.509xI
b ) (0.305 ( 0.021) - log + 0.208 log m -
m
+
for the parameters was made with the help of graphs of Γ as a 1 + 1.5xI
function of pHm. An assumption of s ) 1 was made in the 0.156m + 0.0026m2 (11)
calculation for convenience, as it was found that, when s was a
variable, values very close to 1 resulted (see Table 2b).
Two systems consisting of 1.00 mM acetic acid in 0.10-5.0 was obtained to correlate the calibration parameter b with the
m NaCl and 1.00 mM 8-hydroxyquinoline in the same background concentrations, m (in molality) and C (in molarity), of NaCl. After
electrolyte were investigated. For each titration, at least 150 points correction for liquid junction potential, the proton activity coef-
were taken, and the pHm interval for successive titrant additions ficients calculated with the values of b agreed well with those
was 0.03-0.1 pH unit in the buffer regions of the test solutions. calculated by the Pitzer formalism.16
This allows the value of dV/dpHm to be calculated with high Two tests were made to evaluate the precision of the present
accuracy by eq 10. Table 1 presents a representative set of method. First, the calibration parameters were calculated using
titration data (V0 ) 15.0 mL, 1.0 mM acetic acid in 1.0 m NaCl, 30 the titration data over either the entire pHm region or the HAc/
mM NaOH as titrant). Figure 1a,b shows respectively the Ac- buffer region (3 < pHm < 6). As listed in Table 2b, the
corresponding plots of Γ and Q against pHm. No systematic error maximium deviations in the calibration parameters from the two
was observed between the experimental data and the values sets of data are (0.019, which is within the uncertainty of (0.02
calculated from eqs 5-8. of the electrode for pHm measurements. The agreement for log
Calibration of Glass Electrode for Hydrogen Concentra- (16) Pitzer, K. S. Activity Coefficients in Electrolyte Solutions, 2nd ed.; CRC
tion. The calibration parameters calculated from the acetic acid Press: Boca Raton, FL, 1991; p 86.

Analytical Chemistry, Vol. 68, No. 22, November 15, 1996 3975
Table 2. Values Calculated for Acetate-Sodium Table 3. Protonation Constants of 8-Hydroxyquinoline
Chloride System and Comparison of the Calibration in NaCl Solutions at 25.0 °C
Parameters, the Protonation Constants of Acetate,
and the Ionization Constants of Water at 25.0 °C log K1 log K2
NaCl, m pwa lit. pwa lit.
(a) Values Calculated for Acetate-NaCl Systema
log K pKW 0.02 9.811 ( 0.003 9.82b,c 4.984 ( 0.008 4.94b,c
0.10 9.643 ( 0.034 4.920 ( 0.033 5.10d
NaCl, m b pw lit.b pw lit.(27)
0.30 9.564 ( 0.012 4.923 ( 0.014 5.08d
0.10 0.002 ( 0.001 4.562 ( 0.001 4.56c 13.83 ( 0.02 13.780 0.50 9.563 ( 0.015 4.941 ( 0.010 5.08d
0.30 0.062 ( 0.022 4.511 ( 0.003 4.51c 13.71 ( 0.05 1.00 9.612 ( 0.040 9.60e 5.021 ( 0.008 5.22e
0.50 0.137 ( 0.004 4.499 ( 0.001 4.50c 13.73 ( 0.01 13.701 2.00 9.651 ( 0.014 5.284 ( 0.002
1.00 0.236 ( 0.019 4.515 ( 0.002 4.50d 13.70 ( 0.02 13.718 3.00 9.722 ( 0.016 5.438 ( 0.019
2.00 0.438 ( 0.007 4.623 ( 0.001 4.61d 13.78 ( 0.02 5.00 10.004 ( 0.004 10.05f 5.823 ( 0.019 5.85f
3.00 0.643 ( 0.010 4.770 ( 0.003 4.76d 13.93 ( 0.02 13.966
a 15.0 mL of 1.00 mM 8-hydroxyquinoline (10.0 mM HCl as excess
5.00 1.063 ( 0.004 5.130 ( 0.001 14.30 ( 0.01 14.291
acid) titrated with 30.0 mM NaOH. b I ) 0.01 M. c Reference 28.
d Reference 29. e Reference 30. f Reference 19.

(b) Comparison of Calibration Parameters, Protonation Constants

of Acetate, and Ionization Constants of Water at 25.0 °C
test Ie test IIf
NaCl, m ∆b ∆ log K s ∆b ∆ log K ∆pKw
critical from our observations; i.e., the potential of the electrode
0.10 -0.019 -0.001 0.981 ( 0.002 0.046 -0.019 -0.016 immersed in a standard solution shifted from one experiment to
0.30 0.008 1.001 ( 0.002
another in a range of about 0.1-5 mV. Greater drifting was
-0.006 -0.004 -0.013 -0.02
0.50 -0.011 0.002 0.991 ( 0.003 0.018 -0.006 -0.09
1.00 -0.007 0.001 1.006 ( 0.008 -0.015 -0.001 0.03 observed after the electrode was immersed in highly alkaline
2.00 0.001 -0.001 1.018 ( 0.004 -0.038 0.008 0.16 solution, probably due to the deterioration of the glass surface.
3.00 -0.003 -0.002 1.012 ( 0.004 -0.020 0.007 0.11
5.00 -0.003 -0.001 1.005 ( 0.006 -0.008 0.014 0.06 Therefore, an error can be introduced if the calibration is
conducted prior to the titration for measuring an acid constant.
a 15.0 mL of 1.00 mM acetate (10.0 mM HCl as excess acid) titrated
with 30.0 mM NaOH. Data in the entire pHm region were fitted. The However, a simultaneous determination of the calibration param-
errors given in all the tables are the random errors estimated as the eter and the acid constant from the same set of titration data
standard deviation (1σ). b Interpolated from literature data for NaCl
media. c Reference 17. d Reference 18. e Test I: data in pHm region of presumably eliminates the effect of electrode drift.
3-6 were used. ∆b, ∆ log K, and ∆pKw are the differences between The commonly used approaches, based on a direct pHm-pcH
the values obtained in the test approaches and those listed in Table
2a. f Test II: the slope s was used as an adjustable parameter in the correlation, to calculate the calibration parameter b attach equal
calculations, and data in the entire pHm region were used. importance to each experimental point, which may not be valid.
A solution in the neutral pH region has less buffer capacity, and
the pH can be altered relative easily by many factors, such as
acidic impurities, dissolved CO2, etc. Moreover, the ionization
product, pKw, of water must be known for the calculation of pcH
in the alkaline region. However, it may not be always available
for concentrated electrolyte solutions. In the proposed approach,
through use of the quantity Γ, a weight proportional to the buffer
capacity is given to each data point, resulting in a significant
reduction in the importance of the data in less buffered regions.
Protonation Constants. The protonation constants of acetate
calculated from the 21 sets of titration data are listed in Table 2a.
Robertis et al.17 reported data for 0.04-1.0 M NaCl solutions and
Belevantsev et al.18 for 1.0-4.0 M solutions. The deviations
between our data and those interpolated from the literature values
Figure 2. Ionic strength dependence of the calibration parameters,
are within (0.013. The protonation constants of 1,8-hydroxy-
b.
quinoline are listed in Table 3. The values of log K2 ) 5.823 (
0.019 and log K1 ) 10.004 ( 0.004 for 5.00 m NaCl solution show
K values is also quite satisfactory (within (0.008). In the second
satisfactory agreement with the data from spectrophotometry.19
test, calculations were made to include the slope, s, in the
Our data at low ionic strength are also in reasonable agreement
calibration equation as an adjustable parameter. As summarized
with those in the literature.
in Table 2b, an average value of s ) 1.002 ( 0.018 was obtained
A general equation used to describe the ionic strength
from the 21 sets of experimental data. The deviations between
the log K values thus obtained and those assuming s ) 1 are within dependence of stoichiometric protonation constant has the form20
(0.02. However, larger discrepancies in pKw values were ob-
served. The values obtained for s ) 1 agree better with the (17) Robertis, A. D.; Stefano, C. D.; Rigano, C.; Sammartano, S.; Scarcella, R. J.
Chem. Res. (S) 1985, 42.
literature data for pKw, supporting the approximation s ) 1. (18) Belevantsev, V. I.; Mironov, I. V.; Peshchevitskii, B. I. Russ. J. Inorg. Chem.
A scatter of (0.03 pH unit was observed for the calibration 1982, 27, 29.
parameters obtained from triplicate experiments, which may (19) Xia, Y. X.; Chen, J. F.; Choppin, G. R. Talanta, in press.
(20) Grethe, I.; Fuger, J.; Lemire, R.; Muller, A.; T-Gegu, C. N.; Wanner, H.
reflect electrode drift, variation in liquid junction potential, or an Chemical Thermodynamics of Uranium; NEA-TDB; OECD: Paris, France,
effect of the glass surface. The drift in electrode response seems 1991.

3976 Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
 xI Table 4. Results of the SIT Analysis of the Protonation
log K ) log KT + 0.509∆Z2 + ∆I (12)
1 + 1.5xI Constants

log KT
acid Im -∆a pw lit.b σ
and
HAc 0.1-3.0 0.155 ( 0.005 4.763 ( 0.005 4.757 0.005
∆Z2 ) ∑ i
νiZi2(product) - ∑ i
νiZi2(reactant) (13)
H2Oxn+
0.1-5.0
0.1-3.0
0.167 ( 0.004
0.185 ( 0.013
4.753 ( 0.017
4.865 ( 0.032 4.92
0.016
0.026
0.1-5.0 0.183 ( 0.007 4.867 ( 0.029 0.029

∑ν (i,NaCl)(product) - ∑ν (i,NaCl)(reactant)
HOxn 0.1-3.0 0.117 ( 0.005 9.845 ( 0.012 9.82 0.015
∆ ) i i 0.1-5.0 0.126 ( 0.004 9.837 ( 0.016 0.026
i i
a ∆(HAc) ) (HAc,NaCl) - (H+,NaCl) - (Ac-,NaCl); ∆(H Oxn+)
(14) 2
) (H2Oxn+,NaCl) - (H+,NaCl) - (HOxn,NaCl); ∆(HOxn) )
+ - b
(HOxn,NaCl) - (H ,NaCl) - (Oxn ,NaCl). Reference 31.

where Zi is the electrical charge of species i, νi the corresponding

stoichiometric number in the protonation reaction, and (i,NaCl) Table 5. Protonation Constants of Acetate Calculated
the interaction coefficient between species i and background from Different Methods
electrolyte NaCl. Plots of log K - 0.509∆Z2[I1/2/(1 + 1.5I1/2)]
BEST
versus I are linear for both of the test systems up to 5.0 m. The
NaCl, m pw aa bb lit.c
values of log KT and ∆ were obtained from the linear regression
and are listed in Table 4. The deviations between the calculated 1.0 4.510 ( 0.004 4.72 4.64 4.50
4.518 ( 0.002 4.54 4.64
and experimental log K values are within (0.03. Agreement of
4.520 ( 0.003 4.72 4.64
the protonation constants, log K1T, of acetate and 8-hydroxyquino- 3.0 4.772 ( 0.001 4.86 4.86 4.76
line with the corresponding literature values is satisfactory. Our 4.762 ( 0.007 4.86 4.79
4.769 ( 0.004 4.88 4.87
constant, log K2T, for 8-hydroxyquinoline seems slightly lower than
5.0 5.129 ( 0.024 5.32 5.17
the literature value; however, a curvature was observed in the 5.129 ( 0.011 5.26 5.16
plot of the second protonation constant as a function of ionic 5.129 ( 0.008 5.37 5.18
strength at I < 0.5 m. Applying a curve-fitting to the correspond- a Estimated errors are (0.03. b Estimated errors are (0.01. c Ref-
ing data, the thermodynamic constant of oxine can be calculated erence 18.
to be 4.92 ( 0.01, which is in excellent agreement with the
literature value. These results demonstrate the precision of the
present methodology in processing titration data for different ionic Table 6. Calibration Parameter and Equilibrium
Constant Measured Using Different Concentrations of
strengths.
Acetate (0.10 m NaCl at 25.0 °C)
An algorithm used in numerous computer programs (e.g.
BEST,21 MICMAC,22 PROTAF,23 MAGEC,8 and MINIQAUD24) is CHAc, µM b log K pKw
the minimization of ∑wi(pHi(exp) - pHi(cal))2 or ∑wi(Ei(exp) - Ei(cal))2. 1000a 0.001 4.572 13.806
The method to evaluate the weighing factor wi is, thus, one of 0.002 4.563 13.831
0.003 4.562 13.844
the major differences among these programs. For example, the 130b -0.018 4.581 13.786
BEST program defines wi ) 1/(pHi+1 - pHi)2 to lessen the -0.029 4.578 13.760
influence of less accurate pHm values in poorly buffered regions.21 -0.029 4.579 13.755
60c 4.627 13.792
Such definitions make the weighing factor highly dependent on
-0.010
4.57d
the data structure, and an increase in the interval between two -0.013 4.633 13.786
adjacent points can modify significantly the weighing factor. 4.56d
4.642 13.779
Therefore, equal additions of titrant solution seem necessary to
-0.015
4.57d
obtain a valid result. In the proposed approach, the weight of
a 15.0 mL of acetate (10.0 mM HCl as excess acid) titrated with 30
each data point is proportional to the buffer capacity of the test
mM NaOH. b 15.0 mL of acetate (1.0 mM HCl as excess acid) titrated
solution, so the data in poorly buffered regions have less with 3.0 mM NaOH. c 15.0 mL of acetate (1.0 mM HCl as excess acid)
contribution. titrated with 1.5 mM NaOH. d Value after correction for the protonation
of dissolved CO2. Total concentration of carbonate, 6 µM, log K )
As pointed out by Martell and Motekaities,21 inaccurate 6.12.25
standardization of reagents could be a source of considerable error
in log K calculation when using a program such as BEST. The
BEST calculation becomes increasingly sensitive to the concentra- the table, a random error of roughly (0.006 was obtained from
tions of excess acid and titrant when the ligand concentration our approach, while the BEST program calculation gave a value
decreases, especially if deprotonation occurs in the neutral pcH of (0.1 when the original concentrations of the excess acid and
range. This can be illustrated in Table 5, in which three sets of acetic acid were used. An error of (0.03 was reached with BEST
triplicate titration data serve as examples and the concentrations after the concentrations and the pcH correction parameters were
used in calculation were not standardized. As can be seen from refined to further minimize the standard deviation. However, the
(21) Martell, A. E.; Motekaities, R. J. Determination and Use of Stability Constants;
log K values are higher than those from our approach by an
VCH Publishers, Inc.: New York, 1988. average of 0.12 logarithm unit.
(22) Laouenan, A.; Suet, E. Talanta 1985, 32, 245.
(23) Fournaise, R.; Petitfaux, C. Talanta 1987, 34, 385. (25) Thurmond, V.; Millero, F. J. J. Solution Chem. 1982, 11, 447.
(24) Sabatini, A.; Vacca, A.; Gans P. Talanta 1974, 21, 53. (26) Izaguirre, M.; Millero, F. J. J. Solution Chem. 1987, 16, 827.

Analytical Chemistry, Vol. 68, No. 22, November 15, 1996 3977

Figure 3. Plot of Q vs pHm (15.0 mL of (a) 130 and (b) 60 µM acetic
acid in 0.01 m NaCl medium titrated with 30-1.5 mM NaOH). Dashed
line, without correction for carbonic acid; solid line, with correction
for carbonic acid.
The influence of the inaccurate standardization of reagents on
log K calculation is greatly reduced in the proposed approach. In
the definition equations of Γ and Q, the reagent concentrations
are linear factors of the two quantities and have less influence on
(27) Busey, R. H.; Mesmer, R. E. J. Chem. Eng. Data 1978, 23, 175.
(28) Paljk, S.; Klofutar, C.; Krasovec, F.; Suhac, M. Microchim. Acta 1975, 2,
485.
(29) Mason, J. G.; Lipschitz, I. Talanta 1966, 13, 1462.
(30) Janjic, T. J.; Pfendt, L. B. Talanta 1992, 39 (1), 55.
(31) Martell, A. E.; Smith R. M. Critical Stability Constants; Plenum Press: New
York and London, 1989; Vol. 6, p 273.
3978 Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
the peak position (or log K) in the plot of Q vs pHm (pcH).
Subsequently, an accurate log K value can be obtained using lower
ligand concentrations. Table 6 reports the results of titrations
using 1000, 130, and 60 µM acetate. The values of b and pKw
from the first two sets of experiments are in good agreement
within experimental error. However, the log K values obtained
using 60 µM acetate are higher than those that remained by an
average of 0.06 logarithm unit. This has been ascribed to the
influence of dissolved CO2. In Figure 3, a shoulder is apparent
at pHm ≈ 6 and a weak peak at pHm ≈ 9, which can be related to
the deprotonation of carbonic acid. Significant improvement of
the fit has been obtained by assuming 6 µM carbonate present
and using the log K values at µ ) 0.1 from Thurmond and
Millero.25 The log K values of acetate, after correction for the
deprotonation of carbonic acid, agree well with the first two sets
of data (see Table 6). To assess the lowest ligand concentration
accessible, titration data using 20 µM TRIS in 1.0 m NaCl as test
solution and 1.5 mM NaOH as titrant were processed. A log K
value of 8.278 ( 0.010 was obtained, which compares well with
the value of 8.281 reported by Izaguirre and Millero.26
In conclusion, the proposed method of analyzing potentiometric
titration data has the advantages of graphic representation, internal
pH electrode calibration, and high accuracy. It is particularly
useful in protonation studies of the acids or bases which have
low solubilities or are present in concentrated electrolyte media.
Investigations of the application of this method to polyprotic acids,
such as oxalic acid, citric acid, and EDTA, are underway.
ACKNOWLEDGMENT
This work was performed as part of the Waste Isolation Pilot
Plant (WIPP) Actinide Source Term Program, supported at Sandia
National Laboratories by the United State Department of Energy
under Contract DE-AC04-94AL85000, and at Florida State Univer-
sity under Contract AH-5590.
Received for review February 13, 1996. Accepted August
3, 1996.X
AC960138T
X
Abstract published in Advance ACS Abstracts, October 1, 1996.