0039-9140/85 $3.00 + 0.

00
Talanta, Vol. 32, No. 8A, pp. 645-649, 1985
Printed in Great Britain. All rights reserved Copyright 0 1985 Pergamon Press Ltd

DETERMINATION OF THE DEGREE OF SUBSTITUTION

OF SODIUM CARBOXYMETHYLCELLULOSE BY
POTENTIOMETRIC TITRATION AND USE OF THE
EXTENDED HENDERSON-HASSELBALCH EQUATION
AND THE SIMPLEX METHOD FOR THE EVALUATION

INGRID AGGERYD and L&E OLIN

Department of Analytical Chemistry, University of Uppsala, P.O.B. 531, S-751 21 Uppsala, Sweden

(Received 14 November 1984. Accepted 27 February 1985)

Summary-Determination of the degree of substitution, D.S., of purified sodium carboxymethylcellulose,

CMC, by potentiometric titration has been studied. Different mathematical descriptions of the acid-base
properties of polyelectrolytes were tested on the titration curve. It was found that the extended
Henderson-Hasselbalch equation reproduced the measured data well. It was therefore used as a basis for
evaluating D.S. by the simplex method. The influence of ionic strength, I, on the shape of the titration
cnrve was investigated and the effect of I on the precision of the titrations was evaluated. The titrations
showed a precision of kO.004 D.S.-units and the results were in agreement with those obtained from a
dry-ashing method, within &0.02 D.S.-units. The whole method, including preparation of solutions and
titration, showed a precision of kO.01 D&units.

Sodium carboxymethylcellulose is a bulk chemical
with wide applications in the cellulose and food
industries, for instance. Its trade name is CMC or
cellulose gum. By varying the length of the cellulose
chain and the degree of substitution, D.S., CMC with
different properties can be obtained. The D.S. is
defined as the number of carboxymethyl groups per
glucose unit. The highest theoretical value is 3, but
for commercial products the D.S. is usually much
lower, 0.6-0.8.
A number of methods for the determination of
D.S. have been reported. In some of them CMC is
quantitatively precipitated by multivalent metal ions:
Cu2+, UO:+,2 or Pb2+.3 Two methods have been
proposed by ASTM.4 One is based on a non-aqueous
acid-base titration. In the other, CMC is first con-
verted into the insoluble acid form, separated, and
dissolved in an excess of sodium hydroxide, and
finally the excess of base is determined. The D.S. can
also be determined by conductometric titration,3.5
colorimetryS and by infrared spectrophotometry.3
The methods are applicable only to purified samples
of CMC or include a purification step in the anal-
ytical procedure.
CMC has been used as a model substance in studies
of the acid-base properties of polyelectrolytes by the
potentiometric titration technique, but the method
has not been used for the determination of D.S. The
slightly soluble acid form of CMC reacts slowly with
added base and the shape of the titration curve is
markedly affected by the speed of the titration.6
Hence titration of the acid form cannot be used for
any analytical purpose. The determination of D.S.
must therefore be based on an acidimetric titration of
645
CMC. Since polycarboxylic acids are fairly strong,
evaluation of the equivalence volume directly from
the titration curve can be expected to be imprecise.
Numerical methods have been used to determine
equivalence volumes from titrations of low
molecular-weight acids and bases with unfavourable
pK,-values. A numerical evaluation of the titration
curve of a polyelectrolyte might also be possible, if a
sufficiently accurate mathematical description of the
shape of the titration curve could be found.
THEORY
Equations for description of the acid-base proper-
ties of polyelectrolytes have been proposed by Tan-
ford. For a macromolecule with identical binding
sites and electrostatic interaction between the sites,
the following expression was derived.
k = koeYi (1)
or, in logarithmic form,
In k = In k,, + yi (2)
where k is the mean acidity constant of the sites at a
particular ph, calculated in the usual way from ph
(defined below) and the concentrations of dissociated
and undissociated sites; i is the average charge of a
site, and is proportional to the degree of dissociation,
cc; y is a constant for a given fixed environment for
the reaction, provided that the macromolecule does
not change its conformation. In this equation Ink
and i are both very sensitive to experimental errors;
thus a linear evaluation of equation (2) requires that
both Ink and d are weighted.
646 INGRID AGGERYD and AKE OLIN

Another expression for description of the proto- a proper root, the value of P must not exceed a
lytic properties of polyelectrolytes is the extended critical value given by
Henderson-Hasselbalch eauation. which was nro- .
posed by Katchalsky et al.; P= &(n +3-~)2(J7G-n)
k-1
Ph =pK?+n log[HAl (3)

where [A-] and [HA] are the concentrations of

dissociated and undissociated substituent, re- For a low molecular-weight acid this value is l/27.
spectively, and can be found from knowledge of the For polymeric acids the limiting value of P is smaller
total concentration of the substituent and [H+]; n is because of the larger value of n. The difference
an empirical parameter which is equal to 1 for a low between the inflection point and the equivalence
molecular-weight acid and larger than 1 for a poly- point is shown in Table 1 for a few values of P and
electrolyte. n. The data show that it is more difficult to evaluate
The ph, which is defined by ph = -log[H+], is a titration curve of a polymeric acid than of a
measured by the glass electrode and calculated from monomeric one with the same acidity constant.
E=E,+glOg[H+]+Ej (4) Calculation of the equivalence volume, V,, from a
where E = measured potential (mV), g = @T/F) titration curve obeying equation (3)
In 10 = 59.16 mV (at 25), and E, = liquid-junction The following relationship between the experi-
potential (mV). _I?,is an empirical function of [H+]: mental quantities V and [H+], and the parameters K,,
E, = K,,[H+]. Equation (3) can then be tested by n and V,, can be derived from equation,(3) and the
plotting y = log([A-]/[HA]) against ph. The values expressions for the total concentrations of substitu-
of pK, and n should be evaluated by a weighted linear ents and protons.
regression since the variance of y is dependent on ph.
The following estimate of I* may be used in the
G
= ve(C,-[H+]){l +(K,/[H+I)}
weighting scheme.
vow+1 @)
a*(Y)= {[H;;;_l} +C,-[H]
It can be fitted to the measured data by the simplex
x ;:>$ (AV)* + (&> (LIE)*} (5) method.
U=~(V,,,-
The sum of squared
V,,c)2, is minimized.
residuals,

where [A]* is the total concentration of substituent, EXPERIMENTAL

V, the initial volume, V the volume of added titrant
Chemicals
of concentration Cu, and AV and AE are the esti-
The CMC used in the experiments was supplied by
mated errors in the volume added and the potential,
Billerud Uddeholm AB. Four types of CMC with different
respectively. The influence of [OH-] was neglected in viscosities were titrated (see Table 2). The samples were
the derivation of equation (5).
Equation (3) appears to be applicable only when
Table 1. The difference, A, between the equivalence point
no changes in conformation occur. If this is not the and the inflection point of a titration curve, for some values
case, models using more parameters must be resorted of n and P = KJC.; @ = V/V, at the inflection point
to.10
P
Equivalence volumes from injection points of titration
curves obeying equation (3) 10-r 0.964 0.849 3.6 15.1
lo- 0.997 0.966 0.3 3.4
If the titration curve has an inflection point, the 1o-4 1.000 0.990 0.0 1.0
condition d* (ph)/d V* = 0 must be fulfilled somewhere
along the curve. At the points where the second
derivative is zero the following relationship can be Table 2. Viscosity for 2%
solutions of CMC in water,
shown to be valid for the titration of a polybasic acid according to specifications
obeying equation (3). from the manufacturer
(1 - a)a+ 1 Type Viscosity,
an+ + PnZ(l - ar- =T (6) of CMC CP*

uv 2&50
As before, a is the degree of dissociation, P = KJC, LV 90-200
and C, is the total concentration of acidic groups. MV 4W800
The contributions from [OH-] and from dilution HV 12W2500
have been neglected in the derivation of equation (6). *SI equivalent of the centi-
It can further be shown that for the equation to have poiseis IO-rN.sec.m-*.
 Determination of the degree of substitution of carboxymethylcellulose
,L
0 protective atmosphere of nitrogen.
Volume (ml)
Fig. 1. Titration curve for a CMC sample of type UV. The
curve is calculated from equation (8) with parameters from
the simplex optimization. V, = 4.20 ml and I = 0.20M.
purified according to the method described below. The D.S.
was determined by dry-ashing the CMC and titrating the
resulting sodium carbonate with hydrochloric acid.
All other chemicals were of p.a. grade. Potassium nitrate
was recrystallized before use. Nitric acid solutions were
prepared from the concentrated acid by dilution and stan-
dardized against tris(hydroxymethyl)aminomethane.
Purification of CMC
The CMC samples had a specified purity of 97-99x. The
main impurities were sodium chloride, glycollate and car-
bonate. Carbonate was removed by treatment with acid.
Then the sample was dissolved in sodium hydroxide solu-
tion and small amounts of undissolved material were re-
moved by centrifugation. The CMC was reprecipitated with
ethanol and washed with 70% v/v ethanol on a glass filter.
CMC has very low solubility in this solvent, whereas sodium
chloride and glycollate are quite soluble. Washing was
continued until chloride could not be detected in the wash-
ings. Water was removed by washing with absolute ethanol
and finally ether was passed through the CMC, which was
then dried by sucking air through it. CMC is hygroscopic
and the final product was equilibrated in a hygrostat to have
a moisture content of about lOy& Glycollate and chloride
ions could not be detected in the final product (i.e., less than
0.05% of either was present).
3.75
-t 1
Yl+::
Degree of dlssoclotlon
Fig. 2. Evaluation of a titration according to equation (2). [CMq = 1 g/l., D.S. = 0.79, I = 0.20,~.
647
Titrations
The titrations were done with a laboratory-built auto-
matic titrator. An Ingold type 201 glass electrode was used
as indicator electrode and a Metrohm Ag/AgCl electrode as
reference electrode. The salt bridge contained 3M potassium
nitrate. The emf was measured with a digital voltmeter to
*O.Ol mV.
The titrations were performed in a constant ionic medium
to keep the activity coefficients constant. Before each ti-
tration the glass electrode was calibrated as follows. The
ionic medium was titrated with 25.OOmM nitric acid to pH
2. The parameters E,, and Kh in equation (4) were evaluated
from a plot of E -g log[H+] as a function of [H+], with
E, as the intercept and & as the slope of the straight line.
It was found from these titrations that sodium chloride and
nitrate, which are usually employed as ionic media, often
contain acidic or alkaline impurities. Therefore potassium
nitrate, which is easier to obtain pure, was used as the ionic
medium in all experiments.
CMC solutions (1 g/l.) were titrated with 25.OOmM nitric
acid to pH 2. The experiments were done at 25.0, with a
RESULTS AND DISCUSSION
The titration curve
A typical titration curve of CMC is shown in Fig.
1. No inflection point can be discerned visually and
the equivalence volume must therefore be evaluated
by a numerical method. To test the compatibility of
our data with equations (2) and (3) the relevant
quantities were calculated by using the values of D.S.
obtained by the dry-ashing method (see experimental
section).
Figure 2 gives a plot of -log k versus a. The plot
is linear over most of the range of a values, but
substantial deviations occur at low LX.Since the
calculated values of LYand k are very sensitive to
experimental errors, it is difficult to decide whether or
not Tanfords equation fails to reproduce. the
acid-base properties of CMC. No explicit expression
can be obtained from equation (1) for either [H+] or
V. This makes the equation less suitable as the basis
for numerical evaluation of the equivalence volume
and hence no attempt was made to use it for deter-
mination of V,.
The extended Henderson-Hasselbalch equation,
equation (3), reproduces the data well in the range
0.5
10
648 INGRID AGGWYD and AKE OLIN

from the low molecular-weight salt on the charges on

the polyelectrolyte chain. As a consequence n will
approach unity when Z rises, and pK, will decrease
when the attraction of protons to the macromolecule
is weakened.
An evaluation has also been made of the effect of
the ionic strength on the precision in the titrations. It
was found that the results obtained in 0.1-0.2M
potassium nitrate media best agreed with equation
(3), yielding R*-values > 0.9998. For Z > 0.2M the
solubility of CMC decreases, and at Z below 0.1 M the
liquid-junction potentials are large and difficult to
determine with high precision. Hence 0.2M potas-
log IAI/CHAI sium nitrate was chosen as medium for the deter-
Fig. 3. Evaluation of a titration according to the extended mination of D.S. by titration.
Henderson-Hasselbalch equation. [CMC] = 1 g/l., D.S. =
0.79, I = 0.20M. Determination of D.S. by the simplex method
The D.S. was found by fitting equation (8) to the
titration data by the simplex method with K,, n and
- 1 < log([A-]/[HA]) < 1, as demonstrated by Fig. V, as the adjustable parameters. The values of D.S.
3. Outside this interval the quotient [A-]/[HA], obtained for the four different CMC-samples in-
(where both concentrations are the experimental val- vestigated are presented in Table 4 and compared
ues) becomes increasingly sensitive to errors in E and with the values obtained from the dry-ashing method.
V. The good fit of the data to a straight line led to The difference between the two methods is less than
the use of equation (8) for evaluation of V, by the f0.02 D.S. units.
simplex method. As shown by the results in Table 5, the values of
As will be detailed later the values of n and K, in K, and n obtained by the simplex method agree
equation (3) are about 1.3 and 3 x 10e4M, re- closely with those found by fitting the same titration
spectively. With a CMC concentration of 1 g/l. and data to equation (3) by weighted linear regression. In
a D.S. of 0.8, the value of P in equation (6) becomes the latter calculations the D.S.-values from the dry-
0.08. This P-value is larger than that given by ashing method were used. The final result from the
equation (7) and shows that the titration curve has no simplex optimization was independent of the starting
inflection point, thus confirming the visual obser- values of the parameters. A bad initial guess merely
vation. results in many more iterations being required for the
final values of K,, n and I, to be reached. The method
Influence of ionic strength on the titration curve
was, however, found to be very vulnerable to system-
The ionic strength, Z, affects the shape of the atic errors in the data, e.g., arising from poor cali-
titration curve of a polyelectrolyte. The influence of bration of the electrodes.
Z on pK, and n has therefore been studied. The Titrations of two different total concentrations of
titrations were done with a CMC-sample of type UV CMC, 1 g/l. and 2 g/l., indicated that the values of
with a D.S. of 0.79 and evaluated by weighted linear K,, n, and D.S. are independent of the total concen-
regression. The weights were calculated from equa-
tion (5) with Al= 0.01 ml and AE = 0.1 mV.
Table 4. Determinations of D.S. for different types of CMC;
The results are presented in Table 3. An increase in
I = 0.2M, [CMC] = 1 g/l.
Z leads to a decrease in the values of n and p&, a
result in agreement with other investigations.* An Type
of CMC D.S. from dry-ashing D.S. from simplex
explanation of this behaviour has been given by
Oosawa4 in terms of the screening effect of the ions uv 0.794 0.789
LV 0.716 0.734
MV 0.779 0.780
HV 0.772 0.783
Table 3. The influence of the
ionic strength, I, on pK, and n
for a CMC sample of type UV;
D.S. = 0.79, [CMC] = 1 g/l. Table 5. Comparison between values of pK, and n obtained
by simplex optimization and bv weiahted linear reeression
I. mM PK, n
1.17
Type P% PK, n n
500 3.37 of CMC (simplex) (lin. reg.) (simplex) (lin. reg.)
200 3.46 1.26
100 3.55 1.34 uv 3.47 3.46 1.26 1.27
50 3.68 1.40 LV 3.42 3.44 1.25 1.20
25 3.82 1.43 MV 3.45 3.44 1.27 1.26
12 3.96 1.57 HV 3.43 3.43 1.27 1.26
 Determination of the degree of substitution of carboxymethylcellulose 649

tration. The fit of the evaluation model to the mea- 2. C. V. Francis, ibid., 1953, 25, 941.
sured data can be seen in Fig. 1. The curve is 3. A. Naterovl and J. PolEin, Faserforsch. Textiltech.,
calculated from equation (8) with parameters from 1975, 26, 14.
4. American Society for Testing and Materials, ASTM
the simplex optimization. Standard Dl439-83.
The precision of the titrations was f0.004 D.S. 5. R. W. Eyler, D. E. Klug and F. Diephuis, Anal. Chem.,
units, and the precision for the whole method, includ- 1947, 19, 24.
ing titration and preparation of solutions, was + 0.01 6. B. Philipp and H. Dautzenberg, Wiss. Z. Tech. Univ.
Dresden, 1975, 24, 41.
D.S. units. The substantial increase in imprecision for
7. L. Pehrsson, F. Ingman and S. Johansson, Talanta,
the whole procedure is mainly due to the difficulties 1976, 23, 781.
of handling and transferring CMC samples. The 8. C. Tanford, Physical Chemistry of Macromolecules,
results show that potentiometric titration with sim- Chapter 8, Wiley, New York, 1961.
plex optimization to evaluate the titration curve is a 9. A. Katchalsky and P. Spitnik, J. Polym. Sci., 1947, 2,
432.
feasible method for the determination of the D.S. of 10. U. P. Strauss, Macromolecules, 1982, 15, 1567.
purified CMC samples with high precision. 11. M. C. Caceci and W. P. Cache&, Byte, 1984, 5, 340.
12. J. G. Zadow, Cellulose Chem. Technol., 1975, 9, 353.
REFERENCES
13. Y. Muroga, K. Suzuki, Y. Kawaguchi and M. Nag-
1. A. Z. Conner and R. W. Eyler, Anal. Chem., 1950, 22, asawa, Biopolymers, 1972, 11, 137.
1129. 14. F. Oosawa, Polyelectrolytes, Dekker, New York, 1971.