Coulometric Titrations with Iodine
Titration of Arsenic and Use of a n Amperometric End Point
WILLIAM J. RAMSEY, PAUL S. FARRINGTON,
Calqornia Institute of Technology, Pasadena, calij.
The conditions under which electrolytically generated iodine can be used for
secondary coulometric titrations and the end point determined by an ampero-
metric method have been investigated. Tripositive arsenic in quantities from
64 to 1200 micrograms has been titrated in solutions having a pH value of 8 with
an average error without regard to sign of 0.6 microgram.
T HE advantages of the secondary type of coulometric process,
in which an intermediate half-cell reaction is caused to take
place a t the electrode, have been discussed by Szebelledy and
Somogyi ( 7 ) and by Meier, Myers, and Swift (4).A number of
studies have been made of the use of bromine as an intermediate
oxidizing agent for the titration of reducing agents such as hydra-
zine (Q),hydroxylamine ( f O ) , thiocyanate (8), thiodiglycol ( 6 ) ,
tripositive arsenic ( 5 ) ,tripositive antimony ( 1), and iodide ( 11 ) .
The use of cuprous copper as an intermediate reducing agent for
the titration of chromate and vanadate has also been investigated
(4).
In order to extend the application of such processes, especially
to selective titrations, it seemed desirable to investigate the
application of other intermediate half-cell reactions having
various standard potentials and capable of being used in solutions
over an extended range of p H values.
The criteria establishing the suitability of any proposed inter-
mediate half-cell reaction are the following: The active state of
this intermediate must be capable of being produced electrolyti-
cally with 100% current efficiency. The state thus produced must
then be capable of rapidly and stoichiometrically reacting with the
agent to be titrated. Lastly, there must be a method of determin-
ing an end point, and if the amperometric method is chosen, an
excess of the active state of the intermediate must cause a current
between the two indicator electrodes which for the required con-
centrations and conditions is directly proportional to the concen-
tration of that state.
An investigation of the iodide-iodine half-cell as such an inter-
mediate has been made and the pH range has been established
under which the titration of tripositive arsenic can be made with
iodine and the excess iodine determined amperometrically.
EXPERIMENTAL
Chemicals. Standard solutions of tripositive arsenic were pre-
pared in the following way: A 1-gram sample of Bureau of Stand-
ards arsenious oxide which had been dried for 2 hours a t 120" C.
was accurately weighed out and dissolved in 20 ml. of water con-
taining 2 grams of sodium hydroxide. The resulting solution was
acidified by adding 10 ml. of 3 volume formal ( V F )sulfuric acid
and was diluted to approximately 100 ml. The weight of the re-
sulting solution was determined. Solutions prepared in this way
were used as stock solutions, none of which was kept for more than
4 weeks. Weighed samples of these stock solutions were diluted
to appropriate volumes to provide the standard solutions, and
none of these was kept for more than 2 days.
Solutions of 1 V F potassium iodide were prepared by dissolving
weighed samples of the solid in 0.005 VF sodium carbonate.
A solution buffered to p H 8 was prepared by dissolving solid
sodium hydroxide in a solution 0.25 VF in sodium dihydrogen
phosphate until the correct p H was obtained as measured with a
Beckman pH meter.
Other buffered solutions were prepared in the following way:
The weight of a salt corresponding to one of the ionic species used
in the buffer solution sufficient to give a solution a t least 0.2 VF
in that salt was dissolved in water. The corresponding acid or
sodium hydroxide was added until the dcsired pH was obtained.
332
AND ERNEST 1. SWIFT
Limited range pHydrion indicator paper was used instead of the
pH meter for determining pH values greater than 10.
In all the above solutions, reagent grade chemicals were used.
The laboratory distilled water was often found to contain small
amounts of an unknown oxidizing agent. This oxidizing agent
n as removed by boiling the distilled water for about 20 minutes,
and then bubbling nitrogen through it for about 30 minutes.
Only water treated in this way was used.
Apparatus. The apparatus wm essentially the same as that
described by Aleier, Myers, and Swift (4),with the following
changes: In place of the laboratory direct current supply, the
voltage of which was found to vary slightly, use was made of a
simple voltage-regulated rectifier which was connected to the
alternating current line through a voltage-regulating Sola trans-
former. The generating cathode was enclosed in a shield rather
than the anode. Finally, except where noted, nitrogen gas, after
being passed successively through 6 VF sodium hydroxide and
soda lime, was maintained above the solutions being titrated
Preliminary Adjustment. The current of the generation circuit
was determined by measuring the voltage drop across a standard-
ized 199.87-ohm resistance through which the generation current
was passing. At the low rate of generation, this current vas
found to correspond to the oxidation of 1.0385 X low8equivalent
per second, while the high rate current corresponded to 1.0393
X 10-7 equivalent per second.
The indicator potential was measured with a Gray Instrument
Companv Queen potentiometer.
One yolume formal sulfuric acid was placed in the cathode
shield.
Before each set of titrations the indicator electrodes were
ihorted to the generator anode, a potassium iodide solution n-as
placed in a titration cell (a 40 X 80 mm. weighing bottle), and
iodine was generated in the solution for 100 seconds a t the high
rate.
Titration Procedure. The initial step in carrying out a set of
titrations was the determination of the blank time. In order to
do this, a number of blanks were run. For each blank, 5 ml. of 1
T'F potassium iodide, 20 ml. of the buffer of pH 8, and 25 ml. of
water were placed in a titration cell. The indicator potential was
adjusted to an initial value of 0.15 volt. The generation current,
passing through a dummy resistance, was then adjusted to the de-
sired value. Finally, iodine was generated for short intervals of
time depending upon the rate of generation used (0.5 second for
the high rate, 2.5 for the low). Twenty seconds were allowed
after each interval of generation for equilibrium conditions to be
approached, and the indicator current and the generation time
were recorded, A plot of indicator current us. time of generation
was made, and the linear portion of this plot, from approximatelv
10 t o 40 pa., was extrapolated to zero indicator current. The
average value of the generation time intercepts for a group of
blanks was designated as the blank time.
To maintain the sensitivity of the indicator electrodes, they
were shorted to the generator anode, and iodine lyas generated
for 50 seconds a t the high rate after each blank and every titra-
tion. Then the solution was removed from the apparatus, and the
electrodes were rinsed with water.
After the blank time had been determined, a group of titrations
TI as made. Twenty-five milliliters of a standard arsenic solution
nere pipetted into a titration cell. To this were added 20 ml. of
the buffer of pH 8 and 5 ml. of 1 VF potassium iodide, and the
cell was rapidly attached to the apparatus. The generation cur-
rent was adjusted, and titration was begun. Occasionally during
titrations, small adjustments had to be made on this current.
V O L U M E 22, NO. 2, F E B R U A R Y 1 9 5 0 333

When the indicator current began t o rise, the titration was 0.10 V F in potassium iodide. The desired pH value was obtained
stopped, and a plot of indicator current us. time of generation was by taking 45 ml. of either 0.5 VF sulfuric acid, 0.10 VF sodium
constructed from data obtained in exactly the same manner as in hydroxide, 1.0 VF sodium hydroxide, or one of the following
the case of the blanks. The linear portion of this plot was also buffer solutions: pH 2, sulfate-monohydrogen sulfate; p H 3,
extrapolated to zero indicator current, and the generation time dihydrogen phosphate-phosphoric acid; pH 5, acetate-acetic
intercept was designated as the titration time. The blank time acid; pH 7 , monohydrogen carbonate-carbonic acid; pH 8,
was subtracted from the titration time to give the corrected titra- monohydrogen phosphate-dihydrogen phosphate; pH 9 and pH
tion time. The corresponding weight of arsenic was calculated 10, dihydrogen borate-boric acid; p H 11, carbonate-monohydro-
from the values of the corrected titration time and the rate of gen carbonate; pH 12, phosphate-monohydrogen phosphate.
geneiation.
DISCUSSION
Potential Difference between Indicator Elec-
trodes. In order to minimize effects due to changes
in the internal and external resistance of the in-
dicator circuit, it is desirable to choose an indica-
tor potential which is within the range wherein the :: 2
curve of indicator current us. indicator potential is g
flattest, and which is as high as practicable within
that range. This initial potential value was deter-
m i n d in the following way: Iodine was generated
in a. blank solution until an indicator current of 20
pa. was obtained when the potential difference be-
tween the indicator electrodes was 191 mv. Then
this potential difference was varied, and both it and
the indicator current were recorded. The data 2
thus obtained were plotted, and the resulting graph
is shown in Figure 1. This graph is essentially in
agreement with similar ones obtained by Brunner '
( 2 ) . Because the flattest portion of the curve lies
between 100 and 160 mv., an indicator potential of 10 20
150 mv. was chosen and was used for all subsequent GENERATION TIME (SECONDS)
WOlk. Figure 2. Indicator Current VS. Time of Generation at Various
Hydrogen Ion Concentrations

Figure 2 shows typical data obtained a t p H 9 or less. The plots

have been corrected for the presence of oxidizing or reducing
agents in the reagents used in preparing the buffers, by extra-
polating the linear portions of these curves t o zero indicator cur-
rent, and by calling the intercept on the generation time axis zero
generation time.
That the curves are linear over this range of p H valucs is
evident, and the current readings were stable to *0.2 microam-
pere for a t least 3 minutes. No conclusions can be drawn from
+
z the slopes of these curves because the sensitivity of the indicator
K
W
electrodes varied sufficiently over a period of a few d a y to ac-
a count for the different slopes shon-n.
K
0

0 Table I. Rate of Decrease of Indicator Current

L Indicator Current, Rate of Decreasc
PH pa. ra./Sec.
10 40 0.020
25 0.014
11 48 0.07
37 0.036
12 45 0.09
40 0.07
I 0.1 02
POTENTIAL D I F F E R E N C E
B E T W E E N INDICATOR E L E C T R O D E S ( V O L T S )
In solutions having pH values of from 10 to 12 the indicator
Figure 1. Variation of Indicator Cur- current was not stable, and its rate of decrease could be deter-
rent with Potential Difference between
Indicator Electrodes mined. A t given values of pH and current these rate measure-
Iodine concentration 2 X 10 -8 V F
ments agreed to an accuracy of * 15%, and a series of such measure-
ments is shown in Table I. TThen indicator current readings were
made with only 10 seconds being allowed after each period of
Effect of pH on Indicator Current. In order to determine the generation for stabilization of the current, the curves of indicator
effect of p H on the magnitude and constancy of the indicator current us. generation time were found to be linear. Indeed, a t all
current, blanks were run in solutions of various pH values in the pH values of less than 12 only between 1 and 5 seconds were re-
manner described above, and plots of indicator current us. genera- quired after each period of generation for the indicator current to
tion time were constructed. reach a value within 0.2 microampere of the value taken. In
In all these blanks, a potential difference of 0.15 volt was used, solutions in which the concentration of sodium hydroxide was
the total volume of the solution was 50 ml., and the solutions were greater than or equal to 0.10 V F , the rate of decrease of the indi-
334 ANALYTICAL CHEMISTRY
catoi current was so high that no valid determination of a 1lne:ri solutions buffered to a pH of 8 and with an indicator potential of
dependence could be made-for example, in 1.0 VF sodium 0.150 volt, and then plotting indicator current against generation
hydroxide solution, the rate of decrease of this current was 0 25 time. Though the slopes of the threc lines shown differ noticeably
microampere per second. from one another, it was found that if more blanks were run, the
I t is reasonable to assume that these decreases of indicator rui- slopes of successive blanks would tend to become equal. When a
rent with time a t pH values of 10 or greater are due to decreases in sufficient number of blanks, on the order of three or four, had been
the total iodine-Le., zero oxidation state iodine-concentration,
in view of the fact that the indicator current is linearly propoi-
tional to this total iodine concentration. Three mechanisms are Table 11. Confirmatory Titrations
possible by which this might occur: Iodine vapor may be volatil-
Arsenic, Microgranis
ized from the solution; iodine may hydrolyze, forming iodide and Sllinhel ' Taken round Error % Error
hypoiodite or iodate ions; and iodine may be oxidized to iodatc Ib 1 69,43 69 3 -0.12 -0.17
by atmospheric oxygen, as postulated by hlchlpine ( 3 ) . The 2 69,4l 0.05 0.07
3 69.3: -0.08 -0.12
first of these mechanisms cannot be the rate-determining one, be- AV. 69.38 -0.05 -0.07
cause no comparable rate of decrease of indicator current was oh-
1 ;
served a t higher hydrogen ion concentrations. The conclusion of
3
618,O
E !
618.9
0.6
0.9
0.6
0.10
0.15
0.10
McAlpine that air oxidation of the iodine is responsible for its dis- 4 619..: 1.3 0.24
appearance in solutions of low hydrogen ion concentrations e m - .Iv. 618.9 0.9 0.15
not be accepted because it was found that in a solution buffered at
a pH of 12 the rate of decrease of the indicator current wa, the
same within =t5% when an atmosphere of ovygen was maintained
above the stirred solution as nhen an atmosphere of nitrogen nas
so maintained. Therefore hydrolysis is apparently responsible for
the removal of iodine from solutions having pH values of 10 or
greater.
The above observations shon. that the amperometric method
for the determination of excess iodine may be used in solution., in 203, g 0.0 0.00
which the hydrogen ion concentration is greater than or equal to 111 1 1233.1 12.53. 0.8 0.06
9 1254.: 0.9 0.07
mole per liter and with but little difficulty may be extended 3 1253.6 0.5 0.04
to include solutions in which it is equal to 10-10 mole per liter -i 1263 4 0.3 0.02
1253, R 0 5 0.04
The reproducibility with which the rate of decrease of the indica- .1v , 1253.7 0.6 0.05
tor current could be observed suggests that this amperonietric
IIIC 1 1233 1 1253. 0.4 0.03
procedure could be used to study the rate of hydrolysis of iodine a t
~

2 1252.: -0.3 -0.02

very l o v concentrations. 3 1253, 5 0.4 0.03
4 1252., -0.4 -0.03
.Iv. 1263 .'I 0.0 0.00
a Roman numerals indicate stock solutions used.
b Low rate of generation used for this group only; high rate used for all
o t h e r groups here represented.
C I n this g r o u p only, no atmosphere of nitrogen above solution undergoing
0 titration.
,.
Y
~

a
L run, total concentrations of iodine between 2 X 10-6 and 1 X
B o
C N
equivalent per liter could be determined to an accuracy of *l%.
u However, the slopes of the curves shown have only a second-order
3 effect on the blank time, so that for the purposes of this titration
c
z
W only three or a t most four blanks were run to determine each
C blank time.
: e Effect of pH on Coulometric Titration. The upper limit of the
C
t-
hydrogen ion concentration range a t which arsenic may be
u coulometriexllj- titrated using iodine as the intermediate is deter-
5 mined by the rate of the reaction betx-een iodine and arsenious
0 acid. -At a pH value of 7 , this rate was found to be sufficiently
GENERATION TIME (SECONDS1 hjgh for titrations a t the 1 0 1 rate
~ of iodine generation. However,
the high rate of iodine generation could not be used a t this pH
Figure 3. Indicator Current T S . Generation Time for a value because eycess iodine accumulated in the solution before the
Group of Blanks
equivalcrice point, causing a premature rise in the indicator cur-
rent. At a hydrogen ion concentration of mole per liter
neither rate of generation could be used.
Effect of Iodide Concentration on Indicator Current. I n order I n order to investigate the possibility of carrying out the
to determine whether any significant effect was caused by in- selective titration of other reducing agents in the presence of
creasing the iodide concentration above the usual value, the tripositive arsenic, experiments were made in which it was found
following experiment was carried out: In a blank solution, enough that in solutions 0.1 VF in perchloric acid as much as 1100 micro-
iodine was generated to give an indicator current of 25 pa. Eight grams of tripositive arsenic had no effect on the blank time or on
grams of solid potassium iodide were dissolved in the solution and the indicator current-generation time curve. At a p H of 3 the
the indicator current was again noted. The indicator current was iodine reacted with the tripositive arsenic slowly, but a t such a
found to increase 1 or 2 pa. Thus changes in iodide concentration rate that a stable indicator current could not be obtained.
above 0.1 V F have only a small effect upon the indicator current. The lower limit of hydrogen ion concentration is determined by
Accuracy of Determination of Iodine Concentrations. Figure 3 the effect of the rate of hydrolysis of iodine on the determination
shows the curves obtained by running three successive blanks in of the end point; this effect has been discussed above.
V O L U M E 2 2 , NO. 2, F E B R U A R Y 1 9 5 0
Confirmatory Titrations. Table I1 shows the data obtained
from confirmatory titrations which were carried out as described
above. I t is seen that the titration is accurate to approximately
*0.294 over the range of quantities of arsenic investigated.
I t is believed that the factors limiting the accuracy of these
titrations are: the oxidation of tripositive arsenic by atmospheric
oxygen, the preparation and dilution of the standard solutions,
the determination of the end point, and the measurement and con-
trol of the generation current.
Preliminary titrations of 70-microgram quantities of tripositive
arsenic, over which no atmosphere of nitrogen was maintained,
gave results as much as -0.6% in error. In addition, two groups
of titrations (111) were made with t,he same stock solutions ant1
carried out under as closely identical conditions as possible, ex-
cept that the second group was not done under an atmosphere of
nitrogen; the average value for the first group is 0.6 microgram
greater than that for the second. These effects are attributed to
air oxidation, and in all the other titrations shon-n in Table I1 :in
atmosphere of nitrogen v a s provided.
It is thought that a source of error in the above data was the
accuracy with which standard solutions could be prepared. There
is considerably better agreement between titrations within any
one group than between the average titer of a group and the
calculated value of the titer for that group.
Effect of Ammonium Salts on Determination of Nicotine
C. L. OGG, C. 0. m-ILLITS, AND CONSTANTINE RICCILTI
Eastern Regional Research Laboratory, Philadelphia 18, P a .
Ammonium salts interfere in the determination of nicotine by the silicotungstic
acid method by retarding the precipitation. Graphs are presented to show the
effects of concentrations of nicotine, salt, and reagent and time of digestion on
recovery of nicotine. Procedures are proposed to minimize the effect of the salt.
D ETERMISATIOX of nicotine by the gravimetric silico-
tungstic acid procedure is t,heofficial method of the Associa-
tion of Official Bgricultural Cheniists ( 1 ) . Hon-ever, if other all<a-
loids are present-for esamplr, nornicotine-they are precipi-
tated with nicotine and cause erroneously high results.
Ammonia, another volatile base, has not been consitierrd a pos-
sible source of error in this determination, inasmuch as Chapin
(3) stated that no precipitate was formed \Then silicoturigstic acid
was added to a solution containing ammonium chloridc. In this
laboratory, ammonia \vas assumed to have no effect uiitil an at-
tempt was made to determine nicotine in a solution that con-
tained nicotine and a fairly high concentration of :irnnioriiuni
salts. K h e n silicotungstic acid was added to this solution, no
precipitate formed, either immediately or after the solution had
stood overnight. After i t had stood for 64 hours in a refrigerator,
however, an appreciable precipitate formed, indicating that am-
monium salts have a retarding effect on precipitation of nicotine.
Consequently, experiments were undertaken to prove or disprove
this effect and, if it exists, to determine its magnitude.
METHOD
A stock nicotine solution was prepared by placing a weighed
amount of nicotine in a volumetric flask containing sufficient
dilute hydrochloric acid to assure an acid solution after it was
made to volume with water.
Aliquots containing from 10 to 100 mg. of nicotine were taken
from the stock solution and placed in beakers. Four milliliters of
335
The maximum spread in the values obtained for the blank time
as 0.04 second for the high rate of generation and 0.2 second for
the low rate. Two factors are involved in this spread: the
accuracy with which the measurements and extrapolations could
be made, and the random contamination of blank solutions. I t is
probable that essentially these s:rme factors enter into the detcr-
mination of the end point.
The arcuracy to which the generation cur1Pnt could be deter-
mined n as about * 0.0370, and it could easily be controlled 15 ithin
this limit.
LITERATURE CITED
(1) Brown and Swift, J . Am. Chpm. Soc., 71, 2717' (1949 .
(9) Brunner, Z. p h y s i k . Chem., 56, 391 (1906).
(3) XIcAlpitie, J. Chem. Education, 26, 362 (1949).
(4) Meier, Myers, and Swift, J . Am. Chem. Soc., 71, 2340 [ 1949).
( 5 ) Myers and Swift, Ibid.. 70, 1047 (1948).
(6) Sease, Xiemann, and Swift, -4s.~~. CHEJI.. 19, 197 (194;).
(7) Szebelledy and Somogg-i, Z . arid. Chem., 112, 313 (193R1.
( 8 ) Ibid., p. 385.
(9) Ibid.. D . 391.
i.
( l O j I b i d . ; 400.
(11) Wooster, Farrington, and h i f t , AXAI..CHEM.,21, 14.57 (1949).
RECEIVED August 13, 1949. Contribution 1317 from the Gates and Crellin
Laboratories of Cheriiiatry, California Institute of Technology.
+
hydrochloric acid (1 4) Tl-ere added, and t,he solutions xere di-
luted to make the volume 100 nil. after the addition of the requi-
site amount of 12y0silicotungstic acid. One milliliter of this re-
agent was added for each 10 mg. of nicotine. The beakers xere
placed on a steam bath until the preripitate dissolved, or for 30
minutes, and the solutions were then kept i n the refrigcrator a t
about 5' C. for 18 hours, except \vhPIi the conditions of the experi-
nierit required otherwise. The pwcipitate.+ were collected in
tared. ignited Gooch crucibles cwnt:iining asbestos pads. The
precipitates remaining in tlic Iieakei~c\rere loosened with a rubber
policemm, quantitativelj. rransierled to the crucibles, and washed
well with dilute hydrochloric acmicl ( I $. 1000). Thc precipitates
were dried under infrared lamps atid ignited iii a muffle furnace,
first a t 650" C. for 2 hours arid then at 800" C. for 15 minutes.
The n-eight of the residue n-as multiplied by the theoretical factor
0.1141 to convert the n-eight of the oxides to the weight of nico-
tine.
This procedure differs from the .I.O.h.C. official method in two
respects: The precipitate is digested on a steam bath and then
refrigeratedovernight instead of merely standing a t room tempera-
ture overnight, and ignited Gooch crucibles are used for filtering
and igniting the Precipitate rather than filter paper and platinuni
crucibles. Use of Gooch crucibles and suction speeds the filtra-
tion and eliminates possible errors due to the filter paper.
EXPERIJIENTAL
Esperiments were conducted in which the following condi-
tions were varied: concentration of ammonium salts, time of