Coulometric Titrations with Electrolytically Generated Mercury(1 and II) Determinations of Chloride, Bromide, and Iodide EDWIN P. PRZYBYLOWICZ and L. B. ROGERS Department of Chemistry and Laboratory for Nuclear Science, Massachusetts lnstitute of Technology, Cambridge 39, Mass.
Mercurous ion was generated at 1 0 0 current ~ ~ efficiency from mercury-coated gold and mercury pool electrodes. Potentiometric titrations of halides using this generated reagent were then carried out using similar indicator electrodes. Studies using a mercury-coated silver generator electrode indicated that it could be used equally well, although some silver dissolved. Its endpoint break was nearly the same as that for a mercurycoated gold or a mercury pool electrode and much larger than that obtained with pure silver. Using either an amalgamated gold or silver electrode, the lower limits of an individual halide which were determined with an accuracy of better than 1% in a volume of about 55 ml. were 0.24 mg. of chloride (1 X 10-4.VZ),0.067 mg. of bromide (2 X 10-5.\f), and 0.038 mg. of iodine (6 X 10-6M). Smaller amounts of halide down to 0.013 mg. of chloride, 0.0067 mg. of bromide, and 0.0012 mg. of iodide were determined with better than 5 % accuracy. The coprecipitation errors encountered in argentimetric titrations of halide mixtures are also present in this method, whereas the usual adsorption errors in the titration of iodide alone were not observed.
I
SVESTIGATORS ( 7 , 9, 1 4 ) have reported the use of electrically generated silver ion for the titrimetric determination of halides. Lingane (9) obtained greater sensitivity by titrating chloride in 807, ethyl alcohol, thereby rendering the precipitate less soluble. Since mercurous halides are less soluble than their silver analogs, the generation of mercurous ion was studied in the hope of developing a more sensitive titrimetric method for the halides. After the present paper had been submitted for publication, DeFord (4)made reference to some unpublished work by Horn ( 5 ) using generated mercurous ion as a reagent for the determination of the halides.
b) taking a coil of silver wire (0.070 X 10 inches) cleaning it thoroughly in 1N nitric acid, and dipping i t into metallic mercury. The analogous gold electrode was prepared using a 1.0 X 1.0 inch gold foil. Elertrical contact to the gold foil was made with a platinum wire. These mercury-coated electrodes required only infrequent renewal of the surface after every fourth or fifth titration by putting a drop of mercury on the electrode and removing the excess by shaking. Studies using a mercury-pool generating electrode emplojed a J-type glass tube 10 mm. in diameter, which held a drop of mercury in its upturned tip. Electrical contact to the external circuit was made through a platinum wire which ran the entire length of the glass tube. The generator cathode consisted of a platinum foil in a glass tube connected to the main body of the solution through a fine sintered-glass disk. The cathode chamber contained supporting electrolyte of the same concentration as that in the main body of the solution. The level of liquid in this chamber was always ke t above that of the liquid in the titration cell. 4 h e indicator electrodes n-ere prepared in exactly the same may as the corresponding generator electrodes, except that the amalgamated electrodes were somewhat smaller in size. Titrations in which the J-type mercury pool electrode mas used as a generator electrode employed an identical indicator electrode. These indicator electrodes were used in conjunction with a saturated calomel reference electrode Tvhich was connected to the cell through a saturated potassium nitrate liquid salt bridge. The ends of the liquid bridge contained plugs of Corning unfused Vycor glass No. 7930 ( 1 ) . These electrodes were connected to a Ieeds & Northrup pH indicator (Model 7664) and the titrations were folloived by observing the change in potential with time of generation of mercury ions. Originally the output of the Leeds &Korthrup mas fed into a recording potentiometer, but the recorder made the accurate determination of the elapsed time more difficult and was therefore abandoned after the preliminary survey uork had been completed I n all titrations reported here potential readings were observed periodically on the Leeds & Sorthrup pH indicator and the potential-time curve was plotted manually with the aid of a stop watch. Replicate titrations were carried out by titrating to a definite potential corresponding to the equivalence point. PROCEDURE
Samples. The sample was added to the titration cell and diluted to 50 ml. with distilled water, and 5 ml. of the stock solution of supporting electrolyte were added. Mixing the reagents in this order minimized the danger of oxidation by the perchlorate of either iodide or bromide. The equivalence point potential in these titrations was obtained by determining the potential a t which the slope of the curve went through a maximum and from it subtracting 0.018 volt. The 0 018 volt corresponds to the difference in potential between the point of maximum change of slope in the potentiometric curve and the equivalence point potential as calculated from the solubility product (8). From the equivalence point potential the elapsed time can be determined. I n titrating the smallest amounts of halide reported, the potentiometric curves did not have large breaks and were very flat; hence the titrations mere carried out to a definite potential near the point of maximum slope. An appropriate blank v a s then subtracted. In titration curves of mixtures the end point was determined by inspection and was close to the midpoint of the break. Blanks. A 5-ml. portion of the stock solution of supporting electrolyte was diluted to 55 ml. with distilled water and the solution was titrated. The resulting potentiometric curve was then superimposed on an actual titration curve. The time required to reach the same potential as that selected for the end point in the titration of the sample provided the blank correction. The method of correction compensated not only for the blank but also for any error in picking the end point potential on the potentiometric curve because the curves for the sample and the
EXPERIMENTAL
Reagents. All chemicals used were of analytical reagent grade. Standard millimolar halide solutions were made up by n-eight from their respective potassium salts which had been recrystallized and dried a t 110" C. for 2 hours. These solutions were then checked by potentiometric titration mith standardized silver nitrate. The stock solutions were diluted, and aliquots of these diluted solutions were then used in the analyses. Iodide solutions were prepared fresh for each day's runs. Others were not kept longer than 3 weeks. The supporting electrolyte in all of these titrations was 0.5M sodium perchlorate plus 0.023f perchloric acid. A stock solution of 5 X sodium perchlorate (that from Fisher Scientific Co. was found to contain only a trace of chloride) and 0.2M perchloric acid was prepared and 5 ml. of this solution were added to each sample to be titrated. The total volume of the sample was approximately 55 ml. Apparatus. The current source \$-asan electronically regulated supply ( I d ) . The current was determined during each run by potentiometric measurement of the I R drop across a standard resistance in series with the titration cell. The latter was similar to the one used by Lingane (9). Three different anodes were used to generate mercurous ion. These consisted of mercury-coated gold and silver electrodes and a mercury pool. The amalgamated silver electrode was prepared
799
800
ANALYTICAL CHEMISTRY
blank were essentially identical beyond the end point. However, a blank of this type can be applied only to the more soluble
component in the titration of a binary mixture. RESULTS
Chloride. Aqueous solutions containing between 0.21 and 13.60 mg. of chloride in 55 ml. of solution were analyzed using generating currents from 5 to 50 ma. The results are summarized in Tables I and I1 for the amalgamated gold and silver electrodes, respectively. The results show the accuracy to be superior to argentimetric titrations at high dilutions (9). When generating currents of less than 5 ma. were used, the potentiometric curve became very flat. Using solutions containing 80% by volume of methanol and the same supporting electrolyte, titrations of amounts as small as 0.014 mg. were possible using generating currents of 0.4 ma. These results are also included in the tables. Bromide. Aqueous solutions of bromide were titrated under identical conditions in amounts ranging from 0.027 t o 14.0 mg. Because of the lower solubility of mercurous bromide it was unnecessary to use alcoholic solutions to obtain curves with sharp breaks when using generating currents as small as 1 ma. By using alcohol it was possible to determine 0.007 mg. of bromide Tables I and I1 shotv results of representative bromide titrations using both amalgamated silver and gold electrodes. It was found that no correction was required for the blank. Iodide. As anticipated, the titrations of iodide with mercurous ion were the most sensitive of the reactions studied. Of particular interest was the fact that adsorption errors usually encountered in the argentometric titrations of iodide (9) were not observed. Hence the limiting factor was, as with the other two halides, the solubility of the precipitate. Table I shows that the accuracy of the method is good down to 1.1 y of iodide per 55 ml. in aqueous media. Only a t the lower limit was a blank correction (31seconds, found using a current of 10 pa. Muller and ;larflot (6, 11j found that in the volumetric titration of iodide with mercurous ion, negative errors as high as 2% were observed due to the fact that the reaction, Hg,++ = Hg Hg'-, was favored by the presence of large concentrations of iodide due
+
Table I.
Taken.
Mg.
t o formation of the tetraiodomercurate complex (Hg14--). As the iodide ion was depleted the above reaction was not quickly and completely reversed, thereby producing a small error in the equivalence point. This error was not observed in the present work, though in titrating iodide it was necessary to wait 15 to 20 seconds before reading the potential near the equivalence point. This may have been due to readjustment of the equilibrium suggested above. I n preparing both stock and sample solutions of iodide more dilute than l O - 5 M , care was taken to remove oxygen from the water by degassing with prepurified nitrogen. Mixtures. Adsorption effects nere not as great as in the corresponding argentometric titrations, but there appears to be little advantage in using mercurous ion instead of silver ion for anal) zing mivtures of halides. First, though the mercurous halides are less soluble than their silver analogs, the difference in solubility betn-een any t\vo halides is approximately the same for both metals. Secondly, coprecipitation and mixed crystal formation, which introduce errors into the titration of halide mixtures with silver ( f O j , take place in precipitations of mercurous halides as \Tell. Thus, in equimolar mixtures of 10-4.V bromide and iodide, the amount of iodide found was high by 5y0. When the bromide-to-iodide ratio was increased to 5 to 1, this error increased to +127,. Contrary to expectations, the chlorideiodide system seemed t o be a much worse case. When equimolar mivtures of l O - * M iodide and chloride were used, the iodide error was +7%; when the chloride content of the mixture v-as increased to four times that of the iodide, the error for iodide was +23p0. Thus, there is no apparent advantage in analyzing mivtures with mercurous ion. Effect of Acetone. Because acetone is known to form complexes with mercurous and mercuric ions, it \vas thought that a water-acetone medium could be used to differentiate mivtures of halides by preventing the precipitation of the more soluble of the halides. Actually, hov ever, the mercurous-acetone complev is fairly m-eak, so that the selrctivity of this solvent would depend critically on the concentration of acetone in the solvent. The presence of amounts of acetone up to about 576 by volume gave high iesults in the titration of chloride or bromide, In solutions
Titrations of Halides in about 55 RZ1. of 0.5M Sodium Perchlorate Plus 0.02.M Perchloric Acid Using Mercury-Coated Silver Electrodes Cur-
rent, Ma.
CorrectedQ Time, Seconds
so.
Blank
of Trials
Found, hlg.
Std. Dev.. RIg.
AY. Error,
%
Taken.
Current,
11g.
113,
Chloride 13.60 3.415 1.366 0.683
0 0974b 0 04876 0 0487b 0 0487b 0 0487b 0 02446 0 0244b 0 013Gb 13.98 2.680 1.340 1.621 1.340 8
b
49.46 50.00 20.00 10.00
0 80 1 10 0 80 0 70 0 60 0 70 0 60 0 432 50.00 20.00 20.00 10.00 5.00
747 185 185 185 130 265 132 22 1 79 132 363 89 242 33 1 120 164 188 218 93 109 82 340 161 81 196 324
... ... ...
2 0 4.0 13.0 13.0 24 6 13.0 24.6 57.2 24 6 57.2 91.0
0.07 0.20 0.29 0.29 1.6
3
3 3 3 5 3 5
0.06 0.12 0.49 0.55
4
5
s
0.60 0.00 0.53 0.40 0.55 0.79 0.77 0 89 1.3 1.4 1.1 3.7
5 6
;;;; 5
106.1 124.2 106.1 124.2 150 5
6
4
0
6 3
Bromide 3
... ... .. .. ,.
4 4
3 3
14.04 2.673 1.339 1.629 1.345
0.06 0.013 0.004 0.004 0.001
0.43 0.26 0.08 0.49 0.37
0.670 0.268 0.178 0.134 0.134 0.0809 0,0670 0,0670 0 0268
0,0268 0,0268 0,0670 b 0.0?68 5
0.01346 0 0067Ob
5.00 1.10 1.10 1.10 0.80 1 10 0 80 0 50 0 50
0 0 0 0 0 0
40 30 40 40 10 10
Cor. rected" Time, Seconds
NO.
Blank
of Trials
Found, hlg.
Bromide (Contd.) 4 0.669 161 295 4 0.268 195 3 0.178 4 0 1x4 148 0 135 203 ... 5 88 . , . 4 0 0810 102 ... 3 0 OR77 1 60 ... 5 0 0662 fi4 ... 4 0 0265 88 o 0294 ... 3 123 ... 3 0 0304 0 0666 20 1 80 0 0266 3 0 0136 167 0 00651 3 79
4
Std. Der.,
Ax*. Error
hlg.
%,
0,002 0.0003 0.001 0.0003 0.0005 0 0006 0 0004 0 0006 0 0002 o no02 0 0010 0 0003 0 0002 0.0002 0 00008
0.16
0.11 0 00 0 10 0.4,;
0.12 0 90 1. 2 1.1 8.8 12.0 0 60 0.75 1 .5 4.4
Iodide 18.57 3.496 1.748 0.921 0.380 0.175 0.0921 0,0383 0,00897 0.001 19
49 40 10 00 10 00
no
.5 .. ~
1 10 1 10
0 0 0 0
50 097 051 010
28.5
265 133 140 240 120 139 300 130 85
18.47 3.487 1.751 0.919 0.348
o
176
0.0917 0.0380 0.00878 0.00114
0.05 0.01 0.007 0.002 0.004
n no1
0.0003 0.0002 0.0001 0 00002
Elapsed time was determined t o 1 0 . 1 second. Values of corrected time given merely denote approximate time elapsed for replicate titrations Titrations run i n 80% methanol. No end point was obtained in water alone.
0.54 0.32 0.17 0.27 0.34 0.51 0.43 0 78 2 1 4.2
801
V O L U M E 28, NO. 5, M A Y 1 9 5 6 Table 11. Titrations of Halides in About 55 311. of 0.5.1.3 Sodium Perchlorate Plus 0.02,11 Perchloric Acid Using a JIercury-Coated Gold Electrode Taken, Ng.
Corrected" C n i t c n t , Time. .\I&. Seconds
Blank
1-0. of Trials
Found, Mg,
btd De\..,
A\. Err,ir.
Mg.
Chloride
0.7.50 0.ijOb 0.0750h 0 037'2b
10.00 10.00 1.10 0.55
206 206
183
181
...
7.5 77.4 148 2
5 P 4 4
-0 -0 ' 0 -1
23 01
0.0300
0.001 0.001 0.0007 0.0007
0.934
0 0004
-0 -0
OS 79
0.752 0.751
0.0741
i31 61
Bromide 0 !l,j4 0 0931 o.o4;8 0 0191 0 O191h 0 00:13Dh
9.80 1.10 0.50
0.20 0 20 0.10
118 1OP
115 111 114 114
.
3
. . ..
5
, .
, .
..
4 4
5
0.0952 0.0478
0.0188 0 0190 0.00945
o.oow o.000'
0.000:3
to.^' -1 00
0.000OL - 0 . 2 1 0 00007 - 1 20
because it required large quantities of mercury and because the horizontal electrode surface collected precipitate during the course of a titration, thereby tending to give spurious results. This electrode n-as therefore abandoned and instead amalgamated electrodes, which could be handled in the same fashion as solid electrodes, were adopted. Both silver and gold can be coated easily with mercury and are convenient t o use. Akxording to Sidglyick ( 1 3 ) ,ho\\-ever, the silver electrode will hold three times more mercury than the gold. freshly coated silver wire behaves like a pure mercury surface. On standing, sufficient silver dissolves in the mercury, so that the activity of the mercury at the surface is significantly less than unity. As a result, when a n eek-old amalgamated silver electrode was used to generate mercurous ion, some silver ion x a s also generated. Using this 2.5 amperes per electrode at very high current densities-viz., sq. em.-it wa. found that as much as 3 5 silver ion was produced.
Iodide
... ... ...
3 0 769 0 0011 - 0 18 0.154 1.10 4 0.155 0 0004 -0.0i 0.0771 0.50 4 0 0771 0 0002 -0.18 0 00771 0 05 11, 10 0 3 0 00767 0 0000li - 0 . 5 2 0 00154 0.01 120 57.0 6 0 00160 0 003Otl -3.7'0 a Elapsed time was determined t o 0.1 second. Values of corrected tiine given merely denote approximate t h e elapsed for replicate titrations. b Titrations were run in 8% m e t h a n d N o end point was obtained in i\ater alone.
0 7T1
3.00
117 102 1lz
containiiig about l C $ c acetonc no ptentiometi ic curve was obtained for chloride and results for bromide were still high. Potentiometric titration curves of iodide in solutions containing soyo acetone indicate that the ratio of iodide to mercury at the equivalence point is 4 t o 1, meaning that in this medium mercuric ion is being preferentially generated x i t h the subsequent formation of the tetraiodomercurate complex. As a result of these experiments, it !\as concluded that acetone could not be used to advantage to differentiate quantitatively mivtures of halides. Effect of Alcohol. The effect of increasing percentages of methanol on the sensitivity of halide titrations was studied t o determine the cases for which it a-ould be advisable to use alcoholic solutions to increase the sensitivity of the method. It n as found that the addition of methanol decreased the solubility of individual mercurous halides to different degrees. Mercurous chloride lyas most affected and iodide the least. plot of the logai ithni of the apparent solubility products as calculated from the observed potentiometric curves illustrates the relationships of the three halides in various alcohol mixtures Figure 1 shows that it is most advantageous t o use methanol to increase the sensitivity in determining chloride, n hereas little is gained in titrating iodide in 80yc methanol. It n as noticed that in the titrations of iodide in SOYo methanol solution, a visible precipitate did not form but the titrated solution was greenish yellow and clear. This could be due either to the formation of the more soluble inei curie iodide or to greater solubility of mercurous iodide in alcohol. I n any event, it is apparent that the oxidation state of mercuiy did not affect the stoichiometry. Presumably larger amounts of halide than those used in this study can be determined with equal accuracy and convenience using larger generating currents. However, the piocedure described above is well suited for the determination of microgram amounts of the halides. Because the coulometric procedure is sensitive, accurate, and essentially independent of changes in temperature, the present procedure should compete favorably with polarography for the determination of anions 77-liich produce anodic divolution waves a t the dropping mercury electiode. DISCUSSION
A comparison of the various electrode systems was made to ascertain the most practical type to use for the generation of mercurous ion. The use of a mercury pool proved impractical
30
25 L" c
3cc
2
2 (3
sI
20
15
0
50 VOLUME
5
100
METHANOL
Figure 1. Apparent solubility products of mercurous halides as a function of percentage of methanol
The simultaneous generation of silver did not affect measurably, either the stoichiometry in the titration of halidesor themagnitudo of the potential break in the vicinity of the equivalence point. On the other hand, some investigators may prefer to use a gold amalgam because it is a "cleaner" system, in the sense that the standard potential for gold is sufficiently more noble than that of mercury to obviate the danger of anodically dissolving significant amounts of gold. I n comparing the potentiometric titration curves obtained using a gold amalgam and a mercury pool, it was found that they were almost superimposable. Figure 2 illustrates the sensitivity of the various electrode systems for the determination of chloride. It is evident from these curves that either the mercury pool or the mercury-coated gold electrodes are superior to pure silver and very slightly superior to freshly amalgamated silver for the determination of the halides. Practically speaking,
802
ANALYTICAL CHEMISTRY
the mercury-coated electrodes are to be preferred to the mercury pool as a generator electrode because they can be used in a vertical position in which they will collect little of the halide precipitate. During the preliminary studies for which a recording potentiometer was used, a useful relationship between the generating current and the sharpness of the break at the end point was derived. Several investigators (8, 3) have reported that in using potentiometric indication in coulometric titrations it became apparent that the sharpness of the break a t the equivalence point was a function of the magnitude of the generating current. The magnitude of the generating current may then become a limiting factor which will determine the minimum amount of material that can be analyzed in coulometric titrations.
‘
0.1 0
I 200
I 100
I
400
J
400
Tirne(seconds) Figure 2. Potentiometric titrations of 0.892 mg. of chloride using four different generator electrodes with a current of 10 ma.
0 A
n
Mercury pool Mercury-coated gold Mercury-coated silver Silver
This is particularly true if a recording potentiometer is used a t a single chart speed. For such case, one can use the Xernst equation t o calculate the shape of the potential-time curve for different currents. One can then calculate the current required t o produce a given change in potential within a given time interval near the equivalence point-the change being an arbitrary minimum such as that for a “satisfactory” determination of the end point. Ths simplest function which can be derived for dE/dt as a function of the generating current comes from the first derivative of the Nernst expression. Experimentally, however, the instantaneous dE/dt is difficult to measure accurately. Thus it was decided to select a function which would express the potential E as a function of the generating current over a fixed interval of time, The two potentials chosen for comparison were the equivalence point potential, which is dependent only upon the solubility product of the precipitated salt, and a second potential, t seconds past the equivalence point, which was dependent only upon the rate of generation and the volume of solution. Using the Nernst equation it becomes a simple matter t o derive an expression for A E . 0.0591 A E = -_ _ S X F, X v X n i, x t n log
where S = solubility of the precipitated compound, moles per literLe., the concentration of mercurous ion a t the equivalence point F , = faraday, 96,494 coulombs y = volume of solution, liters z8 = generating current, amperes t = time, in seconds, past the equivalence point a t which the second otential is measured n = number oI)electrons involved in the reaction
After arbitrarily deciding upon the minimum change in potential which, for a given time, will give a fairly sharp break at the equivalence point, a corresponding minimum generating current can be calculated. However, this method is only approximate in that factors, such as ionic strength, which affect the solubility, have not been incorporated for the sake of simplicity. This relationship was tested using the titrations of halides. Using values of 60 mv. for AE and 60 seconds for t , it was found that calculated and observed values of AE in the region of the minimum acceptable generating current differed only by A5 mv. For higher generating currents the agreement was closer. The same relationship should hold for titrations involving complex formation, and only minor changes would need to be made before application to redox systems. One can see that when a recorder is used, another variable, the chart speed, must be considered. As progressiveIy smaller quantities of a substance are determined, one is faced with the dilemma of using smaller generating currents together with slower chart speeds to sharpen the end point breaks, or of using the same generating current with faster speeds to produce longer distances which could be measured more precisely. This dilemma plus the fact that a finite time was required to reach an equilibrium potential were the factors which made continuous automatic recording of the data less desirable than discontinuous generation and measurement. At present a study is under way to extend the application of generated mercurous and mercuric ions t o other systems. ACKNOWLEDGMENT
The authors wish to thank James J. Lingane for his helpful criticisms of the manuscript and for suggesting detailed comparison of mercury-coated gold and silver electrodes. The authors are also indebted to the A4tomicEnergy Commission for partial support of this study. LITER4TURE CITED
Carson, W. N., Jr., Michelson. C. E., Koyama, K., ANAL.CHEM. 27, 472 (1955). Cooke, W. D., Reilley. C. K.,Furman, N. H., Ibid., 23, 1662 (1951). Ibid., 24, 205 (1962). DeFord, D. D., Record Chem. Progr. 16, 165 (1955). Horn, H., Ph.D. thesis. Northwestern University, Evanston, Ill., November 1954. Kolthoff, I. AI., Furman, N. H., “Potentiometric Titrations,” 2nd ed., pp. 144-89, Wiley, Sew York, 1931. Kowalkowski, R. L., Kennedy, J. H., Farringon, P. S., ANAL. CHEM.26, 626 (1954). Latimer, W. M., “Oxidation Potentials,” 2nd ed., PrenticeHall, K’ew York, 1952. Lingane, J. J., A P i A L . CHEM. 26, 622 (1954). Lingane, J. J., “Electroanalytical Chemistry,” pp. 100-4, Interscience Publishers, Sew York, 1953. Muller, E., Aarflot, H., Rec. trav. chim. 43, 874 (1924). Reilley, C. K’.,Cooke, W. D., Furman, N. H., ANAL.CHEM.23, 1030 (1951). Sidgwick, N. V., “Chemical Elements and Their Compounds,” Clarendon Press, Oxford, 1950. Tutundaic, P. S., Doroslovacki, I., Tatic, O., Anal. Chim. Acta 12, 481 (1955). RECEIVED for review September 8, 1955. Accepted February 15, 1956,
