of water to the titration cell, prior to the end point, causes a lowering of the observed potential. Beyond the voltage peak, however, addition of small amounts of water causes a potential rise, indicating a different electrode response before and after the end point. In this method, the glass-silver electrodes are not polarized by an external source. To the authors’ knowledge only one previous method, a nonacidimetric procedure, was reported, where voltage peaks were observed with unpolarized electrodes. Bishop (2) obtained these on titration of halides with thiocyanate in absolute ethyl alcohol using silver-antimony electrodes. He obtained S-shaped curves in argentometric titrations rvith glass-silver electrodes. Voltage peak curves in redox titrations were obtained by Bishop using his technique of differential electrolytic potentiometry (3). I n this method, a minute stabilized current was passed between two platinum electrodes. The behavior of the electrodes was reported to depend on the nature of the ions in solution, magnitude of the current, size, and separation of the electrodes. A simple circuit consisted of the electrodes immersed in the solution connected through a series (ballast) resistor to a battery. Potentials were measured with a p H meter. For ultramicroanalysis, Bishop recommended use of a current of 1 to 5 X 10-9 ampere with 0.07-mm. diameter platinum electrodes which were 1 mm. long and 0.5 mm. apart, a 1- to 2-volt battery and a 400- to 1000-megohm resistor. For macrotitrations he suggested use of a 2- to 120-volt battery with a 200-megohm resistor. Recently, Kirsten, Berggren, and Nilsson (9) used a glass-silver electrode pair with an applied potential of 500 mv. in titrations of certain organic and inorganic bases. Sodium tetraphenylborate served as the titrating agent. With one exception, S-shaped curves were reported. An abnormal curve showing a maximum voltage peak was obtained from the titration of o-phenanthroline hydrochloride. The behavior of the glass-silver electrode system in this investigation can be explained by a combination of polarization and chemical effects. Contrary to popular belief, minute currents with an electronic potentiometer are sufficient to produce polarization in an
1744
ANALYTICAL CHEMISTRY
unpoised system. This was demonstrated by Rogers and coworkers (1.2) in their study of the bromine-bromide potentiometric end point where as little as lo-’* ampere mas sufficient to produce polarization. That polarization is a factor in this system was shown by observing the influence of electrode area and rate of stirring. Three electrodes were studied: an 80-mesh silver gauze electrode, 7.3 cm. X 2.5 cm.; a Beckman silver billet electrode with an area of 3.51 sq. cm.; and a silver wire electrode, area 0.112 sq. cm. The potentials observed were proportional to the area of the electrode. The voltage peak of the titration, however, was not displaced. Small changes also were observed in the potential with marked changes in the rate of stirring. Evidence was obtained which suggests that chemical effects influence the behavior of this electrode system. Silver oxide or silver salt formation on the silver electrode surface may contribute free silver ions for the electrode response. In this respect a silver-silver chloride electrode with the glass electrode produced voltage peak titration curves, as did the clean silver electrode which probably had an oxide film. [Anson and Lingane (1) have shown that oxide films exist even on such a noble metal as platinum.] Depletion of the free silver ions a t the electrode surface would be expected to result in the usual electrode behavior. Through the use of a large excess of halide, therefore, the silver should behave as a normal reference electrode. This was shon-n experimentally by titrating benzoic acid in pyridine saturated with sodium iodide. (Sodium iodide is soluble in pyridine.) This titration gave a normal S-shaped curve for benzoic acid. A mixture of hydriodic and benzoic acids gave two Sshaped curves and the recovery of the benzoic acid was quantitative. Potentially, the voltage peak method offers advantages over the usual potentiometric procedures for titrations of acids in nonaqueous solvents. With the glass-silver electrode pair in the pyridine system, the method is rapid and simple. For most of the acids studied, the electrodes respond quickly
S o . 3jl S o . 42, S o . 43,
118.93 138.35 139.10
3-Methylheptane p-Xylene m-Xylene
to changes in cell potential and the potentials are stable. The approach to the end point is easily anticipated and the necessity for plotting the titration curves is eliminated. Apparatus requirements are simple and no special equipment is needed to perform the titrations. The voltage peak method appears suitable for use in automatic titrations. ACKNOWLEDGMENT
The authors wish t o acknowledge the helpful suggestions by L. B. Rogers, Massachusetts Institute of Technology. Thanks are due to R. N. Peterson, who helped obtain some of the data. LITERATURE CITED
(1) Anson, F. C., Lingane, J. J., J. Am. Chem. SOC.79, 4901 (1957). (2) Bishop, E., Analyst 77, 672 (1952). (3) Bishop, E., Mikrochim. Acta 1-6, 619 (1956); Analyst 83, 212 (1958). (4) CundifT, R. H., Markunas, P. C., ANAL.CHEM.28,729 (1956). (5) Fritz, J. S., “Acid-Base Titrations in
Nonaqueous Solvents,” G. Frederick Smith Chemical Co., Columbus, Ohio,
1952. (6) Fritz, J. S., Lisicki, N. M., ANAL. CHEM.23,589 (1951). (7) Fritz, J. S., Yamamura, S. S., Zbid., 29, 1079 (1957). (8) Harlow, G. A,, Noble, C. M., Wyld, G. E. A., Zbid., 28,784 (1956). (9) Kirsten, W. J., Berggren, A., Nilsson, K., Zbid., 30, 237 (1958). (10) Malmstadt, H. V., Fett, E. R., Zbid., 27, 1757 (1955). (11) Moss, M. L., Elliott, J. H., Hall, R. T., Zbid., 20, 784 (1948). (12) Purdy, W. C., Burns, E. A., Rogers, L. B., Ibid., 27, 1988 (1955).
RECEIVED for review February 25, 1958. Accepted June 9, 1958.
Application of Gas-Liquid Chromatography to Analysis of Liquid Petroleum Fracti ons-Co rrect ion In the article on “Application of GasLiquid Chromatography to Analysis of Liquid Petroleum Fractions” [ (D. H. Desty and B. H. F. Whyman, ANAL. CHEM.29,320-9 (1957)], in Table I the correct data are as follorrs: 10.58 25.72 26.02
1.025 1.410 1.116
9.35 55.63 57.78
0.971 1.745 1.762
