solution of 0.05M potassium chloride was 113 pa., the impregnated electrode, used in stirred solution, should be especially applicable to the determination of very small amounts of dissolved ouygen. This sensitivity to dissolved oxygen emphasizes the need for efficient deoxygenating techniques to eliminate interference with other cathodic waves. Reproducibility of Measurements. I n unstirred solution, t h e reproducibility of measurements made with t h e impregnated electrode is excellent. Replicate determinations on a 10-4M solution of dibutylphenylenediamine showed standard errors for a single run of 0.003 volt for Eli* and 2% for id (Table V). The precision of these measurements is equal to and in some cases better than results reported for other solid electrodes (5-6,8-10, 12, I S ) . Stirring the test solution resulted in positive displacement of Ellz but had no adverse effect on its reproducibility; the standard error in measuring id was, however, increased t o 10%. The magnitude of id was greater b y a factor
F., Conrad, -4. L., Landerl, J. H., ANAL.CHE~LI. 27, 310
Gaylor, V.
of 5.5 when stirring was used. The increased sensitivity resulting from stirring will often justify the reduced measurement accuracy. Similarly, the 6% error in measuring id for the cathodic lead wave (Table V) will usually be satisfactory for determining the low concentrations to which the technique is applicable. The reproducibility of Eli2 values for the lead wave \ m s very poor; in general, the method is not recommended for accurate measurement of cathodic potentials.
(1955); 29, 228 (1957).
Gaylor, V. F., Elving, P. J Conrad, A. L., Ibid., 25, 1078 (1953). Hedenburg, J. F., Freiser, H., Ibid., 25, 1355 (1953).
Kolthoff, I. M., Jordan, J., Heyndrickx, A,, Ibid., 25, 884 (1953). Kolthoff, I. M.:,Lin me, J. J., “Polarography, 2 n 8 ed., Interscience. New York. 1952. Kolthoff,’I. M ., Tanaka, X,, AKAL. CHEI) I . 26, 632 (1954). Lord, S. S., Jr., Rogers, L. B., Zbid., 26, 284 (1954).
Marplle,_. T. L., Rogers, L. B., Zbid., 25, 1351 (1953).
Rogers, L. B., Miller, H. H., Goodrich. R. B., Stehney, A. F., Zbid.,
ACKNOWLEDGMENT
The authors gratefully acknowledge the help and encouragement received from Philip J. Elving throughout this study. The work was performed in the laboratories of The Standard Oil Co. (Ohio) and permission to publish is appreciated. LITERATURE CITED
( 1 ) Fieser, L. F., J. A m . Chem. SOC.5 2 , 5204 (1930).
21, 777 (1949).
Skobets, E. M., Atamanenko, N. N., Zavodskaya Lab. 15, 1291 (1949).
Streuli, C. A., Cooke, W. D., ANAL. CHEM.25. 1691 (1953).
Zbid., 26, 963 (1954). ’ Vlcek, A. K., Mansfeld, V., Krkoskova, D., Collegium 1943, 245.
RECEIVED for review December 23, 1955. -4ccepted October 4, 1956. Pittsburgh Conference on Analytical Chemistry and hpplied Spectroscopy, March 1955.
Polarographic Determination of Antioxidants in Gasoline V. FRANCES GAYLOR, ANNE L. CONRAD, and JEAN H. LANDERL Chemical and Physical Research Division, The Standard O i l Co. (Ohio), Cleveland 6, Ohio
A , polarographic method for the direct determination of antioxidants in gasoline i s based upon measurement of the oxidation waves produced at a wax-impregnated graphite electrode. Analysis i s conducted directly on an alcoholic solution of the sample, thus eliminating time-consuming separation steps. The method has been applied to the determination of two commercial antioxidants, N,N’-di-sec-butyl-pphenylenediamine and N-N-butyl-parninophenol. A single sample can be analyzed in approximately 30 to 45 minutes. Standard deviation of the method i s approximately 9%. Concentrations as low us 2 p.p.m. in the gasoline can be determined.
P
methods for the determination of antioxidants in gasoline have mainly consisted of colorimetric procedures, which usually require the preliminary extraction of the antioxidant so that color can be developed and measured in an aqueous solution-e.g., use of the Folia-Denis UBLISHED
228
ANALYTICAL CHEMISTRY
reagent after alkaline extraction (6), of hydrogen peroxide after acidic extraction (6), and of tungstophosphoric acid after acidic extraction (7). The only direct determination of antioxidants in gasoline described is based on ultraviolet absorption (3) ; the difference in absorption between the base gasoline and the inhibited sample is taken as a measure of the antioxidant content. Unfortunately, in many refineries a representative uninhibited base gasoline is difficult to obtain. Considerable error may be associated with the separation step required for colorimetric procedures. Gasoline dyes or other colored or color-producing materials may be extracted along with the antioxidant. I n addition, the antioxidants tend to be unstable and to drconipose during the extraction procedure. The present paper describes a polarographic procedure which can be applied directly to the gasoline sample and does not require an uninhibited base gasoline for comparison. The method is based on the oxidation waves produced by
gasoline antioxidants a t a suitable indicator electrode. The wax-impregnated graphite rod ( I ) is sufficiently sensitive to determine the very low concentrations involved, is suitable for use in organic solvents, and has the anodic potential range involved. EQUIPMENT
Current-potential curves were automatically recorded a t a speed of 1.24 mv. per second, using a Sargent Model XXI polarograph. The wax-impregnated indicator electrode (1) was prepared from a 0.25-inch graphite rod (special spectroscopic grade, Il’ational Carbon Co.). -4 convenient length (6 to 12 inches) of graphite rod was immersed in melted opal wax (E. I. du Pont de Nemours & Co.) or castor wax (Baker Castor Oil Co.), and allowed to stand for 2 hours a t 100” C. The waximpregnated rod was then withdrawn and allowed to cool to room temperature in a vertical position. The outer surfaces of the rod were covered with a n insulating layer of Seal-All (Allen Products Corp.) and allowed to dry. Approximately 0.25 inch of graphite
was then broken off from each end of the electrode. A short piece of rubber tubing containing mercury served a s a contact between the wax-impregnated graphite and the lead to the polarograph. T h e lower end of the waxunpregnated rod was abraded lightly with a fine grade of sandpaper and then inserted into the test cell. A new electrode surface was exposed for each run by breaking off the used end and sanding the freshly exposed surface. A silver-silver chloride wire electrode was used as a n internal reference cathode. Approximately 10 em. of 20-gage silver wire was wound loosely around a rod 0.5 inch in diameter to form a coil of four or five turns. A straight end of the wire was soldered to a convenient length of Sichrome lvire which was then sealed into a glass mercury well contact into which the polarograph lead was dipped. T h e silver wire was coated with silver chloride b y anodic electrolysis at 1.5 volts in 5 N hydrochloric acid for 10 minutes. T h e silver chloride coating was replenished as needed.
711 1 1-
SS T-D RCRTE R
Ag-AgCI
REFERENCE ELECTsC3E
Cell resistance was measured with a conductivity bridge (Model RC-BC. Industrial Instruments Co.) ; all polarograms were corrected for iR drop. The usual cell resistance was 10,000 to 20,000 ohms and iR corrections as great a s 0.2 volt were encountered. An automatic iR compensator (,$) was used in much of the experimental \Tork. REAGENTS
Commercial forms of two antioxidants were used without further purification. One consisted of a solution which was 50% S-n-butyl-p-aminophenol, 20% anhydrous methyl alcohol, and 30% anhydrous isopropyl alcohol ( S o . 5 antioxidant. E. I. d u Pont de Semours & Co.). N,N'-di-sec-butyl-p-phenylenediamine (Tenamene 2, Tennessee Eastman Co.) was obtained as a liquid claimed to be pure. These antioxidants were stored at 40" F. under nitrogen. Other chemicals used were barium naphthenate (16y0barium, Harshaw Chemical Co.), lithium chloride (Baker, c.P., anhvdrous), isopropyl alcohol (Fisher, c.P.), and iso-octane (2.2,4-trimethy1pentane, Phillips Petroleum, c.P.). Aqueous buffer stock solutions contained 1 mole per liter of the major buffer component; acid or base was added until the pH of a 1 to 1 mixture with alcohol reached the desired value. Buffer components were potassium chloride-hydrochloric acid for p H 1.3; sodium acetate-acetic acid for p H 5.1; and ammonium chloride-ammonium hydroxide for p H 8.5. DEVELOPMENT
Figure 1 .
Cell assembly
The assembled electrodes and test cell are illustrated in Figure 1. The waximpregnated graphite electrode n as inserted into the silver-silver chloride coil so that the tip of the rod extended slightly beyond the coil into the test solution. This arrangement resulted in minimum cell resistance and reduced fluctuations caused b y physical displacement while in use. The electrodes were preconditioned in the test solution before each run b y application of a potential equal to 90% of the total potential to be applied. Two minutes of this treatment was followed b y equilibration a t the initial potential for 3 minutes. Test solutions were stirred by means of two blades welded to a shaft rotated at a constant speed of 300 r.p.m. b y a synchronous motor (Type KYC-23, Bodine Electric Co.). Surfaces of the metal stirrer \+hich came in contact with the test solution mere coated with a thin layer of Seal-All. T h e cell (25-ml. capacity) was jacketed with provision for circulation of water, controlled a t 25" f 0.2" C. The cell could also be equipped with gas inlets for purging the contents and atmosphere with inert gas; removal of oxygen is not essential.
OF
METHOD
Effect of pH. I n buffered 1 t o 1 alcohol-water solution, E I I Pvalues for t h e two antioxidants became less positive with increasing pH (Table I). Ell? values for t h e two compounds were almost identical in either strongly acidic or basic solution; maximum separation occurred at p H 5.1. Limiting currents increased with increasing pH. However, t h e effect of p H on id was not sufficiently great t o influence seriously t h e choice of a medium for quantitative measurement. Selection of Solvent-Electrolyte. For maximum sensitivity, a test solution for t h e determination of antioxidants in gasoline must contain a large concentration of t h e hydrocarbon sample. I n addition, appreciable
Table
I.
quantities of a conducting salt and a polar solvent are required to permit the flow of electrical current. A 0.2.11 solution of lithium chloride in isopropyl alcohol possesses good conductivity properties and is capable of dissolving large amounts of hydrocarbon. .A 1 to 1 niivture of the solution with gasoline resulted in a homogeneous solution; the apparent p H was 5 and the cell resistance was about 10,000 ohms. Effect of Stirring. K a v e splitting occuired when dibutylphenylenediamine n a s polarographed in t h e unbuffered 0.1M lithium chloride medium described; t h e two anodic waves had Ellz values of -0.028 and f0.267 volt. T h e total i d was such t h a t sensitivity of quantitative measurements based upon the sum of the two uaves was very poor. The lower limit of detectability \\as about 10 p.p.m. of antioxidant in gasoline. Stirring the test solution resulted in convergence of the two waves and greatly increased id; a concentration of 5 p.p.m. gave a \
