ANALYTICAL CHEMISTRY
1018 peatability to get poorer as the nitrogen content increases. I n spite of this, the repeatability is much better in the sample containing 1.301, nitrogen than i t is on the sample containing 1.0% nitrogen. More data on intermediate samples, or a more involved statistical analysis, are needed to explain this result. The regular and the modified methods of determining nitrogen are compared in Table XIV. The modified method tends t o give lower nitrogen values. Obviously, the method giving the best value can be determined only by running samples of known nitrogen content. There seems to be little difference between the methods from the standpoint of reproducibility. ACKNOWLEDGMENT
The authors want to thank those who helped in the preparation of this paper. I n particular, they are grateful for the help and advice given by G. V. Dryoff, M. A. Efoymson, C. T. Shewell, and Joseph Stewart.
LITERATURE CITED (1) Davies, 0. L., ”Statistical Methods in Research and Production,” 2nd ed., p. 60, London, Oliver and Boyd, 1949. (2) Dean, R. B., and Dixon, W. J., ANAL.CHEM.,23, 636 (1961).
(3) Dixon, TV. J., and Massey, F. J., “Introduction to Statistical
Analysis,” pp. 145-6, New York, McGraw-Hill Book Co., 1951. (4) (5) (6) (7)
Lake, G. R., ~ ~ N A LCHEM., . 24, 1806 (1952). Mitchell, J. A., Ibid., 19, 961 (1947). Peakes, G. L., A S T M BUZZ., 179, 57-61 (1952). Villars, D. S., “Statistical Design and Analysis of Experiments for Development Research,” p. 120, Dubuque, Iowa, Wm. C. Brown C o . , 1951.
( 8 ) Ibid., p. 276. (9) Wernimont, G., A S T M Bull., 166, 45-8 (1950). (IO) Wernimont, G., ANAL.CHEM.,23, 1572 (1952). (11) Youden, W. J., “Statistical Methods for Chemists,” p. 50, New York, John Wiley & Sons, 1951. (12) Zbid., p. 69. RECEIVED for review May 6, 1953. Accepted hlarch 16, 1954. Presented at Group Session on Analytical Research. 18th Mid-Year Feeting, Division of Refining, American Petroleum Institute. New York, N. Y . , May 11, 1953.
Effect of Aviation Fuel Components on the Accuracy Of the Karl Fischer Electrometric Titration R. W. SNEED, R. W. ALTMAN, and J. C. MOSTELLER Wright Air Development Center, Wright-Patterson Air Force Base, O h i o
The Karl Fischer electrometric titration method for quantitatively determining water has been considered satisfactory for unleaded gasolines, light and dark oils, and, with slight modification, greases and sludges. With the advent of wider specification limits for gasoline and jet propulsion fuels, which permits the inclusion of additional naturally occurring materials, and with new or higher concentrations of fuel additives, the validity of results by this procedure is questionable. It has been theorized that finished fuels now contain materials, natural or added, which influence the accuracy of results obtained by the Karl Fischer procedure. This work was undertaken to determine if this method is practical for determining the water content of aviation fuels. The study conducted determined the effect of as little as 0.001 weight %of tetraethyllead, aromatics, olefins, mercaptans, and oxidation inhibitors on the accuracy of water determinations. Of the compounds investigated, only mercaptans appear to show any appreciable interference.
T
HE Karl Fischer electrometric titration is now a standard
analytical procedure for the determination of moisture content of a large number of materials (8). I n the majority of applications this procedure has been considered satisfactory for the accuracy required. However, there have been considerable disagreement and discussion regarding the applicability of this procedure to aviation gasoline and jet propulsion fuel. This disagreement arises from the fact that these fuels may contain quantities of materials xhich cause reproducible inaccurate results. For example, several members of industry believe that this method causes erroneous results for fuels containing tetraethyllead, because the lead compound reacts with, and hence consumes, the Karl Fischer reagent in much the same manner as water. Data obtained by other organizations indicate that tetraethyllead does not interfere with the accuracy of results. No data have been presented in the literature specifically on the effect of fuel components on the determination of water.
This uncertainty was of particular interest to the U. S. Air Force, which is a major consumer of aviation fuels and must operate in areas where climatic conditions could cause excess moisture in the fuel to crystallize, resulting in malfunction of equipment. Therefore, this program was undertaken t o check conflicting literature and to determine other compounds that may affect the accuracy of results. The project was conducted, keeping in mind the military fuel specifications. It \vas assumed that accuracy to &0.001 weight % ’ water was sufficient for Air Force purposes. The Karl Fischer electrometric titration has been employed in this attempt t o determine specifically the effect of several of the more predominant, and usually disputed, components of fuel blends on .the accuracy of the method. Effects of aromatic, olefin, and mercaptan (thiol) compounds, tetraethyllead, and oxidation inhibitors on the accuracy of results have been investigated. Information obtained shows that fuels need not be specially treated to remove detrimental compounds before the determination of water. METHOD AND APPARATUS
The method of Karl Fischer has been widely employed by research workers and has now become a routine tQ01 for the determination of water in a wide variety of materials or products, including hydrocarbons (1, 3, 4, 11). Previous data indicated that the Karl Fischer reagent was quantitative and specific for small amounts of water. This reagent, a solution of pyridine, sulfur dioxide, and iodine in anhydrous methanol, is normally dark brown in color. A change in this color is usually evidenced at the end point. However, in most instances an electrometric procedure is preferred to the colorimetric technique because of the accuracy in detecting the end point, and because it permits determinations in dark materials where the color end point would be obscure. The electrometric method functions on the basis that an excess of the reagent will conduct current. When a slight excess of water is present, one of the electrodes becomes POlarized, and the current decreases, approaching zero. Thus, the end point may be determined by a dead-stop technique. Karl Fischer reagent, equivalent to about 3.5 mg. of water per
V O L U M E 26, NO. 6, J U N E 1954
1019
ml., was used for the titration of 25 to 40 grams of the sample. The solvent used t o render the system homogeneous consisted of 1 volume of anhydrous methanol plus 3 volumes of chloroform. The specific method, apparatus, and calculations employed have been described in detail (IO). DISCUSSION
Six fuels and an iso-octane were employed in this study. Four gasolines of varying aromatic and olefin content were employed t o study the effect of these materials on the accuracy of the Karl Fischer technique. A single jet propulsion fuel with varying mercaptan content was used. The effect of tetraethyllead on the procedure was investigated with iso-octane as the carrier. A single gasoline containing inhibitors was employed in the investigation of inhibitor effect on the method. This was considered a sufficient number t o allow conclusions to be dran-n from the data obtained.
accuracy of the Karl Fischer method. It is possible that olefinic compounds other than those in the base fuels and those added could have some influence on the data. However, this program indicates that olefins will probably be of little consequence. Because of the presence of more double bonds per molecule, diolefin materials, such as used, probably offer greater difficulty in obtaining accurate results than any mono-olefintype hydrocarbons.
Table 11.
Effect of Olefin Concentration on Water Determination
0.0036 0.0037
Ar.
0.0035 0.0042 0 ,004
Total Water, % CalcuDeterlated mined 0.0075 0.0066 0.0075 0.0076
0.0046 0.0043 0005
0.0036 0.0037
Av.
0.0081 0.0082
0,0078 0.0089
0.0048
0.0043 0.0044
0.0088 0.0089
0,0081 0.0080
Water,
%
Fuela C
D
Table I.
Effect of 4romatic Concentrations on Water Determination
Av.
0 0051 0 0043 0 005
0 0040 0 0037
Total Water CalcuDeterlated mined 0 0087 0 0084 0 0084 0 0079
Av.
0 0052 0 0054 0 005
0 0038 0 0038
0 0091 0 0091
0 0086 0 0088
Av.
0.0072 0.0070 0,007
0.0044 0,0041
0.0115 0.0112
0.0098 0.0107
Water,
c70
Fuel"
4 B
+
A
17 c70 toluene
B
+ 12% toluene
Water Added,
%
0.0064 0,0039 0.0103 0.0111 0.0063 0.0041 0,0105 0.0103 Av. 0 . 0 0 6 a Fuel . I contains 0.0% aromatics and 11.0% olefins. Fuel B contains 14.1% aromatics a n d 0.9% olefins.
Aromatic Compounds. The percentage of water in samples A and B was determined in accordance with the procedure (IO). To each a known amount of water was added, using absolute methanol as a carrier. The water content of each fuel was then determined to see if the difference in aromatic content of the two fuels would affect the results obtained for the total water content. To test further the effect of the aromatic content, 12% by volume of toluene was added to each of the above fuels, and the water content was determined by the Karl Fischer method. The calculated water content, obtained by adding the percentage of water initially in the fuel t o the known percentage of water injected, was compared to the water content as determined by the Karl Fischer method. Fksults of the above tests are tabulated in Table I. The water concentration figures calculated in the manner given and those obtained by employing the Karl Fischer procedure agree within =kO.OOl%, which is the limit of reproducibility of the equipment and procedure employed. Therefore, it appears that in determining moisture in aviation fuels, a high aromatic-containing sample has relatively no more effect on the accuracy of the test than a low aromatic-containing sample. Olefin Compounds. Fuels C and D, typical fuels with a low olefin content, were selected so that the effect of the addition of small quantities of olefins would be amplified. Bfter the initial water content of each of the fuels was obtained by the Karl Fischer method, 12% by volume of diisobutylene was added to each in order to increase the olefin content. The water content was then determined on the newly blended, higher olefin fuel. The results, presented in Table 11,indicate that the concentration of the olefins used in this investigation does not tend to alter the
C
+
12% diisobutylene
Av.
0.0042 0.005
Water Added,
%
D , + 12% diisobutylene
0.0045 0.0040 0.0084 0,0080 0.0042 0,0042 0.0086 0.0089 Av. 0.004 a Fuel C contains 3.57, aromatics and 2.5% o l e f k . Fuel D contains 17.07. aromatics and 0.87, olefins.
Mercaptans. I n vieTT of the fact that jet propulsion fuels in present production allow a maximum of 0.005 weight 70mercaptans, their effects on a typical fuel, JP-3, were investigated. Two mercaptans were used in these tests, a four-carbon and a twelve-carboh compound. Table I11 shows that the addition of these mercaptans to a base fuel increased the figures obtained for its water content proportionately. The addition of 0.005 weight yoof either four- or twelve-carbon mercaptans gave results within the limit of reproducibility of the method, and may not actually be considered as causing an increase in the values.
Table 111. Effect of Mercaptan Concentration on Water Determination Water,
%
Fuel J P - 3 (sample 1691) J P - 3 (1691) tan
+ 0.005%
0.0027 0.0026 tert-butyl mercap-
+ 0.0Z70tert-butyl mercaptan J P - 3 (1691) + 0.005% tert-dodecyl mercaptan J P - 3 (1691) + 0 02% terf-dodecyl mercaptan J P - 3 (1691)
Average 0.003
0.0031 0,0029
0.003
0.0049
0.005
0.0055 0 0029 0,0038
0.003
0 0052 0 0048
0 005
However, in anticipation of higher mercaptan concentrations, a 0.02% addition was tested. Because this gave considerably higher results, which were beyond test reproducibility limits, it may be assumed that the addition of the lower concentrations caused higher results, even though within the limits of the test. There appeared to be no variations between the effect of the high and low molecular weight mercaptans on the method. For fuels in which the mercaptan concentration is known, an empirical correction probably can be applied to the apparent water content. Although not ascertained, it is theorized that other readily oxidizable or reducible sulfur compounds, if present in the fuel, will cause larger or smaller errors, respectively, in the accuracy of water content as determined by this method.
ANALYTICAL CHEMISTRY
1020 Tetraethyllead. A number of test samples were prepared using different concentrations of tetraethyllead in iso-octane, covering the range of tetraethyllead permitted in current military aviation gasoline specifications. I n one series of tests iso-octane, as received from the manufacturer, was used with known quantities of tetraethyllead added. I n a second series of tests the iso-octane, as received, was saturated with
Effect on Water Determination of Addition of Tetraethyllead to 80-Octane Fuel
Table VI.
Tetraethyllead, MI.per Gallon 80-Octane 0
6
Table IV.
Water,
%
Average 0.004
0.0047 0.0047 0.0038 0.0049 0.0052
0,005
Effect on Water Determination of Addition of Tetraethyllead to Iso-octane
Tetraethyllead, MI. per Gallon Iso-octane 0
Water,
70
0 0014 0 0013 0 0021
CONCLUSIONS Average 0 002
1
0 0026 0 0017
0 002
2
0 0030 0 0005
0 002
3
0 0018 0 0020
0 002
4
0 0018 0 0019
0 002
5
0 0012 0 0015
0 001
6
0 0015 0 0014
0 002
S o concentration of aromatics, olefins, tetraethyllead, or oxidation inhibitors, likely to be encountered in the near future in aviation fuels, would interfere to any appreciable extent with the accuracy of water determinations by the Karl Fischer electrometric titration method.
Table VII.
Effect of Inhibitors on Water Determination Water,
%
Fuel Ea
.
0.0058 0.0051
+
Fuel E 1 lb./5000 gal. (2,6di-tert-butyl-4-methylphenol)
+
water prior to the addition of the lead compound. Tables 11and V show the results of these tests, which indicate that tetraethyllead does not affect accuracy in water content determinations in concentrations of 0.0 to 4.6 ml. per U. S. gallon, which is the current specification range used in aviation fuels. A sample of SO-octane unleaded gasoline was obtained and the percentage of water determined before and after the addition of tetraethyllead. The results, shown in Table VI, verify the literature statement that tetraethyllead does not interfere with the Karl Fischer method.
Table V. Effect on Water Determination of Addition of Tetraethyllead to Iso-octane Saturated with Water Tetraethyllead, M1. per Gallon 180-octane 0
Water,
%
Average
0.0067
0.007
Fuel E 1 lb./5000 gal. (2,4dimethyl-6-tert-butylphenol)
a
0.0056 0.0049
0,005
0.0061
0.006
0.0054
+
Average 0.006
Fuel E 1 lb./5000 gal. 0 0059 ( N ,N’-di-sec-butyl-p0.0053 phenylenediamine) Fuel E is uninhibited high-octane aviation gasoline.
0.006
Concentration of mercaptans as high as 0.02% and as low as 0.005% cause higher values for m-ater because of a mercaptan reaction which consumes part of the Karl Fischer reagent. However, since no current military aviation fuel specification allows more than 0.005% of mercaptans, it is unlikely that appreciable errors will arise from this source. These tests indicate that this method is reliable and practical for determining the water content of aviation fuels manufactured under current military procurement specifications.
0,0058
0.0076 3
0.0062
0 0057
REFERENCES 0.006
0.0071 6
0.0057 0.0083 0.0070
0.007
10
0.0085 0.0075 0.0067
0.008
Oxidation Inhibitors. Present fuel specifications permit the addition of up to 1.0 pound per 5000 U. S. gallons of certain oxidation inhibitors. To obtain the ultimate effect of these materials on the method under consideration, the maximum antioxidant concentration was employed. To uninhibited high N,N‘octane aviation gasoline, 2-6-di-tert-butyl-4-methylpheno1, di-sec-butyl-p-phenylenediamine,and 2,4-dimethyl-6-lert-butylphenol were added individually, and the water content Kas determined. I n comparison with tests on the base uninhibited fuel no effect on the accuracy was exhibited. Results of these tests are tabulated in Table VII. All inhibited samples appear to behave similarly under conditions of the test.
(1) Bcker, M. M., and Fredian, H. A., IND.EKG.CHEM.,ANAL. ED.,17, 793 (1945). (2) Aepli, 0. T., and McCarter, W. S. W., Zbid., 17, 316 (1945). (3) Almy, E. G.,Griffin, W. C., and Wilcox, C. S., Ibid., 12, 392 (1940). (4) California Research Corp., Sen Francisco, Calif., “Determina-
tion of Water in Aviation Gasoline Containing Tetraethyllead,” Nov. 12. 1942. (5) Fischer, Karl, Angew. Chem., 48,394 (1935). (6) Krynitsky, J. A., Crellin, J. W., and Carhart, H. W., Naval Research Laboratory, Rept. 3604 (1950). (7) MaoNevin, William, Ohio State University, Final Rept. 24 (Contract W33-038-ac-16679,June 30, 1949). (8) Mitchell, J., Jr., and Smith, D. M., “Aquametry,” New York, Interscience Publishers, 1948. (9) Roberts, F. hl., and Levin, H., ANAL.CHEM.,21, 1553 (1949). (10) Rubin, B., and Burger, R. J., Air Force, Wright Air Development Center, Ohio, Tech. Rept. 5944 (1950). (11) Smith, D.M., Bryant, W. M. D., and Mitchell, J., Jr., J. Am. Chem. Soc., 61, 2407 (1939). (12) Snyder, R. E.,and Clark, R. 0.. “The Karl Fischer Method for
the Determination of Water in Petroleum Products,” Pittsburgh, Pa., Gulf Research and Development Co., July 3, 1947.
RECEIVED for review J u n e 2, 1953.
Accepted February 26, 1954.
