V O L U M E 2 6 , NO. 3, M A R C H 1 9 5 4 I
I
I
i
567 I
I
-
-
+ACONITIC +OXALIC
.4 0
-7
+MALIC +CITRIC +TARTARIC
1
t
*20
I
*05L
‘A.
.35 -
I
I 10 0
Figure 2.
I 200
I
300
x -
I
I
I
400
500
600
ML.
Peak Volume-Concentration Relationships
The values of m and b were determined from the experimental points by the method of least squares. The peak volume-concentration equations for the five acids are‘ Acid Aconitic Oxalic Malic Citric Tartaric
log ( 4 - Y ) - Y) - Y) log ( A - Y ) log fA - Y ) log (a log ( A
= -0.00513X = -0.00103.Y = -0.00308X = -0.00294X = - 0 00270S
++ f ++
0.268 0.225 0.0786 0.162
0.153
Each equation is characteristic of the particular acid and is useful in identification of the acid. The following opposed factors influence the choice of a solvent system for a particular separation:
1. Peaks are higher ( a ) the shorter the column; ( b ) the less the eluent volume required. 2. Separation is better ( a ) the longer the column; ( b ) the larger the eluent volume.
The desired system is the one which will separate the acids and give the highest peaks. A 10-gram, 1-cm. column will separate all the sorghum acids. Since citric and tartaric acids do not separate on a 6-gram column and do separate on a 10-gram column, columns of less than 10 grams would be preferable only if the acids to be separated do not include citric and tartaric. A graph such a3 that of Figure 2 may be used to determine the least volume a t which good separation will occur. The A = 0.iO system, although it requires the least volume, does not separate malic and oxalic acids, as their peak volume-concentrations are too close to each other to allow satisfactory fractionation. The d = 0.35 system gives the best separation, but peaks are flat and spread out. This makes it difficult to determine where elution of the acids began and increases the blank error. -4s the graph indicates, the A = 0.60 system provides separation {rith the least volume and was chosen for the separation of these acids. Such graphs and characteristic volume-concentration equationq can be found for other acids and other systems, thus aiding the investigator in identification of the acids and affording a means of predicting the best system to try. The system of progressively changing solvent concexitration and the resulting graphical and mathematical use of results can be applied to many other separations previously using stepniqe change of eluent. LITERATURE CITED
(1) Cherkin, A,, Martinez, F. E., and Dunn, 11, S., J . Am. Chem. SOC.,75, 1244 (1953). (2) Donaldson, K. O., Tulane, V. J., and Marshall, L. RI., .%NAL. CHEM.,24, 185 (1952). (3) hlader, C., and hlader, G., Ihid., 25, 1423 (1953). (4) I b i d . , p. 1556. (5) Mader, C., and Rlader. G., Proc. Oklahoma Acad. Sci., in press. (6) Marshall, L. AI., Donaldson, K. O., and Friedberg, F., ANAL. CHEM.,24, 773 (1952). RECEIVED for review July 18, 1953. Accepted October 30, 1953. PreSOCIETY, sented before the Trisectional hleeting of the .kMERICAN CHEXICAL Bartlesville, Okla , October 16. 1953.
Spectrochemical Determination of Copper in Turbine Oils J. E. BARNEY, II Research Department, Standard O i l Co. (Indiana), Whiting, Ind.
TUDIES
of the influence of copper on the oxidation of tur-
S bine oils require the determination of microgram amounts of
that element in the oxidized oil. Although the dissolved copper may be present in concentrations up to several thousand parts per million, the usual range of interest is from 0.1 to 50 p.p.m. The American Society for Testing Materiala method ( 1 ) has been standardized as a laboratory procedure for studying the oxidation characteristics of turbine oils. This test, which is also used on white oils, employs copper and iron as oxidation catalysts. Accelerated oxidation procedures for studying turbine-oil life necessitate the analysis of oil samples for copper content as a function of time. Because the life test must be run on a limited amount of oil, often only a few milliliters will be available for each analysis. The problem has been met by extracting the copper with dilute sulfuric acid ( 5 )or an ion exchange resin ( 3 ) and determining it colorimetrically as the diethyldithiocarbamate complex. These techniques are time-consuming and often intro-
duce enough copper from the reagents to require correction for a large blank. Spectrochemical methods are particularly suited for this kind of analysis. Copper in very low concentrations can be detected by the spectrograph, only a few tenths of a milliliter of sample are required, and the analysis can be made rapidly with reasonable accuracy and precision. Because the samples are organic liquids containing minute amounts of just one or two metals, the absence of undesirable matrix effects greatly simplifies the selection of a spectrochemical procedure. A quantitative spectrochemical method, made possible by the simplicity of the system, has been developed in this laboratory. PROCEDURE
Spectrographic equipment includes a Bausch and Lomb Littrow spectrograph, an A.R.L.-Dietert Multisource, and an A.R.L. -Dietert comparator-densitometer. Electrically heated coils
568
ANALYTICAL CHEMISTRY
(2) are used for ashing the samples in semiquantitative electrodes
tained traces of copper that could not be removed without elabo( 4 ) prior to analysis. For excitation of the sample, the Multirate purification techniques. source is adjusted to produce 300 volts and 12 amperes when The accuracy of the method has been evaluated by analyzing qhort-circuited (2). Optical conditions are adjusted to give a background of about 85% relative transmittance. four solutions of copper oleate in turbine oil. Chemical and The oil is ashed in electrodes mounted in the coils. The elecspectrochemical analyses show: trode cups are filled with graphite, and the ashed sample is arced Chemical, p.p.m. 0.74 1.70 3.8 16.7 for 120 seconds to consume i t completely. I n accordance with Spectrochemical, p.p.m. 0.68 1.51 4.1 4.5 14.8 16.7 the amount of copper in the oil, 3 to 10 drops of oil are placed in Sample, drops 10 10 3 10 1 3 the electrodes in 1-drop increments from a medicine dropper, The average difference between the results by the two methods each drop being ashed before the next is added. By holding the was 10%. The amount of sample taken for analysis has little medicine dropper in a horizontal position, drops of sufficiently uniform size are obtained for a given oil a t a given temperature. effect on the accuracy. Thus, the oil can be added to the electrode without weighing; il measure of the precision of the spectrochemical method mas the quantity of sample can be measured later by yeighing one obtained by four analyses of each of t v o unknown samples of or more drops. From the resulting spectrum, readings are taken of the transturbine oil. mission of the background adjacent to the 3274-A. copper line Coooer. Average .. . o.o.m. .. and of the line plus background. To obtain the ratio, Icopper/ 12.5.12.9, 9 , 5 , 1 0 6 11.5 Ibackground, the intensity of the copper line plus background is ob9 3,10.2,10.5,10.8 10.2 tained from the emulsion curve ( 2 ) as a multiple of unit backThe coefficient of variation of the averages xas 10%. ground, and one intensity unit is subtracted from it. The weight of copper corresponding to this ratio is read off Equipment in use in this laboratory permits the analysis of a working curve prepared from standard samples. Standards about 60 samples in 6 hours. covering the desired concentration range are prepared by diluting with turbine oil a chemically analyzed oil solution of a copper LITERATURE CITED compound. Copper naphthenate is satisfactory. Chemical ( 1 ) Am. Sor. Testing llaterials, Philadelphia, Pa., “;I.S.T.M. oxidation of the copper concentrate and analysis of the residues Standards,”Part 5 , p. 1110, Method D 973, 1949. by the method of M o s s and Mellon (6) is convenient. (2) Barney, J. E., and Kimball, UT. A , , ABAL.CHEM.,24, 1548 (1952). Dividing the weight of copper in the sample by the weight of (3) Buchwald H., and Wood, L. G., Ibid.,25, 664 (1953). sample taken for analysis gives the concentration of copper in the (4) Harvey, C. E., “Method of Semiquantitative Spectrographic oil. Analysis,” Glendale, Calif., -4pplied Research Laboratories, DISCUSSION
Spectral background was used as internal standard, according to the method devised by Harvey ( 4 ) for semiquantitative analysis, because suitable oil-soluble compounds of other metals con-
(1947). (5) Kreulen, D. J. W., J. Inst Petroleum, 38, 449 (1952). (6) Moss, M. L., and LIellon, AI. G., ISD. EXG.CHEM.,ANAL.ED.,
15, 116 (1943). RECEIVED for review September 1.5, 1953.
Accepted October 16, 1953.
Colorimetric Determination of GIyce rol H. DARWIN REESE and M A X B. WILLIAMS Oregon State College, Corvallis, Ore. RECENT
colorimetric determination of ethyl alcohol in aqueous
A solution developed by the authors (24) suggested the colori-
metric determination of aqueous glycerol solutions following the same general principles. That the accurate determination of glycerol is of considerable importance may be shown by the fact that over 200,000,000 pounds of glycerol are used annually by dmerican industry and that several hundred articles in the last decade have dealt mainly with the determination of glycerol in pharmaceuticals, cosmetics, synthetic textiles, alkyd rrsins, ester gums, fats and soaps, fermentation products, plant and animal tissues, and a multitude of other substances. Most of the current methods of analysis for aqueous glycerol solutions are based on the oxidation of the glycerol with subsequent titrimetric determination of an oxidation product or the excess of the oxidizing agent. Reagents used include periodates, ceric sulfate, hypobromite. potassium dichromate, potassium permanganate, etc. (15, 19, 26, 25). The periodate and dichromate methods are the most common. The Association of Official A4griculturalChemists, since 1935, has adopted as its official method the potassium dichromate-sulfuric acid oxidation of glycerol with a subsequent estimation of the excess dichromate by titration with standard ferrous ammonium sulfate (2). These volumetric titration procedures usually involve somewhat unstable and troublesome solutions, some of which should be standardized daily. The results of the various methods often do not agree even in the macro range ( I , 20). There are only meager references in the literature concerning simple and accurate microdeterminations, although several attempts have been made to perfect a colorimetric determination ( 4 , 8, 9, 12, 13, 16, 23). One
of the methods, which attempted to measure the concentration of potassium dichromate remaining after oxidation of the glycerol (11, 2f), has the difficulty that Beer’s law is not obeyed in the visible region because one of the oxidation products, chromic ion, has an intense green color which is mixed in varying concentration with the excess yellow-colored dichromate ions that are being determined. Snell and Snell (21) attempt to overcome this difficulty by visually comparing the unknown glycerol sample with a series of glycerol standards after oxidation of both with excess dichromate. They report the accuracy to be approximately 10% with a glycerol range of 1 to 10 mg. per ml. Another method involves multiple wave-length analysis ( 7 ) , and a recent method ( 3 , 6) employs absorbancy measurements in the ultraviolet region. No work has been done with these methods for small concentrations of glycerol. The other colorimetric methods for glycerol have given only fair results and have not been adopted widely. The following method utilized the new technique (2’4) which nullifies the interference of the chromic ions completely, so that the colorimetric determination of the excess yellow-colored dichromate ions is feasible, and an extremely wide range of glycerol concentration can be determined accurately. REAGENTS
s-Diphenylcarbazide. The reagent v a s the same as used for ethyl alcohol ( 2 4 ) ; however, if a reagent of greater stability is desired, a more complicated procedure may be employed (6). Potassium Dichromate and Sulfuric Acid. Analytical reagent grade.