ANALYTICAL CHEMISTRY
1572
is described by Brownlee ( 1 ) for systems of one dependent and three independent variables. The process is essentially that of determining, by the method of least squares, the partial correlaLion coefficients of each variable and the regression equation. The latter is an equation for the dependent variable, in this case apparent heat capacity of the calorimeter, in terms of the three independent variables, average temperature, power, and temperature rise. The regression coefficients are tested for significance by comparing the ratio of the coefficients to their standard error with statistical tables for that ratio. The regression equation thus found is: Heat capacity (tal./' C.) = 10.25 0.01501t (av., C.) - 0 . 1 1 8 0 ~(watts) 0.01846At (" C.)
+
+
O
The coefficients for power and temperature rise were found to be not statistically significant, even a t the 0.10 level. This does not prove that they have no importance, but rather indicates that the effect caused by variation of these quantities is small compared to other errors in the method. In order to minimize variations due to changes in operating conditions, a standard procedure (heating rate of 3 watts and temperature rise of 10' to 15' C.) is used whenever practicable. When other conditions are used, the regression coefficients give the correction to be applied to the apparent heat capacity of the calorimeter.
The equation used for calculating the heat capacity a t any temperature was developed from experimental data taken, under the standard conditions, a t about 68", 141°, and 214' C. The constants were adjusted slightly to fit the data a t 288' C. Heat capacity (ea]./" C.) = 39.62
+ 0.024% - 3.15 X
10-6t'
where t = average temperature, C. This equation, which constitutes the calibration equation of the instrument, fits the experimental data to within *0.2%. Since the procedure and mechanical changes described above were made, the calorimeter has been in regular service for an average of four determinations per day without loss of time for any mechanical or electrical failure. The time required to clean and load the calorimeter was considerably shortened, for it is no longer necessary to remove the shield or the calorimeter or to disturb any electrical connection. The precision equals or excel? that of the unmodified instrument-that is, within *0.5%. O
LITERATURE CITED
(1) Brownlee, K. A , , "Industrial Experimentation," pp. 66-77.
London, H.M. Stationery Office, 1947. International Critical Tables, Vol. V, p. 105, New York, McGrawHill Book Co., 1929. (3) Stow, F. S.,Jr., and Elliott, J. H., ANAL.CHEU., 20, 250 (1948) (2)
RECEIVED January 18, 1949.
Liquid Sampling for Analysis by Mass Spectrometer R. A. FRIEDEL, A. G. SHARICEY, JR., AND C. R. HUMBERT Ofice of Synthetic Liquid Fuels, Bureau of Mines, Bruceton, Pa.
Ifl;ALYSES of liquid mixtures by mass spectrometer require introduction of minute amounts of calibrating liquids (on the order of 0.001 ml.) and measurement of these amounts to about 1%. Sample introduction usually consists of expanding the liquid directly into the high vacuum system of the spectrometer by touching a pipet to a sintered-glass disk under mercury (3). TWO methods for quantitative measurement of small amounts of liquid have been published: The volume of liquid is measured by a micropipet constructed of thermometer tubing ( 3 ) , or the pressure of the expanded vapor is measured by a specially designed micromanometer which operates in the pressure range of 0 to 100 microns of mercury ( 4 ) . The method herein described is a volume-measuring modification which is simple, timesaving, and sufficiently accurate, and involves no reading errors.
6
1 7
rI
1
43,rnm copillory (-0.05 m m I.D.)
Figure 1.
4
4i-f
E @J
Self-Filling 0.001-3Il. Micropipet
Bottom of tip ground on emery cloth if necessary to m a k e good contact with sintered disk
The instrument used is the Consolidated mass spectrometer. In addition to the conventional gas-handling system of this instrument, the method requires only the usual mercury-sintered disk valve and a self-filling 0.001-ml. micropipet (1, 2') (Figure 1 ; available commercially in sizes 0.001 to 0.010 ml.). When dipped in a liquid, the small section of capillary tubing becomes filled only to the top by capillary action; then the liquid is delivered completely to the vacuum system by touching the pipet tip to the sintered disk under mercury. Air is allowed to flow through the capillary following the sample until all visible droplets of liquid are swept into the instrument. As the amount
Table I. Reproducibility of Liquid Introduction into Mass Spectrometer by 0.001-M1. Self-Filling Rlicropipet .lq. %
Av. Peak Height ComAir Mercury flow seal ponent Mass n-Heptane 100 550(5)" 500(5) n-Octane 114 281 (5) 271(5) n-Nonane 128 225 (6) 225 (7) Xylenes 106 213(6) 212(6) a Figures in parenthesis represent number ~
~
RIayimum Spread, % MerAir eury flow seal 1.8
2.8
4.6
8.5 7.3 3.8 8.6 of runs. 4.5
~
Deviation from Average Mer.iir cury flow seal 0.6 1.4 0.7 2.0 1.6 1.2
2.5 2.1
of air thus introduced is immaterial to analysis, the mass peaks for air need not be measured. Quantitative results for hydrocarbons appear to be best if the pressure of sample plus air in the 4liter vacuum system is kept near a Pirani-gage reading of 0.20 * 0.03 mm. Under these conditions, the amount of air represents 20 to 30% of the total; no detrimental effects on the mass spectrometer filament have been noted. An alternative method, sealing with mercury, c o d s t s of filling the capillary with the sample and then pouring enough mercury into the pipet to cover the upper tip. Thus mercury, instead of air, flows through the capillary following the liquid sample.
The methods permit sufficiently good quantitative analysis, as shown by the reproducibility tests in Table I for n-heptane, 12octane, n-nonane, and a xylene mixture, both with air flowing through the capillary and with mercury sealing. Precision of both methods apparently decreases with the vapor pressure. The airflow method is somewhat more precise in all four cases; accuracy of this method is demonstrated by analyses of a typical 6-cornponent synthetic blend, mainly octanes (Table 11). Some inconsistency is expected for the mercury-seal method, inasmuch as droplets of liquid occasionally are trapped inside the capillary by mercury. Use of this method is essential when air must be excluded.
V O L U M E 2 1 , N O . 12, D E C E M B E R 1 9 4 9 Table 11. >lass Spectrometer Analyses of Synthetic Blend Using 0.001-M1. Self-Filling Micropipet (.4ir-flow method used for calibrations and samples) Synthetic Analysis Analysir 1 2 Component Composition Volume 7 ‘, n-Nonane 3.2 3.5 3.9 n-Octane 74.8 73.2 75.3 2-&fethylheptane 7.0 6.9 7.0 3-hlethylheptane 6.9 7.2 6.4 4-Methylheptane 2.9 4.3 2.9 n-Heptane 5.2 4.9 4.5
Calculating time is appreciably decreased by the use of this pipet. It is not necessary to calculate percentage patterns from the spectra of pure calibration compounds because the amount of sample, and, therefore, the spectral peak height, are always the same, within experimental error. For the same reason, sensitivity coefficients (peak heights + sample pressure or volume) are not needed, for the peaks themselves serve this purpose. Because the partial volumes of the constituents add up to near 1.0, approxi-
1573
mate percentages are obtained immediately, although normalization to 1.0, or loo%, is usually necessary. Application with increased accuracy to C, alcohols, Ct acids, and other oxygenated compounds has been obtained by a modification of the mercury seal method, which consists of forcibly squirting sample and mercury through the pipet and onto t,hr sintered disk. ACKNOWLEDGMENT
The authors wish to thank Sidney Katz, Institute of Gas Technology, Chicago, Ill., for describing the use of the self-filling pipet in microchemistry. LITERATURE CITED
(1) Anderson, H.H., ANAL.CHEM.,20, 1241 (1948). (2) Benedetti-Pichler, A. A., “Microtechnique of Inorganic Analysis,” pp. 238-40,257,New York, John Wiley & Sons, 1942. (3) Taylor, R. C.,and Young, W. S., IND.ENO.CEEM.,ANAL.ED.. 17,811 (1945). (4) Young, W. S., and Taylor R. C . , ANAL.CHEM.,19,133 (1947)
RECEIVED July 30, 1948.
2 -Nitro -1,3-indandione Promising Reagent f or Identijication of Organic Bases
’
B. E. CHRISTENSEN, C. H. WANG, I. W. DAVIES, AND DARRELL HARRIS Oregon State College, Corvallis, Ore.
N 1936 Wanag ( 4 ) described a new acidic reagent, 2-nitro-1,3indandione, which formed salts with both inorganic cations and organic bases such as the aliphatic and aromatic amines and the nitrogen heterocycles. These derivatives were reported (4-6) to be crystalline, nonhygroscopic, water-soluble compounds which, with few exceptions, gave sharp melting points. Rosenthaler (3)applied the reagent to the study of alkaloids but other than reporting crystal formation gave no analytical or melting point data. Later Muller ( 1 ) extended the work to a number of miscellaneous compounds such as histamine, tyramine, and arginine. 2-Sitro-1,3-indandione is a strong acid; hence many of its salts hydrolyze in an aqueous medium to yield acidic solutions. The extent of the hydrolysis depends on the ionization constant of the base used in the preparation, which in most cases is so sinall as to permit the titration of the free acid, and thus make it possible to determine the neutral equivalent of the salt. This property, unrecognized by Wanag, and not possessed by the usual amine derivatives, gives unusual promise to 2-nitro-lJ3-indandione as a reagent for the identification of organic bases. For these reasons the work of Wanag was continued in this laboratory, with particular attention t o the acid properties of the salts. Because only a limited number of heterocyclic derivatives were described by U’anag and others, most of the additional preparative work was devoted to the heterocyclic compounds. The salts were originally prepared by the addition of an aqueous solution of the acid to a dilute hydrochloric acid solution of the amine. As 2nitro-1,3-indandione is very insoluble in dilute hydrochloric acid, impure salts were often obtained, which required several recrystallizations. Experiments with other solvents indicated that acetone was a more suitable reaction medium, inasmuch as both the organic bases and the free acid were usually soluble in this medium, whereas the salts either precipitated immediately or slowly crystallized. In a few instances the salts precipitated as oils which failed to crystallize. Except for compounds in which the nitrogen was not functionally basic, such as mracil, xanthine, or acetanilide, all others gave derivatives with this reagent. The alkylamines form
crystalline salts which do not hydrolyze sufficiently to give w u tralization equivalent measurements. The bulk of the nitrogenous bases, however, gave derivatives which with few excep tions (due to color interference) could be titrated. In certain cases, such as the titration of the 4-quinazolone derivative, abnormal amounts of alkali were required, owing to the partial neutralization of the enol form of the quinazolone. ID almost every instance in which the ionization constants of the bases were known to be less than 10-6 the indandione derivative gave a titration value within 2% of theory (with the exceptionc cited above). Although melting point data pertaining to the indandionateh are to be found in the literature, earlier workers failed to point out that these values were for the most part decomposition points which are difficult to reproduce, and hence variations of several degrees often were noted among individual observers. However, the derivatives of 2-nitro-1,3-indandione are very easily prepared, and this makes this reagent extremely useful for isolation as we]) as characterization purposes. 2-Nitro-1,3-indandione was first prepared according to the directions of Pyanag (4, 5 ) by the nitration of 1,3-indandione with cold fuming nitric acid in glacial acetic acid under conditions which gave rather erratic yields. In this laboratory it was discovered that more reproducible results were obtained with ordinary concentrated nitric acid containing oxides of nitrogen These oxides have a marked influence on the course of the reactions, favoring the formation of nitration rather than oxidation products. EXPERIMENTAL
2-Nitro-l,3-indandione. Dissolve 20 grams of 1,34ndandione in 200 ml. of glacial acetic acid and raise the temperature of solution to 48”C. Add 20 ml. of 50 to 60y0nitric acid visibly colored with the oxides of nitrogen. Shake or stir 5 to 10 seconds and place the flask immediately under cold running tap water while continuing the agitation for several additional minutes. After cooling for 30 to 60 minutes, remove as much of the mother liquid as possible by filtration. Dissolve the crude 2-nitro-1,3-indandione in 500 ml. of cold water, decolorize with charcoal a t 10’ to
