V O L U M E 20, NO, 11, N O V E M B E R 1 9 4 8 -------
1019 A /L
’
-.I
-
I
c--- I I
I
10
20
33
40
50
60
70
80
90
100
Volume Recovered ( m l )
Figure 12. Aviation Gasoline
scribed by Method D-86. Installation of such a control would have the possible disadvantage of one more knob for the operator to set, although a record of the actual distillation rate is always printed on the chart. I t has been demonstrated that the distillation rate for certain gasoline samples tested could be increased to 12 or 15 ml. per minute without seriously affecting the curve produced. Further investigation would be reqHired to determine the desirability of using this increased distillation rate for variouq ,amples. Although every effort has been expended on the present apparatus to duplicate the readings of the mercury-in-glass thermometer in order to duplicate manual operation, other uses for this type of equipment may arise wherein the inherent advantage3 possessed by the thermocouple of small heat capacity and no emergent stem error may be a great advantage. One example i. the special solvents distillation used for narrow boiling rang(’ products, for which the present apparatus could be equipped with multiple temperature ranges of 20’ C. span, and suitable thermocouple p d charts. Such an apparatus should solve the probleni of thermometer lag encountered in observation of the initial boiling point when a mercury thermometer iq used. CONCLUSION
.. n
.
1 I
10
20
40
S
!Idurn*
Figure 13.
50
60
70
80
90100
Reeoveitd l m l )
Aviation Gasoline
General ease of operation combined with good accuracy an(l reproducibility when used to run standard AI.S.T.M.Method D-86 distillations recommends this apparatus as a plant control instrument which can be operated at the still by the operator, a replacement for manually operated equipment in the contrnl laboratory, or a “referee” instrument which can be used to settli. disputes arising from varying laboratory and/or operator terhniques. 4CKiYOWLEDGMENT
\Then distilling low boiling materials a boiling chip is required in the flask to prevent superheating, a condition resulting in a relatively large quantity of distillate coming over as the first drop, and causing the automatic heat control to decrease the heat to n point v, here distillation may cease entirely. Although on some samples it has been possible to interpret the curve to show a dry point, there are not yet sufficient data available nith the new thermocouple to state positively that this can alm-a>-sbe done. AiB j C O alarm and light could be easily installed to warn the operator, so that he could observe the dry point and so mark the chart. Another possible modification is the addition of a control to produce, automatically, any desired rate of diqtillation. rit prcsent, o n 1 ~one rate i q provided, the 4.5 ml. per minute prc-
The authors \vi41 to extend their thanks to J. A. Wood for his mechanical ingenuity in construction of the recorder components, and to IT. B. Jlilligan for his suggestion of the special “low-lag” heater. LITERATURE CITED
(1) Josten, G. W., U.
S.Patent 1,953,716 (April 3, 1934). Apparatus
for automatically recording volumetric and temperature distillation data. (2) Leeds & Korthrup Co., private communication. (3) U. S. Technical Oil Mission, Reel 6, Bag 2747, Item 2; German Patent application 14,527 (1943). Continuous measurement, and recording of boiling point and gravity during distillation at constant heat input (pneumatic-hydraulic with photoelectric drive). RECEIVED January 19, 1948.
Determination of Nitrogen in Organic Materials Application of the Mass Spectrometer S. G. HINDIN’ AND A. \-.GROSSEa, Houdry Process Corporation, Marcus Hook, Pa.
W
I T H the usual methods of chemical analysis, one must quantitatively convert the element sought to some specific compound having characteristic and desired properties, isolate this compound in pure form, and then quantitatively measure the total amount of the compound formed. The procedure suggested herein is analogous to t,he “int,ernal standard” method o f emission 1 Present address, American Sugar Refining Co.. Re,!earch and De\-eIopment Division, Philadelphia 48, Pa. 2 Precrnt address, Temple Research I n - t i t i i t ? , P h i l n r i ~ l g h i n ,P a .
spectroscopy. In such analysis, the ehmcnt sought is determined relative to some major constituent of the sample, or to somc element not present originally but added in known amount. This paper describes a similar internal standard method of analysis for nonmetals in organic materials, in which the mass spectrometer is used as a measuring tool. Basically, the procedure entails these steps: known weight of sample, containing the element to he determined, is taken.
1020
ANALYTICAL CHEMISTRY
amount of an internal standard. This internal standard is chosen on the basis of the similarities of its
OG. W Z W C
Figure 1. Diagram of Apparatus
Taking into consideration the fact that the ratio of the amount of internal standard added to the amount of unknoffn element remains constant, a determination of the ratio only of the two elements suffices for calculation of the concentration of unkn’own in the original sample. No quantitative recovery of any of the gases is necessary. To date, this general technique has been applied only to the determination of nitrogen in organic materials, using neon as the internal standard. Nitrogen, formed by copper oxide combustion of the sample and subsequent reduction of the oxides by metallic copper, is determined relative to neon, which has been introduced into the system in known amount before combustion. [A variation of this technique, in which a stable isotope of the element is used as internal standard, has been developed for determination of oxygen ( 2 ) , carbon (S), nitrogen (4),and sulfur]. Keon was selected as internal standard for three reasons:
1. Xeon does not react with the copper oxide under the experimental conditions set forth. 2. Neon and nitrogen show similar inertness to the adsorbents used to scrub the gaseous reaction products. 3. Neon and nitrogen have molecular weights of the same order of magnitude (20 and 28, respectively); this minimizes any possible mass spectrometric error caused by dissimilar rates of diffusion. Briefly, a weighed sample is oxidized with copper oxide in an evacuated silica tube, in the presence of an added, known volume of neon; nitrogen oxides are reduced by metallic copper. After reaction, instead of measuring the volume of free nitrogen formed, as in the Dumas method, only the ratio of nitrogen to neon in a sample of the gaseous products is determined, using the mass spectrometer. No quantitative recovery of any of the gases is even attempted. As the sample weight and the volume of neon added are known, the nitrogen content of the sample is calculated as follows: MNe
= mole ratio (Ne/N2)
e9
Y
gaseous compound, in which state the internal standard is present. A portion of the gaseous reaction product is withdrawn for mass spectrometric analysis, or analysis by other analytical techniques.
% ’
i5g O Z J 4
28.02 weight of sample (in mg.)
x
105 (1)
where mole ratio (Ne/Nz) is determined by the mass spectrometer, and MNeis the moles of neon added. In the determination, carbon and hydrogen are essentially quantitatively converted to carbon dioxide and water, respectively. If the system is set up so as to pick up these materials in the customary absorbents, i t should be feasible to determine these elements, simultaneously. A further possibility is the mass spectrometric determination of carbon dioxide and, perhaps, water vapor, both relative to an internal standard. Because, in theory, most reliable figures are obtained when the number of moles of internal standard added approximately equals the moles of element sought, this entails the addition of relatively large amounts of internal standard in the determination of carbon dioxide. Thus, one might use a 1 to 10 mixture of neon and argon as internal standard: the neon for nitrogen and hydrogen .analyses, and the argon for determining carbon. .
t L 3
m
z w
z
TOEPLER PUMP
To carry the analogy with emission spectroscopy further, the carbon dioxide formed by combustion may be used as internal standard for determination of nitrogen when the carbon content of the sample is known with a fair degree of accuracy. Actually, unless the nitrogen content of the sample is high, the number of moles of the two gases formed would generally be rather far apart, probably causing a decrease in accuracy of the nitrogen analysis. PROCEDURE
K i t h the exception of pure compounds, the authors’ work to date has been in the analysis of high-boiling petroleum fractions. For this reason, the apparatus used and the following discussion relate specifically, a t least in handling procedure, to nonvolatile materials containing more than 0.2% nitrogen. The apparatus used for analysis of high-boiling liquids and solids is shown in Figure 1. Except for the silica tubes, the apparatus is easily constructed by an ordinarily proficient laboratory glass blower. Tube A , approximately 2.5 cm. in inside diameter by 40 cm., tapered down to about 10 mm. a t both ends, is sealed into the system through a graded seal after being packed with about 50 to 100 grams of wire-form copper oxide. The oxide is maintained a t 600” to 800” C. throughout the working day, by means of a furnace around it; the heater extends over about three fourths the length of tube A . (The dimensions given serve merely as a guide; they are not critical and may be altered to suit the convenience of the analyst.) Tube E is about 20 cm. long; its outside diameter is such that it fits snugly inside the tapered end of A . About 2 to 3 grams of finely powdered copper oxide are placed in E , weighed sample is added, 2 to 3 grams more of powdered copper oxide are added, and the tube is rotated to mix the contents. .4 final 2- to 3-gram layer of copper oxide is then placed atop the mixture. A reduced copper spiral is inserted into the open, cool end of A , and then the tivo tubes are joined with de Ichotinsky cement. (During the run, the de Khotinsky seal is water-cooled.) A sample of the order of 20 to 40 mg. is usually taken; the lower limit tends to be set by the nitrogen content of the sample and the volume of A fixes the upper limit to the size of sample that can be conveniently handled. After the two tubes are joined, the copper oxide-sample mixture 1s cooled in a dry ice bath and the system is thoroughly evacuated by a diffusion pump. Neon is then leaked into the system. Depending upon the estimated nitrogen content of the sample, the volume of neon used will vary; a convenient volume to use is 1 to 3 ml. Stopcock 1 is closed, and tightened into place with a rubber band, as there is a slight pressure build-up in the system. The sample is then ignited: A Meker burner is used to heat the copper oxide layer to glowing redness; maintaining this heat, a second burner is used to ignite the sample-copper oxide mixture, gently a t first and finally with full heat. Heat w i maintained for about 15 minutes, during which time the reduced copper spiral is also heated to redness. After reaction, the gaseous products are ex-
V O L U M E 20, NO. 1 1 , N O V E M B E R 1 9 4 8
1021
Nitrogen in organic materials is customarily determined by either a Kjeldahl or a Dumas procedure. With the increasing usage of the mass spectrometer as a laboratory tool, isotopic methods of elemental chemical analysis have been suggested. This paper proposes a modification of the Dumas combustion in which the gaseous products are analyzed by the mass spectrometer, and an inert gas is used as an internal standard. Though the method is still in a developmental state, the application of this type of analysis may be of interest in other fields, particularly in the determination of trace amounts of elements.
panded into the Toepler pump, being scrubbed, in passage, by Ascarite and potassium hydroxide. This scrubbing serves to remove carbon dioxide and xater vapor; this increases the concentration of nitrogen and neon (though maintaining their ratio to each other unchanged) in the gas to be analyzed, thereby increasing the precision and accuracy of the mass spectrometric analysis. If carbon dioxide is used as the internal standard, the scrubbers contain only a desiccant. If the simultaneous determination of carbon and hydrogen is attempted, the absorbers should be placed in series before the Toepler pump. After the scrubbed gas has been taken into the Toepler pump, it is forced over into the gas sample take-off tube. The tube is sealed off in five sections; three are taken for, spectrometric analysis, the remainder held as retains.
the order of 0.2 or 0.3% or higher, the presence of such traces does not materially affect the accuracy of the determination. If, however, the nitrogen content is of the order of 0.1% or less, the contribution of these traces to the mass 14 peak must be evaluated, by the usual mass spectrometer calculation, if the accuracy is not to be impaired. Incidentally, the presence of these traces is the reason for the use of mass 14 peak for nitrogen, rather than the more reproducible mass 28 peak, because the contributions are relatively less than for the mass 28 peak. In view of the erratic nature of the ratio of the mass 14 peak to the mass 28 peak, a correction is made in the following manner:
The spectrometer calculations are simple. -4typical calculation is shown in Table I. Thc spectrometer used is the Consolidated Engineering Corporation instrument model and the data are expressed in terms of units of peak heights. The calculation involves merely a reading of the nitrogen mass 14 and the neon mass 20 peaks, and the multiplication of the 20/14 ratio by a previously determined constant (the sensitivity ratio of the 14/20 peaks). This value then is the factor "mole ratio Se/S*"of Equation 1. Knowing the moles of neon added and the sample weight, the nitrogen content is calculated, as indicated.
First, the sensitivity ratio of the nitrogen mass 28 peak to the neon mass 20 peak is determined by running a group of three or four samples each of nitrogen and neon. This ratio remains fairly constant over a period of 3 or 4 weeks. Then, when a group of samples is to be analyzed, samples of nitrogen are also analyzed before and after the group. From these data, the ratio of the nitrogen mass 14 peak to the mass 28 peak is calculated, and by applying this to the standard mass 28-mass 20 constant, a sensitivity ratio of the mass 14 to mass 20 peak is derived. This value is then used in calculating the group of samples. I t is necessary to make a correction of this sort, because analyses of nitrogen over a period of a week have shown a variation of * 10% in the ratio of mass 14 to mass 28.
Table I.
Analysis of Indole (11.97% Nitrogen Theoretical)
[22.7-mg. sample taken, 1.85 ml. of neon added ( K T P ) = 8.25 X 10-6 mole. Gas sample analyzed in triplicate] hIass 14 Mass 20 Peak Height, L-nits Peak Height, Units 20/14 129.0 158.7 1.23 146.4 172.8 1.18 90.3 107.1 1.19 Av. 1.20 0.02 Sensitivity ratio of 14/20 peaks = 0.724. Therefore, 1.20 X 0.724 =
The actual procedure is rather flexible and may be varied over wide limits to suit the convenience of the analyst. Thus, for example, in most of the authors' work the mole ratio of neon added to nitrogen formed varied rather widely from unity nithout seriously affecting either accuracy or precision. Also, there apparently is nothing critical about the temperature of the large bed of copper oxide. The internal standard may be any gas inert to
f
Table 11. Analysis of Pure Compounds
ACCURACY AND PRECISION OF RESULTS
The accuracy of the procedure 71-as determined by analysis of several highest purity Eastman reagents (Table 11). -4s a check on purity, these materials were analyzed where possible by the Shirley and Becker modified Kjeldahl method (6). Results areindicated in the last column of Table 11. In view of uncertainty as to the exact nitrogen values, it is difficult to assess the accuracy of the procedure. I t is believed, though, that accuracy and precision, as shown in Table 111,are of the same order of magnitude, about 1 to 2% of the nitrogen present. This value is probably determined by the mass spectrometer analyses, which generally show variations in peak height ratios of this same order of magnitude on replicate analyses of the same gas mixture. DISCUSSION
With the procedure indicated, combustion is essentially quantitative; only very small traces of hydrocarbons and carbon monoxide are formed. When the nitrogen content of the sample is of
Compound Pyridine Quinoline Quinaldine Kitrobenzene
Eastman Catalog Description B.P. 113.5115.5' B.p. 110111/14 mm. B.p. 118119/15mm. h1.p. S o
Per Cent Nitrogen Determined b y t h e Shirley Determined and Becker ( 6 ) Theoretb y this Kjeldahl icala procedure procedure 17.71
1 7 . 5 =t 0 . 2
17.1
10.85
1 0 . 5 =t0 . 2
10.7
9.79 11.38
9.7 f 0.1 11.4 f 0.2
9.5 Incomnlete conversion 8.2
Carbazole 4.4-Dinitrodiazoaminobenzene
hI.p.241-244'
8.38
8.6 f 0.2
h1.p. 224-226'
24.39
24.1 i 0.4
Indole
M.p. 52-53O
11.97
11.7 f 0.1
h1.p. 114-113O
18.65
18.7 f 0.2
n-Di m e- t---" h v l----xmr --
inoazobenzene a
Incomplete conversion 11.8 Incomplete conversion
Calculated from compound formula.
Table 111. Precision Obtained in Multiple Determinations Sample Indole Crude A
'
yo Nitrogen 12.3, 11.7, 11.6, 11.9 0.73,0.71,0.73
1022
ANALYTICAL CHEMISTRY
the reactants and absorbents of the system or, as noted previ(~usly, the carbon dioxide formed may be used if the carbon content, of the sample is known. Incidentally, this use of carbon dioxide leads to the interesting possibility that nitrogen could be quantitatively determined without the necessity of xeighing the sample being analyzed. If the mass spectrometric analysis could be carried t o the simultaneous analysis for water vapor, carbon dioxide, and nitrogen, no prior knowledge of the sample would be needed (except that it contained only carbon, hydrogen, and nitrogen) and, as before, it \vould be unnecessary to use a weighed sample for analysis; the mole ratios of the three components would be sufficient for the complete picture. This procedure presupposes, of course, a completely air-free and air-tight system. If air has leaked into the sample, a correction can lie based on the argon niass 40 peak, as argon is present in air to the estent of approsimatdy 10;. After a determination, the reduced copper in tube B is usually a solid, fused mass. Soaking in concentrated nitric acid is effective in removal of the material. Incidentally, it is probably this reduced copper that reduces any nitrogen oxides f o r m d . The spiral is inserted the tube A primarily as a precaution. I n applying this procedure to volatile materials of very low iiitrogen content, a major difficulty is encountered. Aside from the possibility of loss of low-boiling material, the systtlrn d o ~ not s allow for large samples because of pressure build-up. The present approach is to use a large sample, 2 to 3 grams, and reduce the pressure in the system by freezing out the carbon dioxide as i t forms. The authors plan to present these details in the near future. Inasmuch as the error ill the tlt~trrminationis not some absolute value, but is, within reasonable limits, a constant function of the nitrogen content, the procdure becomes most valuable in the range of low nitrogen contents, in contradistinction to the customary prycedures. Contrariwise, the method is relatively less accurate in the range of high nitrogen contents (>20%). As compared to standard procedures f'or nitrogen analysis, this method offers some advantages. Though a revised Dumas train recently reported ( 1 ) appears to be of universal application and to rnsure escellent results, the apparatus required is rather elaborate and it would appear t o require a skilled operator. Thc relative accuracy decreases with decreasing nitrogen content, a t least in the very low ranges. The Kjeldahl method is simple and accurate,' but its applicability is restricted and it is difficult to use with highly volatile materials. Furthermore, a typical Iijeldahl analysis requires a matter of hours for a determination. This method embodies the advantages of thc isotopic dilution method without consuming relatively rare and expensive isotopes. The disadvantage of requiring quantitative conversion to free nitrogen is probably more apparent than real, for all the types of nitrogen
linkages that have been examined, to date, appear to convert quantitatively to free nitrogen under the conditions of the procedure. In the reference noted ( I ) it was found that nitrogen is adsorbed on cold copper oxide, and released slowly on heating. Although the absolute estent of the error so introduced (from the cold copper oxide in tube B ) , has not as yet been determined, it iq probably not very great, as samples of the order of 0.04% nitrogen have shown excellent concordance lxtween results by this method and the Kjeldahl procedure. POSSIBLE I P P L I C I T I O 3
The iiiternal standard method appears to be novel in gas analysis. Aside from its direct use, as indicated, i t might be applied to trace analysis; by proper choice of internal standard, the s u b stance sought could be concentrated (either physically or chemically) until present in sufficient concentration to yield an accuratcx analysis. I t might be of use in tracer work, where the S15/K14 ratio is desired, without addition of an internal standard. From another point of view, the method is a means of accurately measuring small volumes of gases; as such, it might be of value in adsorption measurements. hlthough the method has been written specifically as a means f'or analyzing combustion gases; it is not limited to such determinations, but may be used in conjunction with standard wet chemical procedures. .4CKEOW LEDGMENTS
The authors wish to express their appreciatioii aiid thanks to J. Alexander for his constant criticism and encouragement during both the esperimental work and thcb preparation of this paper; to A. D. Kirshenbaum for his helpful suggestions during the course of the work; to C. S. Pennington, for his help in setting up the system and who performed most of the analyses herein rcport,ed; and to G. -1.Sites, for his suggestions in the design of the system and for his glass blowing. The procedure arose as a result of a conversation with D. Gilmore. LITERATURE CITED
(1) Goiiick, Tunnicliff, Peters, Lykken, and Zahii,
ISD. Ex(;. CHEM.,ASAL. ED., 17, 677-82 (1945). (2) Grosse, Hindin, and Kirshenbaum, J . Bm. Chem. Soc., 68, 3119 (1946). (3) Grosse, Kirschenhaum, and Hindin, Science, 105, 101 (1947). (4) Kirshenbaum, Hindin, and Grosse, S u t u r e , 160, 187 (1947). (5) Shirley and Becker, IXD. ESG. CHEM.,d w a ~ .ED., 17, 437-8 (1945). RECEIVED January 14, 1948. Presented before the Division of Analytical and Miero Chemistry a t the 111th meeting of the AMERICAK CHEMICAL SOCIETT, Atlantic City, N . J.
Modified and Combined Grignard and Quantitative Hydrogenation Apparatus AND
T
HAROLD E. ZAUGG, A b b o t t Laboratories, N o r t h Chicago, 111. W-ALTER RI. L,IUER, University of Minnesota, Minneapolis, M i n n .
\VU pieces of equipment that generally see only intermittent service in the organic microanalytical laboratofy are the quantitative hydrogenation apparatus and the Grignard machine. Yet they are almost indispensable when occasion for their employment arises, the former in the determination of unsaturation, and the latter in the quantitative determination of active hydrogen and other Grignard-reactive groups. lloreover,
in the case of the conventional Grignard apparatus, infrequent use often results in freezing of the main stopcock due to its prolonged contact with the alkaline Grignard reagent. Drastic repairs are sometimes necessary in order to restore the apparatus to working order. The present paper describes a Grigriard apparatus in which the troublesome stopcock has been eliminated through use of a
