Microdetermination of Carbon and Hydrogen RALPH 0. CLARK AND GORDON H. STILLSON Gulf Research & Development Company, Pittsburgh 30, Pa. As the result of considerable routine analyses and some experimental work, a series of observations on the microdetermination of carbon and hydrogen in organic compounds is presented and discussed in detail. The need for a preburner, the interpretation and correct use of blank determinations, combustion time, and the analysis of refractory compounds are covered.
IN GENERAL (H),
unnecessary if the delivery tube of the latter was at least 80 mm. long.) The results in Table I show that three different tanks of oxygen (Linde Air Products Co.), selected from stock at random over a period of several months, contained impurities in amounts sufficient to cause appreciable error in a carbon and hydrogen determination. Therefore, it was concluded that unless a new tank of oxygen is tested for impurities prior to use, a preburner cannot safely be eliminated from the gas purification train.
the literature dealing with the microdetermination of carbon and hydrogen fails to dispel the belief, advanced by Pregl that all compounds should be equal in the eyes of the analyst. Most experienced microanalysts, however, feel that any information to the contrary should be made available to all, so that difficult or unusual situations may be met with suitable procedures. It is with this in mind that the following observations and experiments are presented. Many of the conclusions reached serve to substantiate published information, while othem are in distinct disagreement, or bring out points which have not been discussed adequately in previous reports. A detailed description of apparatus is not necessary for the purposes of this article. Earlier data were obtained with the conventional Pregl apparatus (21), while later determinations were made using an electrically heated, semiautomatic unit (5) with closed-type absorption tubes (4). The differences in design and operation of the Pregl and semiautomatic units have no significant influence on this discussion and no differentiation will be made between the two types of units in presenting the data. The type of absorption tube used need be coqsidered only in the discussion of blank determinations.
ASBESTOS FIBER,
I
N0.26 GAGE COP OXIDE WIRE
3MM. O.D. PYREX NO. 172 IO-MM. O.D.PYREX NO. 172
Y
ki
OXYGEN PURITY
5-MM. O.D. PYREX NO. 172 N0.30 GAGE COPPER WIRE5-MM. O.D.PYREX N0.774 \.
BLANKDETERMINATIONS. The purity of the oxygen or air passed through the combustion train in the course of an analysis is an important factor in the microdetermination of carbon and hydrogen. Purity may be evaluated either by means of blank determinations or by actual analysis of pure samples. Because the latter method favors the introduction of complicating factors, the blank analysis waa chosen for this investigation. If performed by the procedure recommended generally ( $ I ) , blank analyses are not only meaningless but can lead t o erroneous conclusions. Usually the lead peroxide in the combustion tube filling is in equilibrium with an amount of water dependent mainly upon the hydrogen content of the compound previously analyzed. Since the gas passing through the system during a blank determination is relatively moisture-free, it becomes partially saturated through loss of water by the lead peroxide. Therefore, before a valid blank analysis can be obtained, it is essential that equilibrium be established, as indicated by a constant “water blank” in a series of preliminary determinations. Results obtained after taking this precaution constitute a fairly reliable index of the purity of the grts used in the combustion. EFFECT OF A PREBURNER. The advisability of using a preburner for the purification of oxygen and air has been debated (1, 2, 9, IS, 19, 21, 88). Because of these differences of opinion a series of experiments was performed in an attempt to clarify this point, at least for the purposes of this laboratory. Since closedtype absorption tubes were used, only oxygen purity has been considered, although the conclusions apply to air as well. Blank values were obtained on 150 ml. of oxygen, using a preburner (Figure 1) filled with copper oxide wire and heated to 660475” C. by means of an electric furnace. (A cold-water coil for cooling the exit gas from the preburner waa found to be
!
Figure I. Preburner ~~
Table I. Effect of Preburner on Blank Value of 150 MI. of Oxygen Increase, Mg. Without Preburner With Preburner cot Hi0 COI Ha0 Liquid air oxygen, tank 1 0.032 0.153 0.011 0.012 0.039 0.147 0.019 0.019 Tank2 0.050 0.110 0.022 0.010 0.012 0.032 0.047 0.129 Electrolytic oxygen 0.053 0.159 0.013 0.029 0,043 0.147 0.018 0.021
COMBUSTION TUBE FILLING
Pregl (21) makes the statement that his “universal” filling may be used to analyze any organic substance, regardless of type or structure. Flaschentrager (11) infers that this statement can be accepted without reservation. Other investigators (1.2, 17, .22,24, 25) have reported instances in which Pregl’s filling has failed to give satisfactory results. For example, Niederl and Niederl (17) have found that the conventional tube filling gives low carbon values with condensed ring compounds and recom-
520
August, 1945 Table
ANALYTICAL EDITION
II. Effect of Platinum in Combustion Tube Filling yo Carbon
Withoutplatinum Withplatinum
Withoutplatinum
% Hydrogen
Found
Calcd.
Average difference
81.30 81.20 81.41 81.35 81.67 81.55 81.70 81.65 81.47 81.43 81.39 81.53
81.78
-0.48
81.76
-0.12
81.76
- 0 30
Average
Found Calcd. difference 11.04 10.97 t0.04 11.09 IO.96 10.95 10.98 10.97 +0.01 10.98 11.00 10.93 10.99 10 97 +0.04 11.09 11.01 10.97
mend their “combination-band’’ filling as more nearly universal in application. As the result of difficulties encountered in the combustion of refractory compounds a tube filling has been evolved which over a period of several years has proved satisfactory in thib laboratory for the analysis of a wide variety of compounds. A certain organic compound had been synthesized and exhaustively purified a t frequent intervals over a long period of time. Experimental evidence generally considered reliable indicated that it was very pure, b u t the Pregl method of analysis invariably gave carbon results which %ere0.5 to 0.6% lower than the calculated value. Two fsctors might have been jointly or independently responsible: the combustion time and the combustion tube filling. Extension of the combustion time to 20 minutes from 10 minutes increased the carbon values only 0.1 to 0.2%, an indication that this factor, if significant, was minor in effect, Kirner (14) occasionally used platinum gauze as an aid to the complete combustion of compounds which consistently gave low carbon results. As a consequence the following experiment was performed. Three centimeters of copper oxide were removed from the conventional Pregl filling and replaced with 3 cm. of silver wire. Four analyses of the compound in question were performed with this combination, and the customary low carbon values were obtained as expected. After replacing the silver wire with a 3-cm. plug of 80-mesh platinum gauze, followed by a 2-cm. silver wire plug, another series of analyses was carried out. Then, as an assurance that no significant changes had taken place during the experiment, a final series of determinations was made after replacing the platinum with silver wire. The results are shown in Table 11. As a direct consequence of this work, the tube tilling containing a 3-cm. section of platinum gauze has been adopted as standard in this laboratory. The tube filling just described, except for the silver wire plug between the platinum and the sample, is identical with the “combination type” of Niederl and Niederl (18). The Wing described was in routine use before access was had to the Niederl work, and the fact that the two developments were made independently serves to strengthen the case for a “universal” filling of this general type. EFFECTOF PARTIALLY SPENTFILLING.Considerable difficulty was encountered in performing carbon and hydrogen determinations on high molecular weight hydrocarbons, acids, and esters. Diphenylheneicosane, phenylundecylenic acid, and methyl phenoxyphenylstearate are typical examples. It was found, however, that when a freshly filled combustion tube was used, excellent results were obtained without modifying the routine procedure. Further experiments indicated that combustion tubes which had been used for less than fifty analyses generally were satisfactory. Apparently the types of compounds mentioned are very resistant to oxidation and their complete combustion requires high catalytic activity in the tube filling. Highly refined, oxidation-resistant mineral oils are subject to the same precautions.
521
ANALYSISOF ORGANICPHOSPHORUS COMPOUNDS.I t has been reported (24) that the combustion of organic phosphorus compounds has an effect upon the conventional combustion tube filling which influences succeeding analyses adversely. However, it can be shown that the difficulty lies, not in the tube filling, but in the carbonaceous deposit which the combustion of phosphorus compounds invariably leaves on the walls of the combustion tube and in some cases i n the sample container, If this deposit is not ignited to a brilliant red heat, the analysis in question and succeeding analyses will be in error. Unfortunately, this intense heating of the deposit generally causes Supremax and Pyrex 172 combustion tubes to fail after a few additional analyses. USE OF LEADPEROXIDE. Of all the reagents in the standard Pregl combustion tube filling, the lead peroxide represents the most probable source of trouble and various attempts have been made to eliminate it from the tilling ( 3 , 6, 7 , 8, 10, 16, 20). Elving and McElroy (IO),working on a semimicro scale, omitted the lead peroxide and instead introduced, between the two absorption tubes, an absorber filled with a sulfuric arid solution of potassium permanganate or dichromate. The results were satisfactory. Neuworth (16) and Burger Figure 9 . Absorber ( 3 ) reported the successful application of a similar procedure to the micromethod. In an attempt to verify the experiments of Neuworth and Burger, two nitrogen compounds were analyzed with and without lead peroxide in the tube filling. I n the analyses in which the peroxide was eliminated, silver wire plugs were substituted for it, and an absorber (Figure 2) filled with a 0.02 M solution of potassium permanganate in concentrated sulfuric acid was inserted in the gas stream between the two absorption tubes. The results are shown in Table 111. Table
111. Comparison of Lead Peroxide with Permanganate Absorber
% Carbon Acetanilide Anitratedphenol
With PbOt.
With absorber
71.30 71.30 71.31 50.42 50.32
70.93 71.18 71.18 50.41 50.32
% Hydrogen Calod.
71.09 50.00
With PbOs
With absorber
8.94 8.78 6.90 5.30 5.14
7.99 7.77 7.82 7.15 6.58
Calcd.
8.71 594
Consideration was given to the possibility that the high hydrogen values could have been caused by the presence of traces of basic impurities in the Dehydrite used as the water absorbent. That Dehydrite does absorb nitric oxide was readily demonstrated. I n a series of experiments in which nitric oxide was passed through tared absorption tubes, the Dehydrite tube showed a constant gain in weight, and a positive test for nitrate ion was obtained on an aqueous solution of the spent abfiorbent. To take this factor into account, analyses were performed using Dehydrite which previously had been completely saturated with nitric oxide. Again the hydrogen figures were high, and were comparable to those obtained with unsaturated Dehydrite.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
522 Table
IV. Effect of Substitution of Phosphorus Pentoxide for Dohydrite in Use of a Permanganate Absorber
% Carbon
Thiourea Stearanilide Acetanilide Benzenesulfoncyclohexylamide
yo Hydrogen
Average differFound Calcd. ence 15.70 15.78 - 0 . 0 5
15.76 80.02 80.13 71.20 71.19 60.34 60.39 ~
Found
7.27 7.15 80.17 -0.09 12.30 12.26 71.09 SO.11 7,71 7.79 60.23 +0.14 8.25 8.30
~~~
~
~~
%
Average difference
Nitrogen Calcd.
5.29 +1.93
36.80
11.49 f0.79
3.89
6.71 f1.04
10.36
7.15 +1.13
5.86
Calcd.
~~
~~
V.
Effect on Hydrogen Values of Variations in Combustion Rate % Hydrogen Combustion % Carbon
Table
Time, hlin.
Found
Calcd.
Found
Calod.
In the hope that an acid reagent might pass the acidic nitrogen oxides quantitatively, phosphorus pentoxide was substituted for Dehydrite. As before, carbon analyses were satisfactory and hydrogen results were appreciably higher than the calculated values (Table IV). The amount of nitrogen, hydrogen, or both, in the sample, had little effect upon the magnitude of the hydrogen error. It can be concluded, based upon the foregoing experiments, that the elimination of lead peroxide from the combustion tube filling through the use of a permanganate absorber is not applicable to the determination of carbon and hydrogen by the micromethod. Pregl ($1) makes the statement that the purity of the peroxide can be judged by color, and that the black variety only should be used. While the color may be an infallible indication of purity, it does not necessarily constitute an index of the efficiency with which the reagent performs its function as part of the tube filling. Excellent analyses can be obtained using brown or reddish-brown lead peroxide, and it is preferable to base the desirability of a particular lot of this compound on its performance in the analysis of a sample of known composition. COMBUSTION TIME
After sample has been distilled from its weighing container, the rate a t which it is vaporized into the combustion tube filling governs the rate of combustion. This point, of the utmost importance in analyzing certain types of compounds, has not been stressed sufficiently in the literature. The rate of vaporization is not to be confused with “combustion time”. The latter term, strictly speaking, should be used only in referring to the total time of cgmbustion. For the purposes of the discussion which follows, the total combustion time has been divided into three distinct periods: distillation, vaporization, and reburning and flushing. Pregl (81) recommends 10 minutes for the combustion of a sample in a stream of oxygen. Niederl and Xiederl (f7)advise extending this time to 15 minutes. Presumably these time periods include distilling the sample out of its weighing container into the combustion tube. Therefore, assuming 5 minutes as an average distillation period, the time consumed in vaporizing the material into the actual combustion zone would be 5 and 10 minutes, respectively, for the two recommendations. I t was found during the present investigation that 7 to 10 minutes are sufficient for the vaporization of most samples. However, certain types of materials which are characterized by low or erratic carbon analyses with a 7- to 10-minute vaporization time give excellent results when vaporized in 15 minutes. Examples of
Vol. 17, No. 8
such materials are alkylphenyl benzoates; residual petroleum oils, tars, and asphalts; most gasolines; and certain low-boiling aliphatic hydrocarbons. Should the vaporization time be extended longer than is required for quantitative combustion of a particular sample, the carbon and hydrogen results will not be affected, provided the same time is used throughout a series of analyses. Although appreciable changes in the combustion rate from one analysis to the next affect carbon values negligibly, the hydrogen values are influenced to a considerable extent (Table V). This is to be expected if it is recalled that a moisture equilibrium exists between the lead peroxide of the combustion tube filling and the gases passing through it. For a long vaporization period, relatively greater amounts of gases would be passed through the tube filling than for a short period analysis immediately preceding it. Therefore the average moisture content of the greater gas volume would be lower and there would be a strong tendency for moisture to be removed from the lead peroxide. By the same reasoning, when analyzing samples which are known to have abnormally high or low hydrogen contents, it is advisable to burn an unweighed sample first, in order to bring the peroxide into contact with gases of the same moisture content as will be encountered during the ensuing analyses. ANALYSISOF ORGANIC SILICONCoupoums. Ordinarily accurate determination of carbon in organic silicon compounds is extremely difficult because of the formation of silicon carbide in the weighing vessel and on the walls of the combustion tube adjacent to it. This deposit is stable under combustion conditions and will not give up its carbon as carbon dioxide. It has been found that the formation of the carbide is the direct result of distilling the sample from its container too rapidly and that if care is used during the distillation step no further trouble can be expected during the subsequent phases of the analysis. ACKNOWLEDGMENT
The authors wish to thank Lincoln T. Jenkins and Louise Petterson for performing many of the analyses which have been used in this discussion. LITERATURE CITED
Bock, F., and Beaucourt, K., Mikrochemie, 6, 133 (1928) Bruce, W.F., Ibid., 18, 103 (1935). Bilrger, K.,Die Chemie, 55, 260 (1942). Clark, R.O.,and Stillson, G. H., IND. ENG.CHEM.,ANAL. ED., 12,494 (1940). Clark, R. O., and Stillson, G. H., paper presented before Division of Analytical and Micro Chemistry at the 108th Meeting of the AMERICAN CHEMICAL SOCIETY, New York, N. Y. Corwin, A. H., Mikrochemie, 24,98 (1938). Dombrowski, A., Microchaie-Microchim. Acta, 28, 136 (1940). Dubsky, J. V., Z . anal. Chem., 59, 254 (1920). Elek, A., IND.ENG.CHEM.,ANAL.ED.,10, 51 (1938). Elving, P.J., and McElroy, W. R., Ibid., 13, 660 (1941). Flaschentrager, B., 2.angew. Chem., 39, 717 (1936). Futer, M., Mikrochemie, 9, 27 (1931). Kirner, W.R., IND.ENG.CHEM.,ANAL.ED.,6, 358 (1934). Kirner, W. R., private communication, 1938. Kuespert, K. H.,and Whitman, J. B., Mikrochemie, 11, 274 (1932).
Neuworth, M., private conimunication. Niederl, J. B., and Niederl, V., ”Micromethods of Quantitative Organic Analysis”. 2nd ed., New York. John Wiley & Sons, 1942.
Niederl, J. B., and Niederl, V., Mikrochemie-Mikrochim. Acta, 26, 28 (1939).
Niederl, J. B., and Roth, R. T., IND.ENQ.CHEM.,ANAL. ED., 6, 272 (1934).
Niederl, J. B.,and Whitman, J. B., Mikrochemie, 11, 274 (1932). Pregl, F.,and Roth, H., “Quantitative Organic Microanalysis”, 3rd English ed., Philadelphia, P. Blakiston’sSon & Co., 1937. Roth, H., Z . angew. Chem., 50, 593 (1937). Royer, G. L.,et al., IND. ENG.CHEnr., ANAL.ED., 12, 688 (1940). Silbert, F. C., and Kirner, W. R., Ibid., 8, 353 (1936). Verdino, A.,Mikrochemie, 6, 5 (1928); 9, 123 (1931). PRESENTED before the Division of .4nalytical and Micro Chemistry, Symposium on Microdetermination of Carbon and Hydrogen, a t the 108th Meeting of the A t d ~ ~ r c CHEMICAL .4~ SOCIETY, New York, N . Y.
