ANALYTICAL CHEMISTRY
1476 The cell was immersed in carbon tetrachloride and the pressure gradually increased. Extremely small bubbles are usually formed and rise in a fine stream, which can be detected easily in a good light. IVhen a leak was found the cell was removed from the carbon tetrachloride and the nuts were loosened gradually Kith the cell under moderate pressure. The spreading leak gradually separated the window from the spacer (or the gasket from the cover plate) as the confining pressure was relieved. The cell was then taken apart only a t the layer of separation, the faulty surface reamalgamated as before, the cell reassembled and allowed to set, and retested. This process was repeated until no leaks were detected a t 35 pounds per square inch gage. This was the maximum pressure deemed advisable for use with the present loading springs. Stronger springs would permit higher pressures. The maximum safe pressure for salt ~ i n d o w s6.5 mm. thick has not been determined. DISCUSSIOU
It is not advisable to attempt to use a cell that leaks even slightly, for the leak is likely to spread and the sharp local chilling resulting from the evaporation of liquid in the leak may crack both windows. The experience of the authors in this respect has convinced them that the crack results from the leak, not the reverse. Such a crack can liberate enough flammable gas to constitute a fire hazard. After a period of use, the liquid penetrated slightly into cracks in the brittle tin amalgam a t the confining edge of the spacer. The expansion accompanying release of pressure during elimination of bubbles while filling, or when emptying or evacuating, has caused small fragments of the confining edge of very thin spacers to separate slightly. This process may eventually progress to the point where leakage will occur. This behavior is not expected to occur with thicker spacers where there is a core of unamalgamated metal left to hold the brittle surfaces intact. It seems advisable, therefore, to test by immersion from time to time cells which show such behavior, thus minimiz-
ing the possibility of a Iedk breaking through while in use. hfter testing, the thickness of the cells was determined by the interference method described by Smith and hliller (8). The effect of the internal pressure on the thickness of one of the cells was evaluated by determining the thickness before and while the cell was subjected to 25 pounds per square inch gage with oxygen. The increase amounted to 0.0004 mm., about O.l%of the 0.427-mm. cell or 47, of the thinnest cell. When it is desired to use the cells for solutions and to fill thy rells from the bottom, the change of orientation of the connections from a lateral to a vertical position is readily accomplished by removing the nuts and springs, lifting the cover plate and window assembly as a unit from the studs, rotating through go", and replacing them in the desired position without impairing the tightness of the cell. The cells may be cleaned readily without taking them apart b j applying suction to one connection and flushing with solvent from a medicine dropper. When it becomes necessary to repolish the windows, a separation a t the least perfect bond may be accomplished as described above in connection with curing leaks. d tightly adhering spacer or gasket, thus exposed on one side, may be removed without damage to the window by disintegration in excess mercury. LITERATURE CITED
(1) Benning, A. F., Ebert, A . A., and Irwin, C. F., ANAL.CHEM.,19 867 (1947). (2) Creita, E. C., and Smith, Francis A., J . Research .\-atl. Bur Standards, 43,365(1949); RP2031. (3) Dibeler, V. H., and Mohler, F. L., J . Research iVat2. Bur. Standards, 39, 149 (1947); RP 1818. (4) Fry, D. L., Nusbaum, R. E., and Randall, H. hl., J . Applied Phys., 17, 151 (1946). ( 5 ) Gildart, L., and Wright, N., Rev.Sci. Instruments, 12, 204 (1941) (6) Kivenson, Gilbert, J . Optical SOC.Am., 39,484 (1949.) (7) Oetjen, R. A., Ward, W. M., and Robinson, J. A., Ibid., 36, 615 (1946). 18) Smith, D. C., and Miller, E. C., Ibid., 34,130 (1944).
RECEIPEDJune 15, 1949
Determination of Carbon Monoxide in Hydrocarbon Gases Containing Olefins PAUL R. THOMAS, LEON DO"',
AND
HARRY LEVIN
Beacon Laboratory, The Texas Company, Beacon, N . Y .
S
TUDIES of hydrocarbon bynthesis and catalytic cracking made it necessary to determine carbon monoxide in gases
that contain large quantities of olefins. Numerous methods have been published for the determination of carbon monoxide. but they are not suitable for samples containing olefins. Colorimetric methods (4, 8) have been used for the determination of carbon monoxide in concentrations less than 1%. Heated cupric oxide ( 1 ) has been employed for oxidizing carbon monoxide to carbon dioxide. However, hydrogen and some paraffin hydrocarbons are also oxidized a t the temperature necessary to oxidize carbon monoxide completely. Teague (16) determined carbon monoxide iodometrically in motor exhaust gas, employing iodine pentoxide for the oxidation, a i t h a precision of 0.003% for samples containing 0.01 to 0.1%. Larson and Whittaker (10) hydrogenated carbon monoxide over nickel catalyst a t a temperature of 290" to 310" C. McCullough et al. (11) oxidized the gas with red mercuric oxide a t elevated temperature. ' Prefient address, Jefferson Chemiasl Company, S e w York, N Y
Some of the reagents more commonly employed for the absorption of carbon monoxide are acid cuprous chloride, ammoniacal cuprous chloride, and a suspension of 2-naphthol in acid cuprous sulfate. These reagents were reported to dissolve olefins (6), and Table I illustrates this solubility. They must be removed before the determination of carbon monoxide. All the above methods depend on the absence of unsaturated hydrocarbon gases. For the removal of unsaturates, sulfuric acid and bromine water have been commonly used. Fuming sulfuric acid and silver sulfate-activated sulfuric acid (7, 16) were found unsuccessful in this laboratory for the removal of ethene from ethene-hydrogen mixtures. Results in Table I1 show that even after 47 passes not all the ethene was absorbed. Incompleteness of absorption was confirmed by hydrogenation of the gases remaining after sulfuric acid treatment of an ethene-hydrogen mixture. Francis and Lukasiewicz ( 6 ) absorbed ethene in mercuric sulfatesulfuric acid reagent, but Brooks et al. ( 2 ) found this reagent slowly oxidizes carbon mon-
V O L U M E 21, NO. 12, D E C E M B E R 1 9 4 9
1477
h procedure is described for determining carbon monoxide in gases that contain olefinic hydrocarbon gases. All hydrocarbons except methane are removed rapidly by condensation in liqiiid nitrogen from which the uncondensed gases-carbon monoxide, hydrogen, nitrogen, oxygen, and methaneare removed bj a Tspler pump. Orsat apparatus is used for the removal of oxygen from this mixture and for the determination of carbon monoxide.
The major part of the carbon monoxide is absorbed in acid cuprous chloride and the remainder in cuprous sulfate-2-naphthol. Carbon monoxide is determined in concentrations from 0.1 to 100% with an average error of less than 0.5% of the carbon monoxide in samples containing 10% carbon monoxide. The accuracy is somewhat less for lower concentrations. The total time required for the determination is less than an hour.
IV). Other investigators (1 7 , 18) also have reported difficultieb with this procedure. The method described in this paper uses liquid nitrogen to condense and remove all hvdrocarbon gases except methane. The noncondensable portion of the sample contains only hydrogen, nitrogen, oxygen, carbon monoxide, and methane. Orsat technique is applied using potassium pyro.--. -_____ ____ ___I_.__ __^_.___ gallate to remove oxygen and acid cuprous chloride, followed by 2-naphthol in acid cuprous Table I. Absorption of Gases Containing Olefins in Cuprous sulfatr to absorb the carbon monoxide. Reagentsa
8)xidr. Savelli (14) obtained good resulth using one fourth saturated brominc w t e r for the absorption of unsaturatrd hvdrocarhons in butane-butene mixtures. However, this procedure, dven if it be a satisfactory means of eliminating olefins without affecting carbon monoxide, is longer and requires much more , ~ e f i icsontrol l than the procedure describrd in this paper.
Composition of Blend, Volume C/c ~- olefins HydroIso2gen Ethene Propene butene Butenes Butane CO ... 28.1 . . . 71.9 28.1 44.2 24.3 42 .'1 13.7 19.9 30.5 30.5 22.3 47.2 ... , . . 32.6 . . . . . . 67.4 .. ... .. 58.5 30.8 17.7 41.5 ... ... 3 0 . 8 . . . 30.8 . . . 69.2 ..( 31.6 31.6 ... 28.9 39.5 ... ... 28.4 .., 71.6 28.4 ' Acid cuprous chloride followed b y cuprous sulfate-2-naphthol mixture.
+
~
-aiiiplr
CO
..
... ...
~-
.
.-
. .
Hopcalite, a mixture of 50% manganese dioxide, 30% copper g)xide, 15% cobalt oxide, and 5% silver oxide, is reported (9) to oxidize carbon monoxide a t ordinary temperatures, which would be a distinct advantage for the present problem. It is reported to be satisfactory on concentrations up to 0.1% (IS); the oxidation is catalytic and utilizes oxygen from the air or from the sample Tried in this laboratory over a wide range of Laoncrntrations up to 20%, only in the absence of ethene and a t low carbon monoxide concentration were fair results obtainable. Oxidation was incomplcxte on higher concentrations and the presence of ethene causts results to be too high (Table 111). This laboratory obtained erratic results for carbon monoxide,
