October 1951
INDUSTRIAL AND ENGINEERING CHEMISTRY
the ten reported) is used is far less important than how much. Choice of vehicle likewise appears to take precedence over choice of inhibitor. Time is a significant variable and inspection of the data of Table VI11 shows that mold growth on the whole does increase significantly with time. Inspection of the interactions in Table V, however, shows the important qualification that must be laid on the significance of the time variable. Only one of the time interactions (fungicide-time) has any statistical significance whatever. The almost general absence of significance in the time interactions suggests that the mildewing of the wires is not identified with differential rates of, for example, breakdown of the two vehicles. It suggests too that the exposure periods might be curtailed for although mold growth does increase with time, the comparative results of the earlier observations are equally indicative. The highly significant vehicle-concentration interaction indicates the possibility of a difference in solubility of the fungicides in the two vehicles. The work herein reported has several implications applicable to present practices. Most important is the surprising inadequacy of many common fungicides. Laboratory methods capable of discerning these inadequacies are sorely needed. For future work this laboratory plans to continue its explorations of fungicides with emphasis on mixtures of the more active compounds. However, the primary importance of vehicle, as disclosed in this study, will be given its rightful place in the experimental plans. SUMMARY
Approximately 100 different toxic agents a t three concentration levels were incorporated in varnish and lacquer vehicles. The fungus-inhibitive properties of the coatings were determined through natural exposure in a Panama jungle. Most of the toxicants studied, including many of the popular chlorinated phenols, were ineffective, Among the most nearly satisfactory are the following ten, listed in an approximate descending order of effectiveness. o-Hydroxyphenylmercuric chloride Salicylanilide Pyridylmercuric chloride p-Toluene sulfonylamide Uranyl nitrate p-Aminophenylmercuric acetate Uranyl nitrate p-Aminophenylmercuric acetate
2341
Phenylmercuric-o-benzoic sulfimide p-Acetylaminophenylmercuric acetate Hydroxymercurisalicylic acid anhydride Phenylmercuric phthalate The gross evidence of the data is supplemented by a statistical analysis of the observations upon these ten compounds. The following conclusions are drawn from the analysis: Each of the rimary variables (vehicle, fungicide, concentration, and time) f a s a significant influence on mold growth. The varnish vehicle is more resistant to mold than the lacquer. Similarly the toxic varnishes are more resistant than the toxic lacquers. The choice of an individual toxicant (from among the best ten) is of less importance than the choice of vehicle. Retardation of mold varies directly with concentration of toxic agent. Concentration is the most important of the four variables. Study of the time interactions indicates that useful data are obtainable from relatively brief exposures. ACKNOWLEDGMENT
The authors are grateful to Clarence M. Shepherd, Leland B. Shanor, Walter L. Clark, and Ruth M. Little for their interest and assistance in various aspects of this investigation. LITERATURE CITED
(1) Battelle Memorial Institute, Final Report to Air Mathel Command on Contract 33(038)-450 (Nov. 2, 1949). (2) Davies, 0. L., “Statistical Methods in Research and Produc-
tion,” 2nd ed., London, Oliver and Boyd, 1949. (3) Downing, G. V., and Whitmore, L. M., J . A m . Leather Chemists’ Assoc., 36,210 (1941). (4) Findlay, W. P. K., J . Incorp. Brewers’ Guild,26,105 (1940). ( 5 ) Fisher, R. A., “Statistical Methods for Ryearch Workers,” London, Oliver and Boyd, 1936. (6) Hutchinson, W. G., “Studies on Methods of Testing Coating
Materials for Resistance to Fungus Attack,”Officeof Scientific Research and Development, Rept. 5687 (Oct. 31,1945). (7) Rider, P. R., “An Introduction to Modern Statistical Methods,” New York, John Wiley & Sons, 1939. (8) Snedecor, G. W., “Statistical Methods,” 4th ed., p. 175, Ames, Iowa, Iowa State College Press, 1946. (9) Ibid., p. 218. (10) Thompson, J. C., Jr., and Sandford, W. E., Brewers Digest, 16, 27 (1941). (11) Thomson, J. W., Creamery J.,53,10 (1942).
RECEIVED December 9, 1950.
Flammability of BenzeneChlorine Mixtures -
GEORGE CALINGAERT‘ AND WILLIAM E. BURT Research Laboratories, Ethyl Corp., Detroit, Mich.
T
0 ASSESS the hazards involved in the chlorination of benzene on a commercial scale it is necessary to know the conditions under which mixtures of these two materials will sustain and propagate combustion. It is known that certain mixtures of liquid chlorine with hydrocarbons are highly explosive (I), but little information is available on mixtures involving gaseous chlorine, McBee, Hass, and Pierson ( 3 ) give the lower and higher explosive limits of dichloropentane vapor in chlorine as 9 and 6 moles of chlorine per mole of dichloride, and Hass ( 8 ) states that the lower explosive compositions for gaseous propane and butane in chlorine contain one atom of chlorine per atom of carbon. In the present study, data were obtained on the flash point of 1
Present address, Hobarb College, Geneva, N. Y.
benzene in an atmosphere of chlorine, the flammability limits of mixtures of benzene vapor with chlorine, and the underliquid spark-ignition of solutions of chlorine in liquid benzene. FLASH POINT
The flash-point was determined in the Pensky-Martens closedcup apparatus, in the absence of air. The benzene was saturated with chlorine a t a low temperature in the absence of light and the solution in the cup was stirred as the temperature was slowly raised. No additional chlorine was introduced. The vapor phase was periodically exposed to the flame and the flash point was taken to be the tem erature a t which the vapor phase ignited and burned througfout the cup. Duplicate determinations provided values of 13.3’ and 14’ C. At this temperature level, the mole ratio of chlorine to benzene in the vapor is about
2342
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 43, No. 10
Because of their importance in the industrial chlorination of hydrocarbons, the explosion hazards involved in the chlorination of benzene were investigated on a laboratory scale. The flash point of benzene in an atmosphere of chlorine in the Pensky-Martens closed cup apparatus is 13" to 14' C. For gaseous benzene-chlorine mixtures at about 80' C. and atmospheric pressure, with downward propagation of flame, the lower flammability limit is between 8.3 and 9.2%benzene and the higher limit is between 32.6 and 35.0Yo. Exposure to light at the time of ignition widens these limits slightly. For upward propagation, the higher limit is raised to approximately 52%. A liquid benzenechlorine solution at atmospheric pressure is not ignited by
propagation of flame through the saturated vapor above it, nor is it exploded by the discharge of a spark beneath the surface at the lowest temperature tested, 10' C. However, spark discharge in such solutions at temperatures ranging from about 13"to 39"C. creates a bubble of burning vapor which grows as it rises to the surface. On a large scale, this might present a distinct hazard. The data reported indicate that benzene-chlorine mixtures are in general less hazardous than benzene-air mixtures, and ordinary safety devices should be adequate in metal equipment at atmospheric pressure. Even at atmospheric pressure the vapor phase above liquid benzene in an atmosphere of chlorine is potentially hazardous above the flash point of 13' to 14" C., up to at least 60' C.
13 to 1. This composition is near that given below for the lower limit of flammability of gaseous chlorine-benzene mixtures. By comparison, the flash point for benzene in air is -20" C.
for both upward and downward propagation, ab: one tost of a 7% benzene mixture did not show upward propagation. These limits are much higher than those for benzcne in air where, for dry gases a t 60" C., the values reported ( 4 )are 1.48 and 5.55% for downtTard propagation, 1.45 and 7.45% for upward propagation. The burning of mixtures containing high concentrations of hcnBene in chlorine left a solid, carbonaceous deposit on the wall of the tube. However, the burning of the mixtures containing 14.3% or less benzene deposited a yellow solid which proved to be hexachlorobenzene (for which product the etoichiometric niixture contains 14.3% benzene). To learn if the above limits determined in the absence of light would be changed by the simultaneous occurrence of photochlorination, gaseous mixture of 37% benzene in chlorine was prepaied, this coricentration being just above the higher limit for downward flame propagation. Just prior to attempting ignition of this mixture, an 18-inch daylight fluorescent lamp was lighted in front of the viewing port, and an infrared heat lamp was directed a t the top of the tube after removal of the stopper. With application of the flame, ignition took place and the flame traveled very slowly down the tube for approximately 12 inches, then expired. Hence it appears that the presence of light very slightly widens the flammability limits.
FLAMMABILITY LIMITS OF GASEOUS BENZENE-CHLORINE MIXTURES
Wide experience a t the Bureau of Mines ( 4 ) has shown that observations of the flammability of gases and vapors made in a vertical tube 2 inches in diameter and 4 to 6 feet in height are nearly the same as those made in much larger apparatus. The equipment used in this investigation consisted of a stoppered borosilicate glass test tube, 2 inches in diameter and 40 inches long, provided with an agitator made of a perforated cork attached to a wire handle which passed through the stopper. The tube had an electrically heated jacket which was covered to exclude light. A long, narrow viewing port in the cover was opened just before ignition to allow observation of flame travel. For downward flame propagation, the procedure consisted in flushing the tube with chlorine, adding the desired quantity of benzene as liquid, vaporizing and mixing it by repeated slow raising and lowering of the agitator, removing the stopper, and finally applying a flame to the open end of the tube. For upward propagation, after charging and mixing the agitator was removed, the stopper was promptly replaced, the tube was inverted, and a 0.25-inch opening in the stopper was unplugged for application of the igniting flame. At temperatures near the dew point, vaporization of the benzene was so slow that partial chlorination or leakage might have caused the composition to change during mixing. Therefore a level of approximately 80" C. was arbitrarily established for all mixtures. At that temperature, complete vaporization of the benzene and adequate mixing of the gases were obtained in a 10minute period, as evidenced by the uniformity in burning of those mixtures capable of flame propagation. Small variations in temperature in the vicinity of 80' C. were of no concern, since they should have no significant effect on the flammability limits ( 4 ) . The limits of flammability determined with downward propagation of flame a t the prevailing atmospheric pressure (about 750 mm. of mercury) were located between 8.3 and 9.2% benzene and between 32.6 and 35% benzene. Compositions of 8 and 35% benzene correspond to the vapor pressure of pure benzene a t 16' and 49' C., respectively. The rate of propagation varied with composition of the mixtures and, as expected, was slowest near the limits. The only quantitative measurement made mas for a mixture containing 14.3% benzene, for which the rate of downward propagation was 23 cm. per second; mixtures somewhat richer in benzene appeared to burn faster than this. The upward propagation was considerably less uniform; the flame moved with an irregular undulating motion which may have been caused by convection currents. The upper flammability limit wm appreciably higher than for downward propagation, being located between 50 and 52.7% benzene. The latter value corresponds to the vapor pressure of pure benzene a t 61 " C. Apparently the lower limit of flammability is essentially the same
BEHAVIOR OF LIQUID SOLUTIONS
In the flash point determination described above, there was no evidence of reaction in the cold liquid phase. It seemed of interest to determine whether, a t higher temperatures, inflammation of the vapor could set ofi a reaction in a liquid solution in contact with it. One test of this was made in the flame tube a t 44' C. and atmospheiic pressure, with about 50 nil. of liquid benzenechlorine solution in the bottom of the tube and a saturated vapor phase. Do.rmvrard propagation of flame through the vapor was irregular, apparently due to failure to reach homogeneity in preparing the charge. The flame moved slowly down the tube to a point 12 to 18 inches from the liquid surface, then traveled rapidly. There was, however, no evidence of explosion or reaction within the liquid phase, and the flame expired a t the surface of the liquid. Further experiments were made to determine whether the fortuitous occurrence of a flame or spark in a volume of liquid benzene containing chlorine would propagate a flame or a fast reaction through the liquid. The source of ignition used was a highvoltage spark generated by a Delco Remy induction coil of the type commonly used in automobile ignition systems, rated a t 12,000 volts maximum output for a 6-volt input. An open 30 X 190 mm. borosilicate glass test tube, wrapped to exclude light, was approximately half filled with benzene saturated with chlorine a t about 10" C. The tube had sealed-in tungsten wire electrodes with a gap of 0.004 to 0.007 inch between the points, which
October 1951
INDUSTRIAL AND ENGINEERING CHEMISTRY
were usually located a t a depth of 1 or 2 inches below the surface of the liquid. While raising the temperature of the solution (slowly, to avoid turbulence owing to the evolution of chlorine), a momenta ry spark wm discharged periodically. In these tests there was no evidence a t any time of propagation of reaction through the liquid phase itself. However, a t liquid temperatures in the range from 18" t o 37" C., a small, burning bubble of gas formed around the spark and rose t o the surface, expmding as it rose. I t s evolution from the liquid was accompanied by a brief flash of orange-red flame and a puff of dense smoke due to momentary ignition of the vapor above the liquid. Increasing the input t o the spark coil to 8 volts widened the ignition limits to 13.5" and 39" C., but further increase to 10 t o 12 volts had no more effect. At these limits, the calculated solubilities are, respectively, 0.37 and 0.135 mole of chlorine per mole of benzene. These temperatures are not precise, duplicate determinations usually showing variations from 0' to 4" C. This may well have been due to changes in composition resulting from a certain amount of photochemical chlorination induced by the spark or by light entering the open top of the tube during the course of a determination. When a similar ignition was carried out in a 33-mm. tube 44 inches long, which was three quarters filled with solution at 22' C. and had the spark located 28 inches below the surface, the expansion of the rising gas bubble was sufficient to eject one third of the liquid from the tube.
2343
SAFETY HAZARD
The data reported here indicate that benzene-chlorine mixtures are in general less hazardous than benzene-air mixtures. In metal equipment, a t atmospheric pressure, ordinary safety devices should be adequate. Even a t atmospheric pressure, however, the vapor phase above liquid benzene in an atmosphere of chlorine is potentially hazardous a t temperatures above the flash point of 13" to 14" C., up to a t least 60" C. Liquid solutions of chlorine in benzene a t atmospheric pressure and a t temperatures down to a t least 10" C. and perhaps lower, contain insufficient chlorine to be explosive and are not ignited by inflammation of the contiguous vapor phase. Even so, the occurrence of a spark (or, presumably, any other sufficient source of heat) in a volume of the liquid solution may result in an evolution and inflammation of vapor beneath the surface. This, on a large scale, might present a considerable hazard. LITERATURE CITED
(1) Brooks, B. T., IND.ENO.CEEM.,17,752(1925). (2) Hass, H. B., "Science of Petroleum," p. 2788, London, Oxford
University Press, 1939. (3) McBee, E. T.,Hass, H. B., and Pierson, Earl, IND.ENG.CHEM., 33,181(1941). (4) U. S, Dept. Interior, Bur. Mines, Bull. 279 (1939). RECEIVED January 29, 1951
Burley Tobacco J
RELATION OF THE NITROGENOUS FRACTIONS TO SMOKING QUALITY J. M. MOSELEY, W. R. H A R M , AND H. R. HANMER The American Tobacco Co., Richmond, Va. Efforts to evaluate, by chemical means, and to improve the quality of Burley tobacco have engrossed the attention of many research workers in the United States Department of Agriculture and state experiment stations. The authors' laboratory has been engagedin this fieldof research for many years. There is a striking paucity of published information on the subject. These findings provide one means of correlating composition with recognized standards of quality. This paper reports a useful method of evaluation which has not heretofore been published and which is being employed as a guide in the development of new and improved varieties of Burley tobacco. Burley tobacco occupies about 450,000 acres of fertile farm land and provides a cash income of approximately $250,000,000 annually to over 275,000 individual farmers. A contribution to the evaluation, hence improvement, of this economically important crop is significant.
T
H E various qualities by which the leaf expert judges tobacco include color, body, conformation, texture, elasticity, and aroma. The second of these, body, is the most difficult t o define. Theterm as used in the trade appliesin general tostrengthof smoke and is intimately related to smoking quality. Body is recognized principally by the thickness, texture, and gumminess of the leaf, but the exactness with which the chemical characteristics of leaf are related to the physical is subject t o variance. The leaf experts, making use of knowledge acquired from long experience, maintain a remarkable degree of proficiency in the judgment of quality of leaf in terms of manufacturing characteristics, burning
properties, and certain attributes of taste such as sweetness and aroma. The criteria of quality which they employ are not so accurately indicative of strength and blending properties. A crop grown under abnormal conditions may possess unusual smoking characteristics not easily recognized from the physical properties of the leaf. This, with the expanded nature of the industry, lends increasing importance to chemical methods for both evaluation and maintenance of uniform quality. Various investigators have attempted to correlate analytical data with,the observed quality of flue-cured and foreign types (9, 21, 2.2, 16,27, 18). In general, carbohydrates, essential oils, and resins are prominently classed among the compounds which are positive in their influence on tobacco quality. Proteins, pectins, oxalic acid, and ash are among the substances generally considered to have a negative influence. Burley tobacco, the most important of the air-cured types, is widely used in blended cigarettes, pipe mixtureg, and plug chewing products. The cutter grades such as shown in Figure I furnish the bulk of the cigarette leaf. I n popular brands of domestic cigarettes it is usually blended with flue-cured, Turkish, and Maryland. Quality of cigarette grades of Burley is therefore considered only from the standpoint of the combined effect when blended with types of dissimilar composition. I n this respect, the problems of American investigators differ from those in foreign countries, where the a r t of blending of types has not been developed. Detailed work on Burley tobacco has been reported by Shedd (19, bo), who determined total nitrogen, nicotine, nitrate nitrogen, and some ash constituents on good, medium, and common samples. His data showed a higher potash content for the better
