602
INDUSTRIAL AND ENGINEERING CHEMISTRY ACKNOWLEDGMENT
The authors wish to acknowledge the assistance of William R. Musick i n obtaining the experimental data and the helpful suggestions and interest of W. F. Busse. LITERATURE C l T E D
(1) Bayley, C. H., and Weatherburn, A. S., Testile Research J . , 2 0 ,
510 (1950).
(2) Compton, Jack, and Hart, W. J., IND.ENG.CHEM.,43, 1564
(1951).
(3) Hart, W. J., and Compton, Jack, Ibid.,44, 1135 (1952).
Vol. 45, No. 3
(4) Kubelka, P., and Munk, F., 2.tech. Phys., 12, 593 (1931). (5) Lambert, J. M., and Sanders, H. L., IND. ESG.CHEM.,42, 1388 (1950). (6) Mankowioh, A. M., Ibid., 44,1151 (1952). RECEIVED for review December 11, 1951. .bxePrm October 30, 1982. Presented a t the XIIth International Congress of Pure and Applied Chemistry, New York, September 1951. Report of work done under contract with the U. 6. Department of Agriculture and authorized by the Research and Marketing Act. The contract is being supervised by the Southern Regional Laboratory of the Bureau of Agricultural and Industrial Chemistry. The mention of trade products in this paper does not imply their endorsement by the Department of Agriculture over similar products not mentioned.
Relation of Smoke Molecular Structure RUSSELL A. HUNT, JR. Research Department, Standard Oil Co. (Indiana), Whiting, Znd.
S
MOKE points have been used since about 1930 as a measurement of quality of kerosene used as a n illuminating oil. Smoke point is defined as the height in millimeters of the highest flame produced without smoking when the fuel is burned in a specified test lamp. A standard method of test was described ( 7 ) before the World Petroleum Congress in 1933. Kewley and Jackson ( 3 ) showed t h a t the smoke points of organic compounds varied widely and decreased for pure hydrocarbons in the order: alkanes, cycloalkanes, aromatics. Minchen (6) amplified this work and developed a n empirical equation with which he calculated smoke points for some of the members of several homologous series. The Factor lamp, which was developed by Davis during the early 1920's, was modified so that smoke points could be determined with accuracy (6). Clark, Hunter, and Garner ( 1 ) used a modified smoke-point apparatus to screen organic compounds for use in incendiary bombs. This apparatus differed from a smoke-point lamp i n that no wick was employed, but burning occurred a t the surface of a pool of liquid. With this apparatus a group of 25 hydrocarbons and 80 compounds containing oxygen and nitrogen were investigated. On the basis of the work with the 25 hydrocarbons, Clark made several basic observations on the relation of smoke point to molecular structure: Straight-chain alkanes have the highest smoke points. Branching decreases the smoke point markedly, but the position of the branches on the molecule makes little difference. Addition of a n olefinic linkage to an n-alkane appreciably lowers the smoke point. Cycloalkanes have about the same smoke points as highly branched alkanes; apparently the number of carbon atoms in the cycloalkane ring has little effect. As with alkanes, the introduction of a double bond drops the smoke point markedly. Aromatics have low smoke points, irrespective of the configuration of aliphatic side chains. Clark concluded t h a t the compactness of the hydrocarbon molecule is responsible for smokiness. Compounds of oxygen and nitrogen had smoke points as high as or higher than the corresponding hydrocarbons. In a n investigation of combustion processes in a standardized pot-type space heater, the amount of soot formed on the walls of the burner pot when a given amount of fuel was consumed was found in this laboratory to correlate with the smoke point of the fuel. A high smoke point indicates a heating fuel of better quality in the same manner t h a t a high smoke point for kerosene indicates better illumination properties. Smoke point thus provides a general means for studying the combustion properties of individ-
ual compounds which might occur in virgin petroleum distillates boiling in the range of 160' to 330' C. Compounds were selected for test from types known to be present in the distillates. The main constituents of these fuels are alkanes, cycloalkanes, and aromatic hydrocarbons; alkenes and even alkynes are sometimes present in untreated distillates in minute quantities, and compounds of oxygen, nitrogen, and sulfur are generally present as impurities. With the exception of oxygen compounds, which were thoroughly studied by Clark ( I ) , representatives of these classes of compounds were included in this study. The 108 compounds studied are listed in Tables I, 11, and 111, along with observed and literature values for refractive index as a n indication of the purity of the compounds. I n order to study as broad a molecular-weight range as possible, and because of the unavailability of individual isomers of the higher molecular weights, many compounds boiling below the range of the distillates being considered were included, EXPERIMENTAL PROCEDURE
The smoke-point apparatus used was an improved Factor lamp (6) having a 0.25-inch nick tube and a cylindrical glass chimney 7 inches long with an outside diameter of 1 inch. In order to test small samples in the lamp, a 12-ml. glass vessel that could be filled to the depth of oil normally used in the lamp font was placed in the 4-ounce font for this study. The 0.25-inch white felt wicks (No. D-187, Complete-Reading Electric Co., Inc., 100 South Jefferson St., Chicago 6, Ill.) used are superior to either x-oven or sewn wicks, in that they are uniform in quality and easy t o trim. The wicks were cleaned in a Soxhlet extractor -first with benzene or benzene-ethyl alcohol, then with hexane. After most of the solvent was removed by evaporation, the wicks were dried overnight in an oven a t 200' F. A 4-inch length of wick was prepared for use by cutting one end square with a small device resembling a guillotine and singeing from the freshly cut surface small protruding fibers that might cause uneven burning. After the lamp was assembled, filled with material to be tested, and lighted, the wick was adjusted t o give a nonsmoking flame, not over 10 mm. in height, and allowed to burn 10 minutes t o establish eauilibrium. The flame was then turned UTI until a smoky tail appeared a t the top, and was turned down b t i l the smoky tip just disappeared. The height of the flame in millimeters at this point was taken as the smoke point ( 2 ) . Each smoke point was determined by two observers independently. If the readings differed by more than 2 mm., they were repeated until a point agreed upon by both observers was obtained.
As various modifications of the test lamp and procedure have been used, a quantitative comparison of the results of various
March 1953 TABLEI.
INDUSTRIAL AND ENGINEERING CHEMISTRY
SMOKEPOINTS OF ALKANES, ALKENES, ALKYNES, AND CYCLOALKANES
Compound n-Hexane 2-Methylpentane 3-Methylpentane 2,2-Dimethylbutane 2,3-Dimethylbutane n-Hentane ~ ~ . ~ . - ~ ~ 2-Methylhexane 3-Methylhexane 2,3-Dimethylpentane 2,4-Dimethylpentane n-Octane 2-Methylhcptane 3-Methylheptane 4-Methylheptane 3-Ethylhexane 2 2-Dimethylhexane 2'3-Dimethylhexane 2lMethyl-3-eth lpentane 2,2,4-Trimethy&entane n-Decane n-Undecane n-Dodecane n-Tridecane n-Tetradecane 1-Hexene 1-Heptene 2-Heptene 1-Octene 2-Octene 1-Decene 1-Dodecene 1-Tetradecene 1-Hexadecene 1-Octadecene 1-Oct ne 1-Dozecyne C clohexane ethylcyclohexane cis-l,3-Dimethyloyclohexsne Ethylcyclohexane Cyclopentane Methylcyclopcntane n-Decylc yolopentane Bicyclohexyl Decalin Cyclohexene
J
Refractive Index, n%? Observed Lit. 1.3750 1.3750 1.3717 1.3715 1.3763 . 1.3766 1.3688 1.3689 1.3750 1.3750 1.3877 . 1.3877 1.3850 1.3849 1.3885 1.3887 1.3917 1.3920 1.3813 1.3815 1.3973 1.3976 1.3955 1.3955 1.3983 1.3985 1.3979 1.3980 1.4017 1.4021 1.3929 1.3931 1.4013 1.4013 1.4045 1.4046 1.3913 1.3914 1,4119 1.4114 1.4175 1.4173 1.4216 1.4215 1,4254 1.4250 1.4289 1.4288 1.3898 1.3880 1.4011 1.3991 1.4058 1.4041 1.4102 1.4090 1.4125 1.4130 1.4228 1.4220 1.4301 1.4308 1.4369 1.4365 1.4411 1.4441 1.4448 1.4443 1.4159 1.4230 1.4360 1.4351 1.4258 1,4262 1.4228 1.4230 1.4232 1.4310 1,4332 1.4328 1.4060 1.4064 1.4090 1.4097 1.4792 1.4761 1.4478
1,4795 1.475 1.4467
Smoke Point 149 137 135 114 120 147 136 137 123 117 149 137 126 117 117 100 123 102
86
139 142 137 133 137
88 96 91 99 99 94 96 91 94 94 24 36 117 94 70 96 84 64 56 38 47
investigators is difficult. Care must be exercised in interpreting results, unless the test procedure and apparatus are known to be identical. RESULTS AND DISCUSSION
rll
The smoke points of the compounds studied are shown in Tables I, 11, and 111. Aromatic compounds have uniformly low smoke points which do not differentiate satisfactorily between compounds. Various aromatic compounds diluted to the same concentration in an n-alkane solvent might be expected to show different smoke points. To check this theory, varying amounts of sec-butylbenzene and 1-methylnaphthalene were dissolved in n-dodecane, and smoke points were determined on the mixtures. The results are plotted in Figure 1. The difference in the effects of the same concentration of these two aromatics indicates that burning characteristics of a fuel cannot be predicted solely by st determination of aromatic content. The smoke points of 20% mixtures in n-dodecane of the aromatic hydrocarbons studied were determined and are given in Table I1 as blended smoke points. A more detailed treatment of each type of compound is shown in figures that plot smoke points against the total number of carbon atoms in the molecule. ALKANES, ALKENES,AND ALKYNES. I n Figure 2 are shown the smoke points of the unbranched hydrocarbons. All the nalkanes have high smoke points, the several n-alkenes are intermediate, a n d the two n-alkynes have much lower values. The smoke points for the n-alkanes decrease with increasing number of carbons, those of the n-alkenes remain constant, a n d - o n the basis of the limited evidence-those of the n-alkynes increase.
603
TABLE11. SMOKEPOINTS AND BLENDEDSMOKEPOINTS OF AROMATIC HYDROCARBONS Compound Benzene Toluene Ethylbenzene 0-X lene m-Sylene p-Xylene n-Propylbenzene Cumene Mesit lene Mixec?trimethylbenzenes p-Cymene n-Butvlbenzene Isobucylbenzene sec-Butylbenzene tart-Butylbenzene eec-Amylbenzene tert-Amylbenzene Mixed diethylbensenes Mixed triethylbensenes Mixed triamylbensenes Tetralin Naphthalene 1-Methylnaphthalene 2-Methylnaphthalene Mixed dimethylnaphthalenes Diphenyl Styrene Phenylcyclohexane Indene
Refractive Index, n?P Observed 1.5098 1.4955 1.4952 1.5023 1.4969 1.4950 1.4839 1.4912 1.4958 1.4962 1.4899 1.4893 1.4865 1.4893 1,4923 1.4883 1.4911 1.4947 1.4995 1.4897 1.5438
Lit. 1.5012 1.4968 1.4958 1.5054 1.4971 1.4958 1.4919 1.4912 1.4991
1.6131
1: 4633 1,4880 1.493 1.4880 1.4905 1.4852 1.4934
.... .... ....
....
1.5438
.... 1.6007 ....
1.6149
.... ....
1.5472 1.5247 1.5598
1.5470 1.5254 1.5764
....
8
6 6
6 5 5 8 6 6 6 5
e
7 7 5 8 8 7 5 7 6 4 5 5
5
4 7 7
56
66 72
TABLE 111. SMOKE POINTSOF SULFUR AND COMPOUNDS Compound n-Hexyl mercaptan n-Hept 1 mercaptan n-Decyrmercaptan n-Dodecyl mercaptan Benzyl mercaptan o-Thiocresol Thiophenol %-Hexylsulfide n-Heptyl sulfide n-Octyl sulfide n-Decyl sulfide Phenyl sulfide p-Cresolmethyl sulfide Thiophene 3-Methylthiophene Diethylamine Triethylamine +Butylamine Di-n-butylamine Tri-n-butylamine sec-Butylamine Aniline o-Toluidine m-Toluidine n-Butylaniline Pyridine a-Picoline S-Picoline '&Picoline Quinoline Isoquinoline Quinaldine 2.7- Dimethylquinoline
Refractive Index,% 'n Observed Lit. 1.4495 1 4475 1.4518 1.450' 1.4567 1.454 1.4589 1.459 1.5758 1.589 1.5789 I . 5900 1.5861 1.4588 1.459 1.4641 .... 1.4620 .... 1.4655 1.6329 1.635 1.5732 1.573 1.5278 1.5287 1.5199 1.5204 1.3902 1.3873 1.4008 1.4003 1.4027 1.401 1.4188 .... 1.4318 1.3958 1:3950 1.5843 1.5863 1.5711 1.5728 1.5669 1.5711 1.5338 1.5381 1.5080 1.5092 1.5002 1.5029 1.5052 1.5043 1,5008 1.5064 1.6241 1.6245 1.6215 1.6223 1.6108 1.6126 1.6025
....
....
....
Blended Smoke Point 66 48 64 58 56 56 78 71 56 56 61 80 68 76 63 81 66 58 56 70 48 42 37 37 36 44
Smoke Point
NITROQEN
Smoke Point 102 132 119 102 6 3 5 114 104 87 51 4 3 33
8
103 122 104 112 102 114 15 13 11 17 29 19 13 19 6 5
12 4
Smoke points of 16 branched-chain alkanes in the Ce, Cy, and Cs range are shown in Figure 3. These data bear out the findings of Clark (1)that branching markedly decreases smoke point. Contrary t o his findings, the point of substitution on the m o l e cule also affects smoke point, particularly among the methylheptanes and dimethylhexanes. The 2,a-dimethyl substitution has the greatest effect on smoke point among the compounds studied. CYCLOALKANES AND CYCLOALKENES. Changes in molecular structure greatly affect the smoke points of cycloalkanes, as shown in Figure 4. Cyclohexane has a smoke point of 11733 higher than cyclopentane and 70 higher than cyclohexene.
INDUSTRIAL AND ENGINEERING CHEMISTRY
80
% n-DODECANE 60 40
\
15C
Vol. 45, No. 3
0
20 I
U
0 m-BUTYLBENZENE 0 a-METHYLNAPHTHALENE
IOC
-
0 0
I-
z
0
r)
n 0
I2
u
n- ALKENES
a0 . W Y
0
m
50
C
I
2o Figure 1.
I
I 60
IOCARBON ATOMS 15
I
80
40 % AROMATICS
20
100 Figure 2.
Smoke Points of Straight-Chain Hydrocarbons
Smoke Points of Aromatics in n-Dodecane of a methyl group t o either cyclohexane or cyclopentane causes a drop in smoke point of 20; the addition of another methyl group t o methylcyclohexane to form czs-1,3-dimethylcyclohexane causes a further drop of 23 units. The smoke points of the mono-n-alkylcycloalkanes appear t o increase gradually with increasing molecular weight from the low value for the methyl derivative. The lowest value found for a cycloalkane was t h a t of the fused-ring compound Decalin. AROXATICS. When blended smoke points are determined on aromatic hydrocarbons, significant differences in smoke points are found, and there is a definite relation of smoke point to molecular structure. As shown in Figure 5 , the blended smoke point drops from benzene t o toluene and then increases for increafling molecular weight for the n-alkyl-substituted benzenes. For any given monoalkylbenzene, the blended smoke point decreases in the order: n-alkyl, sec-alkyl, isoalkyl, tert-alkyl. All types of monoalkylbenzenes show increased blended smoke points for increased molecular weights. $mong the disubstituted benzenes, the dimethyl compounds all have about the same blended smoke point. p-Cymene has a value somewhat higher than the xylenes but definitely lower than cumene. As in the case of the cycloalkanes, increasing the number of side chains lowers the blended smoke point of aromatics whereas lengthening the chain increases it. The fused-ring compounds had the lowest points observed. SULFUR COMPOENDS. Smoke points of the organic sulfur compounds are plotted in Figure 6. The values for alkyl mercaptans (thiols) and sulfides are somewhat lower than those for the corresponding n-alkanes. Thc aryl mercaptans and sulfides have smoke points equivalent to those of the parent hydrocarbons. Thiophene has a smoke point of 32, a value surprisingly high in view of its unsaturation and its benzenelike behavior in chemical reactions. The drop in smoke point characteristic of methyl substitution is again shown by 3-methylthiophene. NITROGEN COMPOVSDS.D a t a for the organic nitrogen compounds are shown in Figure 7 . The alkyl amines, like the alkyl sulfur compounds, have high smoke points, although somewhat
' 6 0
14
+=ya 2-METHYL
I-
z_
h
2,3-DIMETHYL
x
w 12
8
-
4-METHYL
Y
0
0
0
5
IC
2,2,4-TRIMETHYL 0
801
I 6
I
7
1
8
TOTAL CARBON ATOMS
Figure 3. Smoke Points of Hexanes, Heptanes, and Octanes The addition of a n olefinic bond has a much greater effect on cyclohexane than on a n n-alkane. When the number of carbon atoms in the ring is varied, the difference in smoke point is much greater than would be indicated by Clark's data. The addition
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
March 1953
605
80 -
E-
70-
0
n w Y
2 60-
Y
v)
n W
n
z
50-
40
-
0 5
10
CARBON ATOMS
15
301
20
Smoke Points of Cycloalkanes
Figure 4.
I 6 s
Figure 5.
I
I
I
8 10 12 CARBON ATOMS Blended Smoke Points of Aromatics
I50
100-
100 c
z 8 k2 0
I-
z 2
il
5
50 -
50
C
0' 0 0 Figure 6.
5
10
15
-20
CARBON ATOMS Smoke Points of Sulfur Compounds
lower than the corresponding alkanes. Differences in branching by substitution on the nitrogen atom to produce primary, secondary, and tertiary alkyl amines do not result in lowering of the smoke point, as is the case with branched-chain alkanes. Aromatic nitrogen compounds have appreciably higher smoke points than their hydrocarbon analogs. The addition of an n-butyl group t o aniline does not appreciably raise the smoke point. The increase from quinoline to quinaldine is the only instance
0 Figure 7.
5 10 15 CARBON ATOMS Smoke Points of Nitrogen Compounds
noted in this study of adding a single methyl branch without lowering the smoke point. CONCLUSION
The marked changes in smoke point with molecular structure can probably be best explained in the light of structural retarda-
606
INDUSTRIAL AND ENGINEERING CHEMISTRY
tion factors described by Livingston (Q)in his studies of octane number. The factors he used-for calculating octane numbers would seem t o also apply to low-pressure open flames of the type studied here. H e devised a series of rules t o calculate the effects of various structures on the oxidation rates of hydrocarbons according t o the location of the group in the molecule; methyl, tert-alkyl, vinyl, cyclic methylene, and phenyl radicals are oxidation-retarding groups, and of these the phenyl group has the strongest influence. Certain parallels and extensions of Livingston’s work can be drawn from the present study. Particularly noticeable is the change i n smoke point noted on the addition of a methyl branch. On the basis t h a t a n oxidation-retarding group lowers the smoke point, sec-alkyl and isoalkyl groups should be added t o Livingston’s list. The very low smoke points of the aromatic hydrocarbons would be explained by the strong retarding influence of the phenyl radical. According to the theory of oxidation-retardation and as demonstrated in the present work, the n-alkyl structure is the easiest t o oxidize, and a n y substitution on the molecule causes i t t o resist oxidation in accordance with the position and type of substitution.
Vol. 45, No. 3
ACKNOWLEDGMENT
The author wishes t o acknowledge the valuable assistance of George Hajduk in obtaining the data reported in this study. LITERATURE CITED (1) Clark, A. E., Hunter, T. G., and Garner, P. H., J . Inst. Petroleum, 32, 627 (1946). ( 2 ) Institute o f Petroleum, London, “Standard Methods for Testing Petroleum and Its Products,” p. 456, 1951. (3) Kewley, J., and Jackson, J. S., J . Znst. Petroleum Technol., 13, 364 (1927). (4) Livingston, H. K., IND. ENQ.CHEM.,43,2834 (1951). ( 5 ) Minohen, S. T., J . Znst. Petroleum Technol., 17, 102 (1931). ( G ) Terry, J. B., and Field, E., IND. Esr,. CHEM..Ax.4~.Eo., 8 , 293 (1936). (7) Woodrow, W. A., Secorid World Petroleum Congr., London 1@SS, Proc., 2 , 7 3 2 . RECEIVED for review July 21, 1952. ACCEPTEDOctober 10, 1952. Presented before the Division of Petroleum Chemistry a t the 122nd Meeting of the + x E R I C A N C H E M I C A L S o C I s r Y , .&tlantic City, N. J.
PVT Relations of Nitrogen and ixtures at High Pressure WILLIAM P. HAGENBACH’ AND EDWARD W. COMINGS2 University of Illinois, Urbunu, I l l .
A
LARGE amount of compressibility data covering a significant range of temperatures and pressures is available for pure gases. The amount of data on mixtures of gases is rather meager but has increased considerably in recent years. A knowledge of the compressibilities of mixtures is important in the design of high pressure equipment. Accurate compressibility measurements also provide a basis for calculating the thermodynamic properties. I n the absence o’f P V T data for gas mixtures, their compressibility factors must be estimated from the factors for the component gases. The methods of K a y ( 7 ) and Gilliland ( 5 ) are useful in this respect. The apparatus and procedure used in this investigation were similar t o those employed by Michels ( 1 1 ) who obtained very high precision in his measurements. DESCRIPTION OF APPARATUS
The apparatus consists of four main parts: a dead-weight gage, a glass piezometer, a steel pressure bomb, and a constant temperature bath. Auxiliary apparatus includes oil injectors, Bourdon pressure gages, mercury manometers, steel sample bombs, and high pressure valves and fitting. The dead-weight gage is similar to t h a t described by Keyes (8) and is fitted with five different piston and cylinder combinations to cover a pressure range up to 50,000 pounds per square inch. The piston diameters were measured accurately a t 20” C. by comparing them with standard Johansson blocks in a P r a t t and Whitney Electrolimit gage. The gage constants for the various pistons were calculated by using the empirical formula developed by Beattie and Bridgeman ( 3 ) which indicates t h a t t h e effectivediameter of the piston is equal t o the actual diameter plus 0.00025 cm. This relation gives gage constants accurate within 0.04%. 1 Present address, Yerkes Research Laboratory, E. I. du Pont de Nemours & Co., Ino., Buffalo, N. Y. 2 Present address, Purdue University, Lafayette, Ind.
A glass piezometer similar to those employed by Michels (10) was used to measure the gas volumes. It was made from soft glass and consists ot a series of glass bulbs of ---O 5 MM. v a r y i n g size sepaC A P ILLAR rated by l-mm. capilRESISTANCE METER lary tubing. A 3.8CC. 10-mil platinum wire was fused into each piece of c a p i l l a r y tubing. The total 0.5 C.C. volume of the pir1.1 cc. zometer is a b o u t 68 cc., and the smallest volume measured is about 4 cc. Each successive platinum contact was joined by a small coil of No. 30 Chrome14 wire having a resistance / \\ of a p p r o x i m a t el y 5 ohms. The piezometer is illustrated in Figure 1. The 52.6 CC v o l u m e from each contact to the top of the piezometer was found by weighing the mercury required to occupy this volume a t constant t e m p e r a t u r e . The calibration was Figure 1. Glass Piezometer and carried out a t 50°, Platinum Contact Circuit
11