Specific Refractive Dispersion as a Method for Distinguishing between Different Series of Hydrocarbons A. L. WARDAND W. H. FULWEILER, The United Gas Improvement Company, Philadelphia, Pa.
T
HE analysis of hydrocarbon mixtures is a difficult
problem for which there are no satisfactory methods that are both simple and generally applicable. The experience of this laboratory has been that in general a combination of physical and chemical methods is more satisfactory than either type alone. Among the physical methods that have proved of value is one employing specific refractive dispersion. Dispersion is the difference in refractive indices determined for light of two different wave lengths; specific dispersion is dispersion divided by density, both constants being determined a t the same temperature. In order to express specific dispersion as a convenient whole number, it is multiplied by some power of 10. As employed in this paper, specific refractive dispersion equals nXt
- nX2 X
lo4or An X 104 d
Most of the classical work on the various aspects of refractivity has been concerned with molecular refractivity. This property is of value in establishing the purity or the structure of a compound, but it is not very suitable for the analysis of hydrocarbon mixtures. Waterman (19, 2.9) and his co-workers have used the specific refraction formula of Nz-1 1 Lorentz-Lorentz, X a for hydrocarbon analysis. Several expressions for dispersion have also been used in work on the structure of compounds, but for some reason relatively few serious attempts have been made to employ specific dispersion in hydrocarbon analysis. The most interesting recorded attemptswere made by ~~~~~i~ (6), Di*er (7)) Muthe ( I d ) , Cortot (8), and Waterman and Perquin (21).
Cortot's work was confined to a limited number of cyclic hydrocarbons. Mutte and Dixmier used specific dispersion for the Q and D lines. Waterman and Perquin employed dispersion directly and also the rather unusual expression nG - nC 104. The latter expression does not appear to nD - 1 have as great value as one that includes density. Darmois published constants for the specific dispersion of six series of hYhocarbons. They are apparently intended to be used Only for light gasolines since his value for the aromatics apPears to have been based on benzene, toluene, and xylene only. The higher aromatics have considerably lower dispersions.
EXPERIMENTAL The densities and refractive indices of a number of hydrocarbons and hydrocarbon mixtures were determined, the densities by comparing the specific gravity of the sample at 20" c. with water a t 4" c. using a WestPhal balance, and the refractive indices with a Pulfrich refractometer which had been calibrated as described below. Values were determined at 20" c. for the following five spectrum lines: WAVIU LIUNQTH
LINXI
Nal
A.
6563 5893 5461 4861 4368
C (Ha)
D
a ( H g green)
F
(Hi31
u (Hg blue)
The D h e was supplied by a sodium lamp. The C and F lines were supplied by a Geissler discharge tube containing hydrogen. The e and g lines were supplied by a Geissler tube having aluminum electrodes and having a small amount of mercury in each bulb. Tn each case, instead of using a closed evacuated Geissler tube, an open tube TABLEI. SUMMARY OF REFRACTIVE INDICES AND DISPERSIONS OF DIFFERENT HYDROCARBONS was e m p l o y e d . SAXPLX -REFRACTIVB I N D I C E 5 A T 20-' DIEPER8ION6 x 10' The tube was conNo. SUBETANCE C D e F u F - C V - C 66 105 nected with a 1 n-Hexane 1.3743 1.3762 1.3772 1.3809 1.3848 1.4069 1.4092 1.4104 9 Decane (2 7-dimethyloctane) 77 116 vacuum pump and 1.4146 1.4185 1.3860 1.3883 1.3903 1.3945 1.3983 85 123 also with an auto11 Trimethylkthylene 1.4115 1.4138 1.4159 12 Ootylene 1.4204 1.4252 89 137 m a t i c electrolytic 14 Diamylene 1.4336 1.4363 1.4378 1.4426 1.4475 90 139 1.4227 1.4255 1.4268 88 136 hydrogen genera15 Decenea 1.4315 1.4363 1.4503 1.4532 1.4548 1.4598 1.4652 95 149 t o r . When t h e 17 Triamylene 1.4428 1 4441 1.4393 1.4487 18 Hexadecene 1,4538 94 145 1.4519 103 163 tube Was in use the 1.4404 1.4356 1.4387 1.4459 20 Decadiene 170 270 vacuum pump was 21 Benzene" 1.4967 1.5014 1 5049 1.5137 1.5237 1.4922 1.4966 1.5000 1.5081 1.5180 22 Toluene= 159 258 run constantly and 23 24 25 26
Ethylbenaenea 1.4912 m-Xylene= 1.4931 1 4920 M,esitylene (1 3 5-trimeth~lbeneene)~ 1.4934 Dieth lbeniedeh 27 1,2,4-~imethylethylbenaene" 1.5003 28 p-Cymene 1.4867 1.4952 29 Triethylbensenea 30 or-Phenyl-a-buteneb 1.5200 1.4241 31 Hexahydrobeneene 1.4220 32 Hexahydrotoluene 33 Hexahydro-p-cymene 1.4362 34 Decahydronaphthalene 1.4737 35 Dehydrobeneene 1.4744 36 Dihydro-p-cymene [~-menthadiene-(1,3)1 1.4779 37 Tetrahydrobenzene 1.4548 38 1,2,3,B-Tetrahydrotluene 1.4545 39 Tetrahydro-p-cymene (p-menthene-3) 1.4555 41 Pinene 1.4663 42 1.5685 43 11,2-Dihydronaphthalene4 2 3 4-Tetrahydronaphthalenea 1.5385 45 Abe'ntene 1.4747 47 Indene 1.5655 a Values are average8 of determinations made with two priems. 6 Readings affected by aation of light on samples
1.4955 1.4975 1.4962 1.4974 1.5046 1.4906 1.4990 1.5249 1.4263 1.4243 1.4387 1.4765 1.4783 1.4815 1.4576 1.4574 1.4584 1.4694 1.5748 1.5433 1.4784 1.5719
1.4987 1.5009 1.4995 1.5004 1.5077 1.4936 1.5024 1.5292 1.4276 1.4256 1.4400 1.4779 1.4813 1.4841 1.4593 1.4593 1.4599 1.4711 1.5798 1.5468 1.4805 1.5769
396
1.5065 1.5089 1.5074 1.5080 1.5153 1.5011 1.5090 1.5381 1.4319 1.4299 1.4442 1.4825 1.4884 1.4906 1.4644 1.4645 1.4660 1.4764 1.5913 1.5552 1.4866 1.5887
1.5155 1.5184 1.5163 1.5169 1,5246 1.5097 1.5179 1.5517 1.4362 1.4343 1.4489 1.4876 1.4975 1.4983 1.4706 1.4704 1.4707 1.4823 1.6058 1.5655 1.4936 1.6028
153 158 154 146 150 144 138 181 78 79 80 88 140 127 96 99 95 101 228 167 119 232
243 235 243 235 243 230 227 317 121 123 127 139 231 203 158 159 152 160 373 270 189 373
j u s t sufficient hy-
drogen was admitted t o give the brightest illumination. It was impossible to the monly used G' (4340 A.) line, becaus'e withthe equipment available the 1in e was not sufficientty b r i l l i a n t for accurate m e a s u r e ~
~
substituted. T h e addition of mercury
~
November 15,1934
IN D US T R IA L AND EN G I N EER IN G C H E M ISTRY
397
TABLE11. SUMMARY OF PHYSICAL CONSTAICTS SAMPLE NO.
B. P. SUBBTANCE
nF
Den-
,","z'o'
76;:~.
c.
'","
x
- nC
nu - nC
d 10'
d
:x-.104
B. P.
SAXPLB NO.
AT 760 MM.
SWBBTANCE
=
68.9
0.6602
1.3762
100
159
159.9
0.7240
1.4069
106
160
I M P U R E PARAFFINE F R O M P I T R O L E U M
3 Nonane 5 +Decane 5-A n-Decane after 7 SOz treatments 8 Decane (2,7-dimethyloctane) 10 Dodecane and tridecane mixture
149.9 0.7323 172.1 0.7500
1.4136 1.4200
98 101
159 156
171.0 0.7415
1.4172
100
156
161.5 0.7477
1.4176
100
156
221.3
1.4348
101
162
1.4138 1.4363 1.4255 1.4532 1.4503
122 116 119 117 120
187 181 183 184 184
1.4387
135
214
0.8807 0.8677 0.8682 0.8661 0.8628 0.8640
1.5014 1.4966 1.4955 1.4975 1.4962 1.4974
193 183 178 182 178 169
307 297 280 292 282 272
187.6 0.8792 176.6 0.8582 217.4 0.8655
1.6046 1.4906 1.4933
171 168 160
277 268 262
0.7843
30
0.7217 0.7682 0.7418 0.8087 0.7848
Octylene Diamylene Decene Triamylene Hexadecene
123.2 156.4 160.4 248.0 285.9
20
Deeadiene
162.8 0.7629
DIOLEFIN
31 32 33 34
Benzene Toluene Ethylbenzene m-Xvlene Meai-tylene Diethylbenzene 1,2,4-Dimethyletklylbenzene 28 p-Cymene 29 Triethylbeneene
80.26 110.8 135.8 139.8 164.0 181.8
to a Geissler tube offers a simple and easy method of obtaining the two mercury lines. The e line offers no particular advantage because its wave length is between the wave lengths of the easily obt,ainable D and F lines. The g line is of distinct value since its wave length is almost as low as the G line. The only disadvantage in its use is the necessity of calculations t o make values obtained with it comparable t o those obtained with the G line. However, the closeness of the two wave lengths permits such calculations to be made without introducing significant errors. I
CALIBRATION OF REFRACTOMETER. Suitable precautions to insure accurate readings included the use of apparatus to supply the water jacket of the refractometer with water a t a constant temperature. The instrument is supplied with two prisms to permit determinations to be made over a wide range aad, wherever possible, both prisms were used and the results averaged. The zero point of the instrument was determined by the method described by Firth ( I S ) . Corrections for the zero point were made for each prism. In addition, to calibrate the instrument, determinations of the refractive indices of six substances were compared with determinations made by the classical angle of minimum deviation method by J. W. Barnes. The substances used, which were selected to give a wide range of refractive indices, included water, n-propyl alcohol, chloroform, methyl salicylate, oil of cassia, and monobromonaphthalene. Barnes' measurements were taken as standard and by comparing the results the following sets of corrections were obtained for the two prisms of the Pulfrich: SPECTRUM LINF,
PRIBM 1
C D
-0.00040 -0.00055 -0.00050 -0.00048 -0.00050
$ Q
nF - nC ng - nC ~d d
"D
x
10'
x
lo4
PRIBM 2 -0.00037 -0.00038 -0.00030 - 0.00022 -0.00020
EXPERIMENTAL RESULTS. The refractive indices and dispersions, both a t 20°, for a number of hydrocarbons are summarized in Table I. The more important specific dispersions together with the boiling points, densities, and refractive indices for the D line are summarized in Table 11. The boiling points represent the temperature reading when 50 per cent of a 100-ml. sample had distilled over (A.S.T.M. D-216). The boiling points were determined when the atmospheric pressure was within 5 mm. of 760 mm. and were corrected to 760 mm.
a-Phenyl-a-butene
197.1 0.8927
1.5249
203
356
Hexahydrobenzene Hexahydrotoluene Hexahydro-p-cymene Decahydro naphthalene
100 102 101
156 160 160
99
156
81.0 0.7781 1.4263 100.8 0.7707 1.4243 168.4 0.7950 1.4387 192.4
0.8895
1.4765
UNSATURATED CYCLICS. TWO D O U B L E B O N D S
35 36
Dih ydrobenzene Dihydro-p-cymene [p-menthadiene-
(13)1
0.8417
1.4783
166
274
177.2 0.8502
1.4815
149
239
1.4576
I14
187
1.4574
124
198
1.4584 1.4694
117 116
159 187
85.0
UNSATURATED CYCLICS, O N E DOUBLB B O N D
37 38 39 41
AROMATICS
21 22 23 24 25 26 27
20
BATORATED N A P H T H E N E Q
OLEFINS
12 14 15 17 18
61TY A T 20'
AROMATIC OLEFIN
PARAFFINS
1 n-Hexane 9 Decane (2 7-dimethyldctane)
c.
DEN-
Tetrahydrobeneene 82.4 0.8417 1,2,3,6-Tetrahydrotoluene 103.4 0.8010 Tetrahydro-p-cymene (p-menthene-3) 168.0 0.8098 Pinene 156.5 0.8578
HYDROQENATED N A P H T H A L E N E S
42 43 34
1,2-Dihydronaphthalene 211.0 0.9943 1,2,3,4-Tetrahydrona hthalene 207.0 0.9708 Decatydronaphtha192.4 0.8895 lene
1.5748
230
375
1.5433
172
278
14765
99
156
1.4784 1.5719
140 236
222 380
YIBCELLANIOVB
45 47
Dipentene Indene
176.2
...
0.8512 0.9816
All the hydrocarbons for which constants are given were prepared or purified in this laboratory, and most were highly purified products. In some cases they were mixtures of isomers, the separation of which was impossible by ordinary methods. They were considered to be sufficiently pure for work of this character. In Table I1 the values for a group of impure paraffins separated from petroleum are included. These are believed to be of interest because they prove that the difficulty in purifying paraffins from petroleum is due to presence of naphthenes and demonstrate the impossibility of separating naphthenes and paraffins by the chemical methods normally employed in analyzing hydrocarbon mixtures. The literature is full of physical constants of hydrocarbons isolated from petroleum and considered pure by the men who isolated them. The work of Washburn and his associates at the Bureau of Standards has demonstrated that although it is possible to isolate pure hydrocarbons by special physical methods, the vast majority of hydrocarbons previously isolated have been grossly impure. Shepherd, Henne, and Midgley (17)starting with a petroleum mixture, 70 per cent of which consisted of normal paraffins, isolated some pure individual paraffins with the aid of chlorosulfonic acid. Whether their success was due to the purity of the starting material or to the long treatment with the particular reagent used, the fact remains t h a t practically all attempts to separate pure paraffins by chemical: treatment have failed. Samples 3, 5 , and 8 were from fractionated Pennsylvania petroleum. They had been washed with fuming sulfuric acid; treated three times with nitrating acid; washed with caustic soda, water, and methyl alcohol; washed twice with ten volumes of liquid sulfur dioxide and again with caustic soda and water; and finally distilled over sodium. Sample 5-A was given an additional five washings with ten volumes of liquid sulfur dioxide and was frozen, filtered with suction, and separated from the hydrocarbons of lower melting point. After the fdth washing with liquid sulfur dioxide, further washings continued t o dissolve a portion of the sample without affecting its density or refractive index, and the principal effect of the freezing treatment was to remove 8ome 2,7dimethyloctane from the normal decane which constituted the bulk of the sample.
.
ANALYTICAL EDITION
398
After these treatments, the densities and refractive, indices of the samples were much lower than the values obtained after the original fractionations, but were still larger than the values for synthetic paraffins by 1 or 2 x 10-2. The data show that the hydrocarbons resisting the strenuous chemical treatment were naphthenes, since an increase in density caused by other hydrocarbons would have increased the specific dispersion.
Vol. 6 , No. 6
DISCUSSION OF RESULTS
It is known that for certain classes of organic compounds differences in structure affect dispersion. For example, Eykmann (12)has shown that differences in the distances of side-chain double bonds from the nucleus causes differences in the dispersion of safrole and eugenole compared to isosafrole and isoeugenole, respectively. The object in summarizing a large number of literature values, in addition to the determined values, was to establish generalizations as LITERATURE VALUES free as possible from the influences of differences in structure. The experimental work summarized in Tables I and I1 It was hoped that the hydrocarbons likely to be encountered was completed a number of years ago. At that time, in in practical work would in general correspond to the average order to supplement the determined values with data for a values of their class. The discussion below is limited to longer number of hydrocarbons than it was expedient to specific dispersion for the g-C lines which is more useful than prepare, a search of the literature ivas made for density and the smaller value for the F - C lines. refractive index determinations of synthetic hydrocarbons, SATURATED HYDROCARBONS. The values for paraffins and from which specific dispersions could be calculated. A naphthenes are too close together to permit distinguishing complete list of the references would be too long for inclusion between them. This is evident both from Table I11 and here, but a number of the more valuable ones are given from the data for the paraffins from petroleum summarized (1-6, 9, 11, 16, 16). Numerous values were also obtained in Table 11. from Landolt and Bornstein, International Critical Tables, I n the case of paraffins there was no variation that could be and particularly the monumental work of Eykmann which ascribed to molecular weight or structure, the deviations was collected and published in 1919 by the Hollandsche being irregular and properly ascribed to experimental error. Maatschappi. Two more recent sets (IO, 18) of values for This is also true of cyclics with a six-membered ring. A paraffins have been added to the literature values, but do not total of 23 literature and three determined values were change the older averages. within .t4 of 157. This value is also reasonably accurate for I n general, the usefulness of literature data is limited by the a wide variety of other cyclics ranging from dimethyl cyclofact that only a comparatively few investigators have re- propane to dicyclotridecane. ported densities and indices a t the same temperature. Where UNSATURATED HYDROCARBONS. For olefins with one there was a difference of only a few degrees, temperature double bond, the average of 22 literature values was 186. corrections were made; otherwise the data were not used. The average of 5 determined values was 184. There was no Values that were obviously erroneous were also omitted. consistent variation with molecular weight. Although oleI n calculating specific dispersion, refractive indices for the fins can be readily distinguished from saturated hydrocarbons g line were obtained from curves by plotting wave lengths by this method, they cannot be distinguished from unsatuagainst indices where values for the G' line were available. rated cyclics. For the latter with one double bond, the averThe difference in the index for the Q and G' lines is so small age of 16 literature values for five- and six-membered rings that the errors as a result of using the curves are negligible. was 183. The average of four determined values was someThe values determined experimentally are compared with a what higher, 190. summary of the literature values in Table 111. RepresentaToo few values are available to permit safe generalizations tive values, given for the more important series of hydrocar- for either diolefins or unsaturated cyclics with two double bons, are the approximate averages of the two preceding bonds. Three literature values were available in each case, columns and are believed to be suitable for use on unknown the averages being 222 for the diolefins and 246 for the cyclics. mixtures. For reference, Darmois' values are included. For the former, one determination gave 214; for the cyclics, two determinations averaged 256. TABLE 111, SUMMARY OF SPECIFIC DISPERSIONS OF DIFFERENT Conjugated double bonds produce a marked exaltation, SERIESOF HYDROCARBONS particularly in the case of aliphatic hydrocarbons. Their na - n C presence must be considered in unknown mixtures showing rl x 104 high specific dispersions. CalouG L - C nF - nC' x 1 0 4 lated Saturated cyclics with one double bond in the side chain d from Repre- d have a higher value than that of cyclics with a double bond Litera- Deter- litera- Deter- senta- X 10' T Y POF ~ HYDROCARBON ture mined ture mined tive Darmois in the ring, the average of the values in the literature being Saturated aliphatics (paraf201. On the other hand, unsaturated cyclics with a double 155 find 103 154 I59 167 97 Saturated oyolice (naphbond in both the ring and the side chain, common among the 161 thenea) 155 168 167 101 98 Un&tu&ed diphatics with terpenes, have a lower value than would be expected. Litera194 185 119 188 one double bond. (olefins) 120 184 ture values for ten of these hydrocarbons averaged 214. Unsaturated cyclics with 185 191 183 190 116 119 one double bond Aromatic olefins give high results, the average of 14 literaUnsaturated aliphatics with two double bonds (dioleture for five hydrocarbons being 355, which happens to 222 214 ... 260 cheokvalues 6ns)o 142 131 the determined value for phenylbutene. This is a Unsaturated cyclics with two double bonds 157 159 246 258 260 ... coincidence because the value decreases with increasing moUnsaturated aliphatic8 with two conjugated double weight in a manner similar to that of the aromatics 222 ... 376 . . . . . . . . . lecular bondsa which will be discussed below. The diolefins of this class Unsaturated cyolios with two conjugated double have very high specific dispersion. The conjugated diolefins 184 ... 298 . . . . . . . . . have bonds0 Saturated cyolics with one values over 700. ... 201 . . . . . . . . . double bond in side chain 125 AROMATICS. The determined value for benzene was 307. Unaaturated cyclics +th oue double bond in side The values for the alkyl benzenes are all lower. Since the chain 135 ... 214 . . . . . . Aromatics . . . . . . (See Fig;;; 1) 306 substitution of alkyl groups increases the boiling point, it is Aromatic olefins with one double bond 204 219 355 358 . . . . . . possible to plot the dispersions of the aromatic hydrocarbons against either their molecular weights or their boiling points. a Too few values are available. v
I N D UST R I AL AN D E N G I N E E R I N G CH E M IS T R Y
November 15,1934
The boiling point curve is the more useful. Such a curve is shown in Figure 1, and is a composite of about 80 determinations including those determined here. The purpose of the present paper is to compare the different series of hydrocarbons rather than the effect of individual structure, but in general, it may be stated that the effect of substitution on dispersion is related to its effect in decreasing the benzenoid and increasing the paraffinoid character of the
r
I
I
I
I
I
I
1
399
Some time ago this laboratory was engaged in a detailed examination of the condensable hydrocarbons occurring in different types of manufactured gas (& The I)fractions . boiling between 80" and 200" C. are of such a nature that a complete resolution into their constituents is difficult, partly because the ease with which the unsaturated hydrocarbons are polymerized by heat makes it unsafe t o submit them to a systematic refractionation. The only unsaturated hydrocarbons that could be identified as occurring in significant quantities of these fractions are styrene and indene. From a chemical examination, it seemed probable that the rest of the material consisted principally of aromatic hydrocarbons with but a small percentage of naphthenes, paraffins, or other hydrocarbon types present. A reviously described (&)) bromine method was tentatively aiopted for the determination of the quantities of styrene and indene present in the different fractions. It was desirable to confirm the opinion regarding the aromatic character of the material. Also, in view of the nonselective character of the bromine method and of the fact that styrene and indene are found distributed through a number of fractions, it was essential to demonstrate the truth or fallacy of the previous conclusions: (1) that no unsaturated hydrocarbons other than indene or styrene were present in significant quantities; and (2) that the bromine method used actually gave a true picture of the quantities of these hydrocarbons present. Accordingly, determinations were made of the densities, refractive indices, specific dispersions, and bromine titrations of the different fractions of a number of samples of condensate. The results of this examination of the middle fractions of one sample are given in Table IV.
TABLEIV SPB~CIFIC DIUPERSION
SPECIFIC Dl5PERSlON
FIGURE 1. SPECIFIC DISPERSION OF AROMATIC HYDROCARBONS hydroca,rbon. I n other words, the number of carbon atoms, the length of the side chain, and the character of the chain must aLP be considered. An ethyl group has more effect than two methyl groups, and an isopropyl group has more effect than an n-propyl group. The effect tends to decrease slightly with an increase in the number of hydrogen atoms that have been replaced. For aromatics with molecular weights up to 200, the effect of replacing a hydrogen atom with a methyl group varies from 6 to 13.5, with an average of about 8. Since: dispersion of the aromatic hydrocarbons is a constitutive property, it follows that its use in the case of unknown mixtures containing aromatics can never be strictly accurate. However, in many cases the dispersion of the aromatics is so greatly different from that of the other series with which they are mixed that the use of an average curve such as Figure 1 introduces a relatively small error. I n other cases, it is possible to make a partial chemical separation that will permit a direct or indirect determination of the dispersion of the particular group of aromatics with which one is dealing. NAPHTHALENE AND HYDROGENATED NAPHTHALENES. The best available averages for these hydrocarbons are: Naphthalene AI-Dihydronaphthalene Az-Dihydronaphthalene Tetrahydronaphthalene Decahydronaphthalene
490 402 285
277
156
USE OF SPECIFIC DISPERSION. The possibilities for the use of specific dispersion are varied. Its employment must be governed to a certain extent by the type of hydrocarbon mixture with which one is dealiig. It is not the purpose of this paper to discuss the applications of the method in detail, but rather to provide the constants necessary for its use and to indicate which classes of hydrocarbons can, and which cannot, be distinguished by it. The use of the method is frequently indicated when one wishes to confirm or deny conclusions arrived at by other methods. One example of such use may be of interest.
a. bromine/ml.
c. 8
9
10 11 12 13 14 15 16
105-108 108-1 11 111-114 114-1 19 119-129 129-139 139-143 143-147 147-167
0.8680 0.8657 0.8663 0.8655 0.8678
1.4958 1.4959 1.4957 1.4956 1.4983 1.5048 1.5113 1,5147 1.5191
0.8710 0.8780 0.8834 0.8884
293 293 292 293 304 310 321 343 336
Trace Trace 0.0063 0.0125 0.1250 0.2312 0.3675
0.6375 0.0378
Fractions 8 to 11,inclusive, were but slightly unsaturated. The specific dispersions of these fractions were close to those of aromatic hydrocarbons as taken from the boiling point-dispersion curve. The differences indicated by paraffi-naphthene content of from 3 to 6 per cent. The percentage of styrene in each of the nine fractions was calculated on the assumption that all the unsaturation was due to styrene. The percentage of styrene in each fraction was then calculated from the specific dispersion. (When working with a mixture consisting largely of aromatics and unsaturants with but small quantities of saturated hydrocarbons, approximate values can be obtained by neglecting the small quantities of saturated hydrocarbons. For more accurate values it is necessary to base the calculations on a naphthene- and paraffin-free basis.) The two sets of values are compared in Table V.
TABLEV FRACTION 8 9 10 11 12 13 14 15 16
BY bromine
STYRBNIC 7 By dispersion specific
%
%
Trace Trace 0.5 0.9 9.4 17.3 27.2 47.0 46.7
0 0
0.8 0.8 11.1 17.4 26.4 43.0 40.1
The two methods are in excellent agreement on all of the fractions with the exception of 15 and 16. Since fraction 16 was collected above the boiling point of styrene, it was reasonable to expect some other unsaturated hydrocarbon t o be present. Indene had been identified. An attempt was made to separate the styrene and indene as far as possible. The results indicated that the quantity of indene present wap about two-thirds of that of the styrene. If the results of the two methods are recalrulated on this basis, the apparent discrepancy in this case demonstrates the value of the method more closely than do the agreements in the other fractions.
ANALYTICAL EDITION
400
Recalculated, the unsaturation of fraction 16 is BY SPECIFIC BY B R O M ~ N ~ PDIEPEREION Styrene Indene
%
%
29.2 19.5
27.2
18.1
Fraction 15 also contained some indene. When the data for this fraction were recalculated to allow for the indene, the results of the two methods were in close agreement. The results of the two methods on all nine fractions thus fall within recognized experimental errors. It is difficult to conceive of any other h drocarbon the presence of which is reasonably possible and wKich has both a bromine value and a specific dispersion equal to that of styrene. The results obviously demonstrate that the basic material is toluene and higher aromatics* that the bromine method gave a picture of the degree of unsaturation of this material that was accurate within the limits reasonably to be expected in the analysis of hydrocarbon mixtures; and that, aside from some indene, no umaturated hydrocarbon other than styrene was present in a significant quantity.
LITERATURE CITED (1) Auwers, Ann., 415, 144 (1918); 419, 92 (1919). (2) Auwers and Eisenlohr, J . grakt. Chem., 82, 65 (1910); 83, 429 (1913); 84, 1, 37 (1912).
Vol. 6 , No. 6
(3) Auwers and Ellinger, dnn.. 387, 200 (1912). (4) Auwers and Moosbrugge, Zbid., 387, 167 (1912). (5) Bruhl, J. Chem. Soc., 91, 115 (1907); J . prakt. Chem., 49, 260 (1894). (6) Darmois, Compt. rend., 171, 925 (1920); 172, 1102 (1921). (7) Dixmier, Chimie d? industrie, Special No. 338 (1926). (8) Cortot, Rev. gen. sei., 34, 607 (1923). (9) Cronubert, Ibid., 33, 433 (1922). (10) Dornte and Smith, J . Am. Chem. Soc., 52, 3546 (1930). (11) Eisenlohr, 2. phys. Chem., 75, 585 (1910). (12) Eykmann, Ber., 22, 2736 (1889); 23, 855 (1890). (13) Firth, “Practical Physical Chemistry,” p. 51, D. Van Nostrand Co., N. Y., 1916. (14) Mutte, Chimie & industrie, Special No. 338 (1926). (15) Ostling, J. Chem. Soc., 101, 457 (1912). (16) Redgrove, Chem. News, 102, 3, 13 (1910). (17) Shepherd, Henne, and Midgley, J . Am. Chem. Soc., 53, 1948 (1931). (18) Smith and Stoops, Ibid., 50, 1883 (1928). (19) Vlugter, Waterman, and Westen, Chem. Weekblad, 29, 226 (1932). (20) Ward, Jordan, and Fulweiler, IND. ENO.CHEM-, 24, 969, 1238 (1932); 25, 1234 (1933). (21) Waterman and Perquin, J . Inst. PetroZeum Tech... 13,. 413 (1927). (22) Waterman and Westen, Ibid., 18, 735 (1392). REC~IVE April D 6, 1934.
Nature and Constitution of Shellac IX. Determination of Solubility in Organic Liquids WM. HOWLETT GARDNERAND HARRYJ. HARRIS Shellac Research Bureau, The Polytechnic Institute of Brooklyn, Brooklyn, N. Y.
I
N VIEW of the amount of Methods for accurate determination of s o h can be easily demonstrated that work which has been done the constituents have bilities of resins in organic liquids involve a solon determining the soluvent separation of the resin into its constituents. tion a tendency materialto which carry would intoothersolubility of resins in various solThis paper illustrates how misleading results may wise remaininsoluble and which vents, it is unfortunate that more be obtained with diflerent procedures unless preis retained by the solvent as 8 study has not been given to the methods employed (1, 3). The caution is taken to eliminate two important n o n s e t t l i n g suspension o r lack of careful c o n s i d e r a t i o n Such behavior sources of error: the tendency of insoluble mateof the true nature of these subrial to be carried into solution as a colloidal sol precipitated explain why when resin is solutions sometimes are stances and the factors which and of some qf the soluble resin to be occluded by dilutedwiththesamesolvent(2), may affect the determination the swollen undissolved portion. Consistency in as, for example, a concentrated may a t least in part account for some of the wide variations in results is no criterion of the applicability of a solution of shellac in acetone. values reported. A n o t h e r source of error in method. such determinations results from Resins belong to a very inthe occlusion of soluble constituteresting group of substances which give solutions possessing the properties of both true ents by the insoluble material. Where a complete separation molecular dispersions and colloidal suspensions. Unless this is essential, consistent results can be very misleading. In ceris clearly kept in mind, confusion is apt to arise with regard to tain cases the occlusion may just compensate for the loss of what is meant by the solubility of a resin, and what is meas- insoluble material through colloidal dispersion. ured in quantitatively determining the solubility. Since resins EFFECT OF METHOD do not give saturated solutions even when their average molecUnless precaution is taken to eliminate these sources of ular weights are comparatively small ( 2 , 5 ) ,we are not dealing with solubility in the classical sense. Even with synthetic error in the solubility determination] the values obtained by resins, a determination of solubility involves several molecular different methods will not be the same. Five sets of deterresin species and not single identical molecules such as exist minations with boiling acetone (Table I) show how the above in pure substances. The analyst actually measures the factors can affect the results. amount of those constituents which are soluble; hence, the TABLEI. EFFECTOF METHODON DETERMINATION OF so-called quantitative solubility determination involves a SOLUBILITY separation of resins in exactly the same manner in which the METHOD INSOLUBLE RPISIN % analyst would separate a mixture of two or more totally dif6.48-12.5 f erent species. 0.93-3.00 5.01-5.22 Obtaining consistent results by any method is not in itself 4.79-5.23 a criterion for judging the applicability of that method. It 5.00-5.21
