13C6
INDUSTRIAL AND ENGINEERING CHEMISTRY REACTIONS OF DICHLORODIE'LI OROWITII IODlNE HEPTAFLUORIDL
TABLEI. DATA FOR METHANE
(MeiLuric fluoride catalyst) CCiIFz CCiI, Iodine RecovRecovHeptaDura- Reaction ered. ered, fluoride %yield Based on Run tion, Temp., T'ol. Lit[., Vol. Liq., Used, CChF, No. Hr. C. hfl Mi, G. IFr+Ie useda 3 0 1 6 340 2.0 9.97 8.6 40 1.8 25.95h 2 6 350 1 .? ... 42 6.0 1 2 3 5 360 7.35 7.1 14 5.3 5 5 350 1 0 5.95 4.5 4 15 2 5 5 6 360 1.8 1.95 67 5g 2.5 1.62 6 0.9 67 370 75 6 2.0 0.6 24 7c 6 370 0 9 6 370 1.0 46 8d a T h e apparent pairing of the yields i n this ooliiiiin is t o be largely explained by a similarity i n flow rate for t h e CCLFZ. b A marked increase in the rate of flow of IF, occiirred in t h e last few minutes of this run. C The same sample of HgFz wits ii-ed for this run as was used i n reactions 5 a n d 6. d A fresh sample of HgF2 was used for thiq run.
trifluoromethane divided by the sum of the volumes of dichlorodifluoromethane and chlorotrifluoromcthane obtained. The validity of such a method of estimation is dependent upon the assumption that the two liquids have nearly equal densities
Vol. 42, No. 7
are riot very conclusive. Likewise, in its reaction with dichlorodifluoromethane, iodine heptafluoride appears t'o be even mow potent than fluorine itself; but it is possible that this effect may result from the cat,alyt,ic action of iodine upon the substitutioii react ion, 4. The fact that carbon tetrafiuoride appeared as a produvt in the fluorination of carbon tetrachloride by means of fluorine (6),but was absent in the reaction of fluorine with dichlorodifluoromethane (9), is but one indication of the increasing strength of the C C I bond in the more completely fluorinated members of the series of compounds derived from carbon tet,r:tchloride by successive fluorination, reaching a maximum i n chlorotrifluoromethane, as referred to in the second point abovo. These facts may be interpreted to imply that the mechanism of t.he suhstitution of fluorine for chlorine in this case may not be a simple, st,epwise process, hut ma proceed by way of formation of free radicals, which produce czains of carbon atoms, subsequerit,ly hroken in the fluorination reaction. ACKNOWLEDGMENT
This investigation was materially assisted by it grant-in-aid received from the Harshaw Chemical Company of Cleveland, Ohio. The authors wish to express their appreciation for this assistance. %he authors also wish to acknowledge the aid oi' Kinetic Chemicals, Inc., of Wilmington, Del., from whom the spectroscopically pure samples of Freon-12, CC12Fz, and Freon-13, CCIF,, were obtained.
CONCLUSIONS
From the results of earlier investigators previously referrell to and from the results of the present study, the following conclusions may be drawn: 1. The substitution of fluorine for chlorine i n carbon tetrachloride is accomplished more effectively by fluorine than by iodine pentafluoride. 2. It is recognized that the single chlorine atom in chlorotrifluoromethane is less susceptible to replacement than thoqc of dichlorodifluoromethane. The replacement of chlorine in dichlorodifluoromethane appears t o require a catalyst, such as mercuric fluoride (9). 3. From the facts that iodine pentafluoride has been shov,n able to convert a few per cent of trichlorofluoromethane to dichlorodifluoromethane, while iodine heptafluoride ronverts a few per cent of dichlorodifluoromethane to chlorotrifluoromethane, it may be argued that the heptafluoride is more potent as a fluorinating agent than the pcntafluoride; but the results
LI'I'EHATUHE CITED
H m k s . FmelBus, Haszeldine, and Kerrigan, J . Chem. Soc., 1948, 2188. Booth and Pinkston, Chern. Rets., 41, 421 (1947). Lord, Lynch, Sohumb, and Slowinski, J . Am. C h e m Suc., 72. 522 (1950). JIoiasan, Comp. rend., 135, 563 (1903). Ruff and Keini, 2. anorry. ZL. allyem. Chem., 193, 176 (1930). Ibid.. 201, 245 (1931). Schumb, Young, and Radimer, 1x1).EXG.CHEW,39, 244 (1947). Yharpe and EmelBus, J . Chern. Soc., 1948, 2135. Simons. Bond. and h1cArthur. J . Am. Chmn. Suc., 62, 3477 (1940).
Smith, A n n . R e p f s . on Pioyresu Chem. (Chem. SOC. I J o n d n r l ) , 44, 86 (1947).
'Thompson and Temple, J . CherrL. Soc., 1948, 1422. K E C E I V ENovember ~ 2 5 , 1949.
FRACTIONATOR CALIBRATION SYSTEM Binary vs. Afulticomponent Test Mixtures C. B. KINCANNON AND EARL 3IANNINC.
JH.
Shell Oil Company, H o u s t o n , T P X . T h e effect of the number of components present in the calibration system used in the calculation of plate efficiency of a distillation column was investigated on a 30-plate 1inch inside diameter glass Oldershaw column. Analyses were performed by use of a mass spectrometer. The plate efficiencies were calculated from a multicomponent fraction and compared with those for a binary system. Results from both mixtures show an average efficiency-value of 50% (fifteen theoretical plates) over a load range of 800 to 3000 ml. per hour overhead rate. It appears that the efficiencies of laboratory columns, evaluated with systenls of like molecular weight and type, are independent of the number of components present in the test mixture.
T IS ofteii desirable to l c n o ~the efficiency of a fractionating column when it is io productive service and operating on a multicomponent system. However, the efficiency is usually determined by experimental data and calculations based on a binary test mist,ure. The differences in the number wf degrees of freedom and the rate of approach to thermal and phase equilibrium could make efficiency values based on multicomponent systems appreciably different from those based on binary systems. Therefore, an investigation of the efficiency of a laborat,ory column was iriitiat,ed in \vhicli all variables could be held constant except the number of components in the system. The column selected for use in the study of the effect of calibration mixture was a %plate 1-inch inside diameter glass per-
INDUSTRIAL AND ENGINEERING CHEMISTRY
July 1950
~~~
1387
~
TABLE I. COMPOSITION OF MULTICOMPONENT TESTMIXTURE
1 14 .
M0I.e Fraction Trace Trace 0.05 0.28
Component Cyclopentane Z,Z,-Dimethylbutane 2,3-Dimethylbutane Z-Methylpentane 3-Methylpentane n-Hexane Methylcyclopentme Cyclohexane
0.25
.o , 2 9
0.12 0.01 1.00
~
column temperatures were obtained from data reported by the National Bureau of Standards in API Project 44. The logarithms of the vapor pressures of the pure compounds were plotted against the reciprocal of the absolute temperature. This plot facilitated the selection of individual component vapor pressure values at any desired temperature. Based on these vapor pressure data, values of 01 (relative volatility) for the various components in the multicomponent system were calculated relative to 2-methylpentane over the range of column temperatures and are plotted in Figure 1.
3-met hylpentans
CALCULATIONS AND RESULTS 0.4
L
I
I30
Figure 1.
I
I
I
I40 I50 TEMPERATURE, OF.
160
I
Relative Volatilities of Components i n Multicomponent System
In calculating the data from the studies with the multicomponent system, the number of theoretical plates was obtained by a plate-to-plate method (S), assuming total reflux conditions. Preliminary calculations were made by starting at the kettle and proceeding upward until the top product Composition was reached or until it was obvious that some components were increasing or decreasing to impossible concentrations. In the latter case, minor adjustments were made in the bottoms analyses to permit calculations to the top product composition. Final plate-to-plate calculations were made by starting at the top and proceeding downward until the bottoms analyzed compositions were reached. Cyclohexane, the highest boiling component, was not present in the tops analyses and was introduced a t an assumed plate based on the preliminary calculations. The tops compositions were adjusted by trial and error until a calculation down the column would yield a composition corresponding to the kettle analysis. An example of the adjustments made in tops composition and of the deviations of the determined from the calculated kettle composition is given in Table 11. The calculations of all test runs with the binary system were made with Fenske's equation (2) for total reflux, in which an average 01 value was used. For comparison, two tests were calculated by the plate-to-plate method using (Y values corresponding to the calculated plate temperatures. The results from the two methods of calcuhtion checked within a fraction of a plate. Also, the difference between results from calculations based on actual reflux ratio of 99: 1 and based on assumed total reflux was only a fraction of a theoretical plate. Therefore, the results from the evaluation, assuming total reflux conditions for both series of tests, were used as a basis of comparison and are given in Table 111.
forated-plate column of Shell Development Company (Oldershaw) design (1), manufactured by the Glass Engineering Corporation. The two calibration systems used in these experiments were composed of saturated paraffin hydrocarbons and all components had the same number of carbon atoms per molecule. A multicomponent system was obtained by dearomatizing (with silica gel) a sample of commercial hexanes. The composition of the dearomatized sample as determined by the mass spectrometer method is given in Table I. Two of the components (2,3dimethylbutane and 3-methylpentane), which were present in the multicomponent system, were used for a binary system. These components were obtained from a commercial hexanes cut by precision fractionation. Tests were made a t atmospheric pressure and at rates over the normal operating range. For each series of tests, the column was allowed to come to equilibrium by operating a t total reflux for 6 to 8 hours. A sample of overhead product was obtained over a period of 1 hour. This was done at a reflux ratio of 99: 1. A sample was removed from the kettle 30 minutes after take-off had been started. Thus, the kettle sample was assumed to be of average composition, especially since the ratio of the volume of test mixture in the kettle to that of overhead product was 100 or greater. Measurements were made of the overhead temperature and pressure. The overhead rate was calculated from the volume accumulated in the receiver during the 1-hour period. These rates should be fairly accurate since the autoIENTATIVE DATASHOWING MAGNITUDE OF ANALYSES ADJUSTMENTS matic reflux ratio timer was calibrated TABLE IL ItEPREr (Test run made a t 2680 ml. per hour overhead rate) Ijy timing the interval the bucket was in Tops Composition, Kettle Composition, Mole Fraction Mole Fraction a take-off position. The compositions DeterAdDifDeterCalcuDifof the test samples were obtained from mined justed ference mined lated ference Component an average of duplicate analyses on the 0,1550 0,1490 -0.0060 0.0115 0.0183 4-0.0068 3-Dimethylbu+ane 0.6995 0.6995 0.2285 0.2273 -0.0012 mass spectrometer. 2lMethylpentane VOLATILITY DATA
The vapor pressures for the various components over the range of
0.2425 0.3685 0.1395 0 . m " _ 0.0095 _ 1 OOOO 1.0000 0.0000 1.0000 0.00001 was introduced into fourth theoretical plate from the top.
3-Methylpentane n-Hexane Methylc yclopentane Cyclohexane a
0.1346 0.0100 0.0010 0.0000
0.1385 0,0120 0.0010
+O:b640 +0.0020
.... ...
I
0.2360 0.3675 0.1419
0.0090 1.0000
-0.0065 -0.0010 +0.0024 -0.0005 0.0000
1388
INDUSTRIAL AND ENGINEERING CHEMISTRY
TABLE 111. EVALCATION O F 30-PLATE OLDERSHAW COLGYN WITH MULTICOMPOXENT SYSTEYAND WITH BINARYSYSTEM No.
Components in Calibration Mixture 6 6 6Q
6a 6
Overhead Rate. hfl./Hr. 750 1050 1650 2300
No.
Theoretical Plates 15.8
2680
3000 6a 930 2 1070 2 1260 2 1910 2 2240 2 a I n addition, small traces of oyclopentane detected in these samples.
15.6 14.4 14.6 15.4
Plate Efficiency, % 52.7 52.0 48.0 48.7
51.3
14.7 49.0 15.0 50.0 15.2 50.7 15.2 50.7 15.5 51.7 15.6 52.0 and 2,2-dimethylbutane were
The number of theoretical plates for this column was constant over a loading range of 750 to 3000 ml. per hour. This is in agreement with previous work reported for this type of column (1 ). However, the general level of efficiencies reported herein is slightly lower than those previously reported. CONCLUSIONS
These experiments were made under the same conditions. The test mixtures contained saturated hydrocarbons of like mo-
Vol. 42, No. 7
lecular weight. The calculations were comparable and were based on vapor pressures obtained from the same source. Therefore, the variables inherent in different test mixtures, methods of treating data, and sources of volatility data were minimized. Since the number of theoretical plates remains constant (Table 111) for tests with systems containing 2 or 6 components, it was concluded that the efficiency of the Oldershaw laboratory column was not affected by the number of components present in the test mixture. ACKKOWLEDGMEKT
The authors gratefully acknowledge the assistance of several of their colleagues in the experimental work and calculations. including particularly J. W. Askins, D. M. Bartay, L. C. Carpenter, A. E. Krc, and T. J. McLean. Acknowledgment is especially made to Edward Gelus and Stanley Marple, Jr.., for their valuable suggestions and guidance. LITERATURE CITED
(1) Collins, Franc C., and Lants, Vernon, ISD. ENG.CHEM.,ANAL. ED., 18, 673 (1946). (2) Fenske, M. R.,IND. ENG.CHEJ?.,24, 482 (1932). (3) Lewis, W. K., and Matheson, G. L., Ibid., 24, 494 (1932). RECEIVED September 15, 1949.
Some Physical-Chemical Aspects of Cotton etergency LIMITATIONS OF PRESENT LABORATORY TESTING METHODS JOSEPH M. LAMBERT AND HERBERT L. SANDERS' Central Research Laboratory, General Aniline & F i l m Corporation, Easton, Pa.
A
review of the conventional testing methods has been made which show-ed that the present tests fail in many respects to simulate adequately actual use conditions. I t is pointed out that in practice cotton is soiled by complex mixtures rather than by large amounts of finely divided carbon black. Moreover, the nonlinear relationship between the reflectance and the amount of soil on the fabric indicates that even trace quantities of ingrained soil can reduce appreciably the whiteness of textiles. Also, cotton goods are normally soiled and laundered repeatedly throughout their lifetime in contrast to conventional laboratory wash tests which employ only a single cycle with unused cotton. Preliminary results are described which were obtained with several cotton detergents in multicycle wash tests in which roll towels were soiled in actual use, then washed in a home washing machine and measured in the laboratory. Available field tests made with these detergents essentially substantiated the results of this practical series of tests. Conventional carbon black-type swatches were included in the above washes but in this case the results failed to correlate with the actual performance data. The important distinction between precision and accuracy of laboratory detergency .data is illustrated by a typical example.
T
ECHNOLOGICAL advances in many fields have been dependent mainly on significant laboratory testing procedures. The laboratory tests not only guide in the development of improved products, but also furnish the basis for quality control in production. Standardization of dyestuffs and of most textile auxiliaries is being accomplished on a commercial scale by various empirical laboratory tests. Therefore, one can conclude that fairly indicative results can be obtained in numerous fields related to textile technology. VARIOUS TESTS ON COTTON
The procedures for testing the finishing treatments usually applied to cotton have been described in great detail in the manual of the American Association of Textile Chemists and Colorists ( 2 ) . Tests for dye fastness, waterproofing, shrinkproofing, etc., have been worked out empirically using a strictly practical approach. Also, the testing work related to dye application, where the scientific approach furnished most interesting results on the dyeing mechanism (48), has remained mainly on a technological level. Although there might be room for improvements in experimental procedures and for advances in instrumental techniques, the present test,ing methods have proved of definite value to both manufacturers and users of the related products. 1
Present address, Ninol Laboratories, Chicago, Ill.
