ADSORPTION A Tool in the Preparation of High-Purity Saturated Hydrocarbons ALFRED E. HIRSCHLER AND SENTA =I>\.ION Sun Oil C o m p a n y Experimental Dicision, Xortcood, Pu.
iklixtnres of isomeric 4aturated hydrocarbons, including geometric isomers, can be separated b? adsorption on silica gel or acti\ated carbon; acti7ated alumina gives \ery little separation. Data are gi\en on the separation of many binary sjiithetic mixtures, including paraffin-paraffin, paraffin-naphthene, and naphthene-naphthene types. Rlost of these sjstems gi\e S-tjpe isotherms on silica gel and L-type isotherms on carbon. The effect of hydrocarbon type and molecular weight on adsorbability is discussed. The theoretical basis for the results obtained is considered; the? niay be reasonablj explained in terms of
Freundlich adsorption isotherms. >laterial of high purity may be obtained from many commercially a+ailalde or synthetic hydrocarbons by the use of adsorption. 2,3,3Trimethylpentane and 2,2,3-trimethylbutane mere ohtained in better purity than was previously possible by other niethods. Unsaturated and high-molecular-weight hydrocarbons may also be separated by adsorption. Its rather wide applicability, speed, and the use of Yery simple and inexpensive apparatus make adsorption one of the most powerful and convenient tools available for the separation and purification of saturated hydrocarbons.
D
related saturated hydrocarbon impurity. The second part presents the results obtained in the purification of a number of hydrocarbons by adsorption. . In saturate-olefin or saturate-aromatic mixtures, the olefin or aromatic is adsorbed over the complete concentration range, and plotting the adsorption isotherm over the range of 0 t o 100% unsaturate gives rise to what is commonly termed a U-type isotherm. However, a study of the effect of concentration on adsorbability in binary mixtures of saturated hydrocarbons on silica gel has disclosed the fact that in most cases each component is adsorbed over a portion of the concentration range, giving rise to what is often termed an S-type isotherm. By way of illustration the results of percolating n-heptanemethylcyclohexane mixtures of various initial concentrations through 84 grams of 28- to 200-mesh silica gel in a column 38 X 0.5 inches at 0" to 2" C. are shown in Figure 1. An excess of hydrocarbon mixture (100 ml.) was added in each case. Methylcyclohexane is adsorbed from a mixture containing 10% of this hydrocarbon, confirming the earlier observat,ion (18); however, more n-heptane is adsorbed from a 10% n-heptane solution than methylcyclohesane in the former case.
URIXG the past decade the use of adsorbents, particularly
silica gel, for the separation of hydrocarbons of different chemical types has become rather widespread. Previous work has shown that for hydrocarbons of similar molecular weight, the adsorbability, in general, increases Kith the number of double bonds per molecule. Thus, for example, small amounts of olefins or aromatics are removed from saturated hydrocarbons by percolation through silica gel. Silica gel has also been used where it has been desired to remove small amounts of nonhydrocarbon impurity from hydrocarbons. The possibility that saturated hydrocarbons, or more broadly that hydrocarbons of the same degree of unsaturation, could be separated from each other by means of adsorption has received little attention. [Since this manuscript was prepared, two papers discussing the separation of saturated hydrocarbons by adsorption have appeared ( 7 , 1 7 ) . ] I n 1933, Hofmeier and bleiner (IS) reported that a mixture of paraffins and naphthenes was not fractionated by silica gel. I n 1938, Mair and White (18) percolated mixtures of n-heptane-methylcyclohexane, 2,2,4-trimethylpentane-methylcyclohexane, and n-octane-"nonanaphthene" through silica gel. I n each case the concentration of naphthene v, as 10% by volume. I t was found that in each experiment the naphthene was preferentially adsorbed. Mair and White also percolated a 80-50 volume % mixture of n-nonane and n-tetradecane through silica gel, and found the nnonane to be preferentially adsorbed. However, they reported no attempts to separate from each other tn-o isomeric paraffins or naphthenes. Several years later Willingham (25) reported on the separation by adsorption of hydrocarbons of high molecular weight. While aromatics were separable from paraffins or naphthenes, and a monocyclic aromatic from a dicyclic one, no separation a t all was observed from a mixture of 10 weight % 5-(2-decahydronaphthyl) -docosane with n-dotriacontane. I n the first part of this paper the process of adsorption is extended to the separation of a number of mixtures of paraffin isomers or naphthene isomers as well as to paraffin-naphthene mixtures and to the effect of concentration on relative adsorbability. The results indicate a new general application of adsorption: the purification t o a high purity of saturated hydrocarbons which contain (as is often the case) small amounts of isomeric or closely
THEORETICAL CONSIDERATIONS RELATING TO S-TYPE ISOTHERMS
If an excess of a given binary mixture is poured through a column of adsorbent, the first portion of percolate will be richer in the component less easily adsorbed. After passage of a certain volume, the composition of the filtrate returns more or less sharply t o the initial value and further portions of the mixture pass through unchanged. The volume of hydrocarbon preferentially adsorbed can be calculated by a summation over all percolate fractions which differ in concentration from the charge to the column according to the equation 17 = c(C0 - C)
100
(li
where V is the volume adsorbed, Cothe original solute concentration and C the concentration of the given cut, both expressed in volume per cent, and u the volume of the given cut. This volume is proportional to the area bounded by the line representing the
1585
INDUSTRIAL AND ENGINEERING CHEMISTRY
1586
w
i-
80I
z
a 701 I-
a W
I I
z
...' W
r
I 3 J 0
>
30
2o
______ .
t
ORIGINAL
CHARGE
t I
0'
I
I
I
I
I
IO
20
30
VOLUME
OF
FILTRATE,
ML;
Figure 1. Fractionation of n-Heptane->Iethglc? clohexane 3Iixtures of Various Initial Concentrations b y Silica Gel
initial composition, t h e curve of composition versus volume of filtrate, and t h e composition axis. T h e amount of preferential adsorption a s calculated by Equation l is only a n apparent value and not the true adsorption, as i t neglects the change in volume caused by adsorption of the solute, as well a s considering the amount of "solvent" adsorbed t o be zero. I n systems where the components do not greatly differ in adsorbability, such as those dealt with here, the latter assumption is considerably in error. I n a solution of A and B xhenever the ratio of t h e true adsorption of component A t o t h a t of component B becomes less t h a n t h e ratio of A / B in solution, B xi11 be apparently adsorbed, even t,hough actually there is considerably more A t h a n B in t h e adsorbed phase. Correction for t h e change in volume caused by adsorption of the solute is relatively simple, as has been s h o a n by Bartell and Sloan (2, 3): and may be made bl- replacing the denominator of Equation 1 by 100-Co. However, in a system exhibiting a n Stype isotherm, this correction becomes rather arbitrary unless sufficient d a t a t o permit correction for ''solvent" adsorption are also available. Equation 1 is adequate for the comparison of the relative adsorption in various binary systems and on different adsorbents a n d is therefore employed in this paper. T h e adsorpt,ionisotherm for the n-heptane-methylcyclohesane system calculated as above from t h e data in Figure 1 is shon-n in Figure 2. It is of typical S-shape, with the point of zero selectivity at about 57y0 n-heptane by volume (53.5 mole %). Each component is preferentially adsorbed over t,heregion in rvhich it is present in smallest amount; or stated otherwise, each component is negatively adsorbed in the region in which its concentration approaches 100%. This is a general characteristic oi systems exhibiting S-t,ype isotherms. Bartell and co-workers ( 2 , S) have shown hen-, by combining two modified Freundlich equations, t h e apparent adsorpt,ion isotherm can be calculated over the complete concentration range in systems exhibiting S-type isotherms. If we assume t h a t t'he absolute adsorption of each component of a binary mixture follows the Freundlich isotherm
mV = kc" (where n, < 1) throughout
Vol. 39, No. 12
the n-hole concentration rangc, and further t h a t the two components are equally adsorbable in the sense that the isotherms for each component dissolved in the other are itlrntical, it can easily be shown using t'lit, equation of I h r t e l l and Sloan t h a t thc curve for the appawiit adsorption n-ill be E-shaped, and further t h a t the t,wo loops of t'he S rvill be equal and symmetrical. each component being preferentially adsorbed over the rcgion in n-hich its concentration is 0 t o 50%. Coiisidv~~iiig non- the effect of esponerit 7~ in the Freundlich isotherni upon a systeni in \yhich the tn-o coniponents are equally ads o r l d as defined above, i t is apparent t h a t when TI. = 1 (the socalled linear isotherm) there will be no preferential adsorption a t any conccntraioIi-i.e., the S is reduced t,o a straight line. In general, reducing the value of TI. !Till shift the points of maximum apparent adsorption ton-ard the ext'remes in composition and increase the amount of selective adsorption, The ahore considerations therefore lead to the conclusion t h a t 11-henever tn-o liquids have nearly equal adsorbabilitics, and the adsorpt,ionisot.herms are of the Freundlich t,ype, with 0 < n, < 1, their mixtures should exhibit a n S-type isotherm, and each component will be preferentially adsorbed in the region of low concentration. As a result of this fact, either component can be freed from small amounts of the other by percolation through a column of adsorbent,,the purified material being in the first portion of the percolate. This is the basis for the rather wide applicability of silica gel to the purification of saturated hydrocarbons. If one component is more strongly adsorbed than the other, it xi11 be preferentially adsorbed over more than 50% of the concentration range. T h e greater the disparity in adsorbability, the greater n-ill be the difference in size of the two loops of the S; when the difference in adsorbability beconies sufficiently great one loop will entirely disappear, giving rise to a U-type isotherm. These conclusions are not limited t o systems whose isotherms obey the Freundlich equation. -4ny type of isotherm which is concave to the composition axis will result in a qualitatively similar pict,ure.
O 0
I/
\ N-HEPTANE
(z
0
n METHYLCYCLOHEXANE
I
.5I
0
20 VOLUME
Figure 2.
40 '/e
I
60
80
100
N-HEPTANE
Isotherm (0' C.) for n-Heptane-3Iethylcyclohexane on Silica Gel
In systems whose components differ considerably in density or molecular Lveight, the shape of the isotherm will depend on the choice of volume, weight, or mole fractions as the units for expressing the composition. I n t,his paper volume fraction is employed for convenience. While the percolation of a n excess of a given hydrocarbon mixture through a n adsorption column has the advantage of simplicity, a greater degree of separation m a y be obtained in some cases if the quantity of hydrocarbon mixture is reduced t o a n amount
December 1947
I-N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y
1587
SEPARATIOS OF SYNTHETIC MIXTURES BY .ADSORPTIOX sufEcieiit to n-et pei~liaps50% of the column and, when the liquid level just ri~iclicsthe top of the adsorbent, a suitable desorbing EXPERI\IEST.~L DET.ms. I n order to determine the scope and liquid is added. The desorbing liquid forces the hydrocarbon possible limitations of adsorption as a procedure for purifying portion donm the column, during which process it is fractionated hydrocarbons, a considerable numbcr of synthetic mixtures were according t o the adwlbahility of its components. The first porprepared and percolated through adsorbent columns. tion of 1i~:drocarbonfiltrate is richer in the least adsorbed cornpoFor this work a glass tube 0.5 inch in di:tnicter n-as proritled ricnt than the charge, n-hilc the last portion of the hydrocarbon n-ith a qintered-glass plate near the bottom itnd a rc filtrate, j u c t het‘iirt. tlir clesorbiiig liquid reaches the bottom of the top (column -1).I t i r a s filled with adsorbent to a h column, i.: rirliw in the most strongly adsorbed component,. If a 38 inches; the column and reservoir were provided Tvith a jacket through which ice water was circulated; the temperature of the given binary mistui~eexhibits a L--t,ype isotherm and the separatefflux water xva.5 usually 3 ” t o 6 ” C. The silica gel used vas 28-to ing pon-er of the atlaorbent column i j sufficiently great, a portion 2OO-mesh (Davison Chemical Corp. 659528-2000), 84- grains being of each c o m p o ~ i e n tma>-he obtained in substantially pure form. used to fill the column. The activated carbc’n\vas 60- t o 90-nicsh Thuq, LIair ( I S ) , using this technique, has shown that a mixture (Columbia activated carbon grade L!; 50 grams w r e used t o fill column t o a hf,ight of about 3-1 inches. In the experiments the oi paraffins, Iiaplitliene5, olefins, and aromatics can he separated vith alumina 140 granis of a misture of 3 parts of 20- t o -10-mesh rather sharply i n t o a paraffin plus iiaphthene portion, an olefin (Grade F-1, Aluminum Ore Co.) and 1 part of -SO mesh (grade portion, anti an aromatic portion. .I,hluniinuni Ore Co.) 11-as used. tem which exhibits a n S-type isotherm, hen-The adsorbents Tvere poured into the coliinin and p:icked by tapping t h e outside of the column with a rubber niallc~runtil no ever, it is not pos~it)let o obtain both pure components starting further change in levi.1 occurred. Kith a giwri mixture. regardless of the separating power of the column. I?eCc>rringto Figure 2 , a mixture with coniposition beThe iiiaterials used were for the most part of known high purity t m e n C and B could lie separated into pure n-heptane and a mix(08y0 or better, often 0970+) and were gercolated through silica ture with composition C, while similarly a mixture initially begel prior to use. The first fractions vere exainined by refractive tween ti. 8 C-type, -4 ads. sible methods by v-hich a mix90.0 -1 ?.O ture whose composition corre2,4-DimethyIpentane 10.0 4 . 2,2,3-Trimethylbutaiie b 1.3 S5-100% -4 90.0 .4 0.12 sponds to the point of zero S,4-Diinethylpentanr 1,2,3-Trinietliylbutane C 7 . 0 9 1.7 93-100c; I selectivity on a given adsorb90.0 -1 0.70 ent could be resolved into 2.4-Dimethylpentane 2,3-Diniethylpentane z 10.0 A 0.45 90% A 0 90.0 two portions, each of which 2,4-Dimethylpentane 01..281 96-100$; B 13.0 2,3-~imetliylpentane C B could then be separated on 91.0 B that adsorbent into one or .‘,3,4-TrinietIiylpentaiie ?,3.3-Triinetiiylp~ntane S 10.0 1 1.3 95-100C; .I other of the pure components. 90.0 .i 0.18 Among these are: n-Octane 2,2,4-Trimetliylpentane > 10.0 A 1.4 70rc A
1. Distillation (vAtli or Tit hour entrainer) 2 . Solvent extraction 3. Adsorption, using an adsorbent of different adsorptive properties 4. A change in the temperature of percolation. In many cases it il-ould be expected that the effect of temperature 011 the adsorbability of the tn-o components ~ r o u l d be sufficiently different t o alter the point of zero selectivity 5 . Dilution with a third component separable b y distillation, such t h a t the adsorbabilities of the first t ~ o components are changed by different amounts
n-Octane
2,2,4-TrimetliyIpentane
.\
n-Heptane
if-Pentadecane
3
n-Heptane
n-IIexadecane
C
n-Heptane
n-Hexadecnne
.I
,?-Octane
n-Decane
n-Octane
n-Decane
n -Hexane
2,Z,5-Trimetii~lliexane C
a
S silica pel, C carbon, .1 alumina.
C
51.7 90.0 10.0 86.0 91.9 6.6 86.9 10.0 90.0 10.0 90.0 10.0 90 .o
11.7 90.3 10.0 90.0 9.2 50.4 91.1
A B
.I .4
.1
i .I -1
n
B
B
.i ‘4 B
B A -1 A
0.39 0.36 9 2 3 7 2.2 0.06 0.08 0.97 1.1 0.68 >7.6 0 0.14 0.46 0.17 0.51 1.6 i4 2 5.1
3.1
U-type, 1 ads. -50% C-type,
.i
.Iads.
U - t y p - , B ads.
90SB
C-type, A ads. C-type, B ads. C-type, A ads.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1588
TABLE 11.
S E L E C T I V E .%DJORPTIOS
1d.or-
Component B Cyclopentane Cyclopentane RIethylcyclopentane
bent
n-Hexane
3IethyIcyclopentane
C
n-Hexane
Cyclohexane
9
Component A
IS r . \ R . ~ F l ~ I S - S . ~ P € i T H E ss>->.rl:\Is I.
Eiiuilii>riuiii Concn.. Cumponent (7of .i .idsorbed
u
C
S
16.7
30 0 90.7 10.0 89.i 11.2 50 7 50.0 7.4 90.0 10.0 47.i 90.0
.. .. B B .\
A. A A. 1 B .1
111. .i~liorbcJ ['er 100 (41a111~o! ;irliorbiwil
..,. . .
0.2; 0 ,08 IJ 3 1
3.1 0 , .iO
Co~iipo.ition of Point of Zero Selecrinty, 1-01.fi
(.ipprox.)
Vol. 39, No. 12 no1 was usually emploJ-ed as the desorbing though - liquid, . occasionally bcrizene was used; for activated carbon, twiaene was used for saturated hvdrocarbons anti olefins, and aiiieth~-liia~~htIia~ciie for aroIiiatics. In a icw availability oi thc 1 tmns restricted tlic vuluiiic of mixture to 24 nil.
s
>1 1 1 0.2i
appropriate fractions up t o the point a t which the desorbing .i agent appeared. The composi.A 0.37 5 Cyclohexane tion n-as determined by reZ,?-Dimethylbu tane 0.12 B fractive index, using a Valen1.9 B tincl Abbe rcfractonieter where 0.10 OSc> B .i C 0.5 Cyclohexane 2,?-Dimethylbutane Slinute B 10.0 the difference in 71:: of the 1.9 B 90.0 components was 0.0080 or A 1.210.i E l l e t h l Irycloheyane n-Heptane 0.3, A ?E, more. I n several cases with an 0.29 A 30.0 index difference of 0.0040 a 0.11 B 00.5 0 ,i 2 n 90.0 .special spectrometer-type reC-type, .iads. A 4.2 10.0 C Methylcyclohexane n-Heptane fractometer sensitive to 0.00001 .A 0.i8 40.0 0.22 A 97 . 0 or better was employed. The 0.13 A i5Yc B 10.0 S 2,2,4-Trimethylpentane llethylcyclohexane sample was placed in a metal 0.52 B 49.5 75 ' prism with glass windows. 0.56 B 90.0 In a few cases, the composit,ion 0.42 C-type. B ads. B 10.0 C 2,2,4-Trimethylpentane Methylcyclohexane 2.2 B 90.0 was determined by nieasure0.01 95-100% B B A 9.9 2,2,4-Trimethylpentane Methylcyclohexane ment of freezing point; samB 0.16 90.0 ples as small as 1 ml. could be A 1.1 70% A 10.0 S Cyclohexane 2.4-Dimethylpentane A 0.79 .50.0 analyzed in this waj-, when a B 0.29 90.0 calibrated 5-junction copper0.41 70% B 10.0 s Cyclohexane 2,2,3-Trimethylbutane 0.33 no, 0 constantan thermocouple was 0.75 B 90.5 employed for t e m p e r a t u r e 0.61 A 10.0 Cyclohexane C measurement, 0.20 B 48.0 B 2 0 89,j The rate of percolation in 0.36 70% B 10.0 s E thylcyclohexane n-Octane the silica gel experiments was 0.38 50.6 0.35 B 90.0 about, 1 to 2 nil. per minute. :3,3 A 10 .,o C Ethylcyclohexane n-Octane T h e initial rate under gravity 0.46 .1 go., 0.01 d 95.G flon was appreciably faster 0.13 C-Type, Iads. .I 9.6 A E thglcyclohexane n-Octane than this; as the liquid front A 0 09 89.7 progressed dovin the column 0 90-10OCc B 3 0 .. E 2,2,4-Trirnethylpentane Ethglryclohexane 0 1 B the rate of flon- decreased, so i0.0 0 87 R .a0 . 0 t h a t in most cases slight nitro0.65 B 90.0 gen pressure was used to keep A 0.22 10.0 s Amylcyclohexane Ti-Decane 0.85 50.0 B the rate of travel from falling 0.65 B 8i.7 1 cni. per minute. below 0 , 5 0 C t y p e , A ads. A 9 0 . 3 C Amylcyclohexane n-Decane T h e coarser size of the actiA 0 . G O U - t S pe, A ads s9.7 Amylcyclohexane C n-Dodecane vated carbon resulted in an -4 4.3 C-type, .I ads. 10.0 Dicyclohexyl C n-Dodecane 0.62 A 90.0 average rat.e of about 1.5 to 2 A GO$ 4 10.0 1.4 E: Dicyclohexyl n-Pentadecane cm. per niinute under gravity B 0.88 90.0 flon. The filtrate was collect,ed 0.24 78VCB 10.0 A F Ethylayclohexane 2,2-Dimethylbutane 0.77 B 47.1 a t 1 to 1.5 ml. per minute. 647 A A 0.26 55.5 s Cyclohexane n-Octane The rate for activated alumina n , i x B 80.3 was about the same as for silica B 0.29 50 Cyclohexane n-Decane A 1 08 26% A gel. Dicyclohexyl P n-Dodecane !? JJ A 0 11 RESL-LTS.The results on B 0 50 90.3 the separation of synthetic Estimated from percolation of impure cyclopentane containing s e i e r a l per cent 2,2-dimethylbutiine. mixtures are given in Tables I to 111. In column 6, the extent of preferential adfractional distillation of t,he product. The 2-but~-le~clopentylca.- sorption as calculated Liv Equation 1 is given for each concentration studied. The data permit' a n approximate location of clopentane was synthesized in these laboratories by .I.€'. Stuart. the point of zero selectivity for those systems exhibiting S-type I n most cases 49 ml. of hydrocarbon mixture were added to adsorption curves (column 7). I n many cases the uncertainty in the column, a n d when this had completely entered the adsorbent, this point is of the order of 570, though occasionally it may be a desorbing liquid was added to force the hydrocarbon dovc-n the greatrr. column. When silica gel or alumina was used, met,hanol or ethan-Hexane
Crclohexane
c
I
B" B"
Q
>?.6 0,G l i
December 1947
1589
INDUSTRIAL AND ENGINEERING CHEMISTRY
ing the appearance of desorbing liquid is of increased purity. This is a n esample of a system with a C-type isotherm, and while theoretically it is possible t o obtain pure methylcyclopentane in the last portion of filtrate from a column of sufficient separating power, in practice i t is more difficult than when the purified hydrocarbon is in thc first portion of filtrate. The considerable heat, produced a t the alcohol-hydrocarbon interface disturbs the separation; the alcohol tends to overrun the hydrocarbon somewhat, thus contaminating the purest port’ionof the filtrate; and inipurities desorbed from the gel by t h e alcohol add to t,he contamination. There is also the possibility t h a t some of tho less strongly adsorbed coniponent is held so tightly on the most active centers of the adsorbent t h a t it is not, readily desorbed b y the more strongly adsorbed ciomponent, but is displaced by the alcohol. These difficulties are, of course, proportionately more serious as the amount of preferential adsorption decreases. Where there
.IDSORPTIOS ISSAPHTHESE-SAPHTHESE SYSTEMS TABLE 111. SELECTIVE
Component -1
Component B
Adsorbent
Eq,uilibrium Concn., % of -1 10.0 90.0
Component Adsorbed
A”
Composition 311. of P o i n t of Zero Adsorbed per 100 Selectivity, G r a m s of Vol. % Adsorbent (Approx.) U-type, h a d s >O , :3 0.34
Cyclopentanea
Cyclohexane
S
Cyclopentane”
Cyclohexane
C
10.0 90.0
.1
Jlethylcyclopentane
Cyclohexane
s
10.0 90.0
0.58 0.37
C - t y p e , A ads
hlethylcyclopentane
Cyclohexane
C
10.0 90.9
A A A A
2.1 0.90
C - t y p e , .1 a d s
Cyclohesane
Ethylcyclohexane
S
10.0 56.0 90.0
B B
..
0 0.72 0.10
70%B
Cyclohexane
Ethylcyclohexane
C
14 3 88.6
hlethylcyclohexane
Ethylcyclohexane
S
50.5
A 1 ethylcyclohexane
Ethylcyclohexane
C
10.0 90.0
Jlethylcyclohexane
Ethylcyclohexane
A
10.0 91.2
?+Iethylcyclohexane
Amylcyclohexane
S
9.9 49.6 90.1
cis-Decalin
frons-Decalin
S
..
..
....
.1mylcycloliexane
Dicyclohexyl
S
10.0 90.0
A
>0.48 0.68
2-Butylr?-clopentylcyclopentane
Dicyclohesyl
S
10.0 90 0
Cyclohexane
Dicyclohexyl
C
10.0 90.0
trans-Decalin
Dicyclohexyl
C
11.0
1.1 90-100% Little change
1.7 >3.4
B B A B B
C - t y p e , B adb
.....
0.13
A
B A
i:
A
-1
1.0 2.2
97-1007, B
0.01 0.13
857, B
0.29 0.22 0.25
635 A
0.61 0.10
A B B A
0.73 >4.3
98-100Yc B 98-100% 95-100% A C-type, B a d s
0.27
Contained several per cent of 2,Z-dimethylbutane.
This stated uncertainty applies only t o the particular lot of silica gel used for a given srstem. T h e relative adsorbabilities of various saturated hydrocarbon types vary somewhat from lot t o lot. For example, in t h e system n-heptanemethylcyclohesane t h e point o€ zero selectivity shown in Figure 2 is a t 57%, n-heptane; for tn-o other more recently obt,ained lots t h e point of zero selectivity m-as found t o be a t 50 and 53% ?a-heptane, respectively. The extent of this variation is insufficient t o affect conclusions as to t h e effect of hydrocarbon structure on adsorbability escept in those few cases of nearly equal adsorbabilities.
For a number of systems, t h e composition of the point of zero selectivity is given as “95 t o 1007, -4.’’ I n most of these instances it is probable t h a t the isotherm is of a U-type but in t h e absence of esperimental data a t concentrations near 100% A , the possibility that the curve is P-type, with component B being adsorbed over a few per cent of the concentration range, cannot be excluded. I n some of the esperiment.s, t,he absence of a plateau in the volume-composition curve a t t h e initial concentration indicated t h a t the quantity of hydrocarbon misture was not, sufficient t o saturate the adsorbent. As a result’,the true adsorption a t equilibrium would be somexT-hat greater than the value calculated from the esperimental curve. Such values in column 6 are prefised by “>”. An extreme esample of such lack of saturation is the curve for 107, n-hexane in cyclohesane. Figure 3 illustrates d i e type of volume-composition curves obtained when a desorbing liquid such as alcohol is used to force t,hc hydrocarbon port,ion through the column of silica gel, and compares the adsorbability of cyclohesane n-ith n-hesane, neohesane 12,2-dimcthyIbutane), and methylcyclopentane. K h e n a l o 7, cyclohesane-907, methylcyclopentane misture is percolated through silica gel, t h e first portion of filtrate is of lesser purity than the charge, b u t t h e final portion of hl-drocarbon just preced-
----.----L-L
100
I-, L l
............................
90 -7
w
80
----..........
2
4 X W
I 0 -I V
L,-
(A) (E) (C)
CYCLOHEXANE-N-HEXANE CYCLOHEXANE- NEOHEXANE CYCLOHEXANE- METHYLCYCLOPENTANE
ORIGINAL
* 0
CHARGE
,..’ L-,,
i 0 >
-.I
‘? L,
2ob=-L A-BG_
10
FFKry*r,- .......I ” ’ . . . . . I
1
0
:
I
10 VOLUME
Figure 3.
................................
.
OF
I
I
I
20
30
40
FILTRATE,
ML.
Separations Produced in Several Binary S y s t e m s hj Silica Gel
Column A, 84 grams of gel, 25-m1. charge for cyclohexane-n-hexane at 10 and 89 concentration 49-ml. charge for other systems
INDUSTRIAL AND ENGINEERING CHEMISTRY
1590
f
.-__.... ................ ,.~
00
ETHYLCYC LOHEXANE-N-OCTANE 2,2,4-TRIMETHY LPENTANE-N-OCTANE ETHYLCYCLOHEXANE- METHYLCYCLOHEXANE ORIGINAL
CHARGE
Vol. 39, No. 12
T \ M e for a number of the systems in Tables I t o 111 the boiling points are sufficiently different to make the components easily separable by distillation, in the fullon-ing case" the boiling point differencc ( C.) are small: 2,2,3-trimethylbutane-cyclohexane,0.13 O ; 2,4 - dimethylpentane-cyclohexane, 0.23 '; 2,4-dimethylpentane-2,2,3-trimethylbutane, 0.36"; n-heptane-2,2,40.81 '; 2,2,4triniethylpentane, trimethvlpentanc-nietlivle! clohexane, 1.7 '; and n-heptane-riicth?-lcyclohexane, 2.51 a ( 1 ) . T'ac'TORS . i F F E C T I S G I!EL.ITIVE
AD-
Based on the data in Tables I to 111 the iolloiriiig conclusions appear t o be permisai Me: The majority of the sj-stems studied have S-type isothernis on silica gel, but on activated carbon they have preponderantly C-type isot,hernis. T h e 0 IO 20 30 40 50 data on activated alumina are t,oo incomplete t o permit a generaliz'ntion. VOLUME OF F I L T R A T E , ML Some of the systems n-ifh C-type In Figure 2 . Separations Produced in Sexera1 Binary Systems by curves on silica gel, it v a s somewhat Activated Carbon surprising t o find t h a t the differences Column A. 50 g r a m s of carbon, EO-ml. charge for 2,2,4-trimethylpsntane-n-octane a t 90yo concentration in adsorbability of the txvo components 49-ml. charge for other systems n-ere sufficient to result in this type of iiotherm. Such, for esriniple, were is a considerable difference in adsorbability, the nior(' strongly the systems ci's-deraiiii-t,nr~s-decaliI1 and 2,2,1-rrirni.thylprntane2,3,3-trinietliyl~~entane. adsorbed component can be obtained in substantially pure form in ciiis stutiiid, the cvmponent m o w strongly the last, portion of hydrocarbon filtrate, as has hecn sho~vnby was a l w nioi'e strongly adiorhcd 011 carbon Mair ( 1 6 ) for mixtures of aromatic ant1 saturated hytlrocarbons. and alumina. L7sually,lion-r r>the i-otherm vas S-type on silica I n Figure 4, a fen- curves obtained on activatcd carbon are gel and U-t>-peo n carbon. There i m some esccytions to the first given. T h e most striking feature is the conaiderahle difference in gencralization; t,hus in the system 2,4-dimeth~-lpentaiie-2,3the adsorbability of n-octane and 2.2,4-trinicthylpexitaxi~~, As indicated in Figures 1, 3, and 4, for many of these Raturatc tlimctli!-lpi.iitanc t l i ~formt,r is pdmrhed over !)07cof the concentration ranyc~on silica gel, ~ v l i e ~ on ~ ~ carlion ai thv Iat . sharp; usuallj- thc, volume of systems the separation is not hydrocarbon with composition intermediate 1,ctn.een t h e charge sc~rbcd(~v(br !I5 to 1007, of tlic, concentration ranpc~. TI1 n-octa~ic~-ethy~c.~c~loliexanc, ~i-decane-"aiiiylc~-rlo~iexa and one of the pure components is three or more tinics the volumc ~i-lic~s:~iic~-riirthylcyclo~eritaiie a1.o sho\v this differc,iicc in selecof pure hydrocarbon obtained. If the separation is lair, B. J., J . Research S a t l . Bur. Standards, 34, 435 (1945). l I a i r , B. J., Gaboriault, A. L., and Rossini, F.D . , IND.ENG. CHEM.,39, 1072 (1947). llair, B. J., and White, 3 . O., J . Research Xatl. B u r . Standards, 15, 51 (1935). &lark, H., and Saito, G., Monatsh., 68, 237 (1936). Nederbragt, G. W., and deJong, J. J., Rec. trau. chim., 6 5 , 831-4 (1946). Streiff, A. J., Murphy, E, T., Sediak, V. .4.,Willingham, C. B., and Rossini, F. D., J . Research ,Vatatl. Bur. Standards, 37, 331 (1846).
Tooke. J. W., and =Iston, J. G., J . Am. Chem. SOC.,67, 2275 (1945). Turner, K.C.. Petroleum Refiner, 22, 140 (1943). Ylugter. J. C.. Waterman, H. I., and van Westen. H. A , , J . Inst. Petroleum Tech., 18, 735 (19323; 21, 661 (1935). Willingham, C. B., J . Research T a t l . Bur. Standards, 22, 321
(1939). RECEIVED hlarch 7 , 1947.
Presented before t h e Division of Petroleum Chemistry a t t h e 111th Meeting of t h e AMERICANCHEMICAL SOCIETT -4 t l a n t i c C i t y , N J