1IULTIMOLECULAR ADSORPTIOS
769
MCLTIi\lOLECULBR -4DSORPTIOK FROM BISARY LIQCID SOLUTIOSS R O B E R T S. HASSES,' T I S G FU,
AND
F. E. BARTELL
Department of Chemistry, Cniiersity of Xichigan, Ann A r b o r , Michigan Recezted J u l y 13, 1948 Ih-TR ODUCTION
Prior to 1938 adsorption of single component gases and of substances in binary liquid solution n-as generally discussed in terms of monolayer models of the adsorbed substance. I n that year, holyever, Brunauer, Emmett, and Teller (4) introduced a theory of multimolecular adsorption of gases which immediately found wide application t o previously ill-correlated data for the adsorption of gases below their critical temperatures. This hypothesis that adsorption of such gases is in most cases multimolecular has become widely accepted as an established fact. I n a previous paper (6) the authors presented data indicating that the adsorption of n-butyric acid from aqueous solution by Ion--area carbons is multimolecular. The present research was undertaken for the discovery of much more extensive and convincing evidence of multimolecular adsorption from binary liquid solutions. That the adsorption of a strongly adsorbed solute from solution in a weakly adsorbed solvent is in a sense parallel to the adsorption of single-component gases is indicated by the nidespread application of the Langmuir and Freundlich isotherm equations to both types of behavior. If multimolecular adsorption from the gas phase is visualized as incipient phase formation (of a liquid phase) governed largely by the reduced pressure p/po (where po is the saturated vapor pressure), a parallel behavior might then be expected in the adsorption of a strongly adsorbed solute from a solvent in which this material was of limited solubility. I n this latter case the reduced concentration e/eo (where eo is the s:iturating concentration of the solute) might be expected to be a determinant of a nature parallel to the reduced pressure in the case of gas adsorption. Pursuing this parallel further, it Tvould not be surprising if multimolecular adsorption from binary liquid solution, if found to exist, were representable by BrunauerEmmett-Teller isotherm equations. Finally, small-pore or capillary diameters in the adsorbent are known to be effective in restricting the number of possible adsorption layers in gaseous adsorption, and should be even more effective in this respect in adsorption from boiution, since the molecules commonly studied as adsorbates are larger in the second case. To study the possible multimolecular adsorption and to test the applicability 1 The data reported are from a dissertation submitted by Robert S. Hansen in pnr tial 'ulfillment of the requirements for the degree of Doctor of Philosoph3- in the University of LIichigan. Acknowledgement is made of financial assistance in the form of fellonships teceived from the Board of Governors and the Executive Board of the Horace H. Iinckham School of Graduate Studies and from the Allied Chemical and Dye Corporatlon (1948)
i 70
R. S. HASSES, T. FU
A S D F. E . BARTELL
of Brunauer-Emmett-Teller isotherm equations to adsorption from binary liquid solutions, isotherms for the adsorption of a series of organic compounds from aqueous solution by a series of relatively non-porous carbons were determined. The organic compounds chosen were of limited solubility in n-ater. Isotherms for the adsorption of the same compounds from aqueous solution by a more porous sugar charcoal were determined for comparison. EXPERIlfEiXTSL
Three artificial graphites, tn-o carbon blacks, and a sugar charcoal were selected as adsorbents. The graphites were purified in the manner described in a previous paper (6). The carbon blacks ]yere submitted to two 2-1-hr. elutriations with redistilled reagent ether and were then heated for 2 days in high vacuum (10-5 mm.) a t 980°C. The sugar charcoal was prepared according t o the method of Bartell and Miller (2) ; cane sugar was recrystallized from water three times, charred in a platinum dish, and heated in high vacuum for 3 days at 1100°C. Specific surface areas of these carbons n-ere determined from nitrogen TABLE 1 D c s o iption o j adsorbents ~
ADSORBENT'
Carbon Carbon Carbon Carbon Carbon Carbon
I
h... . . . . . . . . . . . . B . .. . . . . . . . . . . . . E . .. . . . . . . . . . . . . F , .. . . . . . . . . . . . . G . .. . . . . . . . . . . . H . .. . . . . . . . . . . . .
K4TCRE
.Irtificial graphite Artificial graphitc Artificial graphite Sugar charcoal Semi -r el n f ore i 11g f u r n nee 1JIac B Chnnncl hlacli
1
___
ARE1
,
ASH
m?/gram
I
per ceni
1
95.8 18.4 25.5 790. 26.3 124.
'
0 034 0.013 0 04 0 002 0 0% 0.05
* The s y n b o l s used are laborntory designations of these carbons.
adsorption by the method of Brunauer, Emmett, and Teller. The adsorbents are described in condensed form in table 1. The organic compounds used as adsorbates 11-ere n-caproic acid, n-vnleric acid, n-amyl alcohol, n-butyl alcohol, aniline, cyclohesanol, and phenol. Procedures used in the purification of these compounds are given in table 2; ivith the exception of n-caproic acid the starting materials were of reagent grade. K a t e r obtained by redistilling laboratory distilled water from alkaline permanganate solution n-as used as solvent. All solutions were made up of linonn molality and molarity; the dilutc solutions were prepared by dilution of the concentrated. T o determine solubility, an excess of the organic compound n-as added t o redistilled water in a mercury->ealed flask n-hich was shaken mechanically for 48 hr. in an air chamber thermostatted to 25.0"C. =t 0.1". The flask was then allowed to stand for 3 hr. in the air bath, after which a portion of the Tvater-rich phase was removed by means of a hypodermic syringe and Tvas compared interferometrically with the most concentrated solution of that compound 11 hich
771
JIULTIJIOLECULAR .IDSORPTIOS
could be prepared conveniently.? The results given in table 2 are in each case the mean of three such determinations. T'alues of co, saturation concentration in moles per liter of solution, are given to four significant figures not because this is justified by their precision, but because concentrations reported later in this work xere reduced by means of these values; the equilibrium concentrations prior t o reduction were accurate to the fourth decimal place, and may be recalculated to that accuracy by multiplication 11-ith the appropriate co value. Table 2 also gives the solubilities in weight per cent solute in the water-rich phase, together with the deviation of the results. TABLE 2 S l e f h o d s of purification a n d p r o p e r t i e s o j liquids u s e d
~.
I
ORG.4SIC
i,
coypoup;D
PURIFICATIOr
~
, n-Caproic a c i d . , . . . . . . . . .~ n-Valeric acid. . . . . . . . . . n-A%xiiylalcohol.. . n-Butyl alcohol.. . . . . . . . . Aniline.. , . . . . , . . . . . . . , . . Cyclohesanol. . . , . . . . . , . , ' Phenol. . . . . . , , . . . . . . . . . . . I
~
*f
dfffd d dt d: dtl fdt fd
l
I
BOILISG POINT
1
OC.
I 200-01/7-10 mm. '
1 ~
~
1861711 nun. 1361736 nini. 117j745 mm. 183/7-12 mni. 1581730 nim. 180ji40 mm.
0.08759
I ~
,
o.4857 0.2666 0.9680 0.3932 0.3906 0.8955
, ,
~
~
~
,
weight per cent
1.018 k0.006 4.97 zkO.02 2.51 kO.02 7.41 &0.03 3.669 i 0 . 0 0 3 3.92 hO.01 8.412 iO.005
= fr:ictional freezing: d = dist,illation, using 2-ft. glass-packed reflux column.
t The
C.P. n-amyl alcohol purchased \vas found to contain a n appreciable amount of diariiyl ether. T o reiiiove this about 12 gallons of water ix-ere saturated n-ith alcohol, the n-ater phase separated, and tlie alcohol largely separated from the water by boiling the w:tter solution, n-hich caused tlie amyl alcohol t o separate and steam distill. The alcohol in tlie (two-phase) distillate v a s then separated and the treatment repeated. Finally, the alcohol T T ~ Sdried over anhydrous magncsiurii sulfate prior t o distillation. We have sincc found another commercial C.P. grade of n-amyl alcohol t o contain the same impurity, xvliich suggests t h a t purification procedures for generally av:lilable grades of n-amyl alcoliol should be designed t o remove the ether. 2 Refluxed 4 hr. over magnesiuni and iodine prior to dist'illation.
In general, for each determination of adsorption 0.500 g. of adsorbent Tvas n-eighed into an adsorption flask and 10.0 ml. of solution was added to the Bask, Tvhich v-as then sealed and shaken mechanically for 45 hr. in an air bath thermostatted to 25.0"C. =t0.1". The flask was then removed. The solid This technique is open to two objections, both of n-hich are believed t o be minor. First, in transferring the sample f r o m the air b a t h t o the interferometer (also enclosed in m :Lir bath thermostatted t o 25.0'C.) the sample was exposed t o room temperature for Fbout 2 min., and the room temperature deviated from 25°C. by as much as 3°C. It is ieliered t h a t errors from this source are included in the reproducibility of the measurenents. Second, the solubilities a r e based on extrapolations of the interferometric calibruion curves, n-hich appeared t o be justified when checked by volumetric dilution. On the Ither hand, the present work follon-ed the interferometric technique described by Bartell ,nd Sloan (3), and for this reason evaporation from the interferometer cell n-as not 3. source f error. Butler, Thomson, and Naclennan ( 5 ) appear to have encountered some difficulty rising from this source.
772
R.
S. H.%NSEN,
Y. FU AND F. E. BARTELL
phase was caused to separate by brief centrifuging; then a portion of the supernatant liquid n-as pipetted off and analyzed interferometrically. For carbons of higher area it was occasionally necessary t o use smaller amounts of adsorbent for the same volume of liquid in order to obtain as high equilibrium concentrations as were desired for the final points on the isotherms. To prevent evaporation, adsorption flasks of approximately 25-ml. capacity vere constructed from 19,’38 mercury-sealed stnndard-taper glass joints, the bodies of the flasks being constructed from the female portions of the joints, and the caps from the male. After the adsorbent and solution had been added to a flask its cap was placed firmly in position, and the seal gap was filled TI ith analytical reagent mercury and was corked: the system was thus liept in a sealed condition until ready for analysis. An equation similar to that of Ostn-nld and de Izaguirre (9), and deduced from similar considerations, was used for the interpretation of the present data. This equation is
where c is the equilibrium concentration of component -1in moles per liter of solution, Ac is the concentration change due t o adsorption in millimoles per liter of solution, 17 is the initial volume of bulk solution in milliliters, m is the mass of adsorbent in grams, z‘, and Lib are the partial molar volumes of components -4 and B in milliliters at concentration c, and a(c) and b(c) are the numbers of niillimoles of components d and B repoved from the solution a t equilibrium concentration c, per gram of adsorbent. This may be put in the form a =
d - + B(c) m
where
The quantity x’/m may be defined as the apparent adsorption of component h a t equilibrium concentration c. I n the present work equilibrium concentrations were invariably less than 1, ua-5cb, and it was expected that, except for very small concentrations, the quantity a would be greater than the quantity b; under these circumstances z’/m should not differ from a by more than 3 per cent, but it is t o be emphasized that this difference is such that
so that the actual amount of component h taken up by the solid is at least as
great as the apparent amount. Finally, it may be noted that for the systems used x l / m did not differ from vAc/na by more than 10 per cent even a t the high-
TABLE 3 Adsorption of n - c a p r o i c acid from aqueous solution CARBON
c:co
dim
1
0.0364 0.1185 0.2237 0.3252 0.4335 0.5299 0.6665 0,7972 0.8967
,
.k
I
0.165 0.259 0.307 0.366 0.420 0.478 0.576 0.693 0.855
1
, ,
~
c
0 00057
0 934
0 0 0 0 0 0
1 171 1 373 1 632 1 829 1 901 1 935
0035 0169 0612 47 i o Ti90 8839
0.1034 0.2282 0.3543 0.4778 0.6072 0.7232 0.S922 0.9343
'
1
i
I
I
0.051
I
0-067
~
0.0T8
I
0,098 0.115 0.134 0,179 0.209
I 1
~
i
-____ C.\RBOX
~
1
XI,
CARBON
x'lm
E
c 'CJ
x'lm
0.0050 0.2134 0.3355 0,4552 0,5795 0,6920 0.8540 0.9129
0.066 0.093 0.111 0.131 0,163 0.192 0.245 0.285
I
i
m _____I_____i
cil
B
i
C,'CQ
I
~~______~__-__C.ARBOS F _~_____
CARBOS
CARBOX
x' m
c co
0.0033 0.2151 0.3352 0.4539 0.5751 0.6675 0 . 8470 0.9123
-
G
'
0.069 0.090 0.112 0.141 0.171 0.200 0.258 0.2S7
' ~
~
~
H
x'lm
I
0.0360 0.100i 0.1950 0,285s 0.3951 0.4S12 0.6000 0.7595 0.8718
1 1
0.169 0.274 0.353 0.435
~
1 I
0.467 0.563 0.675 0.826 1.0T4
1 , ~
TABI,E 4 d dso r p t ion of n - i'ale 1. ic m i d fro vi arjiieo 14 s sol ii t ion CIRBON
_
_
~
'4
CARBOX
d,'m
c/co
B
CiRBOS
~ _ _ _ _
c co
2'm
i
c
X'.?lt
c3 ~
~~
0.0139 0,0421 0.002s 0,2027 0.3146 0.43SO 0.5416 0.7665 0.8968
0.093 0.164 0.23s 0.321 0.372 0.414 0.465 0.633
' '
0.SOi
~
,
~
~
,
~
E
____~________
- ~-
0.026 0.045 0.054 0.070 O.OS9 0.092 0.0!)6 0.144 0.243
0.020S 0,0543 0.1116 0.22s' 0.3433 0.4i05
0 . 57% 0.bl5O 0.9522
,
, 1
1 ~
1 ,
,
0.036 0.080 0.087 0,109 0.125 0,133 0.15s 0.211 0.326
,
0.0197 0.050; 0.1052 0.2243 0.3394 0.4660 0.5724 0,8085 0.9442
~
~
1
______
I
CARBON
F
1
d/nr
0.00078 0.0090 0.0830
,
0.1815
,
0.3055 0.4079 0.6499 0.7063
,
0.565 1.051 1.486 1.6iO 1.716 1 ,786 1.799 1.816
C/CQ
c;co
_ _ _ ~ _ _ _ _ ___-__
'
' ,
CARBOX
~
,
0.0202 0.0530 0.1092
I I
, ~
x'lm
,
,
0.031 0.058
1
0.077
,
Ti3
0.123 0.140 0.154 0.229 0.33;
C.iRBON C/CO
0.0122 0.0400 0.0%;
1
0.3051 0.4262 0.5295 0.7521
'
O.XSO2
~
, '
H dJJm
-__-_
0.110 0.1%
0,277 0 400
0.1946
0.116
0.2236
0.3309 0.4657 0.5i27 0 . 806T 0.9431
,
G
t
'
, ~
0.466 0.530 0.582 0.779 0.874
7-74
R. S. HA-YSES, T. FU A S D F. E. BARTELL
0.0146 0,OJlO 0.1000 0.2291 0.3662 0.5022 0,6374 0.8245 0.9225
0.083 0,180 0,257 0.362 0,405 0.459 0.520 0 . 678 0.85s
GIGO
x','ni
I
11 I1 [
0 0 0 0 0 0 0 0 0
1
0234 0615 1331 2749 1199 5623 7035 9006 9526
0.032 0.062 0.067 0.088 0.093 0.109 0.134 0.231 0.253
I
' 1
1
,
~
~
1
I
1 1
1
0.0218 0.0582 0.1271 0 2691 0 4123 0 5538 0 6953 0 8S52 0 9386
0.041 0.081 0.101 0.121 0.137 0.159 0.183 0.322 0.330
21 f
x'lm
m ~
I
0.00080 0.0027 0.0225 0.2316 0.3691 0.6268 0.8034 0.9188
0.410 0.814 1.535 2.016 2.071 2.154 2.254
I I
1 ' I 1
2.256
0.0223
1
:E
I
0.2681 0.4118 0.5549 0.6962 0.8911 0,9145
0.030 0.076 0.101 0.12; 0.140 0.153 0.177 0 288 0 300
1 I l
I
1
I
I 1
,
0.0134 0.0395 0.0916 0.2160 0.3498 0.4857 0.6207 0.6084 0.9100
0.090 0.189 0.306 0.427 0 . 499 0 I555 0.617 0.771 1.008
'
TABLE 6 Adsorption of n - b u t y l alcohol f r c m aqueous solution CABBONA
c/ca
0.0302 0.0776 0.1641 0.3419 0.5207 0.7005 0.8759
1
1 i
1 I
X ' h
0.116 0.248 0.337 0.430 0.476 0.517 0.676 ___
1 ~
1 j
I
CARBON
0.0345 0.0S72 0.1i73 0.3587 0.5391 0.7212 0.9006
1 1 I
1 I
1 1
B
CARBON
E ~-
x'/m
0 030 0.057 0.072
O.OS9 0.099 0.114 0.152
~ / G O
0.0339 0.0857 0.1754 0.3564 0.5364 0.7190 0 8983
I
I
x'lm
-~
0.042 0.086 0.110 0.136 0.153 0.161 0.202 __
TABLE 7 Adsorption of aniline froin aqueous solution CARBON
I
A
c cu
2'
CLRBOY
CARBON
m
0,058 0.088 0.111 0.132 0.140 0.160 0.178 0.228 0.327
0.0282 0,0789 0.1650 0.3421 0.5177 0.7000 0.7863 0.8953 0.9596
E
,
c/co
0.240 0,371 0.556 0.733 0.760 0.780 0.826 0.959 1.064 1.277
0,0049 0.0429 0.1087 0.2663 0 4390 0.622s 0.7056 0 . so47 0.86SL' 0 .hen01. . . . . . . . . . . . . . . . . . . . . . . >yclohexanol. . . . . . . . . . . . . . . . .
0,027 0 * 027 0.036 0.036 0.058 (0.045) 0,012 0.040
0.022 0.022 0.020 0.020
0.032 0.032 0.020 0.020
(0.064) 0.013 0.040
(0.059) 0.015 0.036
I
1
0.007 0.012 0.012 0,017
0.010 0.010 0.013
0.042 0.044 0.042 I 0.044 0.020 0.040 0.020 1 0.040 0.076) 0.025 1 0.020 0.025 0.020 0.032 0.030
1
~
* Subsequent d a t a have been calculated both for n = 3 and n = 2 ; values for n = 2 will be ncluded in parentheses. ilms. This means that, even if the adsorbed film contained no water a t all, t would not be possible t o pack all of the molecules adsorbed in a single layer ven in an orientation enabling the maximum possible number of molecules to
778
R . S. HASSES, T. F U .1SD F. E. BARTELL
be in contact n-ith the surface. The fact that there is every reason to expect the adsorbed film t o contain water as well as organic compound only strengthens the a r g ~ m e n t . ~ I t will he observed that for each solute studied the apparent minimum molecular area on the sugar charcoal carbon F is greater than that on each of the other adsorbents by factors of 3 or more. This is easily explained if we suppose that the adsorption of these solutes by carbon F is unimolecular, but by the other carbons multimolecular. Xor is this supposition unreasonable, for this is exactly the manner of behavior of these same adsorbents toward nitrogen; nitrogen adsorption by carbon F followed a Langmuir isotherm, and nitrogen adsorption by the remaining carbons follon-ed B.E.T. isotherms. I t will also be observed that the isotherms for the adsorption of all solutes studied by all carbons except carbon F are sigmoid in form. Such forms in gaseous adsorption have almost invariably been associated with multimolecular TABLE 11 Jfinimum observed area p e r molecule
I
SOLUTE
l
I--
n-Caproic a c i d . . . . . . . . . . . . ' n-Valeric a c i d . . . . . . . . . . . . . ' n-.%rnyl alcohol. . . . . n-Butyl a l c o h o l . . . . . . . . . . Aniline. . . . . . . . . . . . . . . . . . i . Cyclohesanol.. . . . . . . . . . . . , Phenol., . . . . . . . . . . . . . . . . ., ,
.I . I
1
CARBON
A
18.5 19.6 18.5 23.4 12.4 27.1 12.0
-I '
'
B
14.5 I 12.4 12.0 20.0 1 9.3 15.2 10.8 ~
~
'
.i ~
~
'
E
14.8 13.0 12.8 20.9 9.3 16.8 10.0
~
~
~
' '
I
F
83 88 71 66 60 92 59
l
.
~
G
15.2 13.0 I 14.5 1 20.3 ' 11.8 1 21.6 9.2
1
I
H
CLOSE-PACKING ABEAS'
I
~19.1 20.5 ( a ) 21.1 20.5 ( a ) 20.4 21.6 ( a ) 25.4 21.6 ( a ) 15.6 25.8 ( b ) 1 30.5 I 30 ( c ) 11.9 25.0 (e)
1
i ~
~
According t o Adam (1): (a) value for long-chain homolog; (b) value for p-alkylacetanilides; ( e ) value for p-alkyl compound.
adsorption. The parallelism in form is accompanied by a parallelism in reason for multimolecular adsorption; in both cases condensed phases exist in which the concentrations of the solute (or gas component) are much greater than in the media but' in n-hich the chemical potentials, for concentrations approaching saturation, differ but' slightly from those in the media. This means that very 3 This laboratory has published many studies of adsorption from binary liquid solutions over the entire concentration range. I n all systems investigated t h e quantity HAx;rrt \YLS fouiid t o lie :m S-shaped function of x, having positive values over part of the concentratiori r a n g e :ind negative values over the remaining part. To the best of the authors' lmx\-ledge the iimjority of comparable studies in other laboratories have produced ~0111par;~blc>results. These results force a conclusion t h a t any limiting structure of the adsorbed film involves very appreciable quantities of both components, a t least for systems of the type studied. I t would appear strange, therefore, t o assuiiie n-ithout verification t h a t i n the presence oi a binor . solution a n adsorbed film n-ould tend toward a composition of pure solute, >-et this iissu iptiori is basic t o most techniques for the determination of surface areas b>-adsorption froiii binary solution. It is scarcely surprising t h a t such techniques lead t o areas appreciably lolver than those determined by nitrogen adsorption.
779
MULTIJIOLECULAR ADSORPTIOS
little work is needed to accomplish this “phase transition,” and that small forces, such as forces from the solid surface masked by a layer or small number of layers of molecules already adsorbed, may n-ell suffice to do this work. POINTS 0
’”1
CARBON A
N -CAPROIC N-VALERIC
I
0.2
0.4
0.6
0.8
1.0
0
0.2
0.4
0.6
0.8
1.0
0
0.2
0.4
0.6
08
1.0
0 m
2.0 I .5
1.0
0.5 0
REDUCED CONCENTRATION, C/Co
FIG.1. Adsorption of n-caproic a n d n-valeric acids
Figures 1-3 show that the experimental data are well represented by Brunauer-Emmett-Teller curves over reduced concentration ranges a t least as extensive as the reduced pressure ranges over which these curves have been found t o represent gas adsorption data. Table 10 shows that except for carbon
780
R . S. H I S S E S , T. FU A S D F. E . BARTELL
F, and for the adsorption of aniline by the graphites, the value of n (*‘limiting number of adsorption layers”) is greater than 1. Adsorption of all solutes by 6.E.T. CURVES ----------------
Po!NTSN- A M Y L ALCOHOL 0 N - B U T Y L ALCOHOL
00.4
6 F
9
a
K
-
I
”
02
0
E
2 -. W J
0
o.i!?ktT9 02
00
O 0
02
04
06
08
IO
0.4
0.6
0I 8
IO
0.5
z 2 0.4 -
I
-- IE 0 3 0.2 i= a rn $ 0 1
I 1
(1
0.5(,
n 4
2
0 0
0 0.2
0.4
0.6
0.8
w
~
ID
OI
CARBON H I
I
I
0.4
0.3 0.2 0.1
0
0
0.2
0.4
0.8 1.0 REDUCED CONCE1NTRATION C/Co
0.6
I’IC il .Idsorption of n-aniyl and n-but!l alcohols
carbon F (sugar charcoal) appears to be reasonably n-ell represented by Langmuir isotherms for all equilibrium concentrations observed. Let us consider the possible physical significance of the constants obtained from the Brunauer-Emmett-Teller curves representing the experimental data.
78 1
MULTIMOLECULAR ADSORPTION
( a ) The constant 11: There does not appear to be a reasonable interpretation of the values obtained in terms of pore sizes, i.e., n does not appear to represent a limiting number of adsorption layers determined geometrically from capillary POINTS o 0
I
ANILINE CYCLOHEXANOL PHENOL
b o o
I
I
B.E T CURVES ---------------------
0'4rrFJxl
0.3 0.2
0.I
x< $
016 0
02
04
06
08
IO
5 w
0
02
04
06
OR
1.0
I
0 0
0.2
0.4
0.6
1.0
0.8
LT
CARBON G
04
. 0 .a
f4. ! I
i
0.2
0.6
0 I0
0
0.4
n
----
i 0.8 1.0 0 0.2 REDUCED C O N C E N T R A T I O N C/C,
0.4
0.6
0.8
1.0
FIG.3. Adsorption of aniline, cyclohesanol, and phenol
liameters in the adsorbents and dimensional parameters of the adsorbed moled e s . First, the carbon blacks are linon-n to be of very low porosity (electron 2icroscope values for surface areas agree n-ell with gas adsorption values) and
iS2
13.
S. H.ISSES, T. FU AXD F. E . BARTELL
the graphites are believed to be of the same character. Yet the values of obtained correspond t o a relatively small number of adsorption layers. Second, n-ith the exception of the sugar charcoal, carbon F, values of n appear to be characteristic of the solute rather than of the adsorbent. Third, values of n founcl to apply to the adsorption of alcohols were less than those applying to the adsorption of acids, molecule> of 11-hich n-ere as large as or larger than the alcohol molecules studied. I t may I J ~mentioned in passing that there does not appear to be much data substantiating ;I physical significance for the constant n in the gaseous analog for TI hich the Brunauer-Emmett-Teller equations were derived; in this connection it is surpri+ing that carbon F should be able t o adsorb very appreciable quantities of tile large organic molecules studied, yet restrict adsorption of the small nitrogen molecules t o one layer. ( b ) The constant ( ~ ’ / m ) Table ~: 12 phons the area each solute molecule would occupy on each adsorbent, if (s’ n ~ ) millimoles ),~ of that solute per gram of adsorbent were distributed uniformly in a monolayer on the solid surface. If, for a given solute, this area is nearly constant for all adsorbents, the quantity (z’lm),,,is proportional to the specific surface area, and a technique for the measurement of surface area may be based on it. The results show that for a given solute the areas per molecule corresponding to (z‘lm), are in fact nearly constant for the three carbons of lon-ebt area, carbons B, E, and G, but as a rule are systematically higher for the carbons of intermediate area, carbons A and H, and the high-area carbon F. The last three columns show the extent to which this deviation is systematic, the molecular areas being based on areas 27, 12, and 18 per cent lower than the measured areas of carbons F, A, and H, respecti~ely.~ It is seen that, for a given solute, the areas in the last six columns agree quite ne11 among themselves, n-ith a few minor exceptions. The deviations in the values for aniline will be explained later. The authors are unable t o explain the exceptionally high molecular area of cyclohesanol on sugar charcoal. The deviation of 25 per cent betveen molecular areas calculated for n-butyl alcohol on carbons X and H by use of equations nith n = 2 and n = 3, both of n-hich represent the observed data fairly well, indicates a degree of arbitrariness in values cf (.c’ m)nLn h i c h is not a t all satisfying. In short, the results indicate that the specific surface area is a major determinant of ( z ’ / ~ ) ~but , also indicate that the molecular areas corresponding to (.E’ m),>L are substantially greater than those corresponding to any close-packed film of pure solute, indicating the presence of folvent in a completed monolayer if the present use of the B.E.T. model is correct. Khile the results with the present adsorbents indicate the possibility of basing a relatice method of surface area determination on (x’/m),, it is quite likely that the relative amounts of solute and solvent in the complete monolayer, n-hich appear to be nearly the same for the various carbons used in
-
This might Le called a “porosity correction” and justified by arguing t h a t these adsorbents might contain pores of such a size as t o admit nitrogen molecules but not the larger molecules of the organic compounds. As previously mentioned, hoxyever, the grnphites and carbon blacks, of TT hich carbons -1and H are representative, are supposed t o be non-porous.
783
MULTIMOLECCLhR ADSORPTIOX
the present work, might have differed u-idely if results for the carbons had been compared Irith results obtained with an adsorbent such as silica gel, or even a highly hydrophilic carbon. Thus, since both solute and solvent appear to be present in a completed monolayer, the value (s'lm), probably depends to some extent on intensive parameters of the solid surface as n-ell as its area. Langmuir 68) deduced from surfac: tension measurements limiting molecular areas of 34 .k2for phenol and 37 A.2 for aniline a t the aqueous solution-air interface, TF hich he interpreted as indicating close-packed films of these molecules with their planes in the plane of the interface. These values agree well with those cited in table 11 (omitting values of aniline for the graphites). Such a model would also be consistent with the areas found for cyclohexanol. If these models are correct, it is surprising that the solids attract both polar and TABLE 12 Areas p e r solute molecztle, in square .;ngstroms, I
corresponding t o ( x ' l m ) ,
ADSORBEKT, C A R B O S
I
Empirically modified area'
Low area
SOLCTE
!
G
I-
n-Caproic acid . . . . . . . . . . . n-1-aleric a c i d . . n-Butyl alcohol. . . . . . . . . . .~ n-Butyl alcohol.. . . . . . . . . . I Aniline. . . . . . . . . . . . . . . . . .. , ................
Cyclohesanol. . . . . . . . . . . . .~
54
~
51 , 54 ' (41) I 19 ' 44 38 44 4s 51
44
i (41) 49 49 93
I
~
~
~
.
i
20 37 44
~
i ~
.
43
1
i
16 32 42.5
~
44
35 35 41
*Based on specific surface areits of 580,84,and 102 sq. meters per gram for carbons F, A, arid H, respectively, instead of measured surface areas.
non-polar portions of the organic molecules but do not attract any appreciable quantity of water. For each adsorbent, with the possible exception of carbon F, the isotherms :or the adsorption of n-caproic acid and n-valeric acid from aqueous solution i t 25OC. appear to be congruent functions of the reduced concentration within ,he limit of experimental error. Similar apparent congruencies exist between iniline and phenol isotherms for carbons F, G, and H and for the n-amyl alcohol tnd n-butyl alcohol isotherms for carbons B, E, and G, and possibly A and H. longruencies between isotherms for the adsorption of two different solutes by he same adsorbent imply three conditions: ( 1 ) That the solute molecules are similarly attracted and occupy similar reas on the solid surface, or in short that they are similarly oriented, and ossess a group or groups for n-hich the solid exerts the same attraction. Thus, ' the solute molecules were aniline and phenol, one would conclude that the ttractjve forces of the solid n-ere exerted primarily on the phenyl group common
784
R.
S. HASSES, T. F U A S D F. E . BARTELL
to both, and that the amino and hydroxy groups v-ere either not in contact u-ith the surface, or that their areas vere nearly the same and that the attractions of the solid for these tn-o groups were either very similar or negligible compared t o the attraction for the phenyl group; (a) That as the adsorbed films develop in the tn-o cases the solute molecules tend to become surrounded by the same number of water molecules; ( 3 ) That the ~ o r k required s of the adsorbent forces to remove molecules of the two different solutes from their aqueous solutions and transport them to the adsorbed films are practically the same nhen the reduced concentrations of their aqueous solutions are the same. If the adsorbed film were pure solute, considerations of the sort given by Butler, Thomson, and Slaclennan (5) shov that this is true to good approximation, and in general it ~1-ouldbe true if the activities of the solutes In the two films (considered apart from the solid surface), referred to the standard state of pure w l i i t P . nere equal. The three conditions cited above are sufficient to explain the observed data. It is conceivable, but does not appear likely, that the systems could actually deviate from all three conditions in such a \my as to achieve the same results, so that the three conditions would not he rigorously necessary. Such congruencies may be recognized as related t o onc form of Traube’s rule; a somewhat different explanation of this phenomenon given by Langmuir (8) makes use of the first point in the above discussion, but supposes the orientation of the molecules to change as their concentration in the adsorbed film increases, so that the molecules “lie down” at lov concentrations and “stand up” a t high concentrations. The data for the adsorption of aniline by the graphites indicate that the mechanism of this adsorption is very different from that for the adsorption of the same solute by the remaining adsorbents, and of phenol by all adsorbents studied. This suggests that the graphite surfaces may have been appreciably oxidized and become acidic in character, and that they strongly attracted the amino group of the aniline. If this is correct the aniline molecules should be adsorbed by the graphites with their planes normal to the adsorbent surface, and should be very closely packed if the attraction is strong. If the quantities (2m), obtained for the adsorption of aniline by graphites are physically significant, the areasoper molecule in a complete monolayer derived from these values, about 18 A2,indicate that this packing is very close, and that there must be little if any -water not displaced from the surface. SUJIAIARY
1. ,idsorption of a series of slightly soluble organic compounds from aqueous solution by sugar charcoal and a series of graphites and carbon blacks n-as investigated interferometrically. The observed adsorption results for a given adsorbent and solute Ivere plotted against reduced concentration (concentration divided by saturation concentration). 2 . I n every case adsorption of the solute by graphites and carbon blacks ivas demonstrably multimolecular, and the experimental results suggest that this
ABSORPTIOS LISES AND ACTIVATIO?; ENERGIES
785
type of adsorption may be expected from binary liquid solution whenever the adsorbent is of low porosity and the solute of limited solubility. 3. The Brunauer-Emmett-Teller equation was found to represent multimolecular adsorption from binary liquid solution as well as it represents such adsorption in the gaseous analog. 4. The constants n and (z’lm), obtained from the Brunauer-EmmettTeller equations representing the data were examined for physical significance. The data did not support a physical significance for the constant n. The specific surface area appears t o be a major determinant of the constant ( X ’ / W L ) ~ , but use of this quantity for determination of surface area should be adopted n-ith caution, for if interpreted as actually corresponding to a complete monolayer this monolayer must contain solvent as well as solute molecules, and the ratio may vary with varying solid surface compositions. 5 . For each adsorbent, with the possible exception of sugar charcoal, the adsorptions of n-caproic acid and n-valeric acid were congruent functions of the reduced concentration. Adsorptions of aniline and phenol by sugar charcoal and the carbon blacks n-ere similarly congruent, but adsorptions of aniline and phenol by the graphites were not congruent, and the isotherms differed markedly in form. REFERESCES (1) - 4 D A I f : The Physics and Chemistry of Surfaces, 2nd edition. Clarendon Press, Oxford (1938). (2) BARTELL A S D AfILLER: J. Ani, Chem. S O C . 44, 1866 (1922). (3) BARTELL ASD S L O A NJ. : Am. Chem. Soc. 51, 1637 (1929). (4) BRUSAUER, EUIIETT, AND T E L L E R :J . h m . Chem. SoC. 60,309 (1938) (5) BUTLER,THOJISOS, ASD ~IACLEXXAS: J. Chem. soc. 1933,674. (6) Fu, HAXSES,AND BARTELL: J . Phys. Colloid Chem. 52, 374 (1948). ( 7 ) JOYNER, KEINBERGER, AND ~ I O N T G O XJE. R Am. Y :Chem. SOC.67, 2182 (1946). (8) LANGXKIR: J. Am. Chem. Soc. 39, 1848 (1917). \Yo., AND DE I Z A G U I R R Kolloid-Z. E: 30, 279 (1922). (9) OSTWALD,
A LOGARITHMIC FUNCTION EVIDENCIKG A CORRELATIOK BETWEEN ABSORPTION LINES AND ACTIVATION E S E R G I E S DAVID S. IIAHN 149 South 8th East St., Salt Lake City, Utah
Received September 6 , 1948
The techniques of physical chemistry provide several ways of representing the elative energy of a chemical bond in a homologous system of compounds. Three uch variables are used here: dipole moments, atomic refractions, and force onstants. In the context of this paper, dipole moments will be correlated irith uantized discontinuities. However, for homologous compounds, it is to be oticed that there is a linear relationship betn-een the respective force constants