Feb., 1963
ISOSTERIC HE.4TS O F
ADSORPTIOX OF ISOBUTYLESE ox ACTIVATED ALUMIXA
2, we can conclude that C(X) probably lies within the rather large limits given in the previous paper; i e . , between 0.07 and 0.3. Thus, if b is reduced from its initial value of 5.0 to 2 a t the gel point, the weight average molccular weight would have been reduced by degradation a t least to two fifths of its initial value. This would cause G(X) to be 2.5-fold greater than 0.07, or about 0.18. The limits of G(S) also can be w i t t e n as about] 1.2 for its initial value, dropping to 0.27 for doses above the gel point. Acknowledgment.-This research was supported by
3w
the U. S. Air Forcc under Contract No. AF 33(616)-5826 monitored by the Aeronautical Research Laboratory, Wright Air Development Center, and by thc Advancrd Research Projects Agency of the Department of Defense, through Northwestern University Materials Research Centcr. The authors are iiidebted to Dr. Nlitio Iriokuti for the hitherto unpublished eq. 3 of this paper and eq. 10 of thc previous paper, and to Dr. J. H. Elliott and colleagues of the Hercules Powder Company for the samples of polypropylene used in this research and information appcrtainiiig thereto.
ISOSTERIC HEATS OF ,WSOlLI'TIOS 01.' ISOHUTYLESE OS ACTIVATED ALURIISA BY It. D. OLI)E;A-K.~MI' AND G. HOUGHTOS Chemical Engineering Department, Division of hhgzneering Research, University of Pittsburgh, Pittsburgh, Pennsylvania Received J u l y 16, 1968 Adsorption isotherms for isobutylene on activated alumina have been meaaured at temperatures in the range 20-80' and pressures of 10-900 Inm. The surface area calculated from nitrogen adsorption a t - 196' is a factor of 1.58 gre:rter than t h a t obtained from isobutylene adsorption a t 20", which was 1% m 2 / g . Isosteric heats of adsorption for isobutylene on alumina have been calculated a t various coverages in the range 0.06-1.10. A review of existing literature on physical adsorption in comparison with the present nieasurenienta shows that the observed shapes for heat nf adsorption-coverage rurves follow a general pattern consistent with the magnitude of the surface area and the related concentration of surface defects.
and sulfides, relatively lit'tle appears to have been done Introduction on physical adsorption where t,he temperatures are low In the elucidation of the nature of the physical intcrand the heats of adsorption are less than about. 10 kcal./ action between a gaseous adsorbate and a solid adg.-mole. However, ltyerson and Cines30 have shown sorbent, four cases are of interest,: (1) non-polar gases on non-ionic surfaces such as 0 2 , 1 N2,1-4incrt. gase~,~-Y the heat of adsorption of propylene on silica gel to be 6.494.7-1 kcal./g.-mole in the temperature range and hydrocarhons'0-12 on various forms of carbon, -25--35', while Davis, DeWitt, and have another example being the adsorption of incrt gases13 measured adsorption isotherms for 1-butene on A1203 on hloS2; (2) non-polar gases on ionic surfaces such as 0 2 , X2, and inert gases on KCl,''1915C S I , ' ~ and ~ ~ ~ and silica gel and also CHClJ? on ZnO and glass spheres. Crawford and ' I ' o m p k i n ~ ~have ~ , ~ ~ dcterTi0217-22., (3) polar gases on non-ionic surfaces like C2H&l,23-25KH 3, 23,26t27 H20,Z6CH30H,2* CH3NH2,27 mined the isosteric heats of adsorption of SO*, YHR, C02, and NzO on BaF2 and CaF2; they were below 10 and on various forms of carbon; and (4) polar kcal./g.-mole and decreased uniformly with increasing gases on ionic adsorbents. Although considerable coverage, indicat.ing surface non-uniformity. ISETQ, attention has been paid to chemiso~ption~~ on oxides however, appeared to react wit,h CalTs. (1) R. A. Heebe, .J. 13iscoe. W.T. Smith, and C. 13. Wendell, J. Am. Chena. The purpose of the present work was to determine the Soc., 69, 95 (1947). (2) T. L. IIill. P. H. ICinniett, and I,. C. .Joyner. ibid.,73, 5102 (1951). amounts and heats of adsorption of a polar gas, iso(3) R. A. Ueebe, 1%.Millard, and .J. Cyanarski, ibid., 76, 839 (1953). butylene, on a high area ionic solid, activated alumina, (4) E. L. Pace arid A. R. Siebert, J. Phys. Chem., 64, 961 (1960). to ascertain the naturc of the interaction. From tJhe ( 5 ) D. AI. Young. T r a n s . R'araday Sac.. 48, 1164 (1952). (6) A. 1). Crowell a n d D. M. Young, ibid., 49, 1081 (1953). dipole moments of analogous unsaturate^,^^ propylene (7) R. A. Reebo and I). 11. Young, J. I'hus. Chem., 68, 93 (1054). (0.35 D.) and isoprene (0.38 D.), the moment of iso( 8 ) J. Greyson a n d .J. Q. Aston, ibid., 61, 610, 613 (1957). butylene should be about 0.37 D. Although the struc(9) R. .J. Robka, R. E. Dininny, A. R. Siobort, a n d E. L. Pace, ibid., 61, 1646 (1957). ture of at:t'ivat.edalumina is unknown, and indeed there (10) W. I). Shaeffer, hf. 1-1.Polley. a n d W.R. Smith. ibid., 64, 227 (1950). may be no definite st'ruct,ure, most evidence indicates (11) J. W. Ross a n d R. J. Good, {bid., 60, 1167 (1856). (12) A. A. Isirikyan a n d A. W.IGselev, ihid., 66, 210 (1982). that, the disordered lattice is made up of A13+,0 2 - , and (13) P. Cannon, ibid., 64, 1285 (1960). OH- ions with some interstitial wat~r.3503~ C. Orr, I'roc. Roy. Soc. (London), A173, 349 (l9S9). Experimental (15) D. hI. Young, T r a n s . 1"araday Soc., 48, 548 (1952). (16) F. C. Tonipkins and I). .\I. Young, ibid., 47, 77 (1951). (17) IC. Kinnton, R. A. h e b e , .\I. if. Polloy, a n d W. R. Sinitlr, J. Am. Chem. Soc., 73, 1775 (1950). (18) J. M. IIonin and 1,. 11. Iteyerson. J. Phys. Chem., 56, 140 (1952). (19) I,. E. Drain and J . A. Morrison. T r a n s . Faraday Soc., 48, 316 (19Z2). (20) J. H. Singleton a n d G. I). Ifalsey, J . Phys. Chem., 68, a30 (1954). (21) W. A. Steele. ibid., 61, 1551 (1057). (22) W. A. Steele a n d -J. 0. Aston. J. A m . Chem. Soc., 79, 2393 (1957). (23) C. Pierce a n d R. N. Smith, J . Phys. Chem., 6a, 1111, 1115 (1948). (24) J. Mnoi. C. Pierce, a n d K. N. Sinitti. ibid.. 67, 657 (19Xj). (25) C. Pierce and R. N. Sinitb, J. Am. Chsm. Soc.. 75, 816 (1953). (26) C. Pierce and R. N. Smith, .I. I'hys. Chem., 64, 35.4, 705 (1950). (271 R. M. Dell a n d R. .4. Heebs, ibid., 69. 746. 754 (1!155). (28) B. Millard, R. A. I%ecbe.a n d .I. Cyanarski. ibid., 5S, 468 (1954).
The apptratus used to iiiewtire the Ltdsorption of isobutylenc (~-niet,hjlproi,cric:)on idriminit was essentially the sitrno its that (29) " C h c ~ ~ i i s o ~ p t i o nISdited ," by W. IC. Garner. Acadnrnic Press, lnc.. New York, N. Y.. 1957. (30) L. 11. Ryerson and .\I. It. Cines. J . Phys. Chem., 46, 1066 (1942). (81) 11. T. I h v i s . T. W.DeWitt, and P. €1. E m m e t t , ibid., 51, 1232 (1947). Crawfiird unii I:. C. Tompkins, T r a n s . Faraday Soc., 44, 698 (1948); ('rawfiird and 1'. C. l'onigkins, ihid.. 46, 504 (1950). (81) K. 13. Hannar a n d C . 1'. Smith. J . Am. Chem. Sor.. 46, 1931 (1943). ( 3 5 ) E. 13. Ccirneliiis, T. H. Millibean. G. A . Mills, and A. G . Oblad, J. I'hys. Chem.. 66, 641)(1961). (36) J . 1%. I'erri and It. 13. IIannun, ibid., 64, 1526 (1960).
k.D. OLDENKAVP A S D G. IIOUGUTON
304
mcnts. The liqucfird isobutylrno could be vaporized into the volumctric system as describrd earli~r.3~ The same cquipmcnt could tic usrd t o mcasurc the surface area of thc powder by nitrogcn adsorption at liquid nitrogen temperatures. Thc nitrogen for adsorption and t h r helium for dead space determinations were purified by passing thrm over hot copper a t 300' followcd by Dricrite. Adsorption temprratures were measured with an oxygen vapor prctssure thcrmoinctcr.
8
6
4
2
2
M
d o -
VOl. 07
0
50
0
200
150
100
400
600
200
800
Preseure, p , mm
Fig. 1.-Adsorption isotherms for isobutylene (Zmethylpropene) on activated alumina: (a) volume adsorbed (STP) at 0-200 mm.; (b) volurne adsorbed (STI') at 0-900 mm. describcd by IIoughton, et al.,37 for dctcrmining the solubilities of isobutylene in dinonyl phthalate, it being nccrssury only to replace the solubility bulb by one containing i~ wrighed :mount of alumina. I n order to measure the adsorption a t prcssurr's in the range 10-100 mm., a manometer containing glyccrol was incorporated into the system such that one ami measurcd thr pressure in thr adsorption t)ulb ~ l n dthc other wits continuously evacuatrd to about 10-3 mm. IMow I00 mm., i t was ncccssary to determine the dcad space of t h r sample bulb with helimn a t each temperature and prrssure. Abovc 100 mm. the glycrrol manomrtrr could be isolated by a stopcock and the pressurrs mt.ttyurrd on a mercury nianomrter. Siiicc the mercury manometer was s e p a r a t d from the adsorption hilt> by an inclincdtube bdancing mariomrtrbr of constant volurnr, as desrribed p r e v i o i ~ s l y it , ~ ~was possible to ineasiirr :I complete isothwm by ~ prcssuro skirting around 100 nim. and eomI)rc&ng thr g : in increments of about 30 mm. to cnver thr ritngc 100-900 mm. Thc two stopcocks in thr gas volunictric system nrrc lubricatrd with a glycerol-bentoiiitc grritw; g l y r c d has a low vapor prrssure ( < l O - 3 mm. nt 23') and dow nnt d(%cctably absorb hydrocarbons. Isotherms wrm rnciwircd at iritervals of 10' in the range 20-80", the tcmpcmturc. of thv adsorption bulb being maintaincd and i d j u s t d t o within i0.05' liy watcr circulating around it from a constant tcmprraturc bath. Tho pressures in both ranges were obtained within d ~ 0 . 0 5mm. and the volumc$swithin zt0.1 nil. Thc atisorhcnt wits 28-45 mrsh Alcoa F1 activ:bted alumina rated to havr a pore vol;ime of 0.25 in1 / g . and an 2tveragc pore diametrr of abtmt 40 A. 13efore wrighing, w c h sample was hratcd t o 190-200" for 2 hr. itnd cool(sd in a dwiccator. Aftrr weighing into the samplc hilt), the alrirnina was oritg:tssc.d at 1!)0-200° at about 1 0 - 3 mm. b2fore the adsorption measurcmcnts wrrc st:irtd. Samp1t.s rrwrighd :it the rnd of a serics of idsorption rxprrimcnts showed no significant clittnges in weight. Thr weight of the sample WRS actjustrd to givc :L precision of a t Irast 1 in rnrasuring thc volume adsort)cd a t c w h pressure and tcmprmture. The isohutylcnr was RIathwon C.P. grade (D9.0~ominimum purity) distillrd nt low tmnpcrature under vaciium to obtain the middle-third fr:tction used in the cxperi(37) G . IIougliton, A . S. Kesten, ,J. E. Funk, and J. Coull, J . P h y s . Chem.. 66, 649 (Igf31).
Results Figure 1 shows the adsorption isothcrms for isobutylene on alumina as volume adsorbed per gram, V , at STP us. pressure, p , in the range 10-900 mm. for temperatures of 20, 30, 40, 50, GO, 70, and 80'. Since the data below 100 mm. determined nith the glycerol manometer have the same percentage accuracy as those above 100 mm., they hare bcen plotted on an expanded scale in l'ig. l a . The data in I'ig. l a have also bcen plotted to 200 mm. in order to show that the data for the glycerol and mercury manometers and different sample weights join up remarkably well, indicating thc inhrrent accuracy of the cxperimcntal technique as well as the rcproducibility of the surfacc of the activated alumina. Furthcrmorc, no hysteresis efferts were detected and adsorption equilibrium was reached very quickly, contrary to the observations of Davis, DeWitt, and Emmett3I for butrne-1 on alumina. Howewr, the present measurements w r e restrictrd to p / p o < 0.4, whereas the adsorption-desorption hysteresis effects of Davis, ct al., mere observed for p / p o > 0.5, no t.wh hystcrcsis effects being notrd for p / p ( ]< 0.5, although they did rcport that escccdingly long equilibration times were required for thc butene-1-alumina systcm. The shape of the adsorption isotherms in Fig. 1 shows that they are either of Type I1 or Type IV, the range of p / p o not being close enough to unity to distinguish betwcn thew two possibilit ips. Since Coffin and ?tlaassz5give the vapor pressurcs of isobutylene from -80 t o 22' and the liquid densities from -40 to 11.3", it is possiblc to USC the adsorption isotherm a t '20' to determine thc monolayer volume, Vm, and hence the surfacc area by the BET method. A plot of p,'V(po - p ) us. p'p" using po = 1017 mm. at 20" gave a good straight line for 0.16 < p / p O< 0.38, the slope and thc intercept yiclding V,, = 16.2 ml./g. and C = 13.6. Using 0.6058 g./ml. as the density of liquid isobutylene at 11.5' togrther with a icmpcrature cocfi&nt of density of -0.00116 g./ml.-dcg. in the rangc 0-11' from the data of Coffin and !Alaass,J8the density a t 20" is estimated to be 0.39(i g./ml., yielding a crosssection of thr isobutylene molecule of 29.0 x vm.?from which V,,yields a surface area for thc alumina of 126 m.z/g. The low temperature nitrogen adsorption data a t -1106' also gave a straight line BET plot for 0.07 < p / p o < 0.4, yielding V , = 45.7 ml./g. and C = 21.6, so that with a cross-s ion of 16.2 X for the nitrogen molecule thc surfacc area is 199 m.2/g. h Harkins-,Jura plot of log ( p / p " ) US. 1/V2 for thc same nitrogen data gave an arca of 223 m.2,'g., \vel1 within the agrecment expected bctwcen the two mcthods. Hoviever, the ratio of the nitrogen area to the isobutylene area, both by the 131W method, is 1.58. Davis, et u L . , ~ found ~ a similar discwpancy in mcasuring the surface area of alumina Jvith 1-butene at O', xhcre the same ratio was 1.42, similar dcviations being also obscrved for K r , ?z-C4HI0,and CHCIJ?. One rcason (38) C. C. Coffin and 0 hIaaau, Trans. R o y . Sac. (Canada), 21, 33 (1927).
Feb., 1963
ISOSTERIC HEATSOF ADSORPTI~S O F ISOBUTYLENE ON ACTIVATED ALUMIXA
305
given3l for the smaller area with these adsorbates was that the solid surface may have an influence on the packing of the molecules. In thc present case it seems reasonable that if the solid has a large number of edges, corners, pits, or pores, then fetver larger molccules of isobutylcne can pack a t these defects than the smaller nitrogen molec ulc s. The isosterir hcats of adsorption, piso,for isobutylcne on alumina wcrc obtained by plotting log p us. 1/11 a t various constant values of V in the range 1-18 ml./g. using thc experimcntal data of Fig. 1, the slopes of the lincs bcing qlso/2.303R. These plots did not exhibit any dctectablr curvature for the temperature range 20-80’, indicating that the Clausius-Clapeyron equation was applicable. Ilowever, it is to be ~ x p e c t e d ~ ~ ~ ~ ~ that thc Clausius-Clapeyron equation would apply a t the low p / p n ratios usrd in the present expcrirnents. In E’ig. 2 the isosteric heats, qlso, arc plottcd against I V/V,,,usingV, = 16.2 ml./g., yielding a sigmoidal curve. 4 Thc mcan heat of vaporization, AHL,for the temperature 0 0.5 1.0 range 20-80’ was estimated to be 4.5 f 0.2 kcal./g.v/vm. Fie. 2.-Isosteric heats of adsomtion of isobutvlene ( 2mole from the sloprs of thc isobutylcne vapor pressure metl&lpropene) on activated alumina’as a function of -coverage. data of Coffin and Maasss using the method of Watson39 to cxtrapolatc the latent heats beyond 22’. At grain boundaries, pores, roughnesses, and depres2.5’ Rossini, et u ~ . , ~ Ortport the latent hcat scparatcly as ~ i o n s . 3 , 7 , 8 , 1 0 , 1 2 , 1 4 - 1 ~ , 1 8 , 2 2 , ? 4 , 2 5 , 2 7 I-Iowever, a maximum 4.9 kcal ./g.-mole. still exists on the heat of adsorption-coverage plots a t Discussion the same coverage, usually in the range 0.7-1.0. I n the case of a rarbon adsorbent, heterogeneity can be Although thcrc is available a large amount of heat of reduced by graphitizing a t progressively higher temadsorption data in thc literature‘-33 on systems of peratures. However, physical hetcrogcncity on thc various types, littlc appears to have been done to evalusame adsorbent depends upon the nature of the adate thew data to determine features and trcnds that are 1 2 ~ ~ 8In general, when heterogeneity is present, in common. ITowver, inspection of this information the initial heats approaching zero coverage are roughly indicatcs a gcneral consistency in the behavior of the heat of adsorption-coverage plots such that a trend in twice the latent heat of vaporization of the adsorbate. shape with incarcase in surfacc area can bc discerned Pre-adsorption of gases has been obscrved to reduce surthat appears to be more or lcss indcpendent of the parface non-uniformity for the adsorption of a second gas ticular polar-non-polar adsorbcnt-adsorbate combinaon high area powders“ but apparently not for lorn area powdcrs.*0?*1 tion undcr consideration. The isotherms shown in As the arca, and prcsumably the surface defect conFig. 3a-d covcr almost all the shapes reportcd in thc litcraturc. centration, increases, the maximum due to lateral interFigurc 3a illilstratcs thc idcal case of adsorption on a action gradually disappears, bcing rcplaccd by a point of inflection or horizontal region 1, 3, 12,14,1y,21.p2, 27, as in homogcncous surface where, a t low coverages, the heat of adsorption is rclativcly constant, it,s absolute value 2 2 , * 3 , 3 2 , 3 3 as in Fig, 3c or disappearing entirelys,10,’2117,18 being charwterized by the nature of the gas-solid interFig. 3d. Recent measuremcnts12 of the hcat of adact ion which may involve dispersion forces alone, sorption of bcnzene on graphitized carbon have shown c.harge-induc~eddipolc intrractions, chargedipole interthat shapes of thc types shown in Fig. 3d, c, and b arc produced in this order as the temperature of graphitizaactions, or combinations of these possibilitics. As thc covpragc, 8, increases, the adsorbcd specics first cxpcrition is increased from 900 to 3100’, indicating that anence long-range attractivc forces, causing the heat of ncaling a t high temperaturcs produces a more hoadsorption to rise, followed by short-range repulsive mogeneous surfacc for adsorption. Conditionswhcre the forces a t cven larger coveragrs that cause the heat to heat of adsorption does not vary much with covcrage, decrcasc very rapidly. As the adsorbate builds up a as in the horizontal portion of Fig. 3c, have been exmultilayer, the hcat of adsorption approaches the laplained by adsorption in patchesYJ7or by interaction tent heat of vaporization of thc liquid adsorbate. betwen adsorbate m o l e d c s on opposite sides of Itoughly idcal isotherms have hccn obscrvcd experiporcs and ( ~ a ( b k s . ~ JIn J ~ one instance whcre a relamentally in a number of cases?~8~9~1~~23~?7~?8 where the tively non-porous carbon substrate didn ot exhibit a substratr is highly crystalline and has a low area. maximum in hcat as in E’ig. 3a and b, it was postulated As the surface area increases, a sharp initial drop that the fall-off in heat might be due to lateral repulsion in hrat of adsorption a t low coverages is often in patches of adsorbate formed in shallow narrow defound, as in I’ig. 3b, that is usually attributed to pressions a t grain b o u n d a r i e ~ . ~An ~ ~intcresting case is non-uniformitiw in thc surfacc, such as cracks, that of Ar on KCl, where a curvc of thc type shown in Fig. :3b was observed when the (100) planes were prc(39) I< Wntuon, Ittd J;ng. Chem., 86, 398 ( l S 4 3 ) (40) I‘’ I) Kossini K. S. Pltrrr, It. I, Arnett, It 11. I 0.76 Fig. 2 s h o w that the heat of adsorption of isobutylene on alumina is very slowly approaching the latent heat of vaporization of liquid isobutylene. That the approach to the liquid state is quite slow probably is due to the presence of the solid surface, whose influence will only be completely attenuated by several layers of molecules. In conclusion, it should be stated that the use of higharea powders of undefined defect structure makes theoretical calculations concerning the interaction between adsorbate and adsorbent quite difficult. However, it would be worthwhile to investigate such systems further since eventually, if the defect concentration becomes large enough, it might be possible to create a perfectly “random” surface, as opposed to an orderly crystalline surface, such that interaction calculations might then be facilitated by probability methods, as in the molecular theory of gases and liquids. Acknowledgment.-It. D. 0. is indebted to the Westinghouse Corporation for a Graduate Fellowship.