HEATS OF IMMERSION. IV. THE ALUJIIKA-WATER SYSTEMVARIATIOKS WITH PARTICLE SIZE AND OUTGASSIXG TEMPERATURE BY TIT. H.
\VADE AND r\l'ORnfAN
IIACKERMA4N
I>epnrtrtient of Chemistry, The University of Texns, Ausiin, Texas Received March 7,1960
Samples of a-,y- and amorphous-alumina with surface areas from 0.222 to 221 m.2/g. were studied following vacuum outgassing treatments over the temperature range, 100-450". In addition, weight loss measurements were made as a function of temperature. Usually a substantial increase was noted for the heats of immersion with decreasing specific area and increasing outgassing temperature. These studies are consistent with the normal alumina surface being covered with a maximum of 19 OK's/100 A.2.
Introduction
It has been shown in previous communications from this Laboratory's2 that there is a large decrease in heat of immersion per crns2(AH1)with increased specific area of silica samples on immersion in water. Moreover, most silica samples have AH, variations with outgassing temperature which are interpretable in terms of the normal existence on silica surfaces of both physically (reversibly) adsorbed water molecules and chemically bonded surface hydroxyl goups. Silica surfaces once &ripped of their surface -OH groups rehydrate at a rate too low to be observed in most calorimeters.'J The belief is generally held by others that t,he coverage of simple oxides with surface hydroxyl groups is probably the rule rather than the exc e p t i o ~ ~ +With specific regard to alumina, the existence of surface hydroxyl groups can be inferred from infrared spectroscopic measurements,6 dye adsorption studies,' high temperature heat of adsorption measurements3 and from the present heat of immersion measurements. The present investigation was designed to compare alumina to silic,a with respect to the variation of the immersional heats with particle size and water content (both physically adsorbed and chemically bonded) of the surface. Furthermore, the effect of cryst,alliiie modification (a- or 7-1 of the alumina substrate on AHi is raised but iinfortunately can be answered only unsatisfactorily. There is only a single published estimate* of the surface energy of alumina-a value of 560 ergs/cm.2 obtained from heat capacity measurements on y-alumina. Experimental Samples.-The eight samples of A1203 studied are listed in Table I along wit,h their purity, specific surface area and cryst,alline modification. The purities and crystalline modifications are those quoted by the manufacturers. ( 1 ) A. C. hfakrides and N. Hackerman, THIBJ O U R N A L , 63, 591 (1959). (2) A ' .' H. Wade. R. L. Every and N. Hackerman, ibid., 64, 3:: (1960). (3) A. G. Oblad. S. W. Weller and G. A. Mills, ibid., 69, 809 (1955). (4) J. H. de Boer, J. J. Steggerda, J. M. H. Fortuin and P. Zwietering, "Second International Congress of Surface Chemistry," Botterworth Scientific Publications, London, 1957. (5) A. C. Zettlemoyer. Chem. Revs., 69, 937 (1959). (A) A. Babrishkin a n d .\. 1'. Uvarov, Dokladu A k a d . Nauk S.S.S.R., 110, 587 (1956). (7) J. H. de Boer, "The Dynamic Character of SdsorpLion," Chapters VI1 a n d V I I I , Oxford University Press. New York, N. Y . . 1953. (8)R. Fricke, F. Niermsnn a n d C. Fiechtner, Ber.. 70B,2318 (1937).
The B.E.T. surface areas were measured by Kr adsorption in a volumetric adsorption apparatus previously described Above 100' surface areas of all samples were independent of outgassing temperature and, hence, water content of the surface. Samples F and H were furnished with the suppliers Nz adsorption areas of 64 and 210 m.a/g., respectively. This agreement with the Kr areas is good and for uniformity only the Kr areas were used. The samples had been prepared by high (1000-1300') temperature calcination of trihydrates and alums with the exception of sample F, prepared by flame hydrolysis, and sample 11, prepared by precipitation of the hydroxide followed by calcination a t a temperature low enough to prevent collapse of the internal pore structure. Samples D and E had been made from B and C by micronization.
TABLE I Sample
Manufacturer's designation
Purity,
Ares (m.z/g.)
%
Cryst. mod.
T-60 99.8 0.222 CY-ALO~ Alucer MC 99.96 2.72 a-&O3 Alucer HS 99.7 3.04 o-AI~O~ Alucer MCB 99.96 3.12 cy-111208 Alucer HSB 99.7 4.56 a-AlzOa Alon C 99.9 65.2 ~AliOo Alucer MA 99.96 109 T-AltOJ F-20 99.0 221 Amorph. Supplied by Aluminum Company of America. Supplied by Gulton Industries, Inc. Special low-chloride Alon C supplied by Mr. Gregor Berstein of Godfrey L. Cabot, Inc. A"
B* Cb Db E* Fc Gb Ha Q
The samples, before immersion, were outgassed a t mm. Hg for 72 hours in Pyrex bulbs and were sealed off on the outgassing apparatus. The outgassing temperatures listed in Table I1 are accurate to *3O.
TABLE I1 HEATSOF IMMERSION (ERGs/cM.~) t ("C.)
A 0.222
B 2.72
C 3.04
Sample m.Z/g.-D E a 3.12 4.56 7-8
100 656 581 370 552 150 724 676 590 200 821 765 399 673 250 838 866 300 847 931 468 875 350 861 1009 400 870 1076 508 1019 450 a Data of Good, et a1.I2
F
65.2
G
109
-H
221
412 384 323 513 436 361 490 401 901 567 A61 453 714 1011 628 514 728 682 581 719 640 712 537 742 682
532 578 G80
Calorimeter.-The calorimeter has been described previous1y.l It is of the twin adiabatic type with thermistor temperature sensing elements. All measurements were mn\le a t 25 i. 0.1'. Two types of Pyrex sample bulbs w r e used in these studies: (A), a thin-wall-spherical bulb which was shattered completely during immersion (measured heat of breakage is 0.1 i 0.05 joule), and (B), a thick-wall(9) M. J. Joncich and N. Hackerman, TIXISJOURKAL,67, 674 (1953).
Sept., 1960
HEATSOF IMMERSIOK FOR ALUMIK'A-WATER SYSTEM
convoluted cylinder only the tip of which was broken2 (measured heat of tip breakage is less than 0.02 joule). Type (B) bulbs were used for the two samples of lowest area and Type ( A ) for the remaining samples. There were four electrical calibrations for each sample immersion. The average deviation for these runs was less than &l%. All values given in Table I1 are the averages of a measurement from each calorimeter. Sample weights varied from 10 to 0.1 gram depending on the sample area. Total temperature changes during immersion varied from 1 X IO+ to 1 X degree. Differential baseline temperature variations were less than 2 X 10-6 degree over the 20 minute periods required for a measurement. Weight Loss Measurements.-Th! samples, after equilibration with water vapor a t 25 i 1 , were heated in air. Weighings were taken a t 50' temperature int.ervals with intervening 24 hour time intervals. Just prior to weighing, the samples were allowed to cool to room temperature in a desiccator. Weighing operations were performed on a semimicro analyt'ical balance of 0.02 mg. readability.
increase of AH1 with increase in outgassing temperature. The former feature will be discussed first.
AHl* for the 4.56 m.*/g. material, prepared by microiiizatiori of the 3.04 m.2/g. parent, is considerably enhanced. Experience with both the silica-water system and the 2.72-3.12 m.*/g. parent-daughter pair of the present study would dictate a lower AHi for the 4.56 ma2/g. species than for the 3.04 me2/g.sample. All the followiqg discussion mill be presented with no further coiisideration of the 3.04 m.2/g. sample C. It is still startling to the authors how regular this particle-size variation is a t the lower outgassing temperatures. However, this relationship does appear to be more complex a t the higher outgassing temperatures where two t'ypes of behavior exist-one being represented by samples A and E and the other by the remaining samples. Apparently this difference in behavior a t higher t,emperatures is substantially masked by several layers of mater which remain physically adsorbed a t the lowest outgassing temperatures studied. The only alumina-water immersional data for samples of known surface area to be found in the literature are in general agreement wit'h the present work. Stowe's AHi values of 310 and 300 ergs,/cm.2for gel samples of specific area 244 and 340 m.2alg.,respectii-ely, are slightly lower than those found iii the preseiit investigation but due t,o the differelices in sample pretreatment they may be entirely compatible. The work of Puri, et al.,ll is soxnem-!iat anomalous since they observe large vhnnges ~ I Isurface area (determined from water adsorptiolr isotherms) over an outgassing temV. 31 S t o w e , ibid.. 66, 454 ( 1 9 5 2 ) . (11) B. R. Purl, S. hlittal and L. R. S h a r m a , I l e s . Bul. I'anjab Uniu., (10)
111, 309 (192;). (12) C. 4.Guderjahn, D. A. Paynter, P. Good Ttiis J O U R N A L . 63,2066 (1959).
E. Berghausen and R. J.
100o
-
200
1
1197
6 cc
G
-
I
I
1
I
,
,
,
pcraturc range where pore collapse and siriteriiig does not usually occur for alumina gels. They find a rather random variation of AH, with outgassing temperature with an average value of 278 ergs/cm.2. The data of Good, et al., (see Table 11) fit the scheme of AH, 2's. particle size but it was impossible to reproduce their findings of a long time lag for attainment of thermal equilibrium. On the contrary, all the alumina samples studied were a t 95+% thermal equilibrium five minutes after sample breakage. Runs were extended over 30 minute periods in several cases to check this behavior. Since the wetting of as little as 0.5 m.2 of surface is measurable to +2yo reproducibility, the phenomena observed by Good, et al., should have t)ceii detected easily. I n previous studies on the silica-water the hypothesis was advanced that grinding imparted an amorphous substrate character to the substrate surface layers. That an amorphous substrate should show reduced immersional heats compared to a crystalline surface has been explained qualitatively.* This argument cannot be extended bodily to cover the present set of data. Samples -2, B, F and G , although of widely varying specific area, a11 exist as individual single crystals formed during the calcination process. Unless some fundamental but as yet obscure reason exists for the (iccurrence of an increased amorphous character of the surface with decreased particle size, then some other explanation of the large divergence in AHi for these samples must be found. I n fairness to the previous arguments,* it should he noted that sample D, prepared by grinding B has a depressed AHi and that the amorphous gel, G, has the lowest AH, measured other than C.
W. H. WADEAND NORMAX HACKERMAN
1198
OUTGASSING TEMPERATURE FOR V A R I O U S S U R F A C E
C-3 04 m'/g
100
200 t
300
400
("C.).
Fig. 2.
An interesting grouping of the data occurs according to a- or ?-crystalline modification. Whether this grouping is accidental or whether the heats of immersion are to this extent sensitive to crystalline form must be left for a more detailed study where both the a- and y-forms are investigated over a wider range of particle sizes. In view of both the uncertain characterization of the y-alurniiiasl3 and their small but still measurable water content a t the calcination temperatures used, it is perhaps best to say that the lower heats of immersion of the 7-aluminas simply are the results of immersion of samples of greater surface water content than the corresponding a-aluminas. Better crystallographic classification of the yaluminas will be necessary in order to clarify this point. Outgassing Temperature Effect.-There are two types of outgassing pretreatment behavior of the samples. Samples A and E show an initial increase followed by complete independence of A H i a t outgassing temperature above 200" whereas the remaining samples show a continuous increase in immersional heat with outgassing temperature over the entire range (the upper limit of 450" is dictated by collapse of the Pyre2 sample bulbs). Heat of immersion measurements for the silicawater1,2 and ~ a l c i t e - w a t e r ~systems ~ showed the vompletc removal of physically adsorbed water between 200-250" arid, indeed, this would be expected if adsorption energies are less than a liberal 15-20 kcal./mole. The AHi us. outgassing temperature for samples A and E are indicative (13) A. S. Rusaell, Aluminum Res. Lab., Tech. Paper No. 10, Aluminum Co. of America, Pitts., Pennsylvania, 1953. (14) W. H. Wade and N. Hackerman, THISJOURNAL, 63, 1639 (1959).
Vol. 64
of this single mode of existence of water on the surface. The continuous increase of AH; of the remaining samples leads one to postulate the existence of chemically bonded surface hydroxyl groups. Crystallographic datal5 show that the lattice of a-alumina consists of hexagonal close-packed layers of oxygen ions separated by aluminum ion layers where the AI+++ occupy positions in the middle of the O= triangles. Using the closest 0-0 spacing of 2.49 8., one calculates 19 0-/100 A.2 in the oxide ion sheet. Since a surface with this oxide sheet exposed represents the most dense possible packing of oxide ions and if one assumes that each of these oxide ions in the surface is normally converted to a surface hydroxyl group, then there is a maximum of 19 OH's/lOO A.2 on the a-alumipa surface. This can be compared to 8 OH's/100 A.2on quartz.16 For a direct check on the existence of water bonded above 200" (rather arbitrarily chosen as the maximum temperature that HzO would maintain in chemical identity due to the large bonding forces necessary for its surface localization), weight loss measurements were run on samples B, C, F, G and H. The results of these are shown in Fig. 2 . With the exception, once again, of sample C, there is a direct correlation between total weight loss a t a given outgassing temperature with the A H i . Over the temperature range of 200-400' the water loss for samples B, F, G and H corresponds to 14, 5 , 5 and 6 surface OH'S/ 100 A.2, respectively. All these values are sufficiently less than the theoretical maximum to indicate that it would not be exceeded a t still higher temperatures. The correlation between the weight loss and AH, studies is similar to that found in the silica-water system2 in that it clearly seems that low surface area samples have the higher concentration of surface hydroxyl groups. Infrared studies will be needed for a more direct check on the existence of surface hydroxyl groups 011 these samples. As distinct from the silica-water behavior, the A H i us. t("C.) does not pass through a maximum. This indicates a rapid rehydration to the aluminol structure. Recently measured mater adsorption isotherms1' on several of these alumina samples outgassed a t temperatures above 200" all show hysteresis a t coverages below a monolayer indicating lack of reversibility in the desorption of surface hydroxyl groups a t 25". At the present time, it is impossible to explain the discrepancies between samples A and E and the remaining samples with regard to the nonexistence of surface hydroxyl groups 011 just the two samples. In conc1usioii, it should be rioted that at tention has been focused only on relative variations in AH, with a ronsideration of their absolute magnitude being outside the present scope of attack. It has become increasingly obvious that in this (15) R. W G W y ~ h o f f ."Crystal Strurtures. ' Interscience Pub., Inc., New Yorh, N Y , 1948. (le) R . K. Iler, "The Colloid Chemistry of Silica and Silicates," Cornell University Press, Ithaca, N. Y., 1955. (17) R. L. Every, W. €I. Wade and N. Hackerman, unpublished data (1960).
Sept,., 1960
SOlWl‘IOX O F
HYDROGEN CHLORIDE BY DRY LYOPHILIZED p-LACTOGLOBULIN
regard very little can be inferred from the bulk structural and thermodynamic properties of solids. In other words. the Statement that “a solid is in its standard state” has little or no significance with respect to its surface properties. To be specific, “the immersional heat of a substance” a t the present time is a meaningless assemblage of words unless one can describe completely all the parameters of the surface, adsorbent, and adsorbate phases. Acknowledgments.-This work is a contribution from the American Petroleum Institute, Project 47d and the authors thank them for their continued support and interest. Appreciation is expressed to A h . R. L. Ekery aiid Dr. L. Slutsky and hIr. C. L. \Tilliarns, Jr., for their aqsistance in various phases of the study. I~ISCUSSIOX L. ,4.R o m (E. 1. du Pont de Semours & Co.).-Siiice surface crystdlinity appears t o he an important factor, I would suggest that electron diffraction be used to determine variations. Do the srirfacp areas stay constant as a function of degassing temperature? K h a t is the evidence you have for the presence of free vibrating hydroxyl groups on anhydrous &03surfaces? IV, H. WADE-For all samples, the BET arms were found to be independent of outgassing temperature. There arc’ several infrared studies u hich shox discwte surface OH bands. -4. C. ZsTimxouER (IJ(-high I-iiivcrsity) -The question of double layer formation terms not to h a w been considered. Thr sniall amount s of impiiritics prewnt may posbibly product, quite difftwiit dumina-a ater interfaces
1199
from sample to sample. In t,liis laboratory such heat effects have been detected easily for the modd system of graphite immersed into surfactant solutions; even trac: calcium (a few parts per million) gives differences i n heats of immersion. Do you believe such effects might be contributing? IT. H. WADE.-I would say that the regularity of the data is very surprising if your hypothesis is corrwt. GEORGER. LESTER (Universal Oil Products).-You have stated that the calcinations were done by the manufacturer. The outgassing experiments might he very susceptjble t o time and storage between calcination and outgassing. Secondly, my o m calculations, based on a spinel structure for a-alumina, indicate that the averape number of 8- ions in the 100, 110, 111 planes is about 11 ions per 100 L$.2. This may explain the lovier OH loss for the two samples of a-;U,Os. If the “amorphous” sample is really a mixture of true e--41203 and amorphous .1120a, as often has been srigyest,ed,this explanation may apply also.
n.J. C. YATES(Columbia Uiiivereity).-I wish t o congratulate Dr. Wade on his statement that “little can be inferred from the bulk structural and t’hermodynamic properties of solids.” While as to your general stat’ement that the low area materials are more crystalline than high area mat’erials, I agree, but I t’hink t,hat there may be exceptions. For instance, the spectra of the OH groups on .1lon C are quite different from t’hose on the higher area alumina gel (Peri, ACS meeting, Sept., 1959). The peaks on illon C are smeared together, while the bettercrystallized alumina gel showed three distinct peaks. This is what one would expect from the flame process used in manufacturing Alon C. IV. H. \TauE.--ht the present time, I liave found 110 correlation between extent of surface OH coverage and crystallinity.
THE SORPTIOK OF GASEOUS HYDROGEN CHLORIDE BY DRY LYOPHILIZED 8-LACTOGLOBULIN’ BY WASYLS. HNOJEWYJ AND LLOYD H. REYERSON School o j Chemistry, University of Minnesota, Minneapolis, Minn. Receiued March 7, 1960
The sorption of gaseous HCI by dry lyophilized /%lactoglobulin was studied at 27”. HCl is held so firmly on some of the sorption sites that it cannot be removed by pumping to a high vacuum. The results show that the amount remaining sorbed on the protein is some reciprocal function of the temperature.
Earlier work in this Laboratory2 showed that gaseous HC1 was very strongly sorbed by insulin at -78.9 and 20”. The protein used in that study was the zinc crystalline insulin. I n that work it was not possible to determine the total effect of the zinc in the sorption process. However, the amounts of HC1 bound more or less permanently depended on the temperature of the sample during the desorption process. As the temperature was lowered, the protein bound more aiid more HC1. About five times as much was bound at -78.9” as was held a t room temperature. It was felt desirable to repeat the sorption studies of €IC1 by a protein having no metal ions in the structure. Qualitative studies on the sorption of HCl by lyo( 1 ) f3upportt.d by a grant from U.8 P. Health.
(2) (lY3Li)
L H Reyerson and Lowell Peterson, THISJ O U R V I L 69, 1117
philized p-lactoglobulin, carried out in the Carlsberg Laboratorium by one of the authors (L.H.R.) indicated that the amounts sorbed were relatively large. This preliminary study also indicated that the protein was quite stable in that the protein containing adsorbed HC1 was not appreciably hydrolyzed when put into water. This is in contrast, to the marked hydrolysis of the peptide-like bond in nylon when it contained adsorbed HC1. S‘ ince a great deal is known about lyophilized p-lnctoglobulin, it was decided to determine quantitatively the sorption of HC1 by this protein. -4 complete adsorption isotherm was obtained a t 27”. Following desorption the amounts of HC1 retained at I Oe6 mm. pressure were determined a t 27,30 and GO”. Experimental The j3-lactoglobulin used in this study was a fresh sample of the same protein, obtained from the Carlsberg Labors-
