Sept., 1960
NATURE OF THERMAL REGEEERATIOX OF OXIDE-COATED ?\TI, CO AND Cu
geneity lowering the experimental values a t the high coverage studied. In conclusion, the close agreement between urn values for 'TSP graphite as determined from desorption of hydrogen up to 2000", and estimated from adsorption isotherm data is, perhaps, surprising. That is, the virgin nuclear graphite used for the desorption studies had on its surface a subst,aritial amount of oxygen, in a,ddition to hydrogen.32 Upon out'gassing up to 2000", roughly 35% of the total gas desorbed was oxides of carbon. It might be thought that this oxygen would compete with hydrogen for adsorption sites and that' the urn value from outgassing experiments would be lower t'hart from adsorption of hydrogen on (32) .J. P. Iledniond, P1i.D. Thesis, T h e Pennsylvania State Univ., IRR!).
1099
previously degassed samples of graphite. The picture can be complicated, however, by the possibility that oxygen chemisorbed on carbon can act as a promoter for hydrogen chemisorption, as has been suggested to be the case on copper.31 Also it is possible that hydrogen chemisorption was completed in the graphitizer before the graphite came in contact with, and chemisorbed, oxygen. Savagez4shows that at low temperatures niore oxygen than hydrogen is adsorbed on graphite "wear dust," suggesting that some chemisorption of oxygen is possible even when the carbon surface already has reached saturation with regard to hydrogen. Acknowledgments.-The authors wish to thank Jean Rorrall, L. G. austin and E. T. Nelson for their helpful discussions during the progress of this xork.
ADSOKPTIOS STUDIES ON METALS. IX. THE NATURE OF THE THERMAL REGENERATIOS OF OXIDE-COL4TEDNICKEL, COBALT ASD COPPER BY A. C. ZETTLEMOYER,YUKG-FASG YU AXD J. J . CIIESSICK Surface Chemistry Laboratory, Lehigh University, Bethlehem, Penna. Received March 7, 1060
It is well mown that oxidation of nickel, cobalt and copper a t 25' proceeds by two steps, a fast followed by a slow process. Upon heating these oxide-coated metals at modest temperatures such as 300" zn vacuo, oxidation will again proceed multimolecularly by a rapid step a t 25" even though no oxygen was evolved during thermal treatment. The regeneration could be due to (a) migration of metal ions to the metal-oxide interface; ( b ) dissolution of the oxide film in the bulk metal; or (c) crystiillization of the oxide film to reveal bare metal. In this mork, organic vapors were used as a probe for any bare metal surface uncovered during regeneration The heats of adsorption of n-propylamine and acetic acid on the bare, oxidecoated and regenerated metal surfaces support the contention of Dell3 that recrystallization occurs. That is to say, initial heats of adsorption were much higher on the regenerated surfaces, in accord with the values for the freshly reduced metals, and mere much lower on the oxide-coated surfares. The extents of irreversible adsorption followed the pattern to be expected from the heat measurements.
Introduction Russell and Bacon' first observed that reduced nickel saturated with oxygen near 0" and then heated at 300" in z ~ ~ c uwas o , able to adsorb oxygen again a t 0" even though no oxygen was evolved during thermal treatment. Low temperature oxidation and t.herma1 regenerat'ion of nickel, cobalt and copper has been the subject of a number of papers2 from this Laboratory. It was found that the oxidation of reduced nickel, cobalt and copper powders :it 25' (or below) proceeds in two steps: an initial fast process followed by a slow, exponential deca>.process. During the fast process, three apparently non-act'irated processes probably take place: cliemisorption of oxygen on bare metal, oxide forrnat'ion from one to several layers and oxygen chemisorption on the underlying oxide as 0ions. The slow process can be explained semiquan(1) Ty. W. Russell and 0. C. Bacon, J . .4m. Chern. Soc., 64, 5 4 (1932). (2) (a) Y..F. Yu, J. J. Chessick a n d A. C. Zettlemoyer, Advances in Catalysis,9, 415 (1957); (b) J. J. Chessick, Y.-F. Yu and -1.C. Zettlemoyer, "Pmceedings of t h e Second World Congress on Surface Activity," Vol. 11, Academic Press, New York, N. Y., 1957, p. 209; (e)
-4.C. Zettlemoyer, Y.-F. Yu, J. J. Chessick and F. H. Healey. THIS ~ O U X N A L ,61, 1319
(1957).
titatirely by the Mott and Cabrera theory for very thin oxide film growth. A4ccordingly,the electric field created across the thin oxide film, i.e., between the adsorbed oxygen anions on the surface and the metal cations a t the metal-oxide interface, aids the cations to diffuse through the oxide outward to the surface where they react with oxygen to form oxide. Oxidation stops, or at least proceeds a t a very low rate, a t some limiting film thickness. These film-covered surfaces took up oxygen again after heating to a high temperature in vacuo. This process could be accomplished if during thermal treatment: (a) the limiting film thickness is reduced by migration of metal atoms to the metaloxide interface; (b) the oxide film dissolves into the bulk of the metal; or (c) the oxide crystallizes into discrete crystallites covering a part of the surface. Dell3 recently reviewed pertinent work in the literature and concluded that recrystallization occurs during the regeneration process. This hypothesis of Dell's was based on deductive reasoning and needs quantitative data for support. The object of this mork is to obtain such data. Organic vapors were used as a probe for any bare metal surfacae uncovered during regeneration. 13) R. M. Dell, rbad., 6 2 , 1139 (1958).
1loo
h.c. ZETTLEMOYER, Y.-F. YU AND J. J. CHESSICK Experimental
Materials.-The nickel, copper and cobalt powders were prepared by thermal decomposition of C.P. nickel carbonate, basic copper carbonate and hydrated cobalt nit,rate, respect,ively, under reduced pressure a t 400'. The outgassing was continued for 12 hours beyond the point of no apparent gas evolution. These nickel, copper and cobalt samples were then reduced with dry hydrogen a t 400, 300 and 350°, respectively. After reduction the metal powders were degassed a t 10-6 mm. a t 450, 350 and 400°, respectively, just prior t,o adsorption studies. These temperatures are the lowest permissible so that sintering could be minimized. The organic liquids used to provide the adsorption vapors were reagent grade. They were dried with anhydrous magnesium sulfate and then frozen and pumped through several cycles. Only the middle portions were taken for the adsorption nicasurement,s. Adsorption Apparatus.-The basic feature of the adsorption apparatus is the Teflon stopcocks. Besides the simplification these stopcocks made possible, they eliminated possible contamination of the metal samples with mercury from the cut-offs usually employed. These stopcocks would not hold a good vacuiim a t the outset, but they steadily improved when opcrated frequently during the first ten days of use. A liquid nitrogen trap was employed between the adsorption system and the pumps. The adsorption system includes a sample tube, glass break-seal, orgaiiic vapor reservoir and a calibrated doser, and it is terminated wivith a glass Bourdon spoon ga.ge. The gage is balancid wit,li helium pressrire which is measured by an Apiezon B oil manometrr; smdl deflections of the gage from zero are det,ermined by a microscope eyepiece. The equilibrium pressures ran be determined to =t0.02 mm. and the range of the oil manometer is from 0-10 cm. A side arm is attached t,o the gage and immersed in silicone oil to dampen vibr:it,ions. The reduced samples wcre sealed off and transferred to the adsorption apparatus. The seal was broken by a magnetically operated plunger just prior to each run. For studies on oxide-coated surfaces, the reduced powders were first exposed to dry oxygen a t 25" and 1 rm. for 10 minutes so that the oxide films resulting were from 10 to 15 A. thick. After each initid isot;lierni on a given surface, the system was outgassed a t 25' employing a liquid nitrogen trap and high vacuum syst,em for 24 hours to a vaciium of better than 10-6 mm. A good vacuum was obtained in about 4 hours. The isotherms m r e then repeated bo give new BET V,'s. These and the initial V , value were used to estimat,e the amount irreversibly adsorbed. Nickel, copper and cobalt were activated in iiacuo after oxidations a t 350, 250 and :37,5", respectively. These temperatures are 0.37 timrs the melting points of the metals in each case. Since the degree of regeneration is dependent on the temperature and time of act.ivation, these variables were controlled to within 5" and 5 minutes, respectively. Calorimeter .---The basic calorimeter design was that used previously for heats of chemisorption of oxygen on these same metals.* Modification included a glass breakseal so the sample could be transferred from the apparatus on which it was reduced and a 25 jiinction copper-constantan thermocouples with the reference junctions held within 0.1O of the adsorption temperature to minimize heat loss through the thermocouple wires. The e.m.f. was amplified then recorded on a Brown Recorder. The sensitivity of the calorimeter is 0.001" or about 0.01 cal.; the heat capacity of the filled calorimeter ranged from 10 to 15 cal. per degree. It was estimated for the calorimetric heats recorded here that over 90% of the heat'evolution occurred within the first two minutes after the introduction of each portion of organic vapor.
Results Heat curves for the adsorption of n-propylamine on reduced, oxidized and regenerated samples of nickel and cobalt are plotted in Figs. 1 and 2 as a function of surface coverage, e. The high initial heats for the amine on nickel suggest chemisorption, probably through electron transfer to the un-
T'ol. 6-2
filled d-band of the metal. In contrast to the high initial heats obtained on the base metal surface, those for the oxidized nickel surface are much lower. There seems to be little doubt that physical adsorption predominates now. The heats of adsorption on the regenerated sample are the same as those found for adsorption on the reduced metal in the range of 0 from 0 to -0.7; thereafter they fall abruptly to values slightly less than those for the oxidized surface. The initial values suggest that metal atoms exist on the regenerated nickel surface and comprise about 70% of the available area. The heat curve for propylamine adsorption on reduced cobalt is similar in shape to the one obtained with nickel; initial heat values, however, on reduced cobalt are considerably smaller. Again, initial heats of adsorption on the regenerated cobalt sample are in agreement with initial heats on the reduced sample. The data indicate that metallic areas comprising 60 to 70% of the external surfaces are re-exposed on both nickel and cobalt samples during regeneration. This agreement is not unexpected since equivalent members of oxide layers were deposited on both metals prior to regeneration. On the other hand, the total snturation amounts of oxygen taken up by rciduced cobalt and nickel in numbers of layers are 10.2 and 4.4, respectively. Heats of adsorption of n-propylamine on regenerated copper samples were not measured with sufficient precision to present here, but were of the same order of magnitude as on the oxide-coated surface. However, the lack of significant oxygen uptake on regenerated copper samples suggests little alteration in the state of the oxidized surface during regeneration. In another study, similar heats of adsorption were successfully measured for water on the bare, oxidized and regenerated copper surfare. The similarity in the heat data for the aclsorption of water on oxidized and regenerated copper samples plotted in Fig. 3 supports the belief that regenemtion at 250" has little effect on the structure of the oxidized surface. Heat d u e s extrapolated to zero coyerages, per cent. irreversibly adsorbed at, 25" and 10-6 mm. and cross-sectional areas for the adsorpticii of n-propylamine on nickel, cobalt a i d copper saniples are listed in Table I. Heat values arid per cent. irreTABLE I HEATAND ADSORPTIONDAT.4 FOR n-PROPYL I M I N E TION os METALLIC SAnwLm
ADSORP-
Reduced Cll
A H y - 0 ) (kcal./mole)
28
% I r r e v p i b l y adsorbed C.K.4. (A.Z)
19 22
e4 38 27
-)7
Oxide-coated 0 AH(6-0) (kcal./niole) 21 " 7 0 I r r e v y i b l y adsorbed I C S. I.(A.2) 36 Heat values a t siifficiently low 6 values w r e not ured.
versibly adsorbed increase in the order Cu
< Coy
