NOTES
1245
iri the range below room teniperature the activation energy is almost constant, being equal to 10 kcal./ mole, while above room teniperature it increases with increasing desorption tcriiperaturc to 23 kcal./mole. These activation ericrgies of desorption may be assumed to be equal to the heats of adsorption, siricc with platinuni supportod on silica gel both values were found to be equal4 and also adsorption equilibrium was rapidly established on all the platinurn catalysts. From the aniourit adsorbcd a t - 52 O before the desorption experinient, tho ratio of hydrogen atonis cheinisorbed to platinurn atonis in the catalyst was found to be close to 1, suggesting an extrenicly high order of rnetal dispersion in the catalyst as pointed out by several workers.2 Furtherniore, it riiay be concluded that the energy values obtairicd in this work are associated with thc whole range of coverage from very low &values to a nearly fully covered surface. I+oni the rate of desorption R and the activation energy of desorption E , values of log A ( R = Ae--E”RT) were calculated with the results shown in Fig. 2 . It is seen that the plot of log A against the amount adsorbed shows a niiniinurn as well as a niaximuni. Although it is very difficult to predict the behavior of the change in log A with coverage, it seems that such a behavior as represented in Fig. 2 would be uriexpectcd for the case where only orie type of adsorption is opcrating. Seholten, et al.,6 and Taylor and Langmuirfi have investigated the desorption of nitrogen from an iron catalyst and of cesium from tungsten, respectively, in both of which cases one type of adsorption seeins to operate under their experimental conditions. They have found that with decreasing coverage log A decreases monotonically or passes through a rnaximuin and then decreases. In addition, the results of desorption ineasurenicnts carried out by the present ahthors showed that when two types of adsorption are oper-
ating, a mininiuni was always observed in the plot of log A vs. coverage. The considerations described above suggest that in Fig. 2 the type of adsorption is different on each sidCof the ~iiiniinum. Froin the study of infrared spectra of hydrogen chemisorbcd on platiriuni supported on aluniiria, Pliskin and Eischens’ have concluded that the weakly cheniisorbed hydrogen which can be removed by evacuation a t room temperature differs in the type of adsorption from the strongly chemisorbed hydrogen. It might therefore be concluded that in this systcni two types of adsorption are opcrating, one corresponding to the desorption below rooin teniperature with the heat of adsorption about 10 kcal. /niole, and the other to that a t higher temperatures with increasing heat of adsorption up to 28 kcal./ mole. Similar expcrinients were carried out with platinuni black and platiriuni supported on silica gel. I t was found that the mininiuni and maxiinum heats of adsorption are nearly the saine for all three .catalysts and two types of adsorption of hydrogen are exhibited by the three catalysts.
Acknowledgment. The authors wish to express their sincere thanks to Professor P. H. Eninictt of Johns Hopkins University for his helpful discussion. (4) Adsorption equilibrium measurements were carried out only with platinum supported on silica gel. I t was found t h a t the heat of adsorption obtained from the adsorption isotherms between 278 and 226’ is 16 kcal./mole a t 0 = 0.6. being almost equal to the activlction energy of desorption a t the same coverage. (0 = 1 is defined 89 Had,/]% = 1.) ( 5 ) J. J. F. Scholten, 1’. Zwietering, J. A. Konvalinka. and J. 1%.de Boer, Trans. Faraday Soc., 5 5 , 2166 (1959). (6) J. B. Taylor and I. Langmuir, Phys. Rea., 44, 423 (1933). (7) W. A. l’liskin and R. P. Eischens, Z . physik. C h a . (Frankfurt), 24, 11 (1960).
Oxygen Exchange between Chemisorbed Carbon Monoxide on Catalytic Nickel
st
123
215
by J. T. Yates, Jr.1 Department of C h a i s t r y , Antinch College, Yellow Springs, Ohw (Received November 1 , 1.963)
21
0
L
I
2 4 A m o u n t desorbed. c c . STP.
6
Figure 2. Values of log A for hydrogen chemisorbed on platinum supported on alumina.
Studies of the infrared spectrum of adsorbed CO on supported nickel surfaces h a w led to the postulate that two distinct adsorbed spxies exist on the heterogeiieous surface. These are a bridged CO between two adjacent Ni atoms and a linear CO bonded to one surface Ni The i~it~erpretation of tho existence (1) National Bureau of Standards. Washington. D. C.
Volirme 68,Numher h
Ma?/. 1.9G4
1246
of these two species is made by analogy with the spectra of metal cart)onyls. Because of the insensitivity of the infrared method, it is possible that other adsorbed CO species could exist undetected on Xi surfaces. The purpose of this investigation was to examine this possibility by using a technique first employed by Webb and T!iischens5 in their study of CO adsorption on catalytic iron surfaces. A mixture of C1201R and C13016was adsorbed on a Xi surface and portions of the CO were removed with a diffusion pump and analyzed for C13018.A t room temperature oxygen exchange occurred. The data indicate that exchange proceeds most extensively when the equilibrium gas phase is at high pressure and contact time is of the order of many hours. Various possibilities are suggested for an exchange intermediate which forms at high coverage.
Experimental Apparatus. A conventional glass high vacuum adsorption apparatus was employed. The system was evacuated through a liquid nitrogen trap with a twostage fractionating oil diffusion pump using Octoil-S fluid. The system was capable of maintaining a pressure of less than 1 X torr. Baking of some sections of the apparatus was achieved a t 2.50' using heating tape and infrared lamps. The powdered Si sample was held dn a glass support, consisting of three trays each of which had a volume of about 1 The sample and its support were held inside a vertical 2.6cm. diameter Vycor furnace which \vas connected to the vacuum manifold via a high conduct,ance stopcock. The temperature of the support was monitored by a thermocouple which passed up a Vycor finger mounted vertically along the axis of the furnace tube. Carbon monoxide which had been chemisorbed on the Xi surface could be desorbed from the sample through a two-stage mercury diffusion pump which was connected to t,he vacuum manifold via a liquid nitrogen trap and a high-conductance stopcock. The mercury dif'fusion pump was backed by a 6OO-cm. removable bulb which could be pumped out with the oil diffusion pump prior to a desorption experiment. The pressure of the collected gas in the bulb was monitored with a I'irani gage protected from mercury vapor by a Dry Ice trap. Once the bulb had been filled to a desired pressure, it. could be disconnected arid carried to the mass spectrometer for analysis. Desorption could be continued in an uninterrupted manner by using additional tmlbs. hiatwials: CluO'fi was prepared from C'"02 by reduct,ion ovcr Raker 99.8y0zinc dust according to the
S~TES
procedure of Sernstein and Tay10r.~ Residual C02, if any, was condensed using liquid nitrogen on a side arm which was then sealed off. The mass spectrum was obtained showed less than 0.01% CL302.CL2OLY from the Weizmann InstituteR and was used without further purification. In each experiment a mixture of the two enriched CO samples was prepared i n a dosing system connected to the vacuum manifold. A typical analysis of the CO mixture is shown in Table 1.
Sample Z'reparatim. Nickel(I1) oxide obtained from Prof. It. J. Kokes a t Johns Hopkins University was used as a source of Ni in this research. This sample, prepared in t,he manlier of I3est and I i ~ s s e l l was , ~ of high purity and similar to that used by Emmett and Hall in their work on ethylene and benzene hydrogenation.'" This is referred to as sample VI11 in their papers. Approximately 1 g. of the NiO was placed in the weighed glass support which, after reweighing, was lowered into the furnace tube. The furnace was evacuated overnight, and purified hydrogen' was ad(2) It. 1'. Eischena. s. A. Francis, and W. A . I'liskin. .I. Phys. Chem., 60,194 (1956). (3) J. T. Yates. Jr., and C. W. Garland, ibid.,65, 617 (1961). (4) C . E. O'Neill and D. J. C. Yates, ibid., 65, 901 (1961). (5) A. N. Webb and It. P. Eischens, J . Am. Chem. SOC.,77, 4710 (1955); J . Chcm. Phys., 20, 1048 (19.52). (6) Obtained from Isomet Corp.. Palisades Park. N.J. ('02 rrnalysis was 56.7% CL3. After reduction the CO wiw 50.0% C'30'e. (7) It. €3. Bernstein and T. I. Taylor. Science, 106, 498 (1947). (8).Lnheled 89.7% C ' 2 0 1 1 , Our analysis showed 88.5% of the C O to be C " 0 ' ~ .
(9) R. ,J. Best and W. W. Russell, J . A m . Chem. soc., 76,838 (1954). (10) W. K. Hall and 1'. H. Ernmett. J . Phys. Chem.. 62, 816 (19%); 6 3 , 1102 (1959). (11) Hydrogen for reduction of the samples was Airco eleotrolyt,io grade. I t was passed through a Deoxo oat;dytic purifier and then
through a liquid nitrogen trap. In some reductions. a 1)rch:rkcd Spherori charcoal trap. niaintained at liquid nitrogen temperature. wirs employed in the hydrogen &earn ns suggested by ltall trnd ICmmett. No eriharirement of the sperific CO adsorption was whieved with these samples compared to samples prepared witahout the charcoal trap.
1247
NOTES
mitkd. When the pressure reached 1 atm., the furnace \\.as gradually warmed ; reduction began a t about 200'. After thc majority of the water had been liberated, as judged by condensation in the cool part of the exit t,ube, the temperature was raised tjo 350400' and reduction with flowing hydrogen was continued for prriods ranging from 16 to 5 3 hr. a t a flow rate of 1.0 1. S'l'P/hr. After reduction, the sample was outgassed for 20-30 hr. at the reduction temperature using the trapped oil diffusion pump. The dynamic pressure at the end of the outgassiiig period was 2 X lop6 torr or lower; on cooling the dynamic pressure dropped below 1 X IO-6 torr. At the end of an experiment the Ni sample and the holder were rcweighed. The loss in weight observed corresponded to the empirical formula Sil.oOOl.o~, and the variatioii employed i n reduction time and temperature in the various experiments did not affect the weight loss observed per unit weight of sample within experimental error. CO Rdsorplz'on. Adsorption of the known mixture of isotopic CO was carried out a t 25' in stepwise fashion by addition of known quant,ities of CO followed by pressure measurements; the adsorption could be followed with the Mc1,eod gage t,o pressures of about 1 X torr. Up to 50 X IO-$ torr, uptake was measured with a calibrated l'irani gage. Above this pressure, the dead space correction was appreciable and uptake mcasurements became impractical. Analysis of the Desorbed Gas. The gas desorbed from the sample through the mercury diffusion pump was collected in a bulb until the pressure reached about 100 x torr in the bulb. The bulb was then disconnected from the vacuum system and a portion of the gas was transferred by expansion to the sampling system (about 500 cm.3 volume) of a Consolidated Ehgincering mass spectrometer, ;\lode1 21-401. 'I'hc precision in duplicate analyses was +2%. A sample of CO of known composition was retained in the mercury diff usiori pump for approximately 9 hr. as a blank to check that exchange was not occurring on long exposure to the heated mercury vapor. No exchange was detected.
Experimental Results O.rygen Kxchange. Four experiments are reported in Table 11. I n the first portion of the table the highest CO pressure employed in the initial adsorption and the STI' volume of CO adsorbed per gram of Xi a t this pressure are given, as well as the coritact time allowed at this pressure. 'I'he lower entries show the CO volume per gram of
S i as pumped from the sample i n successive fractions. For each of these fractions desorbed, thcre are given two factors related to the increase in the CI3O1*coricentration ratio due to oxygen exchange between CO molecules. In addition, thc C:'201* C 1 9 1 bconcentration ratios in the initial gas arid the desorbed gas are given. It can be seen that in each experiment t,he concentration ratio of Cl30I8in the desorbed gas has increased from that observed in the CO mixture prior to adsorption. This increase varies from about 30% in experiment I to 100yo in the latter part of experiment 111. Satisfactory checks ( f 1.2%) are obtained in thc factor increase in the C I 3 0 1 8 concentration ratio when evsluated either as
or as
The other possible ratio involving the concentration of C L 2 0 lis6 not shown since the accuracy of the mass 28 peak is considerably poorer than peaks at mass 29, 30, and 31 due to a nitrogen background problem in the mass spectrometer. Xo discernible difference in the rate of desorption of C1201* and C1s016is seen, since the value of (nRo/wJd remained the same as (nso/n& within experimental error in each of the four experimeiits. The data in Table 11 support the following qualitative observations relative to oxygen exchange between chemisorbed CO molecules. (a) The highest extent of exchange is achieved by long coritact with the gas phase at relativcly high CO pressure. Thus, experiments 111 and IV show the highest extent of exchange and these two cxperiments employ the highest CO pressure and the longest contact time. Experiments I1 and 111 involve comparable CO pressure and by comparing these two experiments it may be seen that long contact time favors a higher extent of exchange. (b) The extent or exchange observed does not vary appreciably in the successive fractions removed from the surface during desorption (experiment 111). (c) The gas phase a t the higher pressure is in equilibrium with the adsorbed gas after 17 hr. contact insofar as the concentration of C'301xis concerned. This may be seen from experiments 111 and IV where the CI3O1*concentration i n the gas which was pumped away in less than 1 min. is the same as in the desorbed
KOTES
1248
Table 11: Adsorption arid Desorption of C W ' S
+ C1301fi.
Isotope At)undance Ratios in the CO" I
