1114
ROBERT J. L E A K
Ah'D
K. J. HARRICK.-%Stly, the penetration dcpth is considerably larger than one monolayer and thus it does not make much difference whether the reflection is considered to take place on the inside or the outside of the monolayer. Secondly, the index of refraction is a macroscopic quantity and is thus really undefined for a monolayer or a single molecule. In dealing with such thin layers, RIaxmll's theory cannot be used and it is necessary to go t o an atomic model. The condition of total reflrction can be imsatisfied a t the surface of the light pipe yet this technique should work since the radiation is not necessarily lost from the system-see footnote (11i . DON.4LD GRAH.431 (E. I. du Pont de Semours & CO.).To what extent would this technique be applicable to silica
P. TJ'. SELWOOD
T'ol. 64
or alumina as represented by fused qiiartz or sapphire? N. J. HARRIcK.-This technique should work for any transparent material. It is convenient t o have a high index of refraction so that the angle of incidence can be made small and thus increase the number of reflections per imit sample length. DONALD GRAHAM-ISit not possihlr that grinding, lapping, and diamond polishing of hard surfaces could introdurr surface defects making the swfacr ronipirrahle t o that of sninll particlrsP Tu'. J. HARRWK.-yeS. To avoid this i,ffect thfl surface should be etched, cleaved, or annealed. 'l'lie s:tmple mag also be initially grown to the shape desired, e . g . , dendrite?.
THE CHENIISORPTIOX OF OXYGEX ON YICKEI,' BY
ROBERT J. LEAK-4ND P. Jv. SELWOOD
Chemical Laboratory of Northuestern University, Evanston, Illinois Received March 7 , 1960
The chemisorption of oxygen on nickel-silica catalyst systems has been investigated by the low frequency a.c. permeameter method and volumetric gas adsorption techniques. The nickel particles were in the superparamagnetic range or slightly larger. A method has been developed for distinguishing between true chemisorption of oxygen on nickel as contrasted with surface oxide formation. The method makes use of what appears to be a change of magnetic anisotropy energy in particles of nickel in the 100 A. diameter range. A possible explanation is discussedfor theso-called "hydride anomaly" in the chemisorption of hydrogen on nickel a t low temperatures.
Introduction There have been many studies of the interaction of oxygen with nickel surfaces, but there are two basic points which have not been elucidated. First, there is no clear distinction as to where surface chemisorption ends and where bulk oxidation begins. =Ind secondly, there has been some confusion as to the effect of chemisorbed oxygen on the magnetization of nickel. The first point has been further confused by loose terminology. "Chemisorption" is sometimes used when actually bulk oxidation probably occurred. Chemisorption of oxygen and oxidation are easily confused because of the ease with which the former can be converted into the latter. Odaz showed that penetration of oxygen into the nickel lattice occurred even at -183'. His work on evaporated films showed that the oxygen "sorption" proceeded until three molecules of oxygen were taken up per surface nickel atom. Further slow "sorption" also occurred. Scheuble3 reported similar studies a t room temperature on evaporated films. He found a "sorption" of oxygen equivalent to 9.5 times that needed for a monolayer if the surface werc smooth. Beeck and co-~vorkers~ found that oxygen was taken up a t 23" to the extent of two molecules per lattice site. Moreover the oxygen adsorption layer will diffuse into the interior even if no additional oxygen IS amilable from the gas phase. Although riot so stated, this amounts to oxidation, (1) Taken i n p a r t from the thesis of Robert J Leak submitted t o t h e Graduate School of Northnestern Unir ersity in partial fulfillmeiit of t h e requirements for t h e degree of Doctor of Philosophy. (2) Z . Oda, Bull Chem. Soc Japan, 27, 465 (1951). ( 3 ) IT' Scheuble, Z Physak, 136, 125 (1953). (4) 0 Beeck, A. E Smith a n d .4.Wheeler. Proc. Roy Soc (London) 8177, 62 (1940).
not simply chemisorption. Stone and co-workers5 showed the limit of oxygen uptake by nickel powders depends not only on temperature, but also on pressure and porosity. These two factors are important because they determine the efficiency of the dissipation of the heat of react'ion. Further evidence concerning this matter is given by Higuchi, Ree and Eyring6 who shom a wide discrepancy between the calculat,ed and observed heat of adsorption of oxygen on nickel. The complexity of the situation has been shown by Zettlemoyer and eo-workers' mho found four layers of oxygen, namely, oxide, chemisorbed 0-, strongly physically adsorbed 02, and weakly physically adsorbed 02. The reaction was not stopped bet'weerl the transition of the chemisorbed layer into t'he oxide, but at least the existence of two chemically bound layers was demonstrated. There has been some experiment,alevidence shoming a distinction between chemisorpt,ion of oxygen and oxidation. Farnsworth and Schliers showed by electron diffraction experiments on single nickel cryst'als that a monolayer of oxygen is chemisorbed a,t room t,emperatureaft.er a pressure-t,ime exposure of 2 X 10" mm.-min. Above 10-j mm.-min. a nickel oxide layer is formed. Shurmorskaya and Burshteing have shown by contact' potent,ial measiirement,s that, at 35' oxygen ( 5 ) R. hI. Dell. D. F. Rlemperer and F. S.Stone. THISJOURNAL, 60, 1586 (1956). ( 6 ) I. Higuchi, T. Ree a n d H. Eyring, .I. I m . Chem. Soc., 1 9 , 1330 (1957). (7) A. C. Zettlemoyer, Y. F. Yii, J. J. Chessick a n d F. H. Healey, THISJOCPSAL, 61, 1319 (1957). (8) R. E. Schlier a n d I-I. E. Farnsworth, "Advances in Catalysis," Tol. IX, Edited by D. D . Eley, W.G. Frankenburg and V. I. Komarew~ X. Y.. 1937. p. 431. sky, Academic Press, Inc., 1 - e York, (9) N. A. Shurmovskaya and R. K h . Burshtein. Zhur. P i z . K h i m . . 31, 1150 (1957).
Sept., 1960
1115
CHERIISORPTION OF OXYGEY O X NICKEL
increases the work function of nickel and at 100" decreases t,he work function even at a surface coverage of one tenth. This may be interpreted as chemisorption a t the lower temperature and oxidation a t the higher t,emperature. Apparently a real difference exists between nickel with oxygen chemisorbed on the surface and bulk nickel oxide. Some question as t o the effect of relatively electronegative adsorbates on the magnetization of nickel arose when a paper from this Laboratory reported t,hatjhydrogen and water decrease the magnetization of nickel while oxygen and nitrous oxide increase t,he magnet8ization.10v11These results were found with an apparatus that did not, permit thorough wacuation of the sample. Later experiment,s witrh the AC permeameter showed that all gases thus far st'udied produce the same effect on the sign of the magnetization change provided that they are measured a t the same temperature and on the same catalyst,. The purpose of this work was to extend further the study of the interactmionof oxygen with nickel. Experimental All measurements were made with the AC permeameter and volumetric adsorption apparatus that has already been described.12J3 A nickel-kieselguhr catalyst produced by Universal Oil Products Co. was used in an unsintered and sintered form. The nickel content was 52.8%. The unsintered catalyst w m used by reducing the catalyst in situ for 12 houri! a t 350" with flowing hydrogen and evacuating for two hours a t the same temperature. The sintered catalyst was reduced with hydrogen as before and then sintered in helium for one hour a t 500' and one hour a t 600" before weighing arid sealing in the permeameter for reduction and evacuation. Hydrogen and nitrous oxide were obtained from commercial cylinders. Hydrogen was purified by passage through a Deoxo unit and silica gel a t - 7 8 " . Sitrous oxide mas purified by fractional freezing a t -196" to remove nitrogen. Oxygen was obtained by decomposing dried potassium permanganate and passage of the gas through a silica gel trap a t -78". Gas analyses were made with a vapor phase chromatograph with silica gel as the column packing according to the results of Janak."
Results Preliminary Investigations.-Initial experiment's on unsintered catalyst's showed t'hat oxygen produces the same effect as hydrogen on nickel, namely, a decrease of magnetization. The magnetization-v olume isotherms for hydrogen and oxygen in Fig. 1 dlow how the two gases alter the relative magnetization in the same direction for measurements up to one atmosphere pressure at room t.emperature. The oxygen takeup of 59.9 cc./g., which lowered the magnetization to 43% of the original value, was 3.3 times that, of hydrogen chemisorption under t'he same conditions. Evacuattion did not remove any of the oxygen nor did it restlore any of the loss in magnetization. -Inattempt, was made t o use the decomposition of nitrous oxide a,sa source of oxygen to prevent oxida(10) L. E. LIoore and P. W. Selwood, J . A m . Chem. Soc., 7 8 , 697 (1956). (11) J. J. Breeder, L. L. r a n Reijen and A. R. Korswagen, J . chirn. phys., 64, 37 (1957). (12) P. W.Selwood, J . A m . Chem. Soc., 7 8 , 3893 (1956). (13) P. W. Selwood, ibid.,79, 4637 (1957). (14) J. Janak, Collection Czechoslou. Chem. Commun., 2 0 , 343 (1955).
0 10
10
Cc. a t STP adsorbed per g. of Ni. 20 30 40 50 60
0.1) 0.8
.:U . i 0
PRESSURE
YS
cc 0 ~ -
b
0.6
0.5
Fig. 1.-Magnetization-volume isotherms for oxygen and hydrogen (and pressure-volume isotherm for oxygen) on nickel a t 25'.
tion and thus to study the chemisorptive process alone. Analyses of the gases in the system were made by removing samples and transferring to the vapor phase chromatograph. The results showed that nitrous oxide is decomposed instantaneously on unsintered nickel catalysts even at temperatures as low as -78") only 11" above the boiling point of nitrous oxide. The decrease of magnetization accompanying the decomposition of nitrous oxide could have been due to either chemisorption or oxidation. The catalyst then was sintered as described above and used in an attempt to slow don n the decomposition of' nitrous oside and the reaction of the oxygen product n-ith the nickel. This sample was found to show an increase of magnetization upon chemisorption of hydrogen or of nitrous oxide at low teniperatures. Previous work in this Laboratory1.' has shomi that hydrogen adsorbed on nickel a t room temperature sometimes causes a positive change of magnetization as measured at low temperatures It has been surmised (incorrectly) that this indicated some change of nickel-hydrogen bond type as a function of nickel particle size. In x-iew of these results it was felt necessary to make a more detailed study of hydrogen adsorption on nickel. These new results are reported here together with lien- results on oxygen and nitrous oxide. Chemisorption on Sintered Catalysts.-A sample of UOP catalyst reduced for 12 hours at 350" was heated for 1 hour at 500" and 1 hour at GOO" in helium. Xdsorption of hydrogen on this sample gave the results summarized in Fig. 2. In each case the temperature of admission was the same as the temperature of measurement as indicated. It will be noted that on this sample hydrogen a t 22" gave a negative change of magnetization although the isotherm deviated from the straight line fomd with the use of smaller nickel particles. ;It -20" the niagnetizatioii isotherm had a positive slope a t low coverages. followed by a negative slope as coverage increased, while at -78" the hydrogen had a strong positive effect on the magnetization for all but the highest coverages. For all three temperatures evacuation removed a (15) E. L. Lee, 3 . 4 Sabatka and P. 'IT. Pelwood 79, 5391 (1957).
ROBERT J. LEAKAND P. W. SELWOOD
1116 +0.06 $0.04
1
1
I
/7
-0.04 Fig. 2.-Magnetiaation-volume isotherms for hydrogen on sintered nickel-on-kieselguhr a t - 78, -20 : ~ n d22".
200 300 400 500 600 Temp.,OK. Fig. %-Thermomagnetic curve for sintered nickelon-kieselguhr obtained with AC permeameter. Dotted line indicates usual shape for unsintered catalysts. 100
1.04
c
JI
1.03
1
I
I
I
1
4 6 8 1 0 1 2 ec. adsorbed per g. of Ni. Fig. 4.-Rlagnetization-volume isotherms for hydrogen and oxygen (as derived from nitrous oxide decomposition) on sintered nickel-on-kieselguhr a t -78". 0
2
moderate part of the hydrogen and in each case the change of magnetization so produced was positive in sign and reversible on readmission of hydrogen. In each case the pressure-volume isotherms were normal for the temperature and thus gave no indication of the anomalous magnetic changes occurring in the nickel. Further investigation of these effects produced the following information. A sintered sample was treated with 9.9 cc. of hydrogen per g. of nickel a t -78". The sample was then evacuated to remove the trace of hydrogen in the dead-space and some of the more loosely bound chemisorbed hydrogen. This treatment caused a 2.2% increase of magnetization. The sample was warmed to 22" with no further adsorption or desorption and was found to have suffered a 2.4% decrease of magnetization as compared with the sample prior to hydrogen admission a t 22". Recooling showed the change a t -78" to be reproducible.
T'ol. Gi
Several additional samples of sintered catalyst gave results in agreement with those shown in Fig. 2. The highest temperature a t which hydrogen produced a positive magnetic change was found to depend on the nickel particle size. Thus for particles in the 35 A. diameter range no positive effect was found above -253", for 60 A. particles the corresponding temperature was - 78 ', while for 85 A. particles a positive effect was found a t 20", (these diameters are based on X-ray line width broadening, and are approximate only). A sintered sample gave a small positive magnetization change on physical adsorption of nitrogen a t -78", in contrast to the small negative change observed on smaller nickel particles.16 The negative thermal transients which occur when hydrogen is flushed onto smaller nickel particles were found with the sintered samples, but the sign of the effect was reversed when the hydrogen was admitted a t low temperatures. A plot of magnetization versus temperature for a sintered sample was obtained on the AC permeameter for comparison with a thermomagnetic curve obtained by the Faraday method a t above 8000 oersteds. This is shown in Fig. 3. It will be noted that the XC method shows an apparent maximum of magnetization in the neighborhood of 200"K., whereas the high field magnetization approaches the normal for stable single-domain behavior. A magnetization-volume isotherm for oxygen on a sintered catalyst was obtained as follows. Nitrous oxide was admitted in small doses to a reduced catalyst a t - 78". Evacuation and fractionation with a Toepler pump and buret showed only nitrogen in the gas phase. The volume of oxygen (as 02)chemisorbed was half the volume of nitrous oxide admitted. Figure 4 shows the oxygen isotherm obtained in this manner and compared with a hydrogen isotherm a t the same temperature on the same catalyst. The slope of the oxygen isotherm could not be investigated to any great extent because of the competing nitrous oxide chemisorption and physical adsorption a t higher coverages. Admission of a large dose of air to the same system produced a loss of magnetization. Results were also obtained for the chemisorption of molecular oxygen on sintered catalysts. Oxygen was admitted in small doses to the sintered catalyst as above a t -78" after the usual reduction and evacuation. Figure 5 s h o m that molecular oxygen initially increased the magnetization but, after about 4 cc. per g. of nickel was adsorbed, further addition of oxygen decreased the magnetization. The total takeup of oxygen was corrected for physical adsorption by evacuation and readmission to atmospheric pressure. KO significant change of magnetization occurred upon evacuation and readmission, thus showing that no chemically bound oxygen was affected by the evacuation. After correcting for physical adsorption the total takeup of oxygen mas 23.90 cc./g. of nickel, or 1.8 times the hydrogen chemisorption a t the same temperature on the same catalyst. The oxygen-covered nickel was evacuated and (16) P. W. Selaood, J . Am. Chem. SOC.,80, 4198 (1958).
Sept., 1960
CHEMISORPTION OF ~ X Y G E NON NICKEL
hydrogen was then admitted. This, as shown in Fig. 5 , increased the magnetization to near the original value for an uncovered surface. Similar measurements were made a t - 130". Oxygen initially increased the magnetization of the nickel, but further addition beyond 14 cc./g. produced a decrease similar to that shown in Fig. 5. The total takeup of oxygen was 1.9 times that of hydrogen a t the same temperature. Evacuation and admission of hydrogen to the oxygen-covered surface caused an increase of magnetization. Measurements at - 183 to - 196" were complicated by physical adsorption. The initial portion of several isotherms obtained show a lack of quantitative reproducibility, but clearly show an increase of magnetization caused by oxygen chemisorption. After small doses of oxygen had been admitted to produce a total gain in magnetization of 1.6% in one run a ; - 196" a large dose of oxygen was admitted an13 decreased the magnetization to 94.6% of the original magnetization for a bare surface. A similar effect was found by warming the catalyst to near room temperature and re-cooling to - 196".
1117
-0.06 I Fig. 5.-Magnetieation-volume for hydrogen and oxygen or1 sintered nickel-on-kieselguhr and for hydrogen on an oxygen covered surface of the same sintered nickel a t - 78".
oxide system may need to be re-evaluated if gas analyses were not made during chemisorption measurements. The high reactivity of dispersed nickel necessitated sintering the catalyst in order to diminish oxidation and this led to the observation that sintered catalysts give, under certain conditions, a positive magnetization change on chemisorption of hydrogen or of oxygen. Discussion In summary the conditions tending to produce a h dispersed catalytic system such as nickel-onpositive magnetic change rather than a negative kieselguhr is well known to be pyrophoric. The admission of large quantities of oxygen, or even one are the following: (1) nickel particles rather diluted oxygen as air, can produce a complete oxi- larger than those normally produced in nickeldation of the nickel. It was hoped that the admis- silica catalyst and averaging in the neighborhood of sion of srriall doses of oxygen such as used in ob- 100 A. in diameter; (2) low temperatures; (3) low taining mLgnetization-volume or pressure-volume fields; (4) the use of AC rather than DC fields. isotherms mould produce a true chemisorption But these conditions are not sharply defined and instead of bulk oxidation. The decrease of magnet- positive magnetic changes may be produced with ization shown in Fig. 1 could be caused by chemi- DC fields at room temperature on certain samples. Four theories were considered during the course sorbed oxygen decreasing the magnetic moment of metallic n ckel due to covalent bonds formed on the of this work to explain the anomalous results. surface or to the total conversion of some of the These were (1) a temperature dependent equilibsuperpara nagnetic nickel to paramagnetic nickel rium between adsorption on two types of sites; oxide. (Nickel oxide is ordinarily antiferromag- ( 2 ) competing chemical and physical phenomena; netic, but i n a dispersed system it could be paramag- ( 3 ) competition between large metallic particles and netic.) The second explanation seems more plau- small atomic particles; (4)the effect of magnetic sible since the total oxygen uptake was three times anisotropy. An equilibrium between protonic and hydridic that of the hydrogen chemisorption, but other explanations related to the magnetic anisotropy are sites of adsorption on nickel already has been proposed to explain positive magnetic effects a t liquid not excluded. Penetration of the nickel lattice below the sur- hydrogen temperature. lo The positive effects were face may have occurred. An alternative interpre- thought to occur for small particles which became tation is ihen that chemisorption occurred during ferromagnetic only at low temperatures. This couthe initial portion with conversion to the oxide in pled with the relatively low uptake of hydrogen for later stagcs. But since the magnetization-volume extremely finely divided catalysts led to the conisotherm is a straight line with no breaks or change clusion that the direction of electron transfer deof slope, it is more likely that the same process was pended on the particle size. Hydrogen donated being measured in both the initial and later stages electrons to the nickel if the particles were large. of the isotherm, that is, oxidation. Local hot spots Nickel donated electrons to the hydrogen if the probably occurred which took up more than a particles were small. The present work a t first monolayel*of oxygen before all of the surface was supported the notion that there was an equilibrium covered. The total oxygen takeup a t one atm. between two sites. However since the positive effect became more positive with sintering, the pressure corresponded to the production of NiOo.3. The chromatographic analyses during measure- positive effect could not be due to negative hydride ments with nitrous oxide on unsintered catalysts ions on small particles. The positive magnetic showed tkat the dispersed nickel system is such a effect seemed more closely associated with large powerful c atulyst that molecular nitrous oxide can- particles than small particles. Since the sign of not be stably chemisorbed. Immediate decomposi- magnetic change depended on the temperature of 1ion to form an oxide layer and liberation of molec- measurement, it appeared that there might be ular nitrogen t o the gas phase occurred even a t some temperature dependent equilibrium involving - 3". -!ny previous work on the nickel-nitrous desorption and migration between sites. But any
1118
ROBERT J. LEAKAXD P. W. SELWOOD
mechanism proposing two sites, and an equilibrium between t,hem was ruled out by the experiment in which both positive and negative effects were found for the same adsorbed hydrogen when the magnetization was measured a t two different temperatures, the higher of which mas by no means high enough to cause appreciable desorption. The idea of a competit,ioll between a chemical and a physical phenomenon was based on the observation that measurements a t low fractions of magnetic saturat'ion showed a negative magnetic effect,, whereas measurements a t higher fractions of magnetic saturation showed a positive effect. High field st
