KISETICS OF HYDROGEX CIIE~IISORPTION OK NICKEL-JIhGNESIh CATALYSTS BY N.It,~Ji.\sTRi1.\~i.IsI.~N* AND L. M. YEDI)ASAP.LLLI Loyolu College, Madras R c c t w e d August 2'9, 1061
The Klovich plot8 of the data obtained on the system hydrogen on nickel-magnesia show the occurrence of diseontinuities. The two stagcs of adsorption are distinct and both the slow rates are determined only by the initial gas pressure. 'The rapid change in amount adsorbed, *tq, occurring in gas addition or removal experiments is pressure sensitive rind the prcscncbc of an initial chemisorbed state seems to be necessary to explain its occurrence.
Taylor and Thon' suggested that if CY in the equation dq/dt = ae-aq was characteristic of the nature of the sites involwd in the adsorption process, then a break in the p us. log (t to) plot indicattxl a changeover from one kind of site I O another a t a certain stage of the adsorption. The occurrcnre of discontinuities, or breaks, has bccm obscrvcd in quite a number of cases and is believed to be due to different types of adsorption sites. It is expected then that a study of the effect on the cy values, of presdsorbed gas, and of changes in gas pressure brought about before the occurrence of t he break, should provide information regarding the twhavior of the diff ereiit adsorption sites. Thc effect of changes in the prevailing gas pressure on the slow rate of adsorption has been st udicd by Gundry and Tompkins2 and Taylor, et C Z ~ . on , ~ the system hydrogen on nickel. The lattrr, in particular, noted the occurrence of breaks in the q us. log t plots but have confined their attention mostly to the first linear portion. LIoreovcr, in experiment s involving changes in ambient pressure there is observed a sniaI1 rapid adsorption, or desorption, k Aq aiid in order to understand the initial and ambient pressure cff ects more claarly, further work on this topic is considered to be nccessary. I -1preliminary study from this Laboratory of thc offccts of temperature and pressure on the adsorption of hydrogen on nickel-magnesia sho~ved that breaks appeared in plotting the data. There Seems to be no literature data, so far, dealing with hydrogen chemisorption kinetics on iiickel-magnesia. Therefore, the system hydrogen on nickelmagnesia was chosen to carry out a systeniatic study of how the slow rates of adsorption arc influenced by preadsorbcd hydrogen and changes in the prevailing gas pressure. It is felt that such a study would add to the knowledge of the nature of the chemisorption sites and of Aq, which is an expcrimcntal quantity of importslncc in undcrst anding t he mechanism of cheinisorpt ion. Experimental
+
Thv rat(, of adsorption vii1.9 follo\\ed in a coribtnnt volume system by measuring changes 111 pressure of hydrogen w t h The ttdsorberit \\'as prepared by precipitating the hydroxides from a solution of nickel nitrate (A.R. supplied by 13.D.H.) and magneeium nitratc (Scher_ -
* I l i r ision of Applied Cheniistiy, National Rescardi Council, Ottnua, Ontario. (1) H. A. Taylor and N Thon. J A m Chem. Soc., 74, 11b9 (19.52). (2) P. M. Gundry and r C Tornpkins, Trans. Faraday Soc , 62, 1bOO (1956). (3) L. Leibowitz, 31. .J. D. Idow, and If. A. Taylor, J l'hvs. CArm.,
ing Kahlbaum) in about 300 ml. of xatcr, using 400 ml. of 1 IV sodium hydroxide (E. lferck) solution, washing the precipitate over 2 days, drying a t 100' for 12 hr., and finally reducing the dried hydroxides in a stream of pure hydrogen. A samplc of the dried hydroxides was heated at 450 & 5" and the rcsulting oxides were analyzed for nickel. Catalyst I: Thc precipitation T V : ~carricd out from 10 g. of nickel nitrate and 50 g. of niabnesium nitrate in solution and the dried hydroxides were reduced at 475 f 10' for 24 hr. Catalyst 11: 18 g. of nickel nitratc arid 50 g. of niagiiesiurn nitrate were used and the dried hydroxides viere reduced at 390 i 5Ofor 60 hr. The nickel content* of the catalysts obtained from the analysis of the oxides w:re 36.16 arid 79.06y0, respectively. Purified gL3es of hydrogen, oxygcii, and nitrogcn w r e used in the e\perimcnts. In bctnecn any two adsorption cxpcrimcnts the adsorbent was degassed for 8 hr. a t 380 + 5" bv nicans of a cencoinegavac pump coupled with a inerciiry diffusion pump. During an adsorption cqx!rimerit the catalyst w m kept at the desired temperuture to within & l o by an electrically heated furnace.
Results The q us. log t plots for a few typical runs carried out on catalyst I1 are shown in E'ig. 1. Such runs will be referred to as the normal runs in which no interruptions by way of volume or pressure changes or preadsorbing hydrogen were made. It is evident from the linearity of the plots that to or I< is negligibly small. Consequently, the equation5 qnt
- qmt
=
2.303
11
-log 171 -Q
where n aiid ?n are intc.gcrs and qnt aiid gint correspond to amounts adsorbed at times nt and mt, respectively, mas used to calculate CY^ and cyz, the cy values of the two kiiict ic stagm, respectively. In Fig. 2 are presented the Elovich plots of the data obtained from a few experiments carried out on catalyst I to find out the effect of preadsorbed hydrogen on the slow rate of adsorption. The catalyst was exposed to hydrogen a t 252' and 60 cm. and adsorption followed for 2 min. after which the gas phasc was removed coniplet cly by pumping for 2 min. After waiting for a fixed period of time, during which thew was no measurable desorption, LL low pressure run was carried out at the same temperature of 252'. In Fig. 2 the plot for the iiormal low pressure run is included for the sake of comparison. The effect of preadsorbcd undefined amounts of hydrogen is scm to be to decrease the total subseyucwt adsorption of hydrogen. The effect is moro pronounced 011 the first kinetic stage than on the second as is evidenced by the cy \dues. The sccond linear portion is almost parallel to that in the normal run and cy2 shows a slight increase, whereas thcrrl is a threefold iiicrease ( 3 ) N. (19 56).
J. Sai~iioiisakisL i n i i i\;l
I). IAXV. J. Chem P h u s , 25, 178
July, 1962
KINETICS OF HYDROGEN CHEMISORPTION ON NICKEL-MAGKESIA CATALYSTS
1223
the adsorption system in the meanwhile constant. The results of the experiments carried out on catalysts I and I1 are given in Table I where a values .t 6.75G for the normal runs also are included. I n the table, 6 . 6 5 g to represents the time a t which the pressure change 6.55‘was brought about, =kAP the pressure change and f A q the rapid adsorption or desorption accompanying it. Taylor, et u Z . , ~ and Gundry and 6.35 Tompkins2have reported in their work on hydrogen 6.25 d; chemisorption on nickel that the slow rate of adsorption was unaffected by prevailing gas pressure changes. It is evident from Table I that not only 6.05 is a1 uninfluenced by pressure changes but also cyz, irrespective of the interruptions whether made before or after the break. Equally impor’ant is $ the quantity fAy and its dependence on pressure, 4.s2g temperature, and amount adsorbed. In the hy4.72 drogen-nickel work Taylor, et al.,3 did not find 5 any systematic variation of Aq, but the possibility 4.62g of the small rapid adsorptions and desorptions being 4.s2g pressure sensitive mas s u g g e ~ t e d . ~This is found 4 . 4 2 ~to be true in the present study. In the last column 4.32j of Table I are listed the values of the ratios of b; A P to Aq and the constancy of the ratio is readily seen for gas addition or removal experiments, re2 3 4 5 6 8 10 15 20253040506O80 spectivelf, The initial pressure seems to have a t, min. marked influence on this ratio which is strikingly Fig. 1.-Elovich plots; hydrogen adsorption on Xi-MgO temperature independent. I n the gas addition catalyst 11. experiments where the initial pressure is around -. 20 cm. this ratio of AP to Aq is seen to be higher by I 1a factor of more than two than in the gas removal 1.9 experiments carried out at an initial pressure of 62 em. 1.7 The data of a set of experiments on catalyst I1 are plotted in Fig. 3a. For a pressure change of 20 em. and more, the amount rapidly desorbed 1.5 is more than what was taken up slowly between the first minute and to. As a result of the gas removal, p/ 1.3 L the time of occurrence of the break is shifted furz ther and there is an interval of time At, only after which the interrupted runs become parallel to the 1.1 normal one and during which the rate is slower. d After the removal of the gas a t the 12th minute, normally only 0.1 ml. should be taken up for the 0.9 c break to occur a t the 20th minute; actually it is seen that 0.24 and 0.30 ml. are taken up and the 0.7 break occurs a t the 60th and 130th minutes, respectively. There is thus an indication of a siow desorption accompanying the rapid process. In all these respects, the case was similar at the other two temperatures. The plots C and E in the figure 2 3 4 56789 20 3040 6080100 show that the rapid desorption is likely to be not 10 50 70 90 characteristic of the kinetic stage and hence indet, min. Fig. 2.-Fffect of preadsorbed hydrogen; hydrogen ad- pendent of the amount adsorbed. The phenomenon sorption on Ni-MgO catalyst I: A, normal run; EL, pread- of rapid change in amount adsorbed, A A q , seems sorbed hydrogen 1.786 ml. NTP, waiting period 2 min.; to be important in understanding the mechanism of C,. preadsorbed hydrogen 1.751 ml. NTP, waiting period 30 chemisorption and is discussed later. min. I n another type of experiment interruptions in the a1values. It is interesting to note that the were made by cutting off a certain volume while time of occurrence of the break remains almost un- the slow rate was in progress and reconnecting it affected and that the waiting period is without after an interval of time. During this interval influence on the slow rates. the rate could still be followed on the manometer. When the slow adsorption was in progress changes I n such experiments the break occurred a t the same in the prevailing gas pressure were brought about time as in a normal uninterrupted run even though by the addition or removal of a certain quantity the buret’s volume, which was more than two times of gas a t a particular instant, keeping the volume of the volume of the rest of the system, was cut off at 6.95 6.85
6.45g
&
-z
t
62 63 ti5 G4 67 68 69 70
300 300 300 300 260 280 260 260
61.90 61.98 61.94 61.98 62.09 61.98 62.00 62.00
12 12 40 12
-11.50 -20.00 -19.60 -28.20
..
.....
12 12 12
-12.05 -20.40 -29.15
-0.213 ,484 .460 - .657
-
-
.....
-
,203 ,435 ,655
5.333 5.333
...
5.333 7.017 7.017 7.017 7.017
4.473 4 . ‘473 4.473
...
5.795 8.060 6.060
...
1.852‘ 2.170 2.347 2.330 . . . .1.933 i 11% 1.685 2.132 2,247
the Elovich plot are affected to different extents shows the distinctness of the t x o stages of adsorp6.8 tion. In other words, the fact that only the first 6.6 linear portion is affected means that the preadsorbed hydrogen was sufEcient to cover a portion 6.4 of the part of the catalyst surface that is character6.2 istic only of the first stage adsorption. This is 6.0 supported by the work of Taylor, ct aZ.,3 carricd out for a different purpose on the system hydrogen on G5.8 nickel-kieselguhr. An analysis of their data shows 5.6 that the latter course of the adsorption is comparaE n ‘I I D tively less affected than the initial part. The -.K LF ............................................... presence of different stages of adsorption in a _ , I system thus seems to be revealed by difyerent linear 7.2 portions on an Elovich plot and the brmk indicates 7.0 a changeover from one kinetic stagc to another. That at a given ternpcrature only thc initial gas 6.8 pressure determines the slow rate of adsorption a 6.6 already has been shown by Gundry and Tompkins2 %! 6 . 4 and Taylor, et aL3 As the same conclusion is seen to be true for both the linear portions on the 6.2 Elovich plots, it follows that the slow rates of 6.0 adsorption are determined only by the initial pressure even though the adsorption stagcs are distinct. The observation that there is a rapid 2 3 4 6 8 1 0 20 30 50 80 200 400 adsorption or desorption accompanying gas addi25 4060 100 tion or removal is in agrcemcnt with those of the 2, min. workers mentioned previously; it was reported3 Fig. 3.-Ilydrogen adsorption on Ni-MgO catalyst 11: (a) effect of gas removal and (b) effect of cutting off the that no systematic variation of Aq uith A p was burct’s volume. found and that no explanation of the phciiomenon was available. In contrast, in the present study an earlier stage and the rates of the slow adsorption a definite proportionality of Aq to A p is found in remained unaffected. The Elovich plots are gas addition and removal experiments, respectively, given in Fig. 3b. and the ratio is dependent on initial pressure but Discussion independent of temperature. Furthermore, it is In the study of the effect of preadsorbed hydro- seen that in gas removal expcr’mcnts the instnngen, the observation that the two linear portions on taneous desorption can \)e much more t h a n what 7.0
g
2
L
b- 6 1
E
was taken up slowly in that stage and it is not
characteristic of the adsorption stage. These observations indicate every possibility of it taking place from the initial fast adsorption before 'onset of stage oiie, which is considered generally to be temperature independent and strictly proportional to initial pressure. Then the fact that for a 50% change in pressure the -Aq forms only 10% of the amount rapidly adsorbed in the first minute poinls out that it is only a part of the initial fast adsorption that is likely to take part in rapid de-
sorption. This would mean that occurrence of a weak, non-activated chemisorption, Tvhich is considerede to be a precursor of a strong chemisorption, appears to be an essential step. Then, to explain the effect of initial pressure on & Aq, one has to take into account the fraction of surface available for coverage or covered with atoms in the initial chemisorbed state. (6) D. A. Dowden, "Chemisorption," Proe. Symposzum Keele, 1956, Edited by W. E. Garner, Butterworths Scientific Publicatlona, London, 1957, p. 3.
THE THERMAL DECOMPOSITION OF RUBIDIUXl SUPEROXIDE' BY D. L. KRAUSAND A. W. PETROCELLI Department of Chemistry, University of Rhode Island, Kingston, R. I . Received October 11, lQB1
The thermal decomposition of rubidium superoxide has been studied in the temperature range of 280 to 360'. The studies were carried out in a Pyrex glass high vacuum system. A specially designed and con&mctd torsion balance for "zn szlu" vacuum weighings was employed to determine the change in composition of the decomposing superoxide. The results shoa conclusively that the oxide, RbzOa, does not form in the course of the thermal decomposition of the superoxide, and that there is no solid solution formation between the decomposing superoxide and the, peroxide product, The reaction path is, thus, established to be RbOZ(s) = 1/2Rbz03(s) 1/202(g). The thermodynamic properties, AHo, ASO, K,, and AFo have been calculatoed for the above reaction. The thermal decomposition of the peroxide was studied in the temperature range of 300 to 360 , The thermodynamic properties, listed above, also have been determined for the reaction RbzOr(s) = R ~ z Q -
+
(8)
t- 1/202(g).
I n accordance with this view the thermal deIntroduction The higher oxides of t,he alkali metals have been composition would thus follow the path of interest to investigators since their discovery by KO2 +K 2 0 2 +KzO (2) Gay Lussac and Thenard2in 1810. The thermal decomposition processes of rubidium Of particular interest are the alkali metal super- superoxide are not characterized so well as those of oxides. The lon-er molecular weight superoxides potassium superoxide. It is of interest to point are :receiving considerable attention for possible out, however, that the same workers who argue use for the revitalization of sealed breathing atmos- against the existence of Kz03concede the possibility pheres. The need exists for an exact knowledge that rubidium may form the sesquioxide. The concerning the nature of the intermediate solid picture is further complicated by the view of phas'es formed in the course of their thermal decom- Brewer,g who states that potassium, rubidium, and position. I n general it is known that a t elevated cesium probably do not even form h12Q2solid temperatures these compounds release oxy!,sen re- phases. versibly and form lower oxides. Many ~ o r k e r s ~ - ~ The present investigation is concerned with the have claimed that in addition to the expected determination of the reaction path for the thermal peroxide, 1\1202,and the simple oxide, MzO, there decomposition of rubidium superoxide, and the calalso is formed a sesquioxide, hfz03. They propose culation of the thermodynamic properties obtainthat the superoxides decompose thermally in the able from such a study. following steps with the simultaneous libemtion of Experimental oxygen M O ~----j.
-+mo2---3 LLO
(1)
Strong evidence, however, has been uncovered supporting the view that, in the case of potassium a t least, t'he oxide analyzing as K203 is in fact either a solid solution or a mixture of KO2 and K202.+* (1) Based on work performed in partial fulfillment of the requirements of the Ph.D. degree of A. W. Petrocelli. (2) .J. L. Gay-Lussac and L. J. Thenard, Recherche8 Physico-Chinique.3, 8 , 182 (1811). (3) E. Itenrard, Ann,. chim. phys., 11,348 (1907). (4) R. de Forcand, C o n p i . rend., 150, 1399 (1910). (6) R.1. Centerszwer and M. Elumenthal, Bull. intern. acad. polon. B C ~ .Closse sci. math. nat., 19338, 499 (1933). ( 8 ) W. H. Schechter and J. Kleinberg, J. Chem. Educ., 24, 302 (1947) (7) .A. Klemm and H. Sodomann, Z. a n o ~ g .allgem. Chem.. 226, 273 (1935), ( 8 ) I. Xazarnovskii and S. I. Raikhshtein, Russ. J. Phys. Chen., 21, 245 (1047).
The RbOz was prepared in this Laboratory by 1,he direct oxidation of rubidium metal. The metal (99.30%) was supplied by the American Potash Company. The major impurity was cesium, 0.45%. The metal must be handled with extreme caution because of its reactivity. In the liquid state it reacts explosively yith air, and will ignite in the solid state on contact with air. It was shipped in a glass vial under argon gas. The label was scraped off the vial carefully, and the exterior of the vial washed with n-heptane. Then the vial was scratched with a file, broken rapidly into two halves, and submerged in n-heptane contained in a 1 1. beaker. The heptane was heated on a hot plate to above the melting point of the metal (38.5") and the metal was poured out of the two vial halves. The metal was stored under heptane in the beaker, which in turn was kept in a desiccator charged with magnesium perchlorate. Approximately 0.2-0.5 g. of metal, wet with the n-heptane, was transferred quickly from the beaker to a Pyrex (9) L. Brewer, Chem. Rev., 62, l ( l 9 5 3 ) .