J. Chem. SOC., Faraday Trans. 1, 1984,80, 1435-1446 Catalysis by Amorphous Metal Alloys Part 1.-Hydrogenation of Olefins over Amorphous Ni-P and Ni-B Alloys BY SATOHIRO YOSHIDA,* HIROMI YAMASHITA, TAKUZO FUNABIKI* AND TEIJIRO YONEZAWA Department of Hydrocarbon Chemistry, Faculty of Engineering, Kyoto University, Kyoto, Japan Received 12th July, 1983 Amorphous Ni-P (Ni81P19) and Ni-B (Nie2BS8) alloys have been prepared by the rapid quenching method. The untreated alloys are inactive for the hydrogenation of olefins, but successive pretreatments of the alloys with the dilute HNO,, oxygen and hydrogen bring about greater catalytic activity than that possessed by the crystalline alloys. Measurements of the ESCA spectra of the surfaces of the alloys have indicated that the alloys become active by the partial oxidation of the nickel and phosphorus or boron on the surface.Buta-1,3-diene is hydrogenated to give isomeric butenes of constant composition, and butenes isomerise rapidly to give a similar composition to that obtained on the hydrogenation of buta-1,3-diene. The hydrogenation of buta-1,3-diene is first and zeroth order with respect to the partial pressures of hydrogen and buta-l,3-diene, respectively. These results are explained by assuming that the hydrogen and the olefins are not competing for adsorption on the same sites and that hydrogen is activated on electron-deficient nickel atoms formed by electron transfer from nickel to the oxidised species. Much work has been done on the catalytic reactions over crystalline surfaces of metal alloys, but little has been done on the catalytic reactions over amorphous alloys whose surfaces are considered to be homogene~us.~-* It has been reported that such catalysts are formed in an amorphous state when the reduction of Ni-P and Ni-B alloys, known to be hydrogenation catalysts, is performed under mild condition^,^? but the homogeneity of the state and its role in catalysis are uncertain.It is very probable that the catalytic behaviour of amorphous alloys is very different from that of crystalline alloys, because the surface electronic state and the chemical composition may be different. We have previously reported’ that amorphous Ni-P and Ni-B alloys prepared by the rapid-quenching method are highly active for the hydrogenation of olefins in comparison with crystalline Ni-P and Ni-B alloys, and that the pretreatment conditions are very important.We report here the catalytic activity of the amorphous alloys for the hydrogenation of olefins and the relationship between the catalytic activity and the changes on the surface caused by the pretreatment, studied using ESCA (electron spectroscopy for chemical analysis). EXPERIMENT PREPARATION OF CATALYSTS Amorphous Ni-P (NiE1Pl9) and Ni-B (Nie2B3,J alloys were prepared by the rapid-quenching method8 using a single steel roll as reported previ~usly.~ Ni (99.5 wt %), Ni-P (P, 14 wt %) and Ni-B (B, 15 wt %) were used in the preparation and amorphous alloys in the shape of ribbons ca. 5 mm wide and 10-20 pm thick were obtained. An X-ray diffraction analysis (X.r.d.) was used to verify the amorphous state of the alloys2 before and after the reactions.The surface areas of the alloys were measured by the Brunauer-Emmett-Teller (B.E.T.) method using krypton physisorption at 77 Kg 14351436 CATALYSIS BY AMORPHOUS METAL ALLOYS PRETREATMENT OF CATALYSTS AND HYDROGENATION OF OLEFINS Hydrogen was purified by a hydrogen diffusion purifier (Japan Pure Hydrogen Co., LS-OOB) and commercial oxygen and olefins were used without further purification. The alloys were treated with HNO, (1.5 and 6 mol dm-3 HNO, for Ni-B and Ni-P, respectively) washed thoroughly with water and dried under air for 24 h. Cu. 0.5 g of the alloy was placed in a glass vessel which was connected to a conventional closed circulation system (total volume 210 cm3).The alloy was heated under circulating oxygen (6.7 kPa) and evacuated for 0.5 h. The alloy was then heated at 573 K under circulating hydrogen (13.3 kPa) and evacuated for 1 h. Hydrogen (pH2 = 19.3 kPa) and olefin (polefin = 7.3 kPa) were introduced to the reaction vessel and the reaction took placed under the circulating gases. The reaction was followed by measuring the pressure change and analysing the composition of the product by g.l.c., using a 3 m column of dimethylsulpholane on C22. Initial rates were estimated from the initial pressure change. The pretreatments and reactions were performed under different conditions (temperature and time) depending on the purpose of the experiments; the conditions are given in the captions of the tables and figures.The conditions were selected so as to determine the initial rates as accurately as possible. For example, the conditions for the Ni-P amorphous alloys shown in the tables were chosen so as to be able to determine the rate of highly reactive ethene from the change in the levels of a mercury manometer, and the conditions in fig. 3 were chosen so as to be able to determine the slow rates under low hydrogen pressures. The pretreatment conditions are given in the form (O,, 373 K, 2 h; H,, 573 K, 6 h), which denotes heating under oxygen at 373 K for 2 h, followed by heating under hydrogen at 573 K for 6 h. MEASUREMENTS OF ESCA SPECTRA The effects of the pretreatment and the reaction on the surface structure were studied by measuring the ESCA spectra (Shimadzu ESCA-750) using Mg radiation (10 kV, 30 mA). Amorphous and crystalline alloys were placed in the ESCA analyser chamber and spectra were measured either without sputtering or after sputtering the surface with Ar+ to depths of ca.5 and 25 nm (sputtering time 1 and 5 min, respectively). All binding-energy values (accuracy kO.3 eV) were referred to the value of the contaminant carbon (CIS = 285.0 eV) for convenience. RESULTS SURFACE AREA AND CRYSTALLISATION OF ALLOYS Differential thermal analysis indicated that the amorphous Ni-P and Ni-B alloys started to crystallise at 643 and 783 K, respectively. The crystallised alloys exhibited the X-ray diffraction patterns of crystalline Ni, Ni,,P,, Ni2B, B20, etc. However, X.r.d. indicated that both amorphous alloys continued to be amorphous even after treatment with hydrogen at 573 K for 10 h.The B.E.T. surface areas of the amorphous alloys were 0.12 and 0.094 m2 g-l for Ni-P and Ni-B, respectively, and those of the crystalline alloys were 0.12 and 0.1 1 m2 g-l for Ni-P and Ni-B, respectively. This result indicates that the surface areas of the amorphous and crystalline Ni-P and Ni-B alloys are very similar. HYDROGENATION OF OLEFINS The untreated alloys showed no activity for the hydrogenation of olefins even after hydrogen pretreatment of the alloys at 573 K for 6 h, but the alloys became active after successive treatments with dilute HNO,, oxygen and hydrogen. The three pretreatments were necessary to generate the catalytic activity. Fig. 1 and 2 show the effects of the pretreatments with oxygen and hydrogen, respectively, on the catalytic activity for hydrogenation of buta-1,3-diene.As shown in fig. 1, the catalytic activity is dependent on the temperature of the oxygen pretreatment, and oxygen treatment at a high temperature is necessary for activation of the alloys. In the case of the Ni-P alloy, abrupt enhancement of the activity was observed at 513 K, but the activityS. YOSHIDA, M. YAMASHITA, T. FUNABIKI AND T. YONEZAWA 1437 0 373 473 I 573 1.3 x cd 4 PI I z b) 1.2 2 0 x U .- .* U a W u .- 0.1 2 2 0 temperature of the oxygen pretreatment/K Fig. 1. Effect of the oxygen pretreatment on the catalytic activity for hydrogenation of buta-1,3-diene: 0, amorphous Ni-P; A, amorphous Ni-B. Pretreatments: (O,, 1 h; H,, 573 K, 2 h); reaction temperature: 373 K.Relative activities are the initial rates of hydrogenation normalised to that of amorphous Ni-P pretreated with 0, at 493 K. I .o 0 2 4 6 8 10 time/h Fig. 2. Effect of the hydrogen pretreatment times on the catalytic activity of hydrogenation of buta-1,3-diene: 0, amorphous Ni-P; A, amorphous Ni-B. Pretreatments: Ni-P (O,, 573,2 h; H,, 573 K) and Ni-B (O,, 473 K, 1 h; H,, 573 K); reaction temperature: 473 K (Ni-P) and 423 K (Ni-B). Relative activities are the initial rates normalised to those of Ni-P and Ni-B pretreated by hydrogen for 2 and 6 h, respectively.1438 CATALYSIS BY AMORPHOUS METAL ALLOYS Table 1. Product distribution in the hydrogenation of buta-1,3-diene over Ni-P and Ni-B alloysa products (mol %) conversion cis: trans ratio catalyst (mol %)B butane but- 1 -ene cis-but-2-ene trans-but-2-ene of but-2-ene Ni-P, amor.* 73 1.5 57.7 14.9 25.9 0.58 Ni-P, crystc 77 1.4 59.9 12.7 26.0 0.49 Ni-B, amor.d 73 4.1 52.8 13.2 29.9 0.49 Ni"" 75 12 53 5 31 0.16 Ni,Pe 75 3 32 10 55 0.18 Raney Nif 55-75 66 7 6 21 0.29 a Initial pressure: pH2 = 19.3 kPa andpbutadiene = 7.3 kPa; reaction temperature: 373 K (Ni-P) and 423 K (Ni-B).* Amorphous Ni-P (0.404 g) (02, 523 K, 2 h; H,, 573 K, 2 h). Crystalline Ni-P obtained by heating the above catalyst at 723 K for 0.5 h in uacuo (O,, 523 K, 2 h; H,, 573 K, 2 h). Amorphous Ni-B (0.322 g) (O,, 423 K, 1 h; H,, 573 K, 3 h). eData were estimated from the figure in ref. (6). Ni": alumina-supported nickel phosphate catalyst containing 20 wt % Ni,(PO,), was calcinated at 873 K and reduced at 673 K; Ni,P: the above catalyst reduced at 823 K. f Ref.(1 1). B Based on buta-1,3-diene introduced into the reactor. 10 0 Q) 5 Y 3 P a s 0 0- 0-0- 0.5 1 .o 1.5 1 .o lli E + c, P (u 0 0.5 2 E 0 time/h Fig, 3. Reaction of but-1-ene over amorphous Ni-P: A, but-1-ene; 0, cis-but-2-ene; A, trans-but-2-ene; 0, butane; 0, cis: trans ratio of but-2-enes. Initial pressure: pH2 = 19.3 kPa and pbutene = 7.3 kPa; pretreatment: (O,, 523 K, 2 h; H,, 523 K, 2 h); reaction temperature: 393 K. decreased on treatment at higher temperatures. The activity of the Ni-B alloy increased gradually and reached a maximum at 473 K. The decrease in the activity on treatment at higher temperatures was not significant. Fig. 2 shows the effect of different hydrogen pretreatment times at 573 K.The catalytic activity reaches a maximum after 2 h for Ni-P and after 6 h for Ni-B, and then decreased on prolonged treatment.S. YOSHIDA, M. YAMASHITA, T. FUNABIKI AND T. YONEZAWA 1439 Table 2. Product distribution in the hydrogenation of isoprene over Ni-P and Ni-B alloysa products (mol %) conversion 2-methyl- 2-methyl- 3-methyl- 2-methyl- catalyst (mol %I" butane but-1 -ene but-1-ene but-2-ene Ni-P, amor.b 56 2.7 42.4 13.6 51.3 Ni-B, amor.C 71 2.3 44.9 22.2 30.6 Raney Nid 2 41 16 41 a Initial pressure: pH2 = 19.3 kPa and pisoprene = 7.3 kPa; reaction temperature: 403 K Ref. (17). " Based (Ni-P) and 423 K (Ni-B). b* See footnotes b and din table 1, respectively. on added isoprene introduced to the reactor. Table 3. Initial rates in the hydrogenation of olefins over Ni-P and Ni-B alloysa initial rate/kPa min-' catalyst ethene propene cis-but-2-ene buta- 1,3-diene isoprene Ni-P, amor.b 13.89 3.62 0.37 0.23 0.05 Ni-B, amor.d 0.75 0.57 0.12 0.58 0.3 1 Ni-P, cryst.c 8.79 2.47 0.30 0.08 0.0 1 Ni-B, cryst." c 0.01 - - - < 0.01 a Initial pressure: pH = 19.3 kPa andp,,,,,, = 7.3 kPa; reaction temperature: 373 K (Ni-P) Amorphous Ni-B was and 423 K (Ni-B).b-d See footnotes b-d in table 1, respectively. crystallised by heating at 793 K for 2 h in uacuo. Table 1 shows the product composition for the hydrogenation of buta-1,3-diene catalysed by the amorphous alloys and, for comparison, by other nickel catalysts. Amorphous Ni-P and Ni-B alloys gave similar selectivities and crystallisation did not affect the selectivity. Amorphous alloys formed butenes until the level of conversion reached 95 %, showing that no competitive hydrogenation of butenes occurred in the presence of buta- 1,3-diene.The composition was constant throughout the reaction. Formation of but-1-ene as a main product was similar to that for a nickel catalyst prepared from nickel phosphate on alumina by reduction with hydrogen at 673 K, but different from that for a catalyst reduced at 923 K 6 ~ lo and from that for Raney nicke1.l' The cis: trans ratio for but-2-ene was ca. 0.5 for the amorphous and crystalline alloys, unlike other catalysts. In the absence of buta-1,3-diene, the butenes iso- merised rapidly to give a composition similar to that obtained on the hydrogenation of buta- 1,3-diene.Fig. 3 shows the reaction of but- 1 -ene and indicates that isomerisation proceeds more rapidly than hydrogenation. But-2-enes reacted in a similar way. Isoprene was also hydrogenated, yielding methylbutenes with aconstant composition throughout the reaction; the hydrogenation of methylbutenes did not proceed until the conversion level of isoprene had reached 95%. As shown in table 2, the composition of the products was slightly different for the two amorphous alloys, i.e. 2-methylbut- 1 -ene > 2-methylbut-2-ene > 3-methylbut- 1 -ene (Ni-P) and 2-methyl- but-2-ene > 2-methylbut- 1 -ene > 3-methylbut- 1 -ene (Ni-B). The selectivity was not greatly different from that obtained with Raney nickel.1440 CATALYSIS BY AMORPHOUS METAL ALLOYS I I 6.7 13.3 20.0 PH, Or PC,H6 lkPa 0 Fig.4. Dependence of the initial rates of hydrogenation of buta-1,3-diene on the partial pressures of hydrogen and buta-1,3-diene. Dependence on pC4Hs at !HZ = 19.3 kPa: A, Ni-P; A, Ni-B: Dependence on p H z at pCpHs = 7.3 kPa: a, Ni-P; 0, Ni-B. Pretreatments: Ni-P (02, 523 K, 2 h; 573 K, 2 h) and Ni-B (02, 373 K, 1 h; 573 K, 6 h); hydrogenation temperature: 373 K. Relative rates are the initial rates normalised to those under pHz = 19.3 kPa and pCaHs = 7.3 kPa. Table 3 shows the initial rates of hydrogenation of olefins and diolefins over the amorphous and crystalline Ni-P and Ni-B alloys. The crystalline alloys were less active than the amorphous alloys, and crystalline Ni-B exhibited very little activity under our conditions.The reactivity decreased with increasing methyl substitution, the effect of the methyl substituent being more marked for Ni-P than for Ni-B. A kinetic study of the reaction of buta-1,3-diene was carried out by varying the partial pressures of buta- 1,3-diene and hydrogen. As shown in fig. 4 the relative initial rates can be used to show the dependence on the partial pressures as some variations in the activity were observed on repeated runs. Thus, when the reaction was performed repeatedly without pretreatment with oxygen and hydrogen after each run, the initial rate under the same partial pressures was 80-90% for each run in the first 2 or 3 runs and then became constant. When this catalyst was held at room temperature overnight after evacuation, the rate was 30-50% of that of the fresh catalyst, but the activity was restored to that of the fresh catalyst by pretreatment with oxygen and hydrogen.To correct the variation in the activity caused by repeated use of the same catalyst reaction with 19.3 kPa hydrogen and 7.3 kPa buta-1,3-diene was performed as a standard after each run. In spite of the variation of the magnitude of the standard rate, the ratio of the rates was almost reproducible and the results in fig. 4 indicate that the rates are first and zeroth order with respect to hydrogen and buta-1,3-diene, respectively.S . YOSHIDA, M. YAMASHITA, T. FUNABIKI AND T. YONEZAWA 1441 L -- -- binding energy/eV 860 850 I90 180 860 850 190 180 Fig. 5. ESCA spectra of amorphous Ni-B. Successive treatments: (a) untreated, (b) HNO,, (c) 0, at 373 K for 1 h, (d) H, at 573 K for 6 h, (e) hydrogenation of buta-1,3-diene at 373 K, v) crystallisation at 873 K for 2 h in vacuo.Sputtering: ca. 5 nm; scale of peak intensity ( x , ~ ~ ) : x ( N ~ ~ ~ ~ , ~ ) = (a) 4000, (b) 10000, (c) 4000, ( d ) 2000, (e) 2000, cf) 400; x(B,,) = (a-f) 200. ESCA SPECTRA The changes in the surface states of the amorphous alloys following pretreatment, hydrogenation and crystallisation were studied by ESCA. Fig. 5 shows the spectra of the amorphous Ni-B alloy measured after Ar+ ion sputtering of the surfaces to a depth of ca. 5 nm. The spectrum of the untreated alloy [fig. 5 (a)] shows a band from nickel (2p3,& and two bands from boron (Is). The binding energy of the nickel (852.95 eV) is greater than that of pure nickel (852.2 eV)12 and those of boron (188.20 and 192.20 eV) are also greater than that of pure boron (187.8 eV).12 The latter band, which may correspond to boron bound to oxygen,13 was observed even after sputtering the surface to a depth of ca.25 nm. The spectrum after pretreatment with dilute HNO, [fig. 5(b)] shows a nickel band (852.85 eV) and a boron band (188.10 eV), but the boron oxide band has disappeared. When the spectrum was measured without sputtering of the surface the boron oxide and the oxidised nickel bands were observed at 191.95 and 855.80 eV, respectively. The spectrum after pretreatment with oxygen [fig. 5(c)] shows bands from nickel (853.00 eV), nickel oxide (856.05 eV), boron (1 88.15 eV) and boron oxide (192.20 eV). However, when the spectrum was measured without sputtering, the nickel band was very small.When the oxidised alloy was treated with hydrogen, two bands from nickel (853.15 eV) and nickel oxide (856.60 eV) were observed with little change in the1442 CATALYSIS BY AMORPHOUS METAL ALLOYS L L - - 860 8 50 140 130 binding energy/eV Fig. 6. ESCA spectra of amorphous Ni-P. Pretreatment: (02, 513 K, 1 h; H,, 573, 2 h); sputtering: (a) ca. 25 nm, (b) ca. 5 nm, (c) 0 nm; scale of peak intensity ( x , ~ ~ ) : x ( N ~ ~ ~ ~ , ~ ) = (a) 1000, (b) and (c) = 2000; x(P2J = (a-c) 200. relative intensity. The boron band disappeared and only the boron oxide band (192.80 eV) was observed [fig. 5(d)]. However, when the spectrum was measured without sputtering, a small nickel band was observed, indicating reduction of nickel oxide.Fig. 5 (e), which was measured after hydrogenation of buta- 173-diene, showed the decrease in the intensity of the nickel oxide band (856.80 eV) compared with that of nickel (853.05 eV). The boron band (192.80 eV) remained unaltered. When the alloy was crystallised by pretreatment under severe conditions (873 K, 2 h) in vacuo, the nickel band disappeared and the boron band appeared at higher energy (193.75 eV) [fig. 5 d f ) l . Similar spectral changes were observed with the Ni-P alloy, and bands from nickel (853.35 eV), nickel oxide (856.85 eV), phosphine (129.60 eV) and phosphine oxide (1 33.70 eV) were detected. However, the reactivities of Ni-B and Ni-P with oxygen and hydrogen are different. For example, the nickel band was observed after oxidation even when the spectrum was measured without sputtering, and the nickel oxide band was not observed in the spectrum measured after sputtering to a depth of ca.5 nm. Fig. 6 shows the spectra of the alloy pretreated successively with HNO,, oxygen and hydrogen. The ratio of the peak heights of nickel oxide to nickel decreased withS. YOSHIDA, M. YAMASHITA, T. FUNABIKI AND T. YONEZAWA 1443 increasing time of hydrogen treatment (2.13,O h; 0.69,2 h; 0.35,4 h) but the relative intensities of the peaks of phosphine oxide and phosphine were nearly constant. DISCUSSION We expected that untreated amorphous Ni-P and Ni-B alloys would exhibit catalytic activity for the hydrogenation of olefins, but preliminary experiments indicated that the alloys were inactive, irrespective of the reaction conditions.It was assumed that the surface of the alloys might not be clean as the amorphous films were prepared by rapidly cooling the melted mother alloys in air. However, pretreatment with dilute HNO,, in order to clean the surfaces, or with hydrogen, in order to reduce the oxidised species which might be present, were ineffective. On the contrary, we have found that pretreatment with oxygen at fairly high temperatures is essential for catalytic activity and that successive treatments with acid, oxygen and hydrogen are very effective. This indicates that modifications of the amorphous surface by oxidation and reduction are very important. The results in fig. 1 indicate that the effects of the temperature of pretreatment with oxygen are different for the two alloys.Since the effect appears at a lower temperature for Ni-B than for Ni-P, the oxidation must proceed more readily with Ni-B than Ni-P. Ni-B showed nearly constant activity in the range 473-623 K, but the activity is lower than that of Ni-P. On the other hand, the activity of Ni-P is very sensitive to the pretreatment temperature, and this alloy exhibits very high activity in a very narrow range of temperature. Over-oxidation brings about a large reduction in the activity. The oxygen treatment must form oxygen adducts of nickel and metalloids (phosphine and boron), but the oxidised alloys are not active unless reduced by hydrogen. Fig. 2 shows that the reduction proceeds more readily with Ni-P than with Ni-B, and the activities of both alloys decrease on over-reduction, indicating the importance of the partial reduction.The above results suggest that the catalytic activity of these amorphous alloys requires the presence of both oxidised and reduced species on the surface and that it is dependent on the composition of these two species. This is clearly shown by ESCA. The spectrum of the untreated Ni-B alloy indicates the formation of boron oxide on the surface during the preparation of the alloy [fig. 5(a)]. The binding energies of nickel and boron are shifted from those of pure nickel and boron, indicating an interaction between nickel and boron. The positive shift for boron indicates electron transfer from boron to nickel. On the other hand, the spectrum of the Ni-P alloy shows a negative shift of the phosphorus band compared with pure red phosphorus, indicating electron transfer from nickel to phosphorus.Fig. 5(b) indicates that the metalloid oxides of the untreated alloy are removed by the acid treatment. Since the spectra measured without sputtering shows the bands of oxides of nickel and metalloids, after the acid treatment the surface must be very sensitive to oxygen. Fig. 5(c) indicates that Ni-B is oxidised to form oxides of nickel and boron not only on the surface but also inside the alloy. The nickel species on the surface is mainly nickel oxide because the nickel band is negligible in the spectrum obtained without sputtering. On the other hand, the spectra of Ni-P indicate the presence of both of nickel and nickel oxide on the surface and the absence of nickel oxide after sputtering.These results indicate the more facile oxidation of Ni-B than of Ni-P, which is consistent with the results in fig. 1. The spectra of the alloys measured after hydrogen treatment indicate that hydrogen reduces nickel oxide but not metalloid oxides. The reduction of nickel in Ni-P1444 CATALYSIS BY AMORPHOUS METAL ALLOYS proceeds more readily than that of nickel in Ni-B, which is consistent with the results in fig. 2. The more facile oxidation and the less facile reduction of nickel in Ni-B than in Ni-P seem to be related to the different mode of electron transfer between nickel and the metalloids. Further oxidation of the metalloids proceeds during the reduction and shifts in the binding energies of the oxides of nickel and the metalloids are observed.This is because oxygen bound to nickel is removed on the reduction of the Ni-B alloy with hydrogen and the strength of the boron-oxygen bond increases.12 The results shown by the ESCA spectra indicate that the pretreatments modify the amorphous alloys to form nickel, nickel oxide, metalloid and metalloid oxides, not only on the surface but also inside the alloy. The composition of these species affects the catalytic activity, because it determines the number and electronic state of active nickel species on the surface. The reduced nickel adjacent to oxides of nickel or the metalloids must be more electron deficient than that without oxides as neighbours. Thus, it is important to know the composition which gives the maximum activity and the optimum pretreatment conditions which bring about this composition.The spectrum measured after the hydrogenation of olefin indicates the partial reduction of nickel oxide to nickel, as shown in fig. 5(e). It is probable that the reduction takes place during the hydrogenation, and the change in the composition of species on the surface may result in the observed variation of catalytic activity in repeated hydrogenations. A change in the surface composition also occurs when the alloy is treated under severe conditions so as to cause crystallisation. As shown in fig. 6(f), nickel is transferred from the surface to the interior of the alloy, resulting in a low level of reduced nickel on the surface. The positive shift of the binding energy of boron oxide suggests the formation of stable boron-oxygen compounds such as B203,13 which was detected by X.r.d.The low catalytic activities of the crystalline alloys are caused by these changes in surface composition and structure, as the surface areas are essentially the same for the amorphous and crystalline alloys. A structural change in the amorphous alloy starts at a lower temperature than that estimated by differential thermal analysis and alloys pretreated at high temperatures may undergo structural changes which cannot be detected by X.r.d. The drop of the catalytic activity of Ni-P on treatment at > 533 K under oxygen (fig. 1) seems to be correlated with this type of structural change because the crystallisation temperature of Ni-P is lower than that of Ni-B, but we have not obtained any information about the structures of amorphous alloys in precrystallisation states.The presence of different types of active species is also suggested by the kinetic studies. The rate equation for the hydrogenation of buta- 173-diene is r = kPAZ P! (9 where r, k, p H , and p s denote rate, rate constant and the partial pressures of hydrogen and diene, respectively. This result is well explained by the following reaction sequence: K , Hz 2Hads K3 Hads + Sads f SHads (3) k Hads + SHads SH2 (4)S. YOSHIDA, M. YAMASHITA, T. FUNABIKI AND T. YONEZAWA 1445 assuming that (a) hydrogen and diene do not compete for adsorption on the same site, (b) adsorption of hydrogen is very weak compared with that of diene and (c) desorption of product is very rapid.Assumptions (a) and (b) are suggested by eqn (i)14 and assumption (c) by the result that hydrogenation of diene proceeds without hydrogenation of olefin in the presence of diene. The latter result is explained by the stronger adsorption of diene than olefin, which results in the rapid replacement of adsorbed olefin with diene. With NH andN, denoting the numbers of sites for adsorption of hydrogen and diene, respectively, we obtain ~ NH dK P H , + d K I P H z iHadsl = Then, the rate of hydrogenation of olefins is represented by (ii) (iii) Assumptions (a) and (b), i.e. K2ps 9 1 9 K 1 p H 2 , will lead to which is consistent with eqn (i). Eqn (v) indicates that the rate is affected not only by the partial pressure of hydrogen, but also the amount and electronic states of the active sites. Although we have performed kinetic studies only for the hydrogenation of buta-1,3-diene, it is very probable that eqn (v) is applicable to other olefins.The dependence of the rate on the reactant as shown in table 3 is ascribed not only to the different values of k but also to K3 and Ns, which are affected by the electronic and steric factors of the olefins. It seems reasonable to assume that hydrogen is activated by adsorbing on the electron-deficient nickel atoms and the olefins by adsorbing on the electron-rich nickel atoms. It has been reported that in the hydrogenation of styrene on Ni-B and Ni-P catalysts13 the increased (decreased) electron density on nickel caused by electron transfer from boron to nickel (nickel to phosphorus) weakens (strengthens) the adsorption strength of the reactants, resulting in the reduction (promotion) of self-poisoning effects by the adsorbed reactants.However, stable n-olefin complexes are usually formed with electron-rich low-valent metal complexes, and the stability of the complex is increased (decreased) by electron- withdrawing (donating) substituents on the 01efins.l~ The electronic effects are consistent with the substituent effects in table 3, supporting the explanation of reactivity given by eqn (v) rather than by the self-poising effect. The composition of the butenes produced by the hydrogenation of buta-1,3-diene (but-1-ene > trans-but-2-ene > cis-but-2-ene) and the cis: trans ratio of the but-2-enes (ca. 0.5) suggest that the mechanism involves a number of reversible steps which permit conformational interconversion of the di-n-adsorbed buta- 1,3-diene. l6 The mechanism involves interconversions among half-hydrogenated adsorbed species such as Q- and z-butenyl and n-methylallyl species.The interconversion is very rapid since the composition of the butenes is constant throughout the reaction. Interconversion of these intermediates is involved in the isomerisation of the butenes, because the composition of the butenes obtained on the hydrogenation of buta-l,3-diene is similar1446 CATALYSIS BY AMORPHOUS METAL ALLOYS to that obtained on the isomerisation of the butenes. This indicates that the isomerisation proceeds by the abstraction-addition mechanism via adsorbed n-allylic intermediates rather than the addition-abstraction mechanism via adsorbed a-butyl species.This is consistent with the result that isomerisation of the butenes which does not necessarily require the activation of molecular hydrogen proceeds more rapidly than hydrogenation. Kinetic studies of the isomerisation of the butenes have not been performed, but the rate equation may be different from eqn (v) even if the same intermediates are involved in both the hydrogenation of buta-l,3-diene and the isomerisation of the butenes. In conclusion, amorphous Ni-P and Ni-B alloys prepared by the rapid quenching method become active for the hydrogenation and isomerisation of olefins after pretreatment with acid, oxygen and hydrogen. The pretreatments modify the amor- phous alloys by forming oxides of nickel and metalloids on the surface and inside of alloys.The oxides may be able to form electron-deficient nickel species, and the catalytic activity is dependent on the amount of the nickel species in the different electronic states on the surface. Structural changes which are not detected by X.r.d. may occur in the pretreatment procedures, but the relationship between the catalytic activity and the structural changes has not been clarified in the present study. G. V. Smith, W. E. Brower, M. S. Matyjaszcyk and T. L. Pettit, in Proc. 7th In?. Cong. Catalysis, ed. T. Seiyama and K. Tanabe (Elsevier, Amsterdam, 1981), part 1, p. 335. A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto and H. M. Kimura, J. Catal., 1981,68, 355. A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto and H. M. Kimura, Scr. Metall., 1981, 15, 365. A. Yokoyama, H. Komiyama, H. Inoue, T. Masumoto and H. M. Kimura, J. Chem. SOC. Jpn, 1982, 2, 199. R. C. Wade, D. G. Holar, A. N. Hughes and B. C. Hui, Catal. Rev., 1976, 14, 21 1. F. Nozaki and R. Adachi, J. Catal., 1975, 40, 166. 'I S. Yoshida, H. Yamashita, T. Funabiki and T. Yonezawa, J. Chem. SOC., Chem. Commun., 1982,964. * Proc. 3rd In?. Con$ Rapidly Quenched Metals, ed. B. Canter (The Chameleon Press, London, 1978). lo F. Nozaki, T. Kitoh and T. Sodesawa, J. Catal., 1980, 62, 286. R. A. Beebe, J. B. Beckwith and J. M. Honig, J. Am. Chem. SOC., 1945,67, 1554. W. G. Young, R. L. Meier, J. Vinograd, H. Billinger, L. Kaplan and S . L. Linden, J. Am. Chem. SOC., 1947,69, 2046. V. V. Nemoshkalenko, A. I. Kharlamov, T. I. Serebryakova and V. G. Aleshin, Kinet. Katal., 1978, 19, 1567. l3 Y. Okamoto, Y. Nitta, T. Imanaka and S. Teranishi, J. Chem. SOC., Faraday Trans. I , 1979,75,2027. l4 K. Soga, H. Imamura and S. Ikeda, J. Phys. Chem., 1977,81, 1762. l5 E. 0. Fischer and H. Werner, Metal 7r Complexes (Elsevier, Amsterdam, 1966), vol. 1 ; M. Herberhold, Metal 7r Complexes (Elsevier, Amsterdam, 1972), vol. 2. l6 J. J. Phillipson, P. B. Wells and G. R. Wilson, J. Chem. SOC. A, 1969, 1351; P. B. Wells and A. J. Bates, J. Chem. SOC. A, 1968, 3064; B. J. Joice, J. J. Rooney, R. B. Wells and G. R. Wilson, Discuss. Faraday SOC., 1966,41, 223; G. Webb, Catal., 1978, 2, 145. G. C. Bond, Catalysis by Metals (Academic Press, London, 1962), p. 307. (PAPER 3/1205)