812
Langmuir 1986, 2, 812-820
is small are discussed in some detail in ref 14. Appendix Expression for (dI'z/dI'l)pz in the Case of Ideal Surface Mixing. We may define ideal surface mixing as that for which p1 and p2 are given by pl0(T,p,a)+ RT In x1
(Ala)
= ~ ~ ' ( T , P+, T R )T In xz
(Alb)
p1 = 112
where
Consider now eq A3. The quantities rl0and rzo may both be regarded as functions of a. Hence the differential of (A3) may be written as
The next step is to eliminate da from eq AI and AS. After some manipulation we obtain
-
plo and pzodenote the chemical potentials of pure 1 and 2 a t the same T , p , and a as the system of interest. It is
straightforward to show that when eq A1 holds we also have 1=
rl/rlo + rz/r20
(-43)
where rl0and rzorefer t o monolayers of the pure components at the T , p , and a of interest. At constant pz, eq A l b enable us to write dp20
+ R T d In x 2 = 0
(A41
where D denotes the term in brackets in eq AS. In the limit that
rl
- rz- rzo, - rzo 0,
and D
1 drZ0 dT
--
and we find that
Since
dpzo= d7r/rZ0
(A5) In general the rhs of (A10) is clearly nonzero. For example, approaches a limiting "saturation" value when a is if rZo large, dr:/da becomes quite small and limrl+, (dI'2/dI'l)fi2 -(l?zo/I'lo). On the other hand, in the ideal dilute limit the lhs of eq A10 is indeed zero, where 7r = R m ? = RmZo as expected.
and
-
eq A4 becomes
Influence of the Support and of the Preparation on the Surface Structure of the Ru/MgO System Eugenio Guglielminotti Istituto Chimica Fisica, Universitci di Torino, 10125 Torino, Italy Received May 2, 1986. I n Final Form: August 11, 1986 An Ru/MgO catalyst was prepared by impregnation of a Ru3(C0)12solution followed by activation in vacuum at 623 K and H, reduction at the same temperature. The IR spectra of adsorbed CO were utilized to characterize the Ru surface. Two types of Ru sites are displayed: one adsorbing CO linearly (vco = 2040-1960 cm-') and the other as a bridged species (vc0 = 1950-1875 cm-'), the frequency decreasing with the coverage. These absorptions correspond to the formation of Ru microcrystals exposing different faces; treatments of the sample in mild oxidizing conditions favor the formation of microcrystals where the bridged CO species is prevailing, whereas strongly oxidative conditions lead to a RuOz phase inactive to CO adsorption. The possible influences of these surface structures on the activity and selectivity of the Ru/MgO catalyst toward the CO-H2 reaction are considered. Introduction Ruthenium has received in the last years increasing interest both as an unsupported metal adsorbing various gases'" and as a supported catalyst,"" whose activity and (1) Pfniir, H.; Menzel, D.; Hoffmann, F. M.; Ortega, A,; Bradshaw, A. M. Surf. Sci. 1980, 93, 431.
0743-7463/86/2402-0812$01.50/0
selectivity in CO hydrogenation is strongly determined by the support and the type of preparation. In particular the (2) Reed, P. D.; Comrie, C. M.; Lambert, R. M. Surf. Sci. 1977,64, 33. (3) Barteau, M. A.; Broughton, J. Q.; Menzel, D. A p p . Surf. Sci. 1984, 19, 92. (4) Hrbek, J.; De Paola, R. A,; Hoffmann, F. M. J.Chem. Phys., 1984, 81, 2818.
0 1986 American Chemical Society
Surface Structure of the RulMgO System
Langmuir, Vol. 2, No. 6, 1986 813
comparison between the IR spectrum of CO adsorbed on monocrystals and the spectrum of CO adsorbed on supported metal in some cases12 can give useful1 information about the presence of predominant faces in supported metal. In effect the interaction between a metal and the support can be structural or electronic;6the first modifies the concentration of active sites, whereas the second involves charge transfer between the metal and the support: the strong metal-support interaction (SMSI) can be an 250 450 650 850 1050 12 J example of the second type of interaction. However, it is nm difficult to asses the difference between the two effects, especially when the catalytic activity of a “test” reaction Figure 1. UV-vis-near spectra of 1% Ru/MgO sample: (a) is assumed as the main criterium of distinction. R U ~ ( C O ) ~ ~ sample / M ~ Oaged for 6 mo after the impregnation Morris et aL6 have shown that chloride-free Ru/MgO process; (b) activation at 623 K (5 h) (and reduction with H2 at the same temperature); (c) difference spectrum (b) - (a). catalyst is less active than Ru/Si02, Ru/l3X zeolite, and Ru/Ti02 catalysts in CO hydrogenation. In fact Ru/MgO Our previous experiments on R u ~ ( C ~ ) samples ~ ~ / A ~ ~ ~ samples prepared from RuCl, give methane as the prehave shown that the decarbonylation accompanied by the vailing product, whereas those prepared from Ru(acac), support hydroxyls interaction and the conditions of reor R u ~ ( C O are ) ~ ~more selective for propene and higher duction in flowing H, or in vacuo can favor the prevailing hydrocarbons. Because these effects are similar to the promotional effect of alkali metals added to r ~ t h e n i u m , ’ ~ formation of metallic zerovalent particles or of oxidized Ru sites. a similar charge-transfer effect from the magnesia to the In this paper we try to evidence the structural effects Ru particles is invoked in order to explain the catalytic of the support and the sample preparation on the type of behavior. The charge transfer would lead to a weakening sites found on a Ru,(CO),/MgO sample. In a forthcoming of the CO bond and to a strengthening of the Ru-C bond paper17 some results on the reactivity and hydrogenation (the vco frequency of adsorbed CO is strongly shifted to of CO and other molecules on the same catalyst will be lower frequencies) and consequently to a reduced hydrogen reported. adsorption. This effect can explain the lower activity and the shift of selectivity toward higher hydrocarbons and Experimental Section ethene production found on Ru/MgO samples. Several Ru/MgO samples have been prepared by impregnation The effect of MgO support in suppressing the activity of a n-pentane solution of R u ~ ( C Owith ) ~ ~a high-area Mg(OH)2 of Ru for the CO hydrogenation reaction was remarked produced by Carlo Erba. The solvent was than recycled by a also by Schwank and co-workers.14 On the contrary a Soxhlet evaporation technique at low temperature (-330 K) until promoter effect of MgO in the N2dissocation and therefore it distilled nearly uncolored. Due to the low solubility of R U ~ ( C O ) ~ ~ in the ammonia synthesis vas observed by Aika et al.7 and in hydrocarbons, only samples with a low Ru contents (1% Ru/MgO by weight) can be prepared by impregnation and simple by Bossi e t al.15 A h:,lier selectivity in the Fischerevaporation of the solvent. The samples are dried in air at 343 Tropsch synthesis toward oxygenated compounds such as K and during this operation the color changes from orange to alcohols was found also by Bossi15 on Ru/MgO samples green-gray, the color intensity depending on the Ru contents (in prepared from Ru nitroso nitrate and Mg(OH),. The same from 1%to 4.9% Ru/MgO in weight). trend was observed by Pierantozzi et a1.8 for R U ~ ( C O ) ~ ~ /theArange sample containing 2% of Ru is also prepared by impregnation MgO samples yielding significant amounts of oxygenated of a R u ~ ( C Opentane ) ~ ~ solution with an high-area (-200 m2/g) compounds (CH,OH mainly) from synthesis gas. The MgO. Survey experiments show that the results obtained for this presence in X P S experiments of two different Ru peaks sample are the same as those obtained for the Mg(OH), imassigned to metallic and oxidized Ru15 can explain the pregnated samples. The samples are then activated in vacuo (Po 1 x torr) higher selectivity toward oxygenated compounds on at 573 K for several hours. At this temperature the Mg(OH)2 Ru/MgO samples. However, the sample preparation and MgO transformation occurs and most water and surface hydroxyls the thermal and reducing treatments can have an imporare eliminated. tant influence on the structural or electronic interaction A subsequent reduction in a cell with H2 at 623 K for 2 h is effects between the metal and the support. performed, followed by a final evacuation for 30 min at the same temperature. After these treatments the samples (hereafter named, Ru/MgO-A) darkened, without, however, reaching the (5) Chen, H. W.; Zhong, Z.; White, J. M. J. Catal. 1984, 90, 119. usual black color of Ru-supported samples. (6) (a) Morris, S. R.; Moyes, R. B.; Wells, P. B.; Whyman, R. J. Catal. Survey experiments (see Discussion) are performed to compare 1985, 96, 23. (b) Morris, S. R.; Moyes, R. B.; Wells, P. B.; Whyman, R. the above samples, with (i) a sample prepared by an aqueous In Metal-Support and Metal-Additiue Effects i n Catalysis; Imelik, B., solution of ruthenium nitrosonitrate (Ventron product) with et al., Eds.; Elsevier: Amsterdam, 1982; p 247. Mg(OH),, dried at 343 K, activated in vacuo at 623 K, and then (7) &ita, K.; Ohya, A,; Ozaki, A,; Inaue, Y.; Yasumori, 1. J. Catal. 1985, H2reduced, following Bossi et al.,15the final color is gray-green 92, 305. for the 1%Ru/MgO samples, and (ii) a sample prepared by a (8) Pierantozzi, R.; Valagene, E. G.; Nordquist, A. F.; Dyer, P. N. J. Mol. Catal. 1983, 21, 189. Ru02.XH20(an Aldrich product) impregnation with Mg(OH)*, (9) (a) Mc Laughlin; Mc Clory, M.; Gonzalez, R. D. J. Catal. 1984,89, followed by the usual activation and reduction cycle at 623 K; 392. (b) Kiss, J. T.; Gonzalez, R. D. J. Phys. Chem. 1984, 88, 892. the final color is gray-blue for the 1% Ru/MgO. (10) Kellner, C. S.; Bell, A. T. J. Catal. 1981, 71,296. Surface areas are measured by a Sorptomatic 1800 Carlo Erba (11) Ekerdt, J. G.; Bell, A. T. J. Catal. 1979, 58, 170. instruments. TEM micrographs are obtained on a Philips EM300. (12) Guglielminotti, E.; Spoto, G.; Zecchina, A. Surf. Sci. 1985, 161, UV-vis reflectance spectra are carried out with a Varian 2390 202. spectrophotometer. The IR spectra are recorded on compressed, (13) (a) De Paola, R. A,; Hrbek, J.; Hoffmann, F. M. J . Chem. Phys. self-supportingwafers, by means of a Perkin-Elmer 580B spec1985,82, 2484. (b) Weimer, J. J.; Umbach, E.; Menzel, D. Surf. Sci. 1985,
-
155, 132. (14) Shastri, A. G.; Schwank, J. J. Catal. 1985, 95, 284 and references therein. (15) Bossi, A,; Garbassi, F.; Petrini, G.; Zanderighi, L. J . Chem. SOC., Faraday Trans I 1982, 78, 1029.
(16) Zecchina, A,; Guglielminotti, E.; Bossi, A,; Camia, M. J . Catal. 1982, 74, 225, 240, 252. (17) Guglielminotti, E.; e t al., unpublished results.
814 Langmuir, Vol. 2, No. 6, 1986
Guglielminotti 1.c
T oa
0.D
0.D.
0.6
1
00
1
zobo
I
v (cm")
0.4
180
Figure 2. Spectrum in the uco region of the Ru,(CO),,/MgO system after evaporation of n-pentane.
trophotometer equipped with a data station. High purity H2, 02, and CO gases from Matheson are used without further purification and I3CO (90%) is purchased from Prochem BOC Ltd.
Results A preliminary characterization of Ru/MgO-A samples gives the following results: (i) The BET surface area measurements give a value of 230 and 275 m2.g-' for two samples containing 4.9% and 1% Ru in weight, respectively. The isotherms are of type I1 with mesopores. These values of area are exactly in the range found on pure MgO ex hydroxide. Besides, the TEM micrographs carried out with a manification of 103.000 on samples containing from 1% to 4.9% in Ru show the same texture of pure MgO obtained in a similar way: no clear evidence of Ru particles is obtained, thus showing that the Ru particles are very small, i.e., below the instrumental limits of detection (-20 A). (ii) In Figure 1, curve a, the UV-vis near-IR reflectance spectra of a R U ~ ( C O ) ~ ~ / M ~ ((1% O H )Ru) , aged sample is reported. Two absorptions centered at 690 and 350 nm are clearly visible. After activation under vacuum (5 h at 623 K) the spectrum of curve b is found, with a general loss of reflectance of the sample especially in the UV-vis region. No definite maxima are found in the difference spectrum (curve e). The reduction process with H2 at 623 K shows the same spectral behavior as curves b and c. (iii) In Figure 2 is illustrated the spectrum of R U ~ ( C O ) ~ , recorded few hours after the impregnation of MB(OH)~ and pentane evaporation; the same spectrum is shown by a R U ~ ( C O )solution '~ impregnated on MgO obtained from Mg(OH), a t 623 K and then exposed to the atmosphere. The couple of carbonylic bands at 2060 and 1986 cm-' is therefore characteristic of this system and its intensity slowly decreases with the time of exposition to the atmosphere reaching a nearly zero intensity after six months. Figure 3 illustrates the adsorption of CO at increasing coverages 6' up to the maximum on a Ru/MgO-A sample: the growth of two absorptions a t 2020 (low 6')-2040 (Omax) cm-' and at 1950 (br) cm-l is clearly visible. This spectral behavior is typical of CO adsorption on all the 1-5% Ru samples obtained after activation in cell under vacuum at 573-673 K, either followed or not by a reduction cycle with H, at the same temperature. Besides a survey experiment on a sample activated and reduced a t 873 K
-
0.2
0
00
2000
-
181
v (cm-') Figure 3. CO adsorption on Ru/MgOA sample at increasing coverage values. gives a spectrum of adsorbed CO with a slightly reduced intensity, but exactly similar to that of Figure 3. This spectrum of adsorbed CO is therefore a clear fingerprint of a stabilized Ru phase formed on the MgO surface under these thermal and vacuum conditions of preparation. In an experiment of 13C0 (90%) adsorption, the couple of bands is shifted to 1994 and 1905 em-' at "6' as expected from the isotropic ratio rule. However, in order to elucidate the nature and the number of Ru sites which adsorbed CO, the experiments with isotopic 12CO-13C0mixtures are not unambigous in this case, because the isotopic shift (-47 em-') is of the same order of magnitude of the shifts due to the presence of at least a couple of bands and to the dipole-dipole and chemical shift effects.ls Therefore some experiments of l2C0 desorption and "annealing" at 473 K are reported in Figures 4 and 5 . Figure 4 illustrates the CO desorption under vacuum at temperatures increasing from 300 to 623 K, until the complete evacuation is accomplished. The spectral maxima, which remain nearly constant during adsorption (see Figure 3), are in this case strongly red-shifted as the coverage decreases, the peak of higher intensity shifts from 2040 to 1960 cm-', whereas the broad shoulder at -1950 cm-' reveals a composite nature, shifting to 1906 and 1875 em-' for 523 5 T 5 623. The comparison of Figures 3 and
-
(18)Willis, R. F.; Lucas, A. A,; Mahan, G. D. In The Chemical Physics Solid Surfaces and Heterogeneous Catalysis; King, D. A,, Woodruft, D. P., Eds.; Elsevier: Amsterdam, 1983; Vol. 2, p 59. of
Langmuir, Vol. 2, No. 6, 1986 815
Surface Structure of the RulMgO System
0.D. 2200
lsdo
2doo
i (cm-‘) Figure 6. Effect of oxygen interaction on CO adsorbed on an Ru/MgO-A (1%Ru) sample. (a) Sample saturated with CO, evacuated at room temperature, and contacted with 30 torr of 0,; spectrum recorded immediately. (b) After 2 h. (c) After 1 day. (d) Dotted curve, desorption and reduction with H2 at 623 K for 1 h and 40 torr of CO allowed (OD X 2).
!OO
2600
rsbo
i (cm-’) Figure 4. CO desorption from Ru/MgOA (4.9% Ru) at increasing temperatures: (a) CO adsorbed at saturation and then desorbed at 300 K; (b) after 30 min of desorption at 373 K; (c) desorption for 30 min at 423 K; (d) desorption for 30 min at 473 K; (e) desorption for 30 min at 523 K; (f) desorption for 30 min at 573 K; (g) desorption for 30 min at 623 K. 0.4
1
A
2200
2600
ii (cm”J
”
Figure 5. CO equilibration on Ru/MgO-A (4.9% Ru) at 473 K and successive CO adsorption at room temperature: (A) curve 1,CO adsorbed at room temperature (0 = 0.21); curve 2, adsorbed CO annealed 30 min at 473 K. (B) Difference spectrum between CO doses adsorbed at room temperature at increasing coverages (3-5) and spectrum A2 (CO annealed at 473 K). 4 gives evidence of the absence of equilibration in the CO adsorption experiments at increasing coverages carried out a t 300-320 K. In this case CO is immediately adsorbed for its high sticking probability(s)lg on the Ru particles exposed on the external layers of the pellets. These par(19) Menzel, D.; Pfnur, H.; Feulner, P. Surf. Sci. 1983, 126, 374.
ticles are therefore nearly saturated already for admission of low CO doses into the IR cell: successive doses of CO will be adsorbed onto inner Ru particles until CO saturation of all Ru microcrystals is achieved. This “hit and stick” mechanism was already found for CO adsorbed on silica-supported Ru,12 Pd,20and PtZ1and is therefore of general validity for molecules adsorbed with high values of s and heat of adsorption on supported metals. Because at T 423-473 K CO is in general mobile and is desorbed from Ru crystals,22the CO desorption at these temperatures is accompanied also by an annealing process of residual adsorbed CO which spreads uniformly on all the Ru-supported crystals. As a consequence of this equilibration process the residual CO is adsorbed, at higher temperatures (523-623 K), on the more energetic sites and with low 0 values. The results reported in Figure 5 confirm this hypothesis. If CO is adsorbed (0 = 0.21, Figure 5 (curve 1))and then is annealed at 473 K, only the peaks of CO adsorbed on more energetic sites a t -1900 cm-’ remain (curve 2). If successive CO doses are then adsorbed at room temperature on this “equilibrated” sample, the difference spectrum between the overall spectrum and curve 2 of Figure 5A gives curves 3-5 of Figure 5B thus showing that CO is adsorbed nearly exclusively onto the less energetic free sites (VCO = 2030 cm-’). Effects of Oxygen Adsorption. Figure 6 illustrates the effect of O2 interaction at room temperature on CO adsorbed on Ru/MgO-A (1% Ru). A slow process of CO oxidation occurs with preferential erosion of the band at higher frequency: at the same time strong “carbonate” bands between 1700 and 1300 cm-’ are formed. If the sample which had adsorbed oxygen is then outgassed and reduced with H2 at 623 K, the spectrum of CO successively adsorbed at room temperature shows a strong decrease in the component a t high frequency which is also shifted to -2025 cm-’ (Figure 6, dotted curve). The same phenomenon is observed for activated samples exposed to atmosphere: the new cycle of activation and reduction at 623 K leads to a Ru sample which adsorbes CO with an overall reduced intensity in the carbonylic region and where the shoulder at 1950 cm-’ becomes prominent in the spectrum.
-
~~~
~
(20) Palazov, A.; Chang, C. C.; Kokes, R. J. J. Catal. 1975, 36, 338. (21) Hammaker, R. M.; Francis, S. A.; Eischens, R. P. Spectrochim. Acta 1965,21, 1295. (22) Ku, R.; Gjostein, N. A.; B o n d , H. P. Surf. Sci. 1977,64,465 and references therein.
816 Langmuir, Vol. 2, No. 6, 1986
I
Guglielminotti
1
A
O.D.x2 Q3
0.2
0. D
I I
2200
2000
-
v (cm-1)
1
Figure 8. CO adsorption on Ru/MgO-A (1%Ru) sample oxidized at 623 K. (a) Sample oxidized with 30 torr of O2 at 623 K, O2 evacuated, and then contacted with 40 torr of CO; spectrum recorded after 1 day. (b) Sample (a) outgassed for 1 h at 623 K and then contacted for 1 day with 40 torr of CO. (c) Sample (b) outgassed and reduced 1 h in H2 of 623 K and then contacted for 1 day with 40 torr of CO.
2200
2000
2200 1860
i (cm-li Figure 7. CO adsorption on an Ru/MgO-A (1% Ru) sample oxidzed at room temperature. (a) CO adsorbed at full saturation on an Ru/MgO-A reduced surface. (b) Sample (a) outgassed and H2 reduced at 623 K, oxidized for 16 h at room temperature in 30 torr of 0 2 ,and then outgassed 5 min at room temperature; contacted with 40 torr of CO. (c) Sample (b) heated in 40 torr CO for 30 m at 423 K. (d) After heating 30 min in CO at 523 K. (e) After outgassing 1 h at 573 K.
In Figure 7 an experiment of CO interaction with a Ru/MgO-A sample (curve a shows the usurl2045-1950cm-' couple of bands), successively oxidized a t room temperature is illustrated. The chemisorbed carbon monoxide is then completely evacuated a t 623 K, the sample reduced with H2 at 623 K, and cooled up to'room temperature and 30 torr of oxygen are dosed; after 16 h of contact, oxygen is desorbed a t room temperature and 30 torr of CO are adsorbed (curve b). The bands of the CO adsorbed on this oxidized sample decreased in intensity and markedly changed in frequency (u, at 2130,2072, -2000 (br) cm-I). Heating for 30 min in 30 torr of CO a t 423 K (curve c) produced a reduction process of the oxidized Ru phase and the bands at 2130 and 2072 (sh) cm-l strongly decreased in intensity, whereas the bands a t 2040 and 1952 cm-' increased; at the same time a marked growth of the bands in the carbonate range (1700-1000 cm-') is observed. This reduction process is accomplished by heating in CO (curve d) a t 523 K; the 2130-2072-~m-~bands are completely eliminated, and the bands a t 2023, 1950, 1922, and 1875 (sh) cm-' become prominent. Finally an outgassing process of 1 h a t 573 K leaves in the spectrum (curve e) only residual weak bands a t -2000 and 1910-1875 cm-'. In another experiment a reduced Ru/MgO-A sample is heated in O2 a t 623 K: in this case the wafer's color becomes green and CO is not adsorbed as Ru carbonyl a t room temperature. Only after 1 day of contact with the oxidized ruthenium surface, CO is adsorbed in little amounts giving a weak absorption centered at 1950 cm-' (Figure 8a). The outgassing process of 1 h a t 623 K
-
2000
-
v (cm-')
1800
Figure 9. CO adsorption on Ru/MgO-A (4.9% Ru) sample oxidized at 773 K. (a) Reduced Ru/MgO A sample oxidized in O2(50 torr, 2 h) at 773 K, outgassed, and H2 reduced for 2 h at 773 K and then contacted with CO (40 torr). (b) Sample (a) heated in 40 torr of CO 1 h at 473 K. completely eliminates the adsorbed CO; if CO is then dosed (Figure 8b), two weak bands at 1950 and 1875 cm-l appear. A successive cycle of reduction with H2 a t 623 K only partially restores the original adsorbing capacity toward CO with an enhancement of the 1950-cm-l component accompanied by a shoulder a t u > 2000 cm-' (Figure 8c). In order to examine the hypothesis that a RuO, phase is formed a t least during oxygen treatment a t 623 K, a survey expgriment was carried out with RuO, h'igh area 52 m2.g-I, irhpregnated with Mg(OH), (see Experim ntal Section) and then outgassed and reduced with H2 af 623 K: in this case CO is not at all adsorbed as carbonyl and no bands are found in the 2200-1800-cm-' spectral region. In a similar way if a Ru/MgO-A sample is oxidized a t 773 K with O2 (50 torr, 2 h) the Ru is completely oxidized to a stable nearly stoichiometric R u 0 2p h a ~ e ;the ~ ~sam,~~ ple color becomes pale green and the CO adsorption is completely eliminated (no bands are visible in the 22001800-cm-' region). The activation and reduction in H2 at 623 K do not restore any adsorption capacity toward CO. Only a reduction process at 773 K leads to a sample which can adsorb little amounts of CO giving weak bands between 2040 and 1900 cm-l (Figure 9a). These bands are slightly increased by heating in CO a t 473 K (Figure 9b). A Ru/MgO sample prepared from ruthenium nitrosonitrate following Bossi et al.I5 shows similar little adsorption capacity towards CO, and the weak spectrum of adsorbed CO is similar to that obtained on "oxidized" Ru3(23) Duvigneaud, P. H.; Reinhard-Derie, R. Termochim. Acta, 1981, 5 1 , 307 and references therein. (24) Fletcher, J. M.; Gardner, W. E.; Greenfield, B. F.; Holdoway, M. -J.; Rand, M. H. J . Chem. SOC. A 1968,, 653.
Langmuir, Vol. 2, No. 6, 1986 817
Surface Structure of the RulMgO System
p a p e d 6 concerning the characterization of the Ru3o.D. (CO),,/Al,O3 system, it was shown that a low Ru content On
‘
’
2200
2000
W (cm-1)18do
Figure 10. CO adsorption on a Ru/MgO (1%Ru) sample prepared by impregnation of a RuNO(NO&~ aqueous solution with Mg(OH)z. (a) Sample activated in vacuo 4 h at 623 K and Hz reduced 2 h at 673 K; 40 torr of CO adsorbed at room temperature. (b) After heating in 40 torr of CO 1 h at 473 K.
(CO)’,/MgO sample (Figure loa). In particular the couple of weak bands at 1875 and 1907 cm-’ are well evidenced by outgassing a t 473 K (Figure lob) and are clearly due to the same CO species found on our “oxidized” samples. We recall here that the carbonylic bands in the 22001800-cm-l region are not the only ones appearing in our IR spectra. In fact that MgO support outgassed a t 623 K can adsorb CO giving bands assignable to carbonate (at 1667-1288, lo00 cm-’) and to “formate” species (at 2840, 1610,1388,1360 cm-1)25-28 by interaction of CO with Ocw2and OH- ions, respectively. The same bands are found also on Ru/MgO-A samples and can be assigned to the same species. It is, however, possible to observe a relative enhancement of the intensity of the “formate” species and the appearance of a doublet of similar bands; the type of species formed can therefore be influenced by the presence of ruthenium. These results and those obtained for CO Hz adsorption, together with the formation of intermediate oxidized species at room and higher temperatures and in general the problem of reactivity of Ru/MgO system, will be fully discussed in a second part of this work.”
-
+
Discussion
(a) CO Adsorption on Samples Activated and Reduced in Vacuum at 623 K. The Ru/MgO catalyst prepared29by MgO impregnation with a RuC1,.H20 solution, followed by reduction in flowing hydrogen at 573-673 K, yields a “normal” supported metallic ruthenium which adsorbs CO (vmm = 2035 cm-’ shifting to lower frequency upon evacuation, i.e., for lower coverages) a t nearly the same frequency of CO adsorbed on Ru/SiOz5,9,11,12,30,31 and R u / A ~ ~ systems. ~ ~ ~ ~ J ~ , ~ ~ The frequency of CO adsorbed on an Ru(001) monocrystal, shifting from -2060 (0”) to 1980 cm-’, is exaclty in the same spectral region. It can therefore be inferred that CO bands of Ru-supported systems absorbing in this spectral range are the fingerprints of CO species linearly adsorbed on (001) or similar faces. For CO adsorbed on a Ru/MgO sample oxidized at room temperature a predominant couple of bands at 2130-2080 cm-l was found;29these bands were assigned to a RuX+(CO), species (with n = 2 or 3 and x in the range 2-4), in analogy with the assignment given for the same bands found for Ru supported on silica and alumina. In previous (25) Guglielminotti, E.; Coluccia, S.; Gamone, E.; Cerruti, L.; Zecchina, A. J. Chem. SOC., Faraday Trans. 1 1979,75,96. (26) WeiWang, G.; Hattori, H. J. Chem. SOC.,Faraday Trans 1 1984, 80, 1039. (27) Ramsay, J. D. F. In Adsorption and Catalysis on Oxide Samples; Che, M., Bond, G. C., Eds.; Elsevier: Amsterdam, 1985; p 249. (28) Rethwisch, D. G.; Dumesic, J. A. Langmuir 1986,2,73. (29) Schwank, J.; Parravano, G.; Gruber, H. L. J. Catal. 1980,61,19. (30) Brown, M.F.; Gonzalez, R. D. J. Phys. Chem. 1976,80, 1731. (31) Kuznetsov, V. L.; Bell, A. T.; Ermakov, Y. I. J. Catal. 1980,65, 314.
and the static conditions of the decarbonylation process (carried out by outgassing in vacuum) favor the formation of these oxidized RuX+“aluminate” phases by interaction of Ru with atmospheric water and surface hydroxyls of the support. An analogous process cannot occur on the MgO surface, as it is related to the possibility that an analogous ruthenate phase can be formed: in fact the Mg2+ionic radius (0.72 A) is too small to allow the formation of the perovskite-like ABO, (MgRuO,) structure, stable only for A cations with radius >0.9 A. Furthermore the conditions of preparation and activation in vacuum of our Ru/MgO samples are not so oxidizing to obtain a RuOz-MgO supported phase as indicated by the different results (such as the absence of CO adsorption as carbonyl species) obtained during the survey experiments carried out onto the RuOz/MgO system (see Experimental Section). However, an oxidizing effect of hydroxyls interacting with Ruo (which is maximized if Mg(OH), is utilized as support instead of MgO) in the presence of the atmospheric oxygen is certainly occurring on Ru3(CO)1z.This effect is correlated with the so-called “aging” effect as it can be proved by the following considerations. The couple of bands (2060,1986 cm-’) shown by R U , ( C O ) ~adsorbed ~ on a MgO (or Mg(OH),) support can be assigned to the carbonylic stretching vibrations of Ru”+(CO),L, (L = OH-, 02-) surface complexes, formed by the binding of surface hydroxyls on R U ~ ( C Owith ) ~ ~formation of an oxidized and probably monomeric RuX+species by skeletal disruption. The oxidation number of Ru, x+, is different from zero and is probably 11, in agreement with the literature for analogous homogeneous and heterogenized Ru complexes. A Ru carbonyl complex with two CO groups and containing a norbornadiene and two hydroxyls groups, C7H8Ru(C0)z(OH)z,gives vCo bands a t 2047 and 1970 besides [Ru(C0),X2], (X = C1, Br, I) complexes absorb in the same spectral range (vco = 2066-2053 and 1988-1995 ~ m - l ) . ~ ~ The oxidative addition of surface hydroxyls to metal carbonyls of Rh, Os, and Ru heterogenized on oxide supports with CO and Hzevolution is a well-known phenomenum.16,35-37The residual carbonylic groups (adsorbing a t 2060-1986 cm-’) are then slowly eliminated either by interaction with wet atmosphere (aging effect) or by heating treatments in mild temperature conditions (373-473 K). The final results of this process could be the formation of a ruthenium hydroxide phase highly dispersed on MgO with the approximate formula Ru(OH), (n E 3). We recall here that the ruthenium(II1) hydroxide is a poorly defined and characterized compound with an undefined stoichiometry depending from the type of preparation;%it can be easily reduced to Ruo or oxidized to RuN, depending on the gaseous atmosphere condition^.^^ The formation of Ru” and Ru“’ ions can be detected in the (32) Tripathi, S. C.; Srivastava, S. C.; Mani, R. P.; Shrimal, A. K. Inorg. Chim. Acta 1975,15,249 and references therein. (33) Zanderighi, G. M.; Dossi, C.; Ugo, R.; Psaro, R.; Theolier, A.; Chopin, A.; DOrnelas, L.; Basset, J. M. J. Organomet. Chem. 1985,296, 127. (34) King, R. B.; Kapoor, P. N. Inorg. Chem. 1972,11, 336. (35) Brown, T. L. J. Mol. Catal. 1981,12,41. (36) Ugo, R.; Psaro, R. J. Mol. Catal. 1983,20,53. (37) (a) Hucul, D. A.; Brenner, A. J. Phys. Chem. 1981,85,496. (b) Basset, J. M. In Contribution of Clusters Physics to Material Science and Technology; Davenas, J., Rabette, P. M., Eds.; M. Nijhoff Dordrecht, 1986; p 91. (38) Seddom, K. R. Coord. Chem. Rev. 1985,67, 171 and references therein.
818 Langmuir, Vol. 2, No. 6, 1986
Guglielminotti
UV-vis diffuse reflectance spectrum of an aged Ru,(C0)12/Mg0sample (1% Ru, Figure la). The yellow-green color of the sample (weak absorption at 690 and 350 nm) can be ascribed the d-d transitions of Ru" and R u " ' . ~ ~ However, the literature data are ambiguous, changing with the ligands bound to Ru ions, and no certain conclusions about the oxidation number of Ru can be reached on the basis of the UV-vis spectra. Furthermore, the process of activation in vacuo at 623 K (Figure 1, curves b and c) leads to a general darkening of the sample which is typical of the formation of a metallic Ru phase. This effect is already visible before the reduction with Hz a t 623 K, which leaves unchanged the overall UV-vis spectrum. Therefore during the desorption of surface hydroxyls in the 573-673 K range of temperature a complex reaction occurs yielding, at least partially, a reduced metallic Ru phase, as proved by the IR spectra typical of CO adsorbed on Ruo (Figure 3). A possible mechanism could be a disproportionation reaction, under vacuum, of ruthenium(II1) hydrate to Rum + RuO, as suggested by Duvigneaud et aLZ3 2Ru(OH),-H,O
-
-
(Ru2O3); 2(Ru203)
3RuO2 + Ru (1)
The existence of the Ruz03compounds is in fact doubtful. Besides, the very small and dispersed RuO, particles formed during reaction 1are probably oxygen deficient and can be partially reduced to Ruo (at least at the surface) by the contemporary presence of a high vacuum and of reducing conditions (H, evolved during hydroxyl evacuation at 573 K).35-37 Some literature data seem to confirm the hypothesis of instability and of possible reduction of a RuO, phase at least at the surface: (i) Kotz et al.40found by XPS experiments that the highly defective hydrated oxide film formed on RuO, electrodes decomposes to metallic Ru at 583 K in vacuum. (ii) O'Grady et aL41have performed a LEED analysis of a single crystal Ru0,(110) surface; they found that no LEED pattern can be observed, even after annealing in vacuo between 573 and 673 K, because the surface is depleted of oxygen in high vacuum. Only an annealing process in 0, a t the same temperature leads to a LEED pattern. At this point the Auger spectra give a ratio Ru/O 1.4-1.5, i.e., a value lower than the stoichiometric value of 2. In conclusion, the high dispersion of Ru obtained by the R U ~ ( C O )impregnation ~, and anchoring processes on MgO favors the formation of an highly dispersed and nonstoichiometric Ru(OH), phase. This surface phase can be easily decomposed in vacuum at 573-673 K, the final result being the formation of very small and dispersed Ruo microcrystals plus very small amorphous and oxygen-deficient RuO, particles which can be further partially reduced, at least at the surface, to metallic Ru. The presence of several adsorbed CO species and therefore of several uco bands on Ru/MgO-A samples is well evidenced in Figures 3-5. Figure 3 shows at first glance a dual heterogeneity of the Ru sites adsorbing CO: on the first CO is adsorbed linearly; on the second CO is adsorbed as a bridged species (vco I1950 cm-', see the following discussion). (39) (a) Jsrgensen, C. K. Acta Chem. Scand. 1956, 10, 518. (b) HarZion, Z., Navon, G. Inorg. Chem. 1980, 19, 2236. (40) Kotz, R.; Lewerenz, H. J.; Stucki, S. J. Electrochem. Soc. 1983, 130, 825. (41) O'Grady, W. E.; Atanasoska, Lj.; Pollak, F. H.; Park, H. L. J . Electroanal. Chem. 1984, 178, 61.
As discussed in the experimental part, only an "annealing" process of adsorbed CO at T 1 423 K allows to attain true equilibrated coverage. Therefore only the desorption experiments at increasing temperatures, as reported in Figure 4, and not the adsorption at increasing coverages reported in Figure 3 give a true picture of the uco frequency shift with decreasing 6' values. The uco frequency of linearly adsorbed CO is shifted therefore (Figure 4) from 2040 to 2016, 1987, and 1960 cm-' at 373, 423 and 473 K, respectively (it is difficult to say of the =1950-cm-' shoulder observed at 523 K is a residual linear species or if it is due to another type of adsorbed CO species). The overall red shift of the uco frequency with coverage is -80 cm-', in keeping with a similar value found for CO adsorbed on the Ru(001) face. The differences in frequency (Au -20 cm-l) between the CO species linearly adsorbed on Ru/MgO and that adsorbed on Ru(001) or Ru/SiO, systems can be ascribed to (i) an electron-donor effect of the MgO support toward the ruthenium6 or of carbon formed on supported Ru by CO disproportionation,1° (ii) a dipole-dipole effect1J2J8~21 lower than an extended faces as a consequence of the very small sizes of the Ru particles present on our samples (the result should be a lower blue shift of uco at high e), (iii) the prevalence of Ru faces different frcm (001) but again with sites adsorbing CO as a linear species. The experimental data do not allow a choice between these possible effects which are not mutually esclusive; in fact no IR data of CO adsorbed on Ru faces different from (001) can be found in the literature while our samples show Ru heterogeneity. As far as the uco band at lower frequency, i.e., the shoulder at 1950 cm-l of Figure 3 is concerned, the desorption experiments reported in Figure 4 (curves c and d) clearly illustrate a shift to 1900 cm-' at 423-473 K. The annealing process of CO (6' = 0.21) at 473 K reported in Figure 5a and the desorption process of a fully CO-covered sample at 523-623 K (Figure 4, curves e-g) give clear evidence of the presence of at least two distinct types of nonlinearly adsorbed (uco = 1910 and 1875 cm-l, respectively). The remaining shoulder at 1950 cm-l could be assigned to a residual resistant linear species. To our knowledge only few references are found in the literature on carbonylic Ru-CO species with uco absorbing between 1950 and 1850 cm-'. In a pioniering paper Guerra and Schulman4' reported two types of vco bands (2010-1990 cm-l tentatively assigned to linear CO and 1910-1870 cm-' tentatively assigned to bridged CO) on Ru/Si02 samples obtained by reduction at 650 K of a (NH4j2RuC1,silica-impregnated solution. However, the low-frequency band was not confirmed by the following s t ~ d i e s ~on J ' RuC13 ~ ~ ~ solution impregnated on silica. On silica-supported samples obtained from RuC13 and promoter alkali nitrate saltgaa shoulder at 1950 cm-l was found for adsorbed CO together with a concomitant decrease of the hydrogenation activity of Ru by the alkali promoter. In a recent paper concerning the CO adsorption on a Ru(001) monocrystal precovered with potassium,13 a strong shift in the uco frequency was observed (up to 1400 cm-l for high K coverages) which was attributed to a strong charge transfer from the K to the Ru, with a lowering of the bond order of the adsorbed CO. A lowering of vco from 2020 to 1920 cm-l was found by Brown et al.43by adding Cu to a Ru(001) surface. At the same time a fall in the hydrogenation activity occurs, whereas SSIMS studies indicated the presence after CO
-
-
N
-
(42) Guerra, C. R.; Schulman, J. H. Surf. Sci. 1967, 7, 229. (43) Brown, A.; Van Den Berg, J. A.; Vickerman, J. C. R o c . Int. Congr. Cats/., 8th 1984, 4, 35.
Surface Structure of the RulMgO System adsorption of Ru2CO+ ions with CO bridged species. In homogeneous phase only few carbonylic complexes of Ruo such as RU&(CO)l,u and [C5H5Ru(C0),]$* contain bridged CO group absorbing in the 1900-1800-cm-’ range. Therefore the uco bands a t 1875 and 1910 cm-’ (at low 8 values, shifting to -1950 cm-l at high B values) can be assigned with good confidence to (slightly different) CObridged species, uncommon on Ru. These species are more strongly adsorbed than linear species as shown by the desorption experiments reported in Figure 4 and by the annealing experiments reported in Figure 5. In this case the annealing at 473 K of CO adsorbed at 8 = 0.21 (Figure 5a, curve 1) leads to CO adsorbed mainly as bridged species (Figure 5a, curve 2). A t this stage the CO occupies the more energetic sites as a bridged species and successive doses of CO are adsorbed at room temperature mainly as linear species absorbing a t -2030 cm-l (Figure 5b, curves 3-5). The shift of the bridged-C0 frequency from 1875-1906 to -1950 cm-’ at high 8 values may be caused by dipole-dipole interactions and chemical effects, as found for CO linearly adsorbed on Ru(001) and for bridged CO on Pd.45 (b) CO Adsorption on Oxidized Samples. The RuO2system is a very complicated one, the Ru/O ratio depending on the conditions of preparation, relative oxygen pressure, temperature, et^.^^-^^ Since the highest temperature reached in our experiments is 773 K, the RuO, and RuOBspecies, predominant at 1073-1273 K,4’ can be excluded. On the other hand, the Ru02 phase can be easily formed in our oxidative experimental conditions while a perovskite-like structure MgRuO, cannot be formed as previously discussed. In the temperature range 473-655 K Sommerfeld and P a r r a ~ a n ohave ~ ~ shown that the temperature and O2 pressure control the value of x in the surface Ru(O), states. On the MgO-supported samples we can add that the value of x depends also on the dispersion of Ru and on its interaction with the surface hydroxyl groups of the support. Moreover, the studies of oxygen interaction with Ru(001)49,50 and (101)51352monocrystals give evidence of a progressive penetration of the oxygen beneath the surface already a t 300 K with incorporation into the Ru lattice after annealing at 700 K. Madey et al.49have studied by LEED, work function, and flash desorption techniques the oxygen adsorption at 300 K on an Ru(001) surface. A peak at 1400 K in work function can be ascribed to a “RuO,-like” phase decomposed to Ru + 02’, whereas a plateau between 600 and 1200 K can be related to a desordered layer of oxygen adsorbed onto the surface with a stoichiometric Ru/O ratio 300 K. In fact, already after heating in CO at 423-573 K (Figure 7, curves c and d) the couple of bands at 2130-2072 cm-’ is eliminated and the usual bands of CO adsorbed onto the metallic Ru phase at 2040-2023 and 1950-1922-1875 (sh) cm-’ appear. The formation of carbonate species (the bands between 1700 and 1000 cm-’ strongly increase) confirms the reducing effects of CO. Besides, the strong intensity of bridged CO species (curve d) and the residual presence of these species (vco = 1910 and 1875 cm-’) after desorption at 573 K (curve e) confirm that the process of reduction with CO instead of H2 does not change significantly the ratio between the linear and the bridged CO species. On the sample oxidized at 623 K in O2 (Figure 8) the formation of a stabilized and oxygen-rich RuO, phase is strongly enhanced. The 2130-2072-cm-l couple of bands, assigned to CO adsorbed on Ru3+, is therefore inhibited and only a small fraction of the oxidized Ru can be reduced giving CO adsorbing mainly in bridged form [bands at 1950 cm-’ (curve a) and 1920-1875 cm-l (curve b, after outgassing 1h a t 623 K)] as usual on oxidized samples. The effect of H2 reduction at 623 K (curve c) increases the intensity of the adsorbed CO bands, but the usual intensities of CO adsorbed on 623 K activated and reduced samples are too far to be attained (the absorbance is multiplied for 2), especially for the “linear” CO species. The full oxidation process is achieved by heating at 773 K: the possibility of obtaining reduced Ru samples, even by reducing with H2 at 623 K, is almost completely ruled out at this temperature. The hypothesis of a well-dispersed stable and stoichiometric RuO, phase in these oxidizing conditions can be reasonably advanced: this hypothesis is confirmed by the survey experiments on R u 0 2 / M g 0 samples, treated in the same vacuum conditions, which do not at all adsorb CO as carbonyl species. Only a strong reduction process at 773 K leads to CO adsorption in small amounts (Figure 9a, where the absorbance is multiplied by 3!). The process of further reduction in CO atmosphere at 473 K (Figure 9b) leads to a well-resolved spectrum of adsorbed CO with peaks at 2038 and 1952 cm-l, but the intensity of the carbonylic
820
Langmuir 1986,2, 820-823
bands and therefore the amount of reduced Ruthenium is always very small. Finally, Figure 10 shows that by impregnating MgO with a ruthenium nitrosonitrate solution and then reducing at 623 K, a sample which has the imprinting of “oxidized” Ru/MgO is obtained; the band of the linearly adsorbed CO at Y > 2000 cm-’ (curve a) is very weak and disappears almost completely after outgassing at 473 K (curve b). CO is adsorbed mainly as bridged species (vco between 1960 and 1875 cm-I), with peaks a t 1908 and 1875 cm-l prevailing after outgassing a t 473 K. The faces of Ru microcrystals obtained in this case are therefore the same as those obtained from RU,(CO),~/M~O oxidized samples and the hypothesis of an effect of the MgO support on the morphology of dispersed Ru crystals can be advanced.
Conclusions Some final considerations of the structural effect on the reactivity of the Ru/MgO system compared with the reactivity of Ru/SiO, and A1203systems can be made. The bridged CO species, which are typical of the present Ru/MgO system, are desorbed a t higher temperatures (573-623 K) than the linear ones; as a consequence, the lower activity and the formation of a considerable percent of propene and higher hydrocarbons, instead of methane,
found on Ru/MgO samples6 can be explained simply by this surface structural effect. In fact in the bridged CO species the Ru-C bond is stronger than in linearly adsorbed CO and hydrogen adsorption can be partially inhibited: a lower CO hydrogenation activity accompanied by an increased selectivity toward higher hydrocarbons can therefore be expected. Besides, the higher selectivity toward oxygenated compounds such as alcohols, found on Ru/MgO ~ a m p l e s , ~ J ~ can be explained by the easy formation of an oxygenated, nonstoichiometric, RuO, surface phase which can be reduced, in the temperature range of the CO-H, reaction (473-573 K), to metallic ruthenium. In conclusion this paper gives experimental evidence of the importance of the support and of the type of catalyst preparation on the structural surface morphology of Ru microcrystals. The catalytic activity and the selectivity toward same important catalytic reactions might be influenced by these structural effects.
Acknowledgment. The financial support of the Italian Minister0 Pubblica Istruzione, “Progetti Nazionali di Rilevante Interesse per lo Sviluppo della Scienza”, is acknowledged. Registry No. CO,630-08-0; Ru, 7440-18-8; MgO, 1309-48-4.
[ (CH3)5C5]2Th(CH3)2/Silica Surface Chemistry. High-Resolution 13C CPMAS NMR Evidence for Alkylation
of Surface Silicon Sites Paul J. Toscano and Tobin J. Marks* Department of Chemistry, Northwestern University, Euanston, Illinois 60201 Receiued June 18, 1986. I n Final Form: August 15, 1986 The reaction of Cp’,Th(13CH& (Cr’ = q5-Me6Cs)with dehydroxylated silica (ca. 0.4 surface OH/nm2) has been studied by high-resolution C CPMAS NMR spectroscopy. On the basis of I3C chemical shifts, relative signal intensities, and NMR data from model compounds and model reactions, it is proposed that the predominant adsorption pathway involves methyl transfer from thorium to surface silicon sites. The products of this Si-0 cleavage process are surface Si-CH3 and Cp’,Th(CH&siloxide functionalities. The resulting thorium environment appears to be somewhat more electron rich than in the analogous surface complex on dehydroxylated y-alumina. Elucidating the structure and reactivity of the complexes formed when organometallic molecules are adsorbed on high surface area metal oxides and related inorganic materials is of great current interest in catalytic research.’ We have recently studied2 the surface and catalytic chemistry of organoactinides adsorbed on y-A1203both because of properties (e.g., well-defined and restricted actinide oxidation states) which render such systems useful models for technologically significant early transition(1) (a) Lamb, H. H.; Gates, B. C. J . Am. Chem. SOC.1986,108,81-89 and references therein. (b) Basset, J. M.; Chaplin, A. J. Mol. Catal. 1983, 21, 95-107 and references therein. (c) Yermakov, Yu. I. J. Mol. Catal. 1983, 21, 35-55 and references therein. (d) Iwamoto, M.; Kusano, H.; Kagawa, S. Znorg. Chem. 1983, 22, 3365 and references therein. ( e ) Yermakov, Yu. I.; Kuznetsov, B. N.; Zakharov, V. A. ”Catalysis by Supported Complexes”; Elsevier: Amsterdam, 1981. (0 Bailey, D. C.; Langer, S. H. Chem. Reu. 1981,81, 109. (g) Ballard, D. G. H. J. Polym. Sci. 1975, 13, 2191-2212. (2) (a) He, M.-Y.; Xiong, G.; Toscano, P. J.; Burwell, R. L., Jr.; Marks, T. J. J. Am. Chem. SOC.1985, 107, 641-652. (b) Toscano, P. J.; Marks, T. J. J. A m . Chem. SOC.1985, 107, 653-659.
metal/inorganic support interactions3 and because we have found that adsorbed organoactinides can exhibit high and ligation-sensitive catalytic activity for olefin hydrogenation and polymerization. More conventional techniques employed to characterize the organoactinide adsorbate molecules have included evolved product identification/ quantification, isotopic labeling, and reaction kinetics. In addition, we have shown that significant in situ structural and dynamic information on organometallic adsorbates can be obtained with high-resolution solid-state 13C NMR spectroscopy utilizing cross-polarization (CP), high-power (3) (a) Choi, K.-Y.; Ray, W. H. J . Macromol. Sei., Reu. Macromol. Chem. Phys. 1985, (225, 1-56, 57-97. (b) McDaniel, M. P. Adu. Catal. 1985, 33, 47-98. (c) Pino, P.; Rotzinger, B. Macromol. Chem. Phys. Suppl. 1984, 7, 41-61. (d) Karol, F. J. Catal. Reu.-Sei. Eng. 1984,26, 557-595. ( e ) Firment, L. E. J. Catal. 1983, 82, 196-212 and references therein. (f) Gavens, P. D.; Bottrill, M.; Kelland, J. W.; McMeeking, J. In Comprehensive Organometallic Chemistry; Wilkinson, G., Stone, F. G. A,, Abel, E. W., Eds.; Pergamon Press: Oxford, 1982; Chapter 22.5. (g) Galli, P.; Luciani, L.; Checchini, G. Angew. Makromol. Chem. 1981, 94, 63.
0743-746318612402-0820$01.50/0 0 1986 American Chemical Society