1334
J. Phys. Chem. 1992,96, 1334-1340
surfaces at the SMSI state are heterogeneous according to migrated suboxides and ethene is probably adsorbed on at least two different ~ i t e s . ' ~ The J ~ , nature ~ ~ of active sites for these catalysts should be represented in the isotope distributions in reaction products. The profile of the isotope distribution on Pt/V203reduced at 773 K was much different from those on other catalysts, where almost 90% of the isotopes was ethane-d2 (Table 111); this is a typical distribution on oxide catalysts. Thus the active site of 773 K reduced Pt/V203is suggested to be VOXand the surface of Pt particles is fully covered with suboxides. The suppression of the activity is larger than that for the SMSI Pt/Ti02 (Table IV), which may also support this model. Independence of fd of the presence of ambient ethene in Figure 5 may be due not only to the delocalization of the charge but also to little adsorption of ethene. Measured H/Pt is likely to result from the formation of vanadium bronze. Isotope distribution in ethane formed on the 773 K reduced Pt/Ti02 was unique. The deuterium distribution is controlled by the reversible step of the associatively adsorbed ethene and half-hydrogenated state and also the surface H/D ratio during the reaction.I6 Appearance of two peaks of ethane-do and ethane-d2 in the ethane formation strongly suggests two kinds of reaction sites/reaction environments with different H/D ratios.I6 This can be confirmed by controlling the number of each active site. The bare metal area of Pt in Pt/Ti02 decreases as a function of square root of reduction time, as shown in Figure 1, while the peripheral sites of TiO, islands on Pt surface increase with tR1I4 dependence.I6 Ethane-dl and ethaned2showed the tR1/2dependence in Figure 7, suggesting no relation with peripheral site in their formation. Only ethane-do had the tR1/4term in the polynomial above described to fit the curve. Consequently, ethaned, is suggested to be produced at the peripheral sites of TiO, islands
on the Pt surface. This phenomenon is similar to our previous report on Rh/Nb205and Ir/Nb205.16
Conclusion The electronic modification of the density of the unocccupied 5d state of Pt by the reduction temperature strongly depends on the kind of support. Zr02and Y203 donate electrons to Pt without migration of the support oxides upon increasing reduction temperature. The change in the unoccupied d-electron density of Pt for Pt/Y203 and Pt/Zr02 is larger than the cases of TiOz and Nb2O5, which are known as SMSI supports. The density of the unoccupied 5d state of Pt on V203is independent of the reduction temperature but decoration by the support oxide occurs. These variations of the density of state may be due to the reduced region of the supports. Ethene is electron accepting on Pt/ZrO, at every reduction temperature, while it is electron donating on Pt/Y?03 at low-temperature reduction. The increase of the reduction temperature causes the increase of electron-accepting character for both catalysts. These are suggested to be the origin of the different behavior of the kinetic parameters. The active sites for D2-ethene reaction were characterized with kinetic parameters and isotope distributions. The active site of Pt/V203 reduced at 773 K has an oxide character, where the Pt surface may be fully covered with VOX. Pt/Ti02 reduced at 773 K has two kinds of active sites with different surface H/D ratios in D2-ethene reaction. One is the bare metal site of Pt on which ethane-d2 is mainly produced and the other is the peripheral site of the TiO, island producing ethane-do. Acknowledgment. We are grateful to the staff of the Photon Factory for their help with the XANES measurements. R e t r y NO.R,7440.06-4;Y203,1314-36-9;ZrO,, 1314-23-4; V203, 1314-34-7; Ti02, 13463-67-7; C2H4,74-85-1.
Actlve Specles of Molybdenum for Alcohol Synthesis from CO-H,+ Atsushi Muramatsu, Takashi Tatstmi,* and Hiro-o Tominaga Department of Synthetic Chemistry, Faculty of Engineering, The University of Tokyo, Hongo, Tokyo 113, Japan (Received: April 29, 1991)
Molybdenum supported on Si02catalyzed CO hydrogenation to produce higher alcohols as well as hydrocarbons. The alcohol formation was found to require two kinds of Mo species, metallic Mo and Moo2,after reduction with flowing H2. Marked increase in alcohol yield with time on stream suggested that active species for alcohol synthesis were formed during CO-H2 reaction. Because the carburization of Mo resulted in insignificant change in the number and nature of the sites for alcohol formation, Mo carbides were excluded from the active sites for alcohol synthesis. Treatment of Mo catalysts with flowing atmospheric CO or CO-H2 was remarkably effective for the formation of sites producing alcohols. The Mo 3d XPS spectra of the C0-H2 conditioned catalyst showed a shoulder on the low binding energy side of the Mo 3d5,, peak. These findings suggest the formation of CO-reduction-induced defects on MOO,, MOO^-^, during the CO hydrogenation reaction, resulting in the increase in the alcohol synthesis rate. On the other hand, the hydrocarbon synthesis appeared to be solely based on metallic Mo. A dual-site mechanism for the alcohol formation over Si02-supportedMo has been proposed, CO dissociates on metallic Mo to form surface carbide, followed by hydrogenation to carbene and/or methyl species. Addition of methylene unit to alkyl and the following hydrogenation and/or dehydrogenation of alkyl to give hydrocarbons are also catalyzed by species. The mechanism can account for the difference metallic Mo,whereas CO insertion leading to alcohols occurs on in selectivities to branched products between hydrocarbons and alcohols.
Introduction There is a
i n w t in the catalystsfor synthesis of mixed alcohols, since the alcohol mixture has proved effective as number enhancer for motorfuel.i.2 Problems caused by the use of methanol (phase separation, vapor lock, etc.) can be alleviated by incorporating C2+higher alcohols as a cosolvent into Author to whom correspondence should be addressed. Presented in part at the 7th Annual International Pittsburgh Coal Conference, Pittsburgh, PA, September 1990.
the methanol-gasoline mixture. Several investigators reported that an appropriate modification of the methanol synthesis catalyt resulted in the formation of higher alcohols together with methanoi.'4 Institut Francais de Petrole patented catalysts for the (1) Klier. K. In Catalysis oforganic Reaciions; MOM, W. R.,Ed.; Marcel Dekker: New York, 1981; p 195. (2) Xu,X.;Doesburg, E. B. M.;Scholten, J. J. F. Carol. Today 1987,2, 125. (3) Natta, G.; Colombo,U.; Pasquon, I. In Caralysis;Emmett, P. H., Ed.; Reinhold: New York, 1957; Vol. V, p 131.
0022-3654/92/2096- 1334$03.00/0 0 1992 American Chemical Society
Alcohol Synthesis from CO-H2
The Journal of Physical Chemistry, Vol. 96, No. 3, 1992 1335
TABLE I: Influence of Reduction Temperature and K Content on Surface Characteristics of Mo/SiO, Catalysts O2 adsorpn Mo loading, K/Mo reductn XRD phases after reductn uptake," pmol wt 96 ratio temp, K MOO2 metallic Mo (g of catalyst)-' 673 strong not detected 90 5 0.4 773 medium medium 80 5 0.4 strong 70 5 0.4 873 weak 713 weak very strong 530 20 0 773 medium strong 140 20 0.2 773 medium medium 160 20 0.3 773 strong weak 130 20 0.4
apparent oxidn no. 3.4 2.0 1.3
"Measured at 195 K. bEstimated from O2 consumption from 195 to 773 K.
mixed alcohol synthesis, composed of methanol synthesis catalysts (Cu) and Fischer-Tropsch catalysts (CO).~ Molybdenum catalysts have long been recognized as being effective for the methanation and/or Fischer-Tropsch synthesis of light hydrocarbons. Our previous study, however, revealed that silica-supported molybdenum catalysts were active in the production of the C,-C5 mixed alcohols.610 The addition of KCl to Mo/Si02 resulted in a significant increase in the selectivity to alcohol formation. It was also found that the sequence of impregnation profoundly affected the activity and selectivity to alcohols.s Dow and Union Carbide have published a number of patents of the catalysts based on MoS2, promoted by COS and alkali metal salts, for higher alcohol synthesis."-14 Xie et al. investigated the pressure effect on the production of alcohols over MoS2 catalyst^.'^ Saito and Anderson investigated the activity of Mo species for the production of hydrocarbons and reported that zerovalent molybdenum gives higher hydrocarbons with higher activity than molybdenum oxides.I6 Boudart and Levy claimed that Mo carbides exhibited a much higher activity for synthesis of hydrocarbons from CO-H2 mixtures than molybdenum metal.I7 Ranhotra et al. discovered that the CO hydrogenation activity depended on the phase of Mo2C.18 From the study on addition of.probe molecules to CO-H2 catalysts under synthesis conditions, we found that the CO insertion into surface alkyl-metal bond occurred on Mo cataly~ts'~ and proposed a reaction scheme for the formation of C2+higher alcohols.20 The purpose of this study was to carry out a detailed examination of active species for higher alcohol synthesis and/or hydrocarbon synthesis over SO2-supported Mo catalyst. Particular attention was given to identifying the effects of catalyst preparation and pretreatment on the composition and structure of the Mo/Si02 catalysts. Attention was also given to clarifying the specics responsible for the carbon chain structure and carbon number distribution of the products. Experimental Section
Catalysts were prepared by impregnating silica gel (Fuji Davison ID) with an aqueous solution of (NH4)6M070244H20. For promoted catalysts, KC1 or Rb2C03was added to the silica fmt, followed by air calcination at 673 K for 1 h and impregnation (4)Henrici-Olive, G.;Olive S. The Chemistry ofthe Catalyzed HydroBerlin, 1984. (5) Sugier, A.;Freund, E. US.Patent 4122110,1978. (6)Tatsumi, T.; Muramatsu, A.; Tominaga, H. Chem. Lett. 1984,685. (7) Tatsumi, T.; Muramatsu, A.; Fukunaga, T.; Tominaga, H. Polyhedron 1986,5,257. (8) Tatsumi, T.; Muramatsu, A.; Tominaga, H. J. Card. 1986,101,553. (9) Tatsumi, T.; Muramatsu, A.; Tominaga, H. Appl. Carol. 1987,34,77. (10)Muramatsu, A.;Tatsumi, T.; Tominaga, H. Bull. Chem. Soc. Jpn. 1987.60,3157. (11)Quarderer, G.J. EPO-0119609,1984. (12)Kinkade, N. E. EPO-0149255,1985. (13) Kinkade, N.E. EPO-0149256,1985. (14)Stevens, R. R. EPO-0172431,1986. (15) Xie, Y. G.;Naasz, B. M.; Somorjai,G. A. Appl. Coral. 1986,27,233. (16)Saito, M.; Anderson, R. B. J. Card 1978,51, 1. (17) Boudart, M.; Levy, R. Science 1973,181,547. (18) Ranhotra, G.S.;Bell, A. T.; Reimer, J. A. J. Coral. 1987,108,49. (19)Tatsumi, T.; Muramatsu, A.; Yokota, K.; Tominaga, H. J . Mol. Card 1987,I I , 129. (20)Tatsumi, T.; Muramatsu, A.; Yokota, K.; Tominaga, H. J. Coral. 1989,115, 388.
genurion of Carbon Monoxide; Springer-Verlag:
with the molybdenum salt. Unless otherwise noted, the impregnates were dried at 393 K overnight and treated in flowing He at 673 K for 1 h and then reduced in flowing H2 at 773 K for 12 h. The CO hydrogenation was carried out in a stainless steel tubular reactor, the heated section of which was 50 cm long and 0.6 cm inside diameter, containing 1 .O g catalyst in a flow system. Synthesis gas (H2/C0 = 1) was supplied to the reactor through a stainless-steel tube heated to 473 K for decompositionof carbonyl impurities. Hydrogen was purified of oxygen by passage through a DEOXO unit (Engelhard), followed by a molecular sieve trap. Products were analyzed by gas chromatography. Concentrations of CO, C02, and CHI were determined by a TCD gas chromatograph with an active carbon separation column at 323 K using H2 as a carrier gas. The distribution of organic compounds was determined by FID gas chromatographs with the following separation columns using N2 as a carrier gas; 2 m Porapak Q for the analysis of C1-C5 hydrocarbons, 2 m polyethylene glycol 1500 for oxygenates, 2 m silicon SE-30for c6+ hydrocarbons, and 8 m VZ-7 for isomers of C3, C4, and C5. The catalytic surface areas of reduced Mo catalysts were measured by O2 adsorption at 195 K.21 After the O2 adsorption was measured, each sample was subjected to oxygen titration; oxygen was admitted to the chemisorption cell at about 300 Torr and the sample was heated stepwise up to 773 K. The final equilibrated O2pressure was >50 Torr. The original oxidation number of the Mo could then be calculated from the total amount of O2consumed, assuming that Moo3 was the final product. The XPS spectra of supported catalysts were recorded on a Shimadzu electron spectrometer ESCA 750 with Mg-Kar excitation radiation. Sample preparation of XPS measurement was performed in a glovebox filled with nitrogen. X-ray powder diffraction measurements were performed using a Rigaku Denki diffractometer RU-2OOA with CuKa radiation.
Results and Discussion Mo Species on Freshly Reduced Catalysts. The oxidation state of molybdenum naturally lowers with increasing the reduction temperature. Saito and Anderson reported that small peaks of Mo metal were found in an X-ray diffraction pattern following reduction of unsupported Moo3 with H2 at 673 K and that Mo metal was obtained by the reduction at 873 K.16 The effect of temperature for H2 reduction on the characteristics of Si02-supported molybdenum catalysts has been invtstigated. Table I lists O2 adsorption uptakes, phases observed by X-ray diffractograms (XRD), and average oxidation numbers of Mo after H2 reduction (before exposure to synthesis gas). Strong X-ray lines corresponding to metallic Mo and weak lines characteristic of Mo02 appeared on the catalyst reduced at 873 K. On decreasing the reduction temperature, the intensity of the Moo2 lines increased at the expense of metallic Mo lines. On the catalyst reduced at 673 K, metallic Mo lines could be hardly observed. Average oxidation number estimated by O2titration was consistent with the XRD observation. The influence of K content on the surface characteristics of the fresh catalyst are also shown in Table I. The presence of K (21)Muralidhar, G.;Concha, B. E.; Bartholomew, G. L.; Bartholomew, C. H. J . Card 1984,89,274.
1336 The Journal of Physical Chemistry, Vol. 94, No. 3, 1992
Muramatsu et al.
TABLE E Influence of Reduction Temoeratwe and K Content on Performan& of Mo/SiO%CaWwts Mo activity loading, K/Mo reductn co co, alcohols, g wt %ratio temp, K convn, % yield,-% (kg of catalyst)-! HC,b h-' 5 0.4 673 1.9 0.96 2.6 5.3 5 0.4 3.8 17 25 773 8.2 6.5 14 34 5 0.4 813 12 18 1.4 140 20 0 773 31 5.0 22 34 20 0.2 773 11 20 0.3 773 5.8 2.5 20 13 20 0.4 173 3.4 1.4 14 5
C2+/CI molar ratio alcohols HCb 1.1 1.2 2.1 1.3 2.9 1.8 0.49 1.4 1.1 1.4 2.2 1.7 2.3 1.7
olefin content in C,, % 39 65 81 11 51 72 78
"Reaction conditions; 573 K, 1.6 MPa ( H 2 / C 0 = l), W/F = 10 g of catalyst h mol-'. bHydrocarbons.
effectively prevented the complete reduction of molybdenum to metal. O2adsorption uptakes for the catalysts suggest that the Mo dispersion decreases sharply on addition of K but is insignificantly affected by the reduction temperature. The interpretation of O2adsorption uptakes, however, may not be straightforward, because the stoichiometry of adsorption of oxygen may depend on the Mo species (Moo2, metallic M O ) . ~ ' ~ ~ ~ The influence of the reduction temperature of K content on activity in CO hydrogenation is shown in Table 11. As expected from the water gas shift activity of Mo c a t a l y ~ t s ,these ~ . ~ catalysts yielded C 0 2 almost exclusively instead of water as the coproduct (vide infra). In Table 11, specific activities in g/(kg of catalyst h) were used to compare alcohol synthesis activities. Clearly there are marked difFerenm between these catalysts, the low conversion of CO on the catalyst reduced at 673 K beiig the most noticeable. CO conversion increased as the reduction temperature increased. The activity for hydrocarbon formation and the ratio of alkene to alkane also increased. The catalyst reduced at 873 K exhibited the highest activity and selectivity for C,+ hydrocarbons. Best results in terms of the activity and selectivity for alcohol synthesis were obtained for the catalyst reduced at 773 K. The C2+ alcohols/methanol ratio increased with increasing the reduction temperature. The effect of K content on the catalytic performance of Mo/Si02 will be summarized as follows. On increasing the K content, the selectivity for alcohols increased at the expense of CO conversion, the C2+/C1 ratios for alcohols and hydrocarbons increased, and the alkene content increased. Taking into account the findings of the XRD observation, the higher CO conversion seems to be related to the higher metallic Mo content in the catalysts.16 There was no clear relationship between O2uptake and CO conversion. Reduction of Mo at higher temperature or in the absence of K resulted in the evolution of large amounts of metallic Mo, leading to the catalyst producing hydrocarbons predominantly. This is in good agreement with our previous findings on the effect of Mo precursors' and the order of impregnation with molybdenum and potassium salts.* The catalyst reduced at 673 K and with high K content, for which XRD patterns of metallic Mo phase could be hardly observed, showed very low activity for CO hydrogenation. Thus it is suspected that the presence of both metallic Mo and Moo2 species on the catalyst surface before exposure to synthesis gas is required to effect the production of alcohols and that there is an appropriate content of metallic Mo. Activity chpoge with Procmsing Time. It has been shown that the presence of an adequate amount of metallic Mo and Moo2 on the freshly reduced catalyst is a prerequisite for the superior performance for alcohol synthesis. However, it is inconceivable that the species over the freshly reduced catalysts are directly responsible for the alcohol synthesis from synthesis gas because we found a rather long induction period for alcohol formation. The change in yields of alcohols, hydrocarbons, and C 0 2over Mo (10wt 9%)-K (1.63 wt 9%)/Si02at 523 K, 1.6 MPa, and W/F = 10 g of catalyst h mol-', is illustrated in Figure 1. In the initial (22) Concha, B. E.; Bartholomew. C. H. J. Caral. 1983, 79, 327. (23) Shultz J. F.;Kam, F.S.;Anderson, R.B.US.Bur. Mines Rep. 1967, No.6974, 1. (24) Dim, J. W.;Gulari, E.;Ng, K. Y.S.Appl. Carol. 1985, 15, 247.
5
\
I\
Hydrocarbons 0l
5
10
i
15
20
:25
Processing Time (h)
F v 1. Change in product yields with processing time. Catalyst: Mo (10 wt %)-K (1.63 wt %)/SO2; reaction conditions; 523 K, 1.6 MPa ( H 2 / C 0 = l), W/F = 10 g of catalyst h mol-'.
stage of the C0-H2 reaction the yield of alcohols was surprisingly low. The activity for alcohol formation remarkably increased with time on stream and reached a steady-state level after ca. 15 h. This implies that active species for alcohol production are formed during the CO-H2 reaction. In contrast, the yield of C 0 2 was remarkably high at the initial stage of the reaction and decreased to reach a nearly steady state after ca. 15 h on stream. The yield of hydrocarbons gradually decreased and leveled off after 8 h on stream. As mentioned above, Mo catalysts are active in CO shift reaction (eq 5 ) to form C02;under conditions employed in this study, the equilibrium conversion of eq 5 was estimated at 97% at 573 K. Almost all of the oxygen which had been included in CO left the reactor as C 0 2 (eqs 3 and 4)rather than as H 2 0 (eqs 1 and 2). Figure 1 shows that at the steady state the yield of C 0 2 is found to be in good agreement with the combined yield of hydrocarbons and alcohols, assuming the equilibrium in the CO shift reaction (eq 6). In the initial stage of the CO-H2 reaction, however, the yield of C02noticeably exceeded the yield of alcohols and hydrocarbons. The formation of excess of C 0 2 should be ascribed to Boudouard reaction (eq 6) or reduction of Moo2 by CO (eq 7). It should be noted that there was a coincidence between the times required to reach the steady state of alcohol synthesis and C 0 2 formation. nCO + 2nHz -(CH2),- + nHZO (1) nCO + 2nH2 CH3(CH2),10H + ( n - 1)H20 (2) 2nCO + nH2 -(CH*)"- + nCO2 (3) (2n - l)CO + (n + 1)Hz CH,(CH2),1OH + ( n - 1)COz
-
4
+
+
---
CO + H2O nMo + 2CO Moo2 + CO
(4) (5)
COZ + H2 Mo,C + C 0 2 (6) M002, + CO2 (7) CO,C0-H2 pnd C d m r M q Treatment,A control experiment was carried out in which Mo ( 5 wt 5%)-K (0.83 wt W)/SiO,
The Journal of Physical Chemistry, Vol. 96, No. 3, 1992 1337
Alcohol Synthesis from C0-H2
I
I
pq
Fresh
30
20
10
50
40
Processing Time (h)
Figure 2. Effect of CO pretreatment on change in alcohol yields with processing time. Catalyst: Mo (5 wt %)-K (0.83 wt %)/SO,; reaction conditions: 573 K, 1.6 MPa ( H 2 / C 0 = l), W/F = 10 g of catalyst h
I '
245
I
I
1
240 235 230 Binding Energy (eV)
I
225
Figure 4. Mo (3d) ESCA region of fresh and used Mo (10 wt 7%)-K (1.63 wt %)/SiO,.
40
80
120
Processing Time (h)
Figure 3. Effect of CO-H, and n-C4Hlotreatment on change on alcohol yields with processing time. Catalyst: Mo (10 wt %)-K (1.63 wt %)/ SO,; reaction conditions: 573 K, 1.6 MPa (H2/C0 = l), W/F = 10 g of catalyst h mol-'.
catalyst reduced in H2 at 773 K for 12 h was further treated in flowing CO at 523 K,0.1 MPa, and W/F = 10 g of catalyst h mol-' for 20 h, left under H2 stream at 773 K until CH4 formation virtually ceased, and subjected to C0-H2 reaction. The change in the yield of alcohols over the untreated catalyst and the catalyst treated by a flow of CO is illustrated in Figure 2. The initial rate of alcohol formation is markedly higher over the CO-treated catalyst than over the catalyst resulting from H2 reduction alone. Over the former catalyst, the rate of alcohol synthesis was virtually constant during the processing time up to 26 h, while over the latter the rate after 40 h on stream was 20 times as high as that in the initial stage. Hence sites for alcohol synthesis appear to have been formed on the catalyst during the CO treatment. The amount of C 0 2 produced from the CO treatment was 117%based on the loaded Mo. We expect that the carbide formed according to eq 6 would react with H2 to form CH4. The CH4 formed by the successive H2 treatment accounted for only 46%; a significant portion of C 0 2 seems to have been formed according to eq 7, resulting in the reduction of MOO, to form oxygen defects according to eq 7. The amount of O2adsorption on the catalyst treated in CO and the untreated catalyst was 46 and 79 mmol (g of catalyst)-', respectively. Thus it is speculated that the lower activity of the CO-treated catalyst at the steady state would be explained by the lower dispersion of molybdenum. A linear correlation between O2adsorption uptake and CO hydrogenation activity was established for unsupported As shown in Figure 3, pretreatment of the reduced Mo-K/Si02 catalyst with atmospheric C0-H2 at 673 K resulted in similar enhancement of initial alcohol formation activity. Mo carbides have been known to be very active in methanation of C0.17J8Hence one might wonder that the enhancement of
alcohol synthesis activity during the C0-H2 reaction and also by the CO or C0-H2 treatment is due to the carburization of Mo. The Mo (5 wt 7%)-K (0.83 wt 5%) catalyst reduced with H2 was treated with n-C4H,o (10 vol %)/H2 according to the methods of Leclercq et alaz5and then subjected to the reaction. The activity change with processing time is shown in Figure 3. The initial activity for hydrocarbon formation was increased by a factor of 3 by the treatment. This increase is in agreement with the X-ray powder patterns for the catalyst, which contained lines characteristic of Mo2C. Although the initial yield of alcohols was also considerably increased with the n-C4Hlo-H2treatment, the alcohol yield over the treated catalyst was only a half of that over the CO-H2 treated catalyst. Moreover, on the n-C4Hlo-H, treated catalyst the alcohol yield was gradually increased with time on stream and continued to increase after 30 h on stream, in contrast with the rapid attainment of the steady state on the latter. In this regard it should be noted that the CO/H2 ratio, temperature, and pressure proved crucial in the C0-H2 pretreatment; we have found that by the pretreatment with CO-rich synthesis gas with CO/H2 = 3 at 573 K and 1.0 MPa, the activity for hydrocarbon formation was increased by 1 order of magnitude and at the same time the probability of C-C chain growth for hydrocarbons was markedly increased.26 This activity increase may result from the formation of molybdenum carbide species MoC, which was revealed by X-ray diffraction measurement. On the other hand, the activity for alcohol formation and the probability of chain growth for alcohols remained virtually unchanged after this treatment. In contrast, there was no XRD evidence for Mo carbide species in the catalyst activated for alcohol formation. From these findings, we can exclude the possiblity that Mo carbides are directly related to the active site for alcohol synthesis. It may be concluded that the increase in alcohol production rate during the CO-H2 catalytic run is ascribed to the formation of CO-reduction-induced defects on Moo2 (eq 7). It has been established that CO reduction induces defects in ZnO faces which cannot be reduced with H2 and that these are specific sites for methanol synthesisan It has been recently proposed that the active site for hydrogenation of CO to methanol over Zr02 is an oxygen anion vacancy.28 Figure 4 shows the Mo 3d XPS spectra of fresh and used Mo-K/Si02 catalysts. On the used catalyst there appeared a shoulder on the low binding energy side of the Mo 3d5,, peak, supporting that Mo oxides on the catalyst surface was partially (25) Leclerq, L.; Imura, K.; Yoshida, S.; Boudart, M. In Preparation of Catalysrs II; Delmon, B., et al., Eds.; Elsevier: Amsterdam, 1978; p 627. (261 Tatsumi, T.; Muramatsu, A.; Tominaga, H. J . Jpn. Pet. Inst. 1989,
32,'43: (27) Bowker, M.; Hyland, J. N. K.; Vandervell, H. D.; Waugh, K. C. Proceedings 8th International Congress on Catalysis; Verlag Chemie: Weinheim, 1984; Vol. 2, p 35. (28) Silver, R. G.; Hon, C. J.; Ekerdt, J. G. J . Catal. 1989, 118, 400.
Muramatsu et al.
1338 The Journal of Physical Chemistry, Vol. 96, No. 3, 1992 COIH2IH20 = i i l i l
COIH2
= 1/1
/"t
Alcohols
0.2 -20
0
20
40 60 Processing Time (min.)
80
100
120
Figure 5. Effect of H 2 0 addition to CO-H2 reactant gas on product yields. Catalyst: Mo (10 wt %)-Rb (3.6 wt %)/Si02; reaction conditions: 523 K, 1.6 MPa ( H 2 / C 0 = l), W / F = 10 g of catalyst h mol-'; start of the H 2 0 addition at processing time = 0.
reduced during the CO-H2 reaction. Addition of H 2 0 to C0-H2. Figure 5 illustrates the influence of the addition of H 2 0 to CO-H2 reactant gas upon the yields of alcohols and hydrocarbons over Mo (13 wt %)-Rb2C03 (Rb = 3.55 wt %)/SiOz. After the start of H 2 0 addition, the space-time yield of alcohols enormously increased, while that of hydrocarbons gradually decreased. This marked transient increase in the alcohol yield was ascribed to change in the character of surface Mo; supposedly surface metallic Mo was partially oxidized, resulting in the formation of active species for alcohol synthesis. The yield of alcohols reached a maximum at ca. 40 min on stream and then gradually decreased. The steady-state activity for CO hydrogenation during the H 2 0 addition was lower than that before the exposure to H20, while the shift reaction was not retarded. It seems that the site necessary for the activation of hydrogen (presumably metallic Mo) has been fully blocked or oxidized by excess water. In fact, the XRD of the catalyst after the addition of H 2 0 showed no lines corresponding to metallic Mo. Product alcohols consist mainly of methanol, possibly because the content of metallic Mo species, on which C-C chain growth reaction could occur (vide infra), was quite low on the addition of H20. This is consistent with the much lower yield of hydrocarbons than without addition of H20. The addition of C 0 2 (5-20 ~ 0 1 %to ) synthesis gas on alcohol production has been found to produce a similar effect on that of the H 2 0 addition. Significant increase in the initial alcohol yield on C 0 2addition29could be ascribed to the increase in the content of oxidized active state for alcohol synthesis. However, the synthesis rate at the steady state was lowered by addition of C02. M e d d s m for Branching. As has been described elsewhere,20 isomer distributions in alcohols and alkanes obtained with Mo/ Si02catalysts were different. Although almost all of hydrocarbons were straight-chain ones, alcohols consisted not only of isomers with straight alkyl chain but also consisted considerably of those with branching in the 2-position. It is also pointed out that alcohols are exclusively primary. From the study of addition of probe molecules to CO-H2,20 it has been clarified that the alcohol formation from CO-H2 proceeded via the same intermediate as the alkene carbonylation; a mechanism including CO insertion into an alkyl-metal bond as a key step has been proposed as the main reaction path for the higher alcohol formation from CO-H2 (eq 8) on Mo/Si02 catalysts. The hypothesis of the insertion of CO into a growing alkyl-metal bond has already been proposed by many authors to explain the formation of oxygenated prodThe formation of alcohols with branching in the 2-position can be accounted for by eq 9; the insertion of CO into the secondary alkyl-metal bond leads to branched alcohols. Secondary
ratio
Figure 6. Effect of K content on branched/straight ratio in C4 and C5 alcohols. Catalyst: Mo (20 wt W)/Si02; reaction conditions: 573 K, 1.6 MPa ( H 2 / C 0 = l ) , W/F = 10 g of catalyst h mol-'.
alkyls are thought to be formed through u v interconversion of coordinated primary alkyls. Equations 10 and 11 show the formation of straight-chain and branched alkanes, respectively. The difference in the selectivity to branched products between hydrocarbons and alcohols suggests that alcohols are formed at sites different from those for hydrocarbon formation, that is, the sites, M', in eqs 8 and 9 differ from the sites, M, in eqs 10 and 11. From the findings which have been described above, there is good reason to suppose that M' is M a 2 - * and M metallic Mo. It has been found that the metal electron density determines the secondary/primary ratio of coordinated alkyls?I Electron-riches of metal gives rise to more anionic character of coordiiated alkyls, favoring the normal type of alkyl species. If alkyl species has radical or cationic character as a result of less electron-rich character of the metal, secondary alkyl species should be more stable than primary alkyl species. MOO,, site should be less electron-rich than metallic Mo, allowing the formation of secondary alkyl groups (eq 8). In contrast the pathway eq 11 is supposed to be infrequent since secondary alkyl groups should be disadvantageous on electron-rich sites of metallic Mo. co H RCHpCH# RCHpCHpC-M' RCHpCHpCHpOH ( 8 ) II
c "3
-
RCH-M' I
RCHpCHpM CH3
C
CHI
I RCH-M
II
0
RCH&H&Hfl
CHI
0
CH3 O RCHGM' I
CH3
I RCHCHpM
CH3 RCHCH,OH I
H
H2
Hz
(9)
RCH2CHpCH3 (IO) CH3
I
RCHCH,
(11)
Figure 6 represents the branched/normal ratio in C4 and Cs alcohols plotted against the content of K added to Mo (20 wt %)/SO2 catalysts. With increasing K content, the selectivity to branched alcohols significantly decreased. In contrast with this observation, Vedage et al. found that the yield of 2-methyl-lpropanol over Cu/ZnO was enhanced by addition of Cs and explained its formation by assuming aldol condensation of aldehydic intermediates with Cl h~termediates.~~ The decrease can be accounted for by K-induced increase in electron density of molybdenum in Mo02,(M') species, which makes the reaction eq 8 more favorable than eq 9. It should be also pointed out that the significant formation of 2-ethyl-l-butanolzocan be accounted for by the CO insertion mechanism (eq 12) but is inconsistent with the aldol condensation mechanism. C2HSCH2CHZCH2-M -.c C2HSCHz(CH3)CH-M +
~~~
0.4
0.3
K/Mo
(CzH3)ZCH-M
-+
(CzHS)zCHCH20H
(12)
~
(29) Tatsumi, T.; Muramatsu, A.; Tominaga, H. Chem. Len. 1985,593. (30) Rabo, J. A.; Risch, A. P.; Poutsma. M.L. J . Catal. 1978, 53, 295. Takeuchi, A.; Katzer, J. A. J . Phys. Chem. 1982,86,2438. Sachtler, W . M. H.; Ichikawa, M. J . Phys. Chem. 1986, 90,4752. Favre, T. L. F.; Van der Lee,G.; Ponec, V. J . Chem. SOC.,Chem. Commun. 1985, 230.
(31) Tamao, K.; Kiso, Y.;Sumitani, K.;Kumada, M. J. Am. Chem. Soc. 1982, 94, 9268. (32) Vedage, G. A.; Himelfarb, P.; Simmons, g. W.; Klicr, K.ACSSymp. Ser. 1986, 279, 295.
The Journal of Physical Chemistry, Vol. 96, No. 3, 1992 1339
Alcohol Synthesis from CO-H2
TABLE Iu: Dependence of Alkene Conversion on Carbon NumbeP K=O C, paraffin
C,,+, alcohols yield (A), % 3.4 0.53 0.031
olefin C. ethylene propylene 1-pentene
yield (P), % 30
ratio
Alp 0.1 1
Cn+,alcohols yield (A), %
3.3 0.20
0.027 0.0053
20
5.8
K = 1.63 wt % C, paraffin yield (P), % 8.8 6.3
ratio AIP 0.38 0.032
@Catalyst:Mo (20 wt %)/Si02. Reaction conditions: 473 K, 1.6 MPa (H2/C0 = 1, olefin 1 mol %), W/F = 10 g of catalyst h mol-'. R l cno l o l
1
2 )
CO
i
01c f i n
' 0 C'xEy C E y, CO, k0
H2
Mo lnchl si02
Figure 8. Schematic model of CO hydrogenation to alcohols and hydrocarbons on the surface of Si02-supported Mo catalysts. 3
2
1
5
4
Carbon Number, n
F g u e 7. Anderson-Schulz-Flory plots for hydrocarbons and alcohols. Catalyst: Mo (10wt %)-K (1.63 wt %)/SO,; reaction conditions: 573 K, 1.6 MPa (H2/C0 = l), W/F = 10 g of catalyst h mol-'.
Anderson-Schulz-Flory Distributions for Hydrocarbons and Alcohols. Figure 7 shows Anderson-Schulz-Flory plots for hydrocarbons and alcohols produced over Mo (10 wt %)-K (1.63 wt %)/Si02 catalysts. Over Mo catalysts, the carbon number distribution obeyed an Anderson-Schulz-Flory type distribution for hydrocarbons and alcohols, respectively, except for methanol. Because the yield of methanol was far below the equilibrium value, the deviation cannot be caused by thermodynamic control. This deviation could be due to a pathway for methanol formation different from other products; methanol is the only product that does not involve the cleavage of carbon-oxygen bond of CO. Over our Mo catalysts, the probability of the chain growth, a, was 0.48-0.32 for hydrocarbons and 0.32-0.14 for alcohols; although the a values were dependent on the reaction conditions such as temperature and pressure and K content of the catalysts, the a value for alcohols was always smaller than that for hydrocarbons. One might wonder that the difference in the a value between hydrocarbons and alcohols is attributed to the difference in the carbon-chain growth step between these products. However, it is thought that this is not the case. The chain propagation of hydrocarbons is represented by the CH2 insertion steps in eqs 10 and 11. These steps are followed by hydrogenolysis to give alkanes (eqs 10 and 11) or 8-elimination to give I-alkenes. We consider that the alkyl groups thus formed on the metallic site, M, migrates to the Moo2-, site, M', and is subject to CO insertion to give alcohols (eqs 8 and 9). If the chain propagation of alklyl groups of alcohols occurred on MOO^-^, we should observe alcohols branching in the 3 or further positions according to the following equation y
3
RCH-M'
C
O
CH3 I RCHC-M'
H A
II
CH3 I
RCHCHd'
0 y
3
RCHCHztM'
H2
4
CH3
I
RCHCHZCHZOH (13)
0
Table I11 shows the dependence of the yields of carbonylation and hydrogenation products on the carbon number of alkenes which were added to the reactant CO-Hz gas. The yields from added alkanes were determined from the increase in the yields
of corresponding products as a result of addition of each alkene.
As the carbon number of the alkene increased, activities for both hydrocarbonylation and hydrogenation decreased. This was associated with the marked decrease in the ratio of hydrocarbonylation to hydrogenation, suggesting that the insertion of molecular CO to alkyl-metal bonds is more retarded for the longer alkyl chain. It is conceivable that this difference in the carbon number dependence between hydrocarbonylation and hydrogenation results in the difference in the a between alcohols and hydrocarbons even if alcohols and hydrocarbons follow a common chain growth step, namely, the insertion of a methylene unit to an alkyl group. Dual-SiteMechanism for Alcohols and Hydrocarbons Formation. Figure 8 depicts a mechanistic model of CO hydrogenation reaction on the surface of a Mo catalyst. As decribed above, oxidized Mo species, MOO^-^, should be responsible for the formation of alcohols, whereas metallic Mo species is necessary for the formation of both hydrocarbons and oxygenates. It is known that metallic Mo readily dissociates CO at room temperature and that oxidation of the surface inhibits the dissociative component of adsorption and increases the associative component.33 Presumably C2+ alcohols are formed via a dual-site mechanism as follows. First, CO adsorbed dissociatively on metallic Mo to form surface carbide, which is hydrogenated to give carbene and/or methyl species, while carbon monoxide is nondissociatively adsorbed on sites. The alkyl chain growth mechanism must be identical for hydrocarbons and alcohols. The chain growth is likely to occur on the same metallic Mo site, where carbene insertion into an alkyl-metal bond leads to a longer alkyl group. It appears that the termination steps for alcohols and hydrocarbons occur on different sites. Hydrocarbons should be formed via hydrogenolysis of alkyl-metal bond and/or 8-H abstraction on metallic Mo sites, whereas the insertion of CO giving rise to alcohols occurs on MOO,,; alkyl groups formed on metallic Mo are assumed to migrate to a molecular CO on Moo2-, species. A similar dual-site mechanism has been proposed for Rh catalysts, yielding Cz oxygenates ~electively.'~In a mechanistic study on CO hydrogenation over Fe catalysts, Huff and Satterfield found two sites over Fe catalysts with one site producing 1-alkenes, alkanes, and oxygenates, and the other not producing o~ygenates.'~ Anderson and Ekerdt claimed a similar mechanism for the carbon chain-growth, occurring on one site to produce alcohols and on (33)KO,E.I.; Maddix, R. J. Surf. Sci. 1981, 109, 221. (34)Watson, P.R.;Somorjai, G. A. J. Carol. 1981, 72, 347;1982,74,282. Driessen, J. M.; Poels, E. K.; Hindermann, J. P.;Ponec, V. J. Caral. 1983, 82,26. van den Berg, F.G. A.; Glezer, J. H. F.; Sachtler, W. M.H. J. Caral. 1985. - - - - , -93. - , -340. .- . (35) Huff, Jr., G. A.; Satterfield, C. N. J . Caral. 1984,85,370.
J. Phys. Chem. 1992, 96, 1340-1343
1340
SCHEME I: Scbeme of Elementary Steps for CO Hydrogenation to Alcohols and Hydrocarbons on Mo/SiO, M-CO
G
MI-CO -CEJOH
M;C
M-~CE
co
M-MO ~ 1 1 1 MJ-MoOZ-r
the other site to produce hydrocarbon^.^^ These sites may differ only in the type of termination step allowed, alkyl hydrogenation/dehydrogenation to hydrocarbon products vs CO insertion leading to oxygenates. They suggested the difference of Fe valence between sites forming hydrocarbons and sites forming alcohols. Methanol is the only product that does not involve the carbon-oxygen bond of CO and assumed to be formed solely on MOO,-,. The activity change with processing time is consistent (36) Anderson, K. G.; Ekerdt, J. G. J . Catal. 1985, 95, 602.
with this mechanism. In the initial stage of the CO-H, reaction the yield of methanol was noticeably low compared to C,+ alcohols. As the CO-H2 catalytic run was continued, the increase in the yield of methanol was more remarkable than that of C,+ alcohols, resulting in the decrease in the C,+/Cl ratio for alcohols. This is interpreted in terms of the conditioning of the catalyst surface; Mo02-, was increased through reduction of M a z by CO according to eq 7. Consequently, a reaction scheme for hydrogenation of CO over the Mo catalysts producing hydrocarbons and alcohols consists of the following steps as shown in Scheme I. 1. Dissociation of adsorbed CO on metallic Mo to form CH3/CH2eventually is followed by growth of alkyl chain via CHI insertion. 2. Migration of alkyl group to partially reduced Mo02 followed by insertion of nondissociatively adsorbed CO results in acyl species. Their hydrogenation leads to higher alcohols. 3. Hydrocarbons are formed by way of hydrogenolysis or @-hydrogenelimination of alkyl group on metallic Mo. 4. Nondissociatively adsorbed CO is hydrogenated to methanol on partially reduced MOO,. Regisby NO. CO,630-08-0; MOO,, 18868-43-4; Mo, 7439-98-7.
Near- Infrared Surface-Enhanced Raman Scattering from Metal Island Films C. A. Jennings? G. J. Kovacs,t Xerox Research Centre of Canada, 2660 Speakman Drive, Mississauga, Ontario, Canada L5K 2Ll
and R. Aroca* Department of Chemistry and Biochemistry, University of Windsor. Windsor, Ontario, Canada N9B 3P4 (Received: May 7, 1991; In Final Form: August 26, 1991)
‘
Surface-enhanced Raman scattering (SERS) of molecular species on metal island films of Au, and Au films roughened with underlayers of CaF,, was obtained using the near-infrared excitation line of the Nd:YAG laser at 1.064 pm. The molecular species were vanadylphthalocyanine and 3,4,9,1O-perylenetetracarbylicdianhydride, which were coated onto the Au surfaces as thin vacuum-evaporated layers. An enhancement factor of about 100 was obtained for the electromagnetic enhancement. Demonstration of the infrared SERS effect for the two macrocycle molecules is presented and comparisons are made with previous SERS results obtained with visible light excitation. This is the first reported observation of SERS from metal island films with excitation in the infrared region.
Introduction Fourier transform (FT)-Raman spectroscopy is the most recent addition to a long list of Raman spectroscopic techniques.’., It uses a Michelson interferometer for detection of the inelastic light scattering excited with a Nd:YAG laser operating at 1.064 Hm. A ‘‘fluoreacencefree” Raman technique has been one of the main advantages of FT-Raman spectroscopy for practical applications. However, there are fundamental and technical limitations to sensitivity in FT-Raman spectroscopy. The fourth-power frequency dependence of the inelastically scattered light is a fundamental limitation; other limitations are technical, and improvements are under investigation.’ Enhancing the Raman signal in the infrared region will make FT-Raman spectrosoopy an even more powerful analytical technique and will extend its present applications to thin solid films and LangmukBlodgett monolayers. Since most materials have no electronic absorption in the near-infrared region, which prevents a resonant Raman enhancement, surface-enhanced Raman scattering (SERS) is the ‘This work is part of the doctoral thesis of C.A.J. ‘Present address: Xerox Palo Alto Research Center, 3333 Coyote Hill Road, Palo Alto, CA 94304-1314. ‘Author to whom correspondence should be addressed.
0022-3654/92/2096-1340$03.00/0
only approach that could produce a nonconventional enhancement of the Raman signal. SERS3 was the first of a number of new techniques, encompassed under the general name of surface-enhanced spectroscopy. In the visible region of the spectrum, SERS is now a common analytical technique routinely used, for instance, in biochemist$ and electrochemistry? Experiments are generally carried out on rough surfaces with the appropriate dielectric constant and shape, as required by the electromagnetic enhancement$ Ag, Au, and Cu are the surfaces largely used. Recently, infrared FT-SERS of pyridine on Cu and Au electrodes’ (1) Chase, D. B. ICORS 1990 (International Conference on Raman Spectroscopy); Durig, J. R., Sullivan, J. F.,Eds.; John Wiley & Sons: Chichester, 1990; p 11. (2) Spectrochim. Acta 1990, 46A. Special edition on FT-Raman spectroscopy, P. J. Hendra, Ed. (3) Surface Enhanced Raman Scattering; Furtak, T. E., Chang, R. K., Eds.; Plenum: New York, 1982. (4) Cotton, T. M.In Spectroscopy of Surfaces; Clark, R. J. H., Hester, R. E., Eds.; J. Wiley & Sons: Chichester, 1988; Vol. 16, p 91. (5) Chang, R. K.;Laube, B. L. CRC Crir. Rev. Solid State Mater. Sci. 1984, 12. (6) Moskovits, M. Rev. Mod. Phys. 1985, 57, 783. (7) Angel, S.M.; Katz, L. F.;Archibald, D. D.; Lin, L. T.; Honigs, D. E. Appl. Spectrosc. 1988, 42, 1327.
0 1992 American Chemical Society
