4542
J . Phys. Chem. 1992, 96, 4542-4549
Acknowledgment. This work was performed in the department of Prof. M. Kahlweit. We are indebted to him for support. We thank K.-V. Schubert and J. Winkler for assistance with the SANS experiment as well as our local contact B. Farago at ILL, Grenoble. M.J. thanks the Swedish Board of Technical Devel-
opment (STU) for financial support during his stay at the MPI in Gottingen. Registry No. C,H,,, 1 1 1-65-9; BuOH, 71-36-3; H3C(CH2),0H, 7141-0; H ~ C ( C H ~ ) ~ O 1H1 ,1-27-3; H ~ c ( c H ~ ) , O H , 1 1 1-87-5; H3C(CH2)90H, 112-30-1; CBES, 19327-40-3.
Roles of Chemisorbed Oxygen and Zinc Oxide Islands on Cu(ll0) Surfaces for Methanol Decomposition Sabrina S. Fu*it and Gabor A. Somorjai Materials Science Division, Center for Advanced Materials, Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720, and Department of Chemistry, University of California, Berkeley, California 94720 (Received: April 15, 1991)
The interaction of methanol with Cu-Zn-O surfaces is studied on a model system produced by depositing zinc oxide islands on Cu(ll0) surfaces. Various amounts of oxygen are then adsorbed onto the exposed copper part of the surface to form ZnO,/y ML O/Cu( 110) surfaces. The role of O/Cu( 1 10) in methanol decomposition on our model Cu-Zn-O surfaces was discerned by examining the interaction of methanol, adsorbed at 150 K in ultrahigh vacuum, with these ZnO,/y ML O/Cu( 110) surfaces using temperature-programmed desorption (TPD), low energy electron diffraction (LEED), and Auger electron spectroscopy (AES). Methanol reduces the zinc oxide and removes oxygen chemisorbed on copper. The decomposition temperatures of formate and methoxy species formed upon methanol decomposition on the components of oxygen covered Cu( 110) and ZnO, are unaffected by the presence of the other component. However, the amount of formate species decomposed on the ZnO, component of ZnO,/O/Cu( 110) surfaces is controlled by the amount of chemisorbed oxygen on copper. We show that there are cooperative effects between O/Cu and ZnO which increases the total amount of formate formed; for Zn0,/0.2 ML O/Cu( 110) surfaces, the concentration of surface formate doubles on the three-component system as compared to the sum of the concentration of formate species on the separate components. This produces a change in the relative surface formate to methoxy ratio from 1:9, for the case of 0 . 2 ML O/Cu(l lo), to 1:3, for the case of Zn0,/0.2 ML O/Cu(llO) surfaces.
Introduction Cu-Zn-0 catalysts are important for the production of methanol from CO/CO2/H2 gas mixtures. Hence, it is not surprising that numerous studies have been carried out on CuZn-0 catalysts [l and 2 and references therein]. Several investigator~~,~ have found that after reaction in CO/C02/H2 gases, the more active Cu-Zn-O catalysts have 30-60% of their copper surface area covered with oxygen. Ren and co-workers5 have shown that COz has an inhibiting effect on methanol synthesis catalysts that do not contain copper but has a promotional effect on catalysts containing copper. This, along with studies which have shown that C 0 2 produces chemisorbed oxygen on the copper component of the catalysts: suggest that chemisorbed oxygen on the copper component has a promotional effect on methanol synthesis over Cu-Zn-0 surfaces. It has been proposed by Chinchen, Spencer, Waugh, and Whan6 that the role of chemisorbed oxygen on supported copper catalysts is to promote C 0 2 adsorption on the copper component of the catalysts. In this paper, we will show that chemisorbed oxygen on the copper component also plays a role in promoting the total amount of formate formation. In addition to studies of Cu-Zn-0 catalysts, various investigations concerning methanol interaction with zinc oxide and oxygen modified Cu(110) have been reported. Wachs and Madix’ monitored methoxy and formate formation upon chemisorption of methanol on oxygen modified Cu( 110) (henceforth denoted O/Cu( 110)) surfaces by observing their decomposition into C H 2 0 and C 0 2 , respectively. They found that both the amount of [CH,0],,,h,,,/CH30H (to emphasize the relationship between the various species, we have used the notation [X],/Z, where X is the decomposed product from intermediate Y, produced by ‘Present address: Code 61 14, Department of Chemistry, Naval Research Laboratory, Washington, DC 20375-5000.
0022-3654/92/2096-4542$03.00/0
adsorption of Z) and [CO,] formatc/CH30H production increased and then decreased with increasing oxygen coverage. Investigation on the interaction of methanol with ZnO powders, oriented thin films, and single crystals*-14have shown that methanol decomposes sequentially into methoxy and formate species. Chan and Griffnl3 have observed methanol decomposition over copper overlayers on zinc oxide oriented thin films. They observed a new C02 desorption peak (from methanol decomposition) which they interpreted as evidence for dispersed copper cation sites. In this paper, we take a new approach to the modeling of Cu-Zn-0 catalysts: We use ZnO, islands on Cu(ll0) that is covered with various amounts of oxygen as a model for studying the interaction of methanol with Cu-Zn-0 surfaces. Electron microscopy studies of Cu-Zn-0 methanol synthesis catalysts15 have shown copper and zinc oxide components in separate phases; hence, the appropriativenes of using ZnO, islands on copper as our model. Our previous studies have shown that ( I ) Chinchen, G. C.; Denny, P. J.; Jennings, J. R.; Spencer, M. S.; Waugh, K. C. Appl. Catal. 1988, 36, 1. (2) Klier, K. Adu. Catal. 1982, 31, 243. (3) Denise, B.; Sneeden, R. P. A.; Cherifi, 0.Appl. Catal. 1987,30, 3 5 3 . (4) Bowker, M.; Hadden, R. A.; Houghton, H.; Hyland, J. N. K.; Waugh, K. C. J . Catal. 1988, 109, 263. (5) Ren, Z. X.; Wang, J.; Jia, L. J.; Lu, D. S. Appl. Catal. 1989,49, 83. (6) Chinchen, G. C.; Spencer, M. S.;Waugh, K. C.; Whan, D. A. J . Chem. SOC.,Faraday Trans. 1 1987, 83, 2193. (7) Wachs, I.; Madix, R. J . Catal. 1978, 53. 208. (8) Bowker, M.; Houghton, H.; Waugh, K. C. J . Catal. 1983, 84, 252. (9) Roberts, D. L.; Griffin, G. L. J . Catal. 1985, 95, 617. ( I O ) Vohs, J. M.; Barteau, M. A. Surf. Sci. 1986, 176, 91. ( 1 1 ) Akhter, S.;Cheng, W. H.; Liu, K.; Kung, H. H. J. Catal. 1984,85, 437. (12) Cheng, W. H.; Akhter, S.; Kung, H. H. J . Coral. 1983, 82, 341. (13) Chan, L.; Griffin, G . L. Sur5 Sci. 1985, 155, 400. (14) Akhter, S.; Liu, K.; Kung, H. H. J . Phys. Chem. 1985, 89, 1958. (15) Dominquez E., J . M.; Simmons, E., D. W.; Klier, K. J . Mol. Catal. 1983, 20, 369.
0 1992 American Chemical Society
Interaction of Methanol with Cu-Zn-0 Surfaces ZnO, overlayers on Cu( 110) can be characterized by a combination of C02and CO TPD (along with AES and LEED), since CO preferentially chemisorbs on copper while C02 chemisorbs on Zn0,.16 We use the same characterization techniques to follow changes in our model Cu-Zn-0 surfaces after methanol decomposition. We show that the role of chemisorbed oxygen on the copper component of our Cu-Zn-0 system is to promote the total amount of surface formate produced by methanol decomposition. The role of ZnO, islands is to increase the surface formate: methoxy ratio, from 1:9 in the absence of ZnO,, to 1:3 in the presence of ZnO,.
Experimental Section The experiments described in this paper were performed in an ultrahigh vacuum (UHV) chamber which has been described elsewhere.I6 The UHV chamber is equipped with a Varian four-grid LEED optics for LEED and AES, an ion gun, a UTI-lOOC quadrupole mass spectrometer, leak valves and a zinc source. All of the ZnO, overlayers described in this paper were prepared in the same manner so as to obtain approximately “2.0 ML” (AES calibration) of ZnO,. The procedure used was as follows: Zinc vapor was deposited onto Cu( 110) which had been predosed with 0.5 ML oxygen (henceforth denoted 0.5 ML O/ Cu(ll0)) at 150 K in an oxygen ambient of 1 X Torr of 02. After zinc deposition, the surface was further oxidized with 300 langmuir O2at 250 K. This last step determines the difference between oxygen deficient zinc oxide, which cannot adsorb C 0 2 under our conditions, and zinc oxide, which can adsorb CO,. Finally, the surface was annealed in vacuum at 710 K for 2 s to produce threedimensional ZnO, islands on essentially oxygen-free Cu( 110). (The presence of oxygen adsorbed on Cu( 110) can be detected by CO, production from C H 3 0 H decomposition.) The techniques employed for the characterization of ZnO, islands on Cu( 110) have been described elsewhere.I6 Briefly, CO TPD was used to titrate the exposed Cu( 110) surface area and CO, TPD was used to titrate the ZnO, islands. After the preparation described above, the three-dimensional ZnO, islands covered about 20-30% of the Cu( 110) surface area as shown by CO titration, adsorb (5-6) X lo1, molecules of C02, and produce a (1 X 1) LEED pattern (due to long-range order of the copper). After the ZnO, overlayer is formed, various amounts of oxygen can be deposited onto the bare copper by exposure to 0,.The exposure to O2does not change the CO, adsorption characteristics but does change the CO adsorption characteristics between oxygen on Cu(ll0) blocks CO adsorption sites. In this paper, monolayer (ML) is defined as the atoms of oxygen adsorbed on the surface divided by the atoms of free surface copper. Hence, there are 20-30% less oxygen atoms on the copper for a Zn0,/0.5 ML O/Cu(llO) surface (Cu(ll0) surface covered with ZnO, islands on 20-3076 of its surface area and 0.5 ML oxygen adsorbed on the rest of the copper) than on a 0.5 ML O/Cu(llO) surface, as ZnO, cover 20-30% of the Cu(ll0) surface. A typical experimental procedure is as follows: The copper single crystal is cleaned by cycles of sputtering with 5 X Torr of argon at 300 and 910 K and then annealed a t 910 K for 15 min. Sample cleanness is then checked by AES and surface order by LEED. Once the sample is cleaned, ZnO, and/or oxygen overlayers are produced and characterized as described above, and then TPD experiments begin. The sample is cooled to 150 K, positioned 2 mm in front of the mass spectrometer, and dosed with a known amount of gas, and then the sample temperature is ramped linearly at 30 K/s with the mass spectrometer tuned to a particular mass. After each surface is characterized by AES, LEED, CO, and C02 TPD,I6 the interaction of methanol with the model Cu-Zn-0 surface was determined. The methanol TPD experiments ended at 710 K because that is the temperature where oxidized zinc begin to desorb. AES spectra and LEED patterns were obtained after TPD experiments to correlate composition and structure of the surface to methanol decomposition. By the (16)
Fu, Sabrina S.; Somorjai, G. A. Surf.Sci. 1990, 237, 87.
The Journal of Physical Chemistry, VO~. 96, NO.11, 1992 4543 end of each day of methanol TPD studies, the surface accumulated -5% of a monolayer of carbon. This does not alter the data as shown by performing the same set of experiments over again on this 5% carbon contaminated surface. As methanol reduces zinc oxide and takes away oxygen from Cu( 110) (see results), each ZnO,/O/Cu( 110) (or O/Cu( 110)) surface changes after each methanol TPD. The same procedure was used for each methanol TPD reported in this paper: We dose the surface with 1 X Torr methanol at 150 K for 200 s (2.0 langmuirs of CH30H), wait 200 s to pump out the methanol, and then ramp the temperature of the crystal a t 30 K/s while monitoring a particular mass. It takes 300-400 s (variation in cooling rates from day to day) to cool back down to 150 K after each TPD ending a t 710 K. We have chosen to work with the subsaturation coverage produced by a 2.0 langmuir dose of methanol because the pumping speed of our ion pump was not fast enough to allow clean experiments at higher exposures of methanol. unit at a time was During TPD experiments, only one followed in order to obtain the best signal to noise ratio possible. Hence, many surfaces had to be prepared to check reproducibility. As an extra check on reproducibility, in some of the experiments, two masses which produced signals in different temperature ranges were followed. For example, after methanol adsorption, = 31 (methanol) would be followed up to 450 K, after which the center was switched to center as switched to = 44 (CO,) in order to obtain the products from 450 to 710 K. All these experiments indicated that these model Cu-Zn-0 surfaces are very reproducible. The calibration for molecules of CO and C02 was done by determination of the area under the C O desorption peak from a ~ ( 2 x 1 CO ) overlayer on Cu(l10) at 130 K. This ordered structure corresponds to 0.5 ML CO.I7 The amount of CO molecules desorbing from a ~ ( 2 x 1 CO ) overlayer was taken to be half the number of atoms on a Cu(l10) surface. To determine the amount of CO,, we took into account the different mass spectrometer sensitivity to C O and C 0 2 . The products observed were identified by comparing their observed cracking pattern in the mass spectrometer with those in the literature. Once the products were identified, the parent fragment was followed except for CH30H, in which case = 3 1 methoxy signal was followed as it is 50% greater in signal = 32. than The methanol used was 99.9% pure (Aldrich) and used from a glass vial fitted with a Teflon stopcock. The methanol was taken through several freeze-pumpthaw cycles before use. The C O was 99.5% pure (Matheson), the C 0 2 was 99.99% pure (Matheson), and the O2 was 99.99% pure (Matheson).
Results 1. Methanol Decomposition on Cu(ll0). The products from methanol TPD on Cu(ll0) are shown in Figure 1. On clean Cu(I IO), the decomposition of methanol produces simultaneous formaldehyde and hydrogen desorption a t Tp = 370 K. Wachs and Madix also obtained simultaneous desorption of formaldehyde and hydrogen at Tp = 365 K and have assigned this result to the decomposition of methoxy surface species.’ Our results agree well with those of Wachs and Madix except for the desorption of undissociated methanol. This discrepancy is due to a difference in adsorption temperature between our studies and theirs: our adsorption temperature is 150 K while their adsorption temperature is 180 K. The C H 3 0 H desorption peak centered at 190 K is not observed in the work of Wachs and Madix, but is observed when the adsorption temperature 5150 K,as shown by Bowker and MadixI8 and by us. 2. Methanol Decomposition from O/Cu(llO). In addition to methanol decomposition on clean Cu( 1lo), we investigated how oxygen on Cu( 110) influences methanol decomposition. The products from methanol TPD from 0.5 ML oxygen on Cu(l10) surfaces are shown in Figure 2. This was the highest coverage (17) Horn,K.; Hussain, M.; Pritchard, J. Surf.Sci. 1977, 63, 244. (18) Bowker, M.; Madix, R. J. Surf.Sci. 1980, 95, 190.
4544 The Journal of Physical Chemistry, Vol. 96, No. 11, 1992
Fu and Somorjai TABLE I: Products from Methanol Adsorption on 0.5 ML O/Cu(llO) and on ZnO,/O.5 ML O/Cu(llO)
II
(a) On 0.5 ML O/Cu( 110)' CH7O H7 co7 (a) On 0.5 ML O/Cu(llO)" 200 490 410 410
CHXOH
I'
183 220 215 420
ico.3
D 2
500.3
420.:
803.2
i
20 atm), there is a direct correlation between CO2/CO feed ratio and the amount of chemisorbed oxygen on the copper surface.6 Hence, although the source of surface oxygen under industrial conditions is C02, we have used O2 due to the inability of C 0 2 to dissociate on ZnO,/Cu(l 10) surfaces at low pressures.
Discussion Wachs and Mad” have shown that [C02]~-e, [CH201,~xy, and CH30H production from C H 3 0 H exposure are all affected by preadsorbed oxygen on Cu(l10). We show in this paper that for the threecomponent system of ZnO,/O/Cu( 1lo), O/Cu( 110) affect [C02 CO]fo,at,/CH30H production on the ZnO, component as well. By following the decomposition products of C02, CO, and C H 2 0 , we have shown that the three-component system of ZnO,/O/Cu( 110) has a promotional effect on the production of formate species, but the amount of methoxy species decomposed remain unperturbed. Wachs and Madix’ have suggested that formate species are formed from methoxy species on O/Cu( 110) surfaces. In addition, investigations on the interaction of methanol with ZnO powders, oriented thin films, and single crystals&14have shown that methanol decomposes sequentially into methoxy and formate species. This would suggest that the three component system of C u - Z n 4 does form greater amounts of methoxy species than the addition of the separate components of O/Cu and ZnO,, but the additional methoxy species formed react with the O/ Cu( 110) component to form either formaldehyde or formate species, which then spills over to the ZnO, component and decomposes during TPD experiments. The relative ratio of [CO2lfonnak/CH3OH to [CH20],hxy/CH30H changes from 1:9 on 0.2-0.3 ML O/Cu(llO) to 1:3 on ZnOJ0.2-3.0 ML O/Cu(1 10) surfaces, indicating that the three-component system has a higher concentration of surface formate. All this suggests that formate formation is increased on the three-component system, maximizing on a surface that contains 0.2-0.3 ML oxygen on the bare copper component of ZnO, on Cu( 1 10).
+
The Journal of Physical Chemistry, Vol. 96, No. 11, 1992 4549 Detailed studies by various investigator^^.'^ have shown formate species to be the common and most long-lived intermediate on Cu/ZnO/A1203 catalysts, ZnO catalysts, and polycrystalline copper. They argue that methanol is produced from carbon dioxide and hydrogen reacting to form formate species. These investigators4along with other^,^ have also found that the more active Cu-Zn-0 methanol synthesis catalysts have 3 6 6 0 % of their copper surface area covered with oxygen after reaction in CO, C02, and H2. Assuming each oxygen atom blocks two copper sites (this is a good approximation in the O/Cu( 110) case), these methanol synthesis catalysts are covered with 0.15-0.30 ML of oxygen. We have shown above that the maximum in total formate production ( [ C 0 2 + COlrormate/CH3OH production) on the three-component system occurs when the Cu( 110) component contains 0.2-0.3 ML oxygen. Though the maximum in formate production from the addition of the separate components is also 0.2-0.3 ML oxygen on Cu( 1lo), its maximum is only half of the maximum produced in the three-component system. We propose that one of the roles of oxygen on the copper component is to increase the amount of formate formation. This suggestion is consistent with recent findings that Cu/ZnA1204 catalysts produce 20 times the amount of formate than ZnA1204catalysts from CO, + H2 feed.2O These Cu/ZnA1204catalysts were shown to contain chemisorbed oxygen on the copper component. Hence, both the model ZnO,/O/Cu( 1 10) surfaces and the working Cu/ZnAl20, catalysts show a promotional effect for formate formation with chemisorbed oxygen on the copper component.
Conclusion We have shown that the amount of surface oxygen on the Cu( 110) part of the model catalyst determines the amount of formate produced on the ZnO, component of the surface as well as the O/Cu(110) component of the surface. The threecomponent system of ZnO, islands on Cu( 110) with 0.2-0.3 ML oxygen on the Cu(ll0) part produced twice the amount of formate than the maximum reached with the addition of the separate components. In contrast, we saw no promotional effect on the threecomponent system for the amount of surface methoxy species decomposed. The relative ratio of formate species to methoxy species increases from 1:9 in the case of 0.2-0.3 ML O/Cu(llO) to 1:3 in the case of Zn0,/0.2-0.3 ML O/Cu( 110). Acknowledgment. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Material Sciences Division, U.S.Department of Energy under contract No. DE-AC03-76SF00098. Registry No. Zn,7440-66-6; Cu, 7440-50-8; 02, 7782-44-7; CH30H, 67-56-1. (19) Bowker, M.; Houghton, H.; Waugh, K. C. J . Chem. Soc., Furuduy Tram. I 1981, 77, 3023. (20) Chauvin, C.; Saussey, J.; Lavalley, J. C.; Idriss, H.; Hindermann, J. P.; Keinnemann, A.; Chaumette, P.; Courty, P. J . Catal. 1990, 121, 56.
