Multicomponent surface analysis system combined with high pressure reaction cells for studying metal oxide model

catalysts
gl W. Weiss,a) M. Ritter, D. Zscherpel, M. Swoboda, and R. Schlo
Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany

Received 12 June 1997; accepted 12 September 1997 A three chamber ultrahigh vacuum UHV surface analysis system combined with high pressure reaction cells has been designed. It can be used for the preparation and characterization of ordered metal oxide model catalyst surfaces and for the investigation of heterogeneous catalytic reactions there on also under high pressures. The reaction cells are completely separated from the UHV analysis chambers. They can be used for oxidation treatments and reaction studies at total pressures up to 1 bar. A new sample transfer mechanism together with sapphire sample holders and sample heatingcooling stations on the analysis chamber manipulators provides electron beam and resistive sample heating, liquid nitrogen sample cooling, and precise thermocouple temperature control. First experimental results obtained on single crystalline iron oxide dehydrogenation lms grown onto Pt111 substrates are presented. These results demonstrate the capability of the system for in situ investigation of a sample surface using scanning tunneling microscopy, low-energy electron diffraction, Auger electron spectroscopy, photoelectron emission microscopy, and temperature programmed desorption before and after catalytic reactions at high pressures. This single crystal surface science approach provides new insight into the atomic scale surface chemistry of metal oxides under real catalysis conditions. 1998 American Vacuum Society. S0734-21019801701-1

I. INTRODUCTION Many important heterogeneous catalytic reactions are performed over metal oxide based catalysts, but still little is known about the reaction mechanisms on an atomic scale. To improve this situation a better understanding of the surface chemistry on metal oxides is required. The rst step towards this goal is the preparation and characterization of ordered metal oxide surfaces with dened structures and compositions. The higher structural and chemical complexity of oxides when compared to metals makes this a challenging task, which has resulted in increasing numbers of metal oxide surface science studies during the last ten years.1,2 In the second step catalytic reactions have to be performed on single crystalline model catalyst surfaces also at high pressures, that means under conditions similar to the technical catalytic reactions. This combined approach allows one to study the dependence of the catalytic activity on surface structure, surface stoichiometry, and surface defects. The adsorption kinetics of reaction educts and products determined in ultrahigh vacuum UHV adsorption studies on ordered surfaces that are catalytically active allows one to predict reaction kinetics also under high pressures. Together with pre- and postreaction UHV surface analysis insight into the atomic scale chemistry on active metal oxide surfaces can be obtained. The combination of traditional high pressure catalytic reaction studies with UHV surface analytical techniques was pioneered in the 1970s by Somorjai and co-workers3,4 and substantiated later by Goodman et al.5 These authors showed
a

Author to whom correspondence should be addressed; Electronic mail: weiss w@fhi-berlin.mpg.de J. Vac. Sci. Technol. A 161, Jan/Feb 1998

that methanation and FischerTropsch low area metal single crystal catalysts are directly comparable to high area supported catalysts with respect to their activities and selectivities. The theoretical prediction of reaction kinetics at high pressures with the help of kinetic parameters obtained from low pressure UHV single crystal studies was demonstrated by the groups of Ertl6 and Norskov7 for iron ammonia synthesis catalysts. Few attempts to apply this concept to single crystalline metal oxide catalysts have been made so far. This requires the preparation of dened and ordered metal oxide surfaces and the combination of high pressure reaction experiments with in situ UHV surface analysis including temperature programmed desorption spectroscopy. Ordered oxide surfaces can be obtained by using metal oxide single crystals or by growing thin epitaxial metal oxide lms. Such lms can be prepared either by oxidizing the surface region of the corresponding metal single crystals as done for example by Freund and co-workers,8 or by growing epitaxial oxide lms onto dissimilar substrates via metal deposition and simultaneous or subsequent oxidation.9 Depending on the thermodynamic stability ranges of metal oxides as a function of ambient conditions like oxygen partial pressure and temperature, the preparation of surfaces with dened structure and stoichiometry may require high pressure oxidation treatments. After the high pressure preparation at an elevated temperature these structures can be quenched by cooling down to room temperature and studied afterwards in UHV, as demonstrated recently for an -Fe2O3 0001 surface.10 A sample transfer system with heating and cooling capability and sample temperature control is needed to perform high pressure oxidation treatments and to combine high pres1998 American Vacuum Society 21

21

0734-2101/98/161/21/9/$10.00

22

Weiss et al.: Multicomponent surface analysis system

22

sure reaction studies with pre- and postreaction surface analysis in UHV, including temperature programmed desorption for studying the adsorption kinetics of molecules involved in the catalytic reactions. The sample holder also should be inert against oxidation or other chemical reactions. We have designed such a transfer system connecting three UHV analysis chambers and two high pressure reaction cells. They are equiped with a scanning tunneling microscope STM and low energy electron diffraction LEED optics for determination of the atomic metal oxide surface structures and surface defect structures, an Auger electron spectroscopy spectrometer AES for determination of the chemical lm compositions, a quadrupole mass spectrometer QMS for performing temperature programmed desorption TPD measurements and for analyzing product formation during the catalytic experiments. A photoelectron emission microscope PEEM is used for in situ investigations of surface reactions, for imaging surface defects and surface deposits on a mesoscopic scale before and after catalytic reaction experiments, and for imaging the desorption of reactants during TPD experiments. In Sec. II we describe the UHV analysis system and its individual components: the pumping systems, the sample transfer mechanism, the sample holder design together with the sample heatingcooling stations on the UHV manipulators, the radiation heater in the high pressure cells, and the setup of the high pressure catalysis experiments. In Sec. III rst experimental results obtained with this UHV system are presented. The technological important dehydrogenation reaction of ethylbenzene EB to styrene carried out over iron oxide based catalysts is studied on single crystalline iron oxide model catalyst lms grown onto Pt 111 substrates. Several iron oxide phases with dened stoichiometries and surface structures were prepared by repeated cycles of iron deposition and subsequent oxidation at different oxygen partial pressures. The measurements demonstrate the in situ application of all UHV surface analysis techniques mentioned above on one sample, before and after catalytic reactions under high pressures have been carried out on the same sample. A summary is given in Sec. IV. II. INSTRUMENTATION
A. Surface analytical part

FIG. 1. Perspective side view a and top view b of the UHV surface analysis system combined with high pressure reaction cells. 1 turbomolecular pump, 2 ion getter pump, 3 titanium sublimation pump, 4 64 mm gate valves, 5 high precision x , y , z off-axis manipulators.

Figure 1 displays a perspective side view a and a top view b of the whole system with all three chambers being connected. The sample transfer is performed with the magnetic transfer rods indicated, which will be described in detail in Sec. II C. The STM analysis chamber on the right side is equiped with a commercial Burleigh Instruments STM head mounted horizontally on a 150 mm ange, an Omicron backview LEED optics mounted on a 150 mm ange and a cylindrical mirror analyzer Auger spectrometer mounted on a 100 mm ange. Several anges are available for the installation of evaporation sources, ion bombardment guns, gas inlet leak valves, or a quadrupole mass spectrometer. A full rotatable Fisons off-axis manipulator with an x y motion stage allowing 25 mm movements and a z movement of
J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

200 mm is mounted on top of the chamber. The sample heatingcooling station on this manipulator is described in the next section. The high pressure reaction cell attached to this chamber consists of a 64 mm double cross and is completely separated from the analysis chamber by a 64 mm gate valve. It is described in more detail in Sec. II D. The pumping system consists of a titanium sublimation pump, a 450 l /s ion pump Varian and a 180 l /s turbomolecular pump Balzers. The ion pump and the turbomolecular pump can be separated from the analysis chamber by 150 and 100 mm gate valves, respectively. The reaction cell can be pumped via a bypass with the turbomolecular pump. The analysis chamber on the left side is equipped with a PEEM microscope Staib Instruments mounted horizontally on a 150 mm ange and two 35 mm ange windows for illuminating the sample with light sources from outside the chamber. UV discharge lamps utilizing xenon, deuterium, and mercury are used. The channelplate region in the microscope is pumped by a 50 l /s turbomolecular pump via a 35 mm ange, which allows PEEM measurements at total gas pressures up to 10 5 mbar in the analysis chamber. A cylindrical mirror analyzer Auger spectrometer is mounted horizontally onto a 100 mm ange Omicron. Several anges for mounting evaporation sources, ion bombardment guns, and gas inlet leak valves are available. The geometry and

23

Weiss et al.: Multicomponent surface analysis system

23

FIG. 2. Top view photograph of the sapphire sample holder with a sample mounted on a. Schematic top- and sideview of the sample holder without a sample b.

pumping system as well as the off-axis manipulator of this chamber are identical to those of the STM chamber on the right side. A high pressure reaction cell can also be attached directly to this chamber if needed. The central chamber has a smaller volume and is used for TPD experiments. An off-axis manipulator VAB with an x y motion stage allowing 12.5 mm movements and a z movement of 100 mm is mounted on top. The sample heatingcooling station on this manipulator is identical to those on the off-axis manipulators in the STM and PEEM chambers. Furthermore, three sample parking stations are mounted on this manipulator. A Balzers quadrupole mass spectrometer is mounted horizontally onto a 64 mm ange. For the TPD measurements the sample is placed 1 mm in front of a stainless steel tube with a diameter of 5 mm, smaller than the sample diameter of 10 mm. This tube is connected to the ionization region of the QMS which is shielded from the surrounding by a tantalum basket, so that only molecules desorbing from the sample surface are detected. This chamber is also equipped with an Omicron backview LEED optics mounted horizontally onto a 150 mm ange. Several anges for mounting evaporation sources, ion bombardment guns, and gas inlet leak valves are available. The chamber is pumped by a 450 l /s turbomolecular pump Balzers and a titanium sublimation pump. A high pressure reaction cell identical to the one of the STM chamber is attached to this chamber. It is separated from the TPD chamber by a 64 mm gate valve. In Fig. 1 all three chambers are connected with stainless steel bellows between two 64 mm gate valves, which allows one to transfer samples into all three chambers and to apply all techniques in situ on the same sample. The chambers can also be disconnected from each other if desired. The STM and PEEM chambers are mounted onto tables with pneumatic air legs to provide vibration damping.
B. Sample holder design

ments, which can be performed in a separate chamber equipped with the same sample holder and sample transfer design not presented in this article. The high elastic modulus of sapphire makes this material mechanically more stable than quartz or macor. Its high thermal stability and low thermal conductivity above room temperature allows high temperature sample treatments without heating up the sample environment very much, and its high thermal conductivity below room temperature allows fast liquid nitrogen cooling down to temperatures of 100 K. Furthermore, sapphire is chemically inert and therefore it is suitable for long time high pressure oxidation treatments and catalytic experiments in reactive atmospheres. It turned out that heating from 100 up to 1300 K with 5 K/s does not damage the sapphire because of thermal expansion if it has been tempered with a heating treatment at 2100 K before. The sample holder has an 8 mm diameter hole in the center, which provides space for an electron beam heater behind the sample as discussed in the next section. For the transfer, the sample holder can be grabbed at the two upper holes by a wobble stick with two bolts. The four holes around the central hole are for thermocouple and electrical heating feedthroughs. The metal single crystal substrates are 10 mm in diameter and 12 mm thick with two parallel slits at their sides. They are placed above the central hole of the sample holder as shown in Fig. 2, where two tungsten wires slipped through the side slits x the crystal onto the sample holder. These tungsten wires are mounted with the molybdenum screws 3 and 4, which provide electrical feedthrough contacts at the backside of the sample holder. Currents up to 30 A can be passed through these wires heating the crystal up to 1200 K or more depending on its thickness. If the tungsten wire is in good thermal contact with the platinum crystal this sample heater works well up to 1 mbar oxygen partial pressure and oxidation times up to several hours. For higher oxygen pressures and longer oxidation times the tungsten wire eventually gets brittle and breaks. In this case we use the radiation heater described in the next section. A chromel-alumel thermocouple spotwelded to the side of the crystal measures the sample temperature and is connected to the cromel and alumel screws 1 and 2, which also provide electrical feedthrough contacts at the backside of the sample holder.
C. Sample transfer and sample heatingcooling stations

The sample holder is made from single crystalline sapphire which has been tempered at temperatures of 2100 K after its fabrication. The geometry is adapted to the Burleigh Instruments STM head, a thickness of 5 mm, and the basal trapezoidal shape shown in Fig. 2. Sapphire provides electrical insulation needed for STM measurements and for applying bias voltages to the sample during photoemission experiJVST A - Vacuum, Surfaces, and Films

In Fig. 3a a front-view photograph of the sample heatingcooling station with the sample holder and the wobble stick is shown. In Fig. 3b the off-axis manipulator with the sample heatingcooling station, the transfer rod and the wobble stick are displayed in a schematic side view at the transfer position. The sample holder is placed in a vertical orientation with the short side down into the magnetic transfer rod, where it is held by spring clamps. In the transfer position the magnetic rod is connected to the off-axis manipulator by the centering bore indicated in Fig. 3b, which provides mechanical stability during the sample transfer. The

24

Weiss et al.: Multicomponent surface analysis system

24

FIG. 4. 5 K/s temperature ramp and subsequent cooling period obtained with the electron beam heater in the liquid nitrogen cooled sample station.

FIG. 3. Front view photograph of the sample heatingcooling station on the off-axis manipulators in the three UHV analysis chambers a. In b a schematic sideview of the magnetic transfer rod, the wobble stick, and the heatingcooling station in the transfer position is shown.

wobble stick approaching from above grabs the sample holder with two bolts into the upper two holes and pushes it down into the heatingcooling station. Afterwards the wobble stick and the transfer rod are removed. In the same way the vice versa transfer from the off-axis manipulator up to the transfer rod is performed. Due to the mechanical connection between the off-axis manipulator and the magnetic rod this transfer is very safe and a sample loss is not possible. The transfer into the STM head is performed in two steps. First the sample is transferred from the off-axis manipulator into a transfer station at the end of a transfer arm connected to the rotatable manipulator rod, from there it can be transferred with a wobble stick directly into the STM station. As can be seen in Fig. 3, the heatingcooling station on the off-axis manipulator consists of a liquid nitrogen reservoir made of copper. For sample cooling liquid nitrogen is passed continuously through this reservoir, which is led in and out of the chamber through spiral formed stainless steel tubes. The sample holder backside is pressed with two stainless steel plates against the liquid nitrogen reservoir, which creates a good thermal contact allowing fast cooling down to 100 K. In the center of the heatingcooling station the electron beam heater is placed see Fig. 3a. It consists of a ceramic tube with a lament inside. This lament leads into
J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

the base of a commercial 20 W halogen lamp from which the glass head has been removed. The sample holder has a slit at the backside below the central hole see Fig. 2b, and therefore it can be moved over the ceramic tube of the electron beam heater during the transfer. The lament is located about 1 mm behind the sample when the sample holder is placed into its nal position. This allows electron beam heating above temperatures of 1500 K using relatively low voltages and emission currents. The electric thermocouple contacts to the chromel and alumel screws at the backside of the sample holder are made by chromel and alumel bolts connected to cromel-alumel wires leading outside the chamber. These bolts are isolated by ceramic tubes from the liquid nitrogen reservoir as can be seen in Fig. 3a, and they are pressed by springs against the sample holder backside. The electric contact to the molybdenum screws at the sample holder backside is made in the same way with sprung molybdenum bolts. These heatingcooling stations in combination with the sapphire sample holders provide fast cooling and heating from temperatures of 100 up to 1500 K. The samples can be heated by electron bombardment from the back or by the resistive tungsten wire heater on the sample holder. With the electron beam heater heating rates of 15 K/s can be reached. The molybdenum and chromel-alumel sprung bolts provide electric contacts for resistive heating and thermocouple temperature control after the sample has been transferred. Figure 4 shows a 5 K/s heating ramp generated with the electron beam heater and the subsequent cooling down period to T 100 K again.
D. High pressure reaction cells

The high pressure cells consist of 64 mm double crosses which are separated from the TPD and STM chambers by 64 mm gate valves. They are pumped via bypasses with the turbomolecular pumps of the analysis chambers. The gas pressures are measured with Baratron gauges. The sample heating station displayed in Fig. 5 is positioned on a platform mounted onto the bottom 64 mm ange of the double cross

25

Weiss et al.: Multicomponent surface analysis system

25

III. EXPERIMENTAL RESULTS AND DISCUSSION

A. Substrate and iron oxide lm preparation

The lm preparation is described in detail in Ref. 10. A clean Pt 111 surface is obtained by repeated cycles of argon ion bombardment and annealing to 1400 K. Iron oxide lms are prepared by repeated cycles of iron deposition and subsequent oxidation at temperatures between 900 and 1000 K in different oxygen partial pressures. One to two monolayers ML thick FeO lms and Fe3O4 multilayer lms are obtained after oxidation in 10 6 mbar oxygen partial pressure, -Fe2O3 multilayer lms are obtained after oxidizing a Fe3O4 multilayer lm in 1 mbar oxygen partial pressure. This is done in the high pressure reaction cell descibed in Sec. II D.
B. Growth and structure of iron oxide lms studied by STM and LEED
FIG. 5. Sideview of the sample heating station in the high pressure cell. A stainless steel tub is placed over the sample for the batch reactor experiment, a cappilar leads to the QMS for product detection.

which also contains all electric and liquid feedthroughs. With a wobble stick on the top ange of the double cross the sample is transfered from the magnetic transfer rod into this station in the same way as done for the off-axis manipulators described in the previous section. The sample heating station in Fig. 5 is similar to the heating stations on the off-axis manipulators, but without an electron beam heater and without a liquid nitrogen reservoir. Again sprung molybdenum and chromel-alumel bolts press against the corresponding screws at the backside of the sample holder and make electric contacts for thermocouple temperature control and resistive sample heating with the two tungsten wires. This resistive heater is used for short oxidation treatments up to oxygen pressures of 10 mbar. For longer oxidation treatments in oxygen pressures up to 1000 mbar a 250 W halogen lamp can be positioned in front of the sample and used as a radiation heater not shown here. It is mounted onto a water cooled nickel platform and it is covered at the backside by a ceramic shield. Sample temperatures up to 1000 K can be reached with this simple radiation heater. In the reaction cell attached to the TPD chamber high pressure catalysis experiments are performed. The educt gases are leaked into the cell and with the wobble stick a stainless steel tub is put over the sample thus forming a batch reactor with approximately 1 ml volume. For product detection during the batch reactor experiments, reactor gas is leaked via a stainless steel tube directly into the ionization region of the QMS, which is located in the neighboring TPD chamber. At total reactor pressures between 0.1 and 10 mbar this allows online product and educt detection during the batch reactor experiments.
JVST A - Vacuum, Surfaces, and Films

Figure 6a displays a constant current 50 50 atomic resolution STM image of an iron oxide lm grown onto Pt 111 about 1 ML thick. It exhibits a hexagonal surface structure with a short periodicity of about 3.1 and a superimposed large hexagonal periodicity of about 26 . The corresponding LEED pattern is shown in Fig. 6d. The main spots correspond to the 3.1 surface periodicity, and each main spot is surrounded by six satellite spots which are generated by kinematic double scattering between the platinum substrate and the oxide overlayer.11 This lm consists of a hexagonal close-packed iron-oxygen bilayer with oxygen on top, as has been shown by photoelectron diffraction12 and a STM image simulation applying electron-scattering quantum-chemistry theory.13 Because of the lattice mismatch between the oxide layer 3.1 periodicity and the platinum substrate 2.77 periodicity about eight FeO 111 lattice spacings t onto nine Pt 111 lattice spacings. A model was proposed where the iron atom rows are rotated by 0.6 with respect to the platinum atom rows on the underlying Pt 111 surface, which leads to identical iron sites on the platinum surface after going nine platinum lattice spacings along the rst main direction from left to right in Fig. 6c and one platinum lattice spacing along the second main direction on Pt 111 120 relative to the rst main direction.14,11 This rotational mismatch creates the 26.4 coincidence superstructure cell indicated in Fig. 6c, which is rotated by 5.8 with respect to the FeO 111-(1 1) cell. We observe a superstructure rotated by about 10, which can be moiree seen in the STM image in Fig. 6a where the brightest atomic features about 26 apart do not line up with the short periodicity atom rows. This larger rotation angle is due to a slightly larger lattice spacing in the submonolayer FeO 111 lm as discussed in more detail in Ref. 11. The Auger spectrum of the FeO monolayer lm is shown in Fig. 7. Considering the O KLL and Fe MNN Auger transition cross sections the oxygen-iron intensity ratio corresponds to a FeO stoichiometry. In Fig. 6b a constant current 2000 2000 STM scan of an iron oxide lm about 1.5 ML thick is shown. The

26

Weiss et al.: Multicomponent surface analysis system

26

FIG. 6. a 50 50 STM constant current image of a FeO 111 monolayer grown onto Pt 111, U B 0.9 V, I t 0.3 nA. b 2000 2000 constant current STM image of 1.5 ML FeO 111 on Pt 111, U B 1.0 V, I t 0.2 nA. c Schematic model for the FeO 111 monolayer lm. The coincidence superstructure cell is indicated. Large white circles are Pt atoms, small grey circles Fe2 ions, and black large circles O2 ions. d LEED pattern of the FeO 111 monolayer lm at E 60 eV.

morphology of the second layer exhibits hexagonally shaped holes exposing the rst iron oxide layer underneath, which forms the same atomic surface struture as shown in Fig. 6a.14 The step height from the rst to the second layer is 2.5 which corresponds to the distance between subsequent iron-oxygen 111 bilayers in the FeO bulk structure. Small islands or clusters about 10 in size are distributed randomly on top of the second layer. On the left-hand side of this image a feature can be seen that looks like a third layer island. But the surface of this island also exhibits the triangular shaped superstructure. It is located on an almost circular region about 250 in diameter which exhibits the honeycomb surface superstructure and which is located 2.2 above the other regions of the rst FeO monolayer darkest areas in Fig. 6b. The same step height is measured between the highest island mentioned above and the second layer regions. Since this value corresponds to the step height on Pt 111 2.26 , we interpret this as the rst and second iron oxide layer on top of a platinum island about 250 in diameter and one layer high. With further increasing lm thickness, the satellite LEED spots due to the rst iron oxide coincidence layer disappear
J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

and new LEED spots evolve. Figure 8a shows the LEED pattern of an iron oxide multilayer lm prepared by oxidation in 10 6 mbar oxygen partial pressure at T 1000 K. A dynamic LEED intensity analysis performed previously on this lm revealed a strongly relaxed iron terminated Fe3O4 111 surface structure.15 When this Fe3O4 multilayer lm is oxidized in 1 mbar oxygen partial pressure at T 1000 K the LEED pattern in Fig. 8b is obtained, which corresponds to an unreconstructed -Fe2O3 0001 surface. Thermodynamic calculations revealed this oxide phase to be stable only in oxygen partial pressures above 1 mbar at the lm preparation temperature of 1000 K,10 which is necessary for obtaining an ordered lm. A dynamic LEED intensity analysis of the -Fe2O3 0001 surface structure is underway.14
C. Adsorption of ethylbenzene studied by TPD

Different amounts of EB were adsorbed at T 120 K onto the FeO monolayer lm shown in Fig. 6. Figure 9 shows EB-TPD spectra recorded with a heating rate of 5 K/s. For low initial coverages we observe rst order kinetics molecular desorption at T 210 K. Above 1.5 L exposures a sec-

27

Weiss et al.: Multicomponent surface analysis system

27

FIG. 7. AES spectrum of a FeO 111 monolayer lm on Pt 111.

FIG. 9. TPD spectra of ethylbenzene adsorbed onto a FeO 111 monolayer lm.

ond desorption peak at T 155 K evolves which continously increases in intensity with increasing EB exposure. It exhibits a zero order desorption kinetics and is due to a thick layer of condensed EB that forms above the rst layer of weakly chemisorbed EB. A threshold analysis which analyzes the low temperature leading edge of the TPD spectra16 reveals desorption energies of 58 and 50 kJ/mol for the rst layer of chemisorbed EB and for the condensed EB multilayer, respectively. For EB exposures above 10 L we also observe small amounts of EB desorbing above room temperature not shown here, which presumably is due to EB chemisorbed at surface defect sites. On Fe3O4 and -Fe2O3 multilayer lms we observe different EB-TPD spectra which will be published elsewhere.
D. Ethylbenzene dehydrogenation studied in reactor and PEEM experiments

formed by a FeO monolayer and a -Fe2O3 multilayer lm as a function of time, measured with the QMS connected via a capillar to the batch reactor as described in Sec. II IV. No signicant styrene production is observed on the FeO lm,

The catalytic dehydrogenation of EB to styrene was studied in batch reactor experiments. An EBwater mixture with a molar ratio of 1:10 and a total pressure of 0.6 mbar was dosed into the high pressure cell, and then the sample was heated to the technical catalyst reaction temperature of 870 K. Figure 10 shows the educt and product concentrations

FIG. 8. LEED pattern of a Fe3O4 111 a and a -Fe2O3 0001 b multilayer lm on Pt 111 at E 60 eV. JVST A - Vacuum, Surfaces, and Films

FIG. 10. Ethylbenzene, styrene, and hydrogen QMS signals measured during the batch reactor experiment.

28

Weiss et al.: Multicomponent surface analysis system

28

PEEM image than the clean oxide lm and which presumably deactivate the lm in Fig. 10. These hydrocarbons also can be removed from the oxide surface by heating it up to 1000 K in the presence of oxygen, producing water, CO, and CO2.

IV. SUMMARY A three chamber UHV surface analysis system equipped with STM, LEED, PEEM, AES, and TPD has been designed. It is combined with separate reaction cells where metal oxides can be prepared in oxygen pressures up to 1 bar, and where batch reactor catalysis experiments under total pressures up to 1 bar can be performed. These reaction cells are completely separated from the UHV analysis chambers, which eliminates signicant contamination of the UHV analysis chambers during and after high pressure experiments. A novel sample transfer system in combination with sapphire sample holders and sample heatingcooling stations on the manipulators of the analysis chambers provides resistive and electron beam heating, liquid nitrogen cooling, and precise thermocouple temperature control. This allows one to apply all surface analysis techniques listed above onto the same sample before and after high pressure catalysis experiments have been carried out on this sample. In particular, elementary reaction steps and the adsorptiondesorption kinetics of the molecules involved can be studied by TPD experiments on a wide variety of structurally and chemically well dened surfaces. Presently we study the dehydrogenation of ethylbenzene to styrene on single crystalline iron oxide model catalyst lms. Well ordered FeO 111 lms 12 ML thick and Fe3O4 111 multilayer lms are grown onto Pt 111 substrates by repeated cycles of iron deposition and subsequent oxidation at 10 6 mbar oxygen partial pressure, -Fe2O3 0001 multilayer lms are obtained by oxidation in 1 mbar oxygen. All lms exhibit LEED patterns that are characteristic for their corresponding crystal bulk structures, which allows one to identify these oxide phases by inspection of their LEED patterns. For small exposures, weakly chemisorbed EB molecules desorbing at T 220 K are observed on the FeO 111 monolayer lm, at higher exposures a second desorption peak at T 155 K evolves which is due to desorption of condensed EB multilayers. The EB dehydrogenation was studied in batch reactor experiments in an EBwater mixture with a total pressure of 0.6 mbar, and the model catalyst surfaces were imaged with PEEM before and after the catalysis experiments. No styrene production is observed on the FeO monolayer lm, and no carbonaceous deposits were observed on these lms after the reaction. The lm is partly removed after the reaction and platinum surface areas covered with carbonaceous deposits were observed. On -Fe2O3 0001 multilayer lms a styrene production and the forma-

FIG. 11. PEEM images of clean FeO a and -Fe2O3 c lms before the reaction experiments 200 m diameter. PEEM images after the reaction experiments: Pt 111 with carbonaceous deposits b after partial removal of the FeO monolayer lm, Fe2O3 lm with carbonaceous surface deposits d 40 m diameter.

whereas on the -Fe2O3 lm a signicant production of styrene and hydrogen accompanied by a simultaneous decrease of the EB concentration is detected after the sample has reached the reaction temperature. This indicates that Fe2O3 is the active phase for the dehydrogenation reaction. After approximately 50 min the -Fe2O3 lm is deactivated and no styrene is produced anymore. With the PEEM microscope we studied the iron oxide surfaces on a mesoscopic scale before and after the reactor experiments. The clean FeO and -Fe2O3 lms on the left side in Figs. 11a and 11c exhibit no visible structures in their PEEM images, beside a bright spot due to a dust particle or some other mesoscopic defect. After the reactor experiment, we observe a partial removal of the FeO monolayer lm, and the PEEM image in Fig. 11b is observed. It corresponds to a Pt 111 surface covered with carbonaceous deposits, which is deduced from the carbon signal in the AES spectra. The features in Fig. 11b presumably are due to steps or step bunches on the Pt 111 surface, as the decoration of steps on Pt 100 surfaces by carbon has been observed earlier in similar PEEM images.17 These carbonaceous deposits were removed by heating the sample up to 1000 K in the presence of oxygen, which resulted in the formation of water, CO, and CO2. This shows that hydrocarbons have formed on the surface that can be oxidized to above product molecules. On some regions on the sample we still observe the FeO lm exhibiting the same PEEM image as before the reactor experiment in Fig. 11a, that means no carbonaceous deposits have formed on the remaining FeO lm. Figure 11d shows the PEEM image of the -Fe2O3 lm after the catalysis experiment. It is completely covered with carbonaceous surface deposits which produce a brighter
J. Vac. Sci. Technol. A, Vol. 16, No. 1, Jan/Feb 1998

29

Weiss et al.: Multicomponent surface analysis system

8

29 H.-J. Freund, H. Kuhlenbeck, and V. Staemmler, Rep. Prog. Phys. 59, 283 1996. 9 W. Weiss, A. B. Boffa, J. C. Dunphy, H. C. Galloway, M. B. Salmeron, and G. A. Somorjai, Adsorption on Ordered Surfaces of Ionic Solids and Thin Films, Vol. 33, Springer Series in Surface Science, edited by H.-J. Freund and E. Umbach Springer, Berlin, 1993. 10 W. Weiss, Surf. Sci. 943, 377379 1997. 11 M. Ritter, W. Ranke, and W. Weiss, Phys. Rev. B in press. 12 C. S. Fadley, M. A. Van Hove, Z. Hussain, and A. P. Kaduwela, J. Electron Spectrosc. Relat. Phenom. 75, 273 1995. 13 H. C. Galloway, P. Sautet, and M. Salmeron, Phys. Rev. B 54, R11 145 1996. 14 H. C. Galloway, J. J. Benitez, and M. Salmeron, Surf. Sci. 298, 127 1993. 15 A. Barbieri, W. Weiss, M. A. Van Hove, and G. A. Somorjai, Surf. Sci. 302, 259 1994. 16 gl, Surf. Sci. 382, 326 1997. D. Zscherpel, W. Weiss, and R. Schlo 17 M. Mundschau, M. E. Kordesch, B. Rausenberger, W. Engel, A. M. Bradshaw, and E. Zeitler, Surf. Sci. 227, 246 1990.

tion of carbonaceous surface deposits was observed, indicating this oxide phase to be catalytically active. ACKNOWLEDGMENT We thank the company BESTEC GmbH for fruitful collaboration.
H.-J. Freund, Angew. Chem. 109, 444 1997. V. E. Henrich and P. A. Cox, The Surface Science of Metal Oxides Cambridge University Press, Cambridge, 1994. 3 B. A. Sexton and G. A. Somorjai, J. Catal. 47, 167 1977. 4 D. J. Dwyer and G. A. Somorjai, J. Catal. 52, 291 1978. 5 D. W. Goodman, R. D. Kelly, T. E. Madey, and J. T. Yates, J. Catal. 63, 226 1980. 6 G. Ertl, Catalytic Ammonia Synthesis, edited by J. R. Jennings Plenum, New York, 1983. 7 J. K. Norskov and P. Stoltze, Phys. Rev. Lett. 55, 2502 1985.
1 2

JVST A - Vacuum, Surfaces, and Films