Al2O3 catalyst by XPS and TPD

Applied Surface Science 189 (2002) 59±71

Hydrogen production by methanol reforming: post-reaction characterisation of a Cu/ZnO/Al2O3 catalyst by XPS and TPD F. Raimondi, K. Geissler, J. Wambach*, A. Wokaun General Energy Research, Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland Received 6 September 2001; accepted 2 December 2001

Abstract Post-reaction X-ray photoelectron spectroscopy (XPS) and temperature-programmed desorption (TPD) experiments have been carried out after exposure of a commercial Cu/ZnO/Al2O3 catalyst to methanol reforming conditions in the presence of O2 and H2O. A dependence of the copper oxidation state on the temperature was observed, with low temperatures favouring a higher oxidation state. In the presence of water, Cu(II) was formed at T  485 K, giving rise to a very low catalytic activity. Under methanol dry partial oxidation conditions at T  530 K a decrease of the Cu surface concentration was observed. Adsorbed methoxy species were found on the surface after all catalytic treatments, with the amount decreased by the presence of water in the feed. The presence of O2 favoured the formation of surface formate. # 2002 Elsevier Science B.V. All rights reserved. PACS: 82.65.J; 79.60; 82.65.M; 82.65.J Keywords: Methanol reforming; Partial oxidation; Cu/ZnO catalysts; X-ray photoelectron spectroscopy; Temperature-programmed desorption

1. Introduction Hydrogen is expected to play a major role in the future as a carbon free energy carrier. Although it can be burnt with very low emissions in traditional heat engines, its use in fuel cells is especially advantageous, due to their high ef®ciency and the complete absence of toxic emissions [1]. However, ef®cient and safe storage of hydrogen with a reasonable energy density, especially for non-stationary applications, is still an unsolved issue. One solution currently under discussion is the use of liquid organic hydrogen carriers, such as methanol. Methanol can be produced via synthesis gas from basically any organic carbon *

Corresponding author. Tel.: ‡41-56-310-2111; fax: ‡41-56-310-2199. E-mail address: [email protected] (J. Wambach).

source, including biomass. Another synthesis route is the reaction of renewable hydrogen with carbon dioxide from the air or with recycled carbon dioxide from power plants using fossil fuels [2±4]. In this way carbon dioxide is recovered from the environment and used to store hydrogen in the form of methanol. When the stored hydrogen is liberated by methanol reforming, only the carbon dioxide previously removed from the atmosphere is emitted, giving rise to no additional carbon dioxide production. Methanol as a liquid energy carrier offers a high energy density and can be reformed into hydrogen directly on a fuel cell vehicle [5]. In principle, various reactions can be used for the production of hydrogen from methanol. Probably the most widely applied one is the steam reforming reaction (1) [6]. However, due to its endothermicity it has drawbacks if used for hydrogen production in non-stationary reactors

0169-4332/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 ( 0 1 ) 0 1 0 4 5 - 5

60

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

working under strongly varying loads and if short start-up times are required. CH3 OH ‡ H2 O ! CO2 ‡ 3H2 ;  DR H298 K ˆ 49 kJ mol

1

(1)

Dry partial oxidation of methanol (2) is strongly exothermic and therefore the reaction is very dif®cult to control for scales bigger than microreactor size. Furthermore, the heat loss due to the necessary cooling signi®cantly lowers the overall ef®ciency of a partial oxidation reformer. CH3 OH ‡ 12 O2 ! CO2 ‡ 2H2 ;  DR H298 K ˆ

192 kJ mol

1

(2)

For non-stationary applications autothermal reforming (3) is considered to be a promising option. Since it has a net reaction enthalpy of zero, a reactor does not need to be heated after it has reached its reaction temperature. Moreover, in order to shorten start-up times or improve transient behaviour, the stoichiometry (amount of oxygen in the feed) can be varied. 4CH3 OH ‡ 3H2 O ‡ 12 O2 ! 4CO2 ‡ 11H2 ;  DR H570 K  0 kJ mol

1

(3)

All three reactions (1)±(3) can be carried out over Cu/ ZnO type catalysts as well as over noble metal catalysts. Although recently very high selectivity towards carbon dioxide was claimed for Rh/Zn-alloys, noble metal catalysts favour the decomposition of methanol rather than the steam reforming, leading to high CO concentrations in the mixture of products compared to Cu-based catalysts [7]. Since CO is a strong poison for the electrocatalyst in the fuel cell, the fuel cell feed needs to be virtually CO-free [8]. Therefore, Cu-based catalysts that ensure lower CO content in the reformate, provide the opportunity of simplifying the necessary gas-cleanup stage. The interaction of methanol with various Cu- and Zn-containing surfaces has been intensively studied during the last 35 years, mainly in connection with methanol synthesis from synthesis gas and its selective oxidation to formaldehyde. On clean Cu(1 0 0), Cu(1 1 0) and Cu(1 1 1) single-crystal surfaces, methanol adsorbs without signi®cant dissociation due to the insuf®cient proton-af®nity of the exposed Cu atoms [9±11]. However, dissociation of the O±H

bond and formation of adsorbed methoxy groups was reported for the more basic Cu(2 1 0) surface [12]. More commonly, dissociative adsorption of methanol is achieved on O2-predosed Cu surfaces, with the formation of adsorbed methoxy and hydroxy groups [13]. The investigation of Cu-clusters supported on oxygen-terminated ZnO…0 0 0 1† has revealed that the interaction of methanol with metallic Cu is dependent on the cluster size. Even after O2-predosage, atom-thin Cu islands are incapable of dissociating methanol, because the basicity of the adsorbed O-atoms is insuf®cient [14]. The interaction of methanol with ZnO single-crystals depends on the nature of the exposed surface. The O-terminated ZnO…0 0 0 1† surface is non-reactive due to the absence of accessible acid±base pairs necessary for dissociative adsorption [14,15], whereas on Zn-terminated ZnO(0 0 0 1) and on non-polar ZnO…1 0 1 0† methoxy groups are formed upon methanol adsorption [15,16]. The dissociation of the O±H bond in the methanol molecule is a fundamental prerequisite for its activation. Adsorbed methoxy groups on Cu surfaces decompose at 330±400 K to form formaldehyde and adsorbed H atoms [13,17]. If O adatoms are available, competitive formation of formate groups is observed [18±20]. Also with respect to the reactions of adsorbed methoxy groups, the different ZnO surfaces behave differently. Temperature-programmed desorption (TPD) showed that methoxy groups adsorbed on the Zn-terminated ZnO(0 0 0 1) decompose mainly forming formate and only a small amount of formaldehyde [15]. Under the same conditions no surface species other than methoxy groups could be detected on ZnO…1 0 1 0† [16]. Investigation of methanol adsorption on model and industrial Cu/ZnO-based catalysts by in situ vibrational spectroscopy con®rmed the relevance of the surface species found in surface science studies [21±23]. In spite of the vast amount of data published on methanol synthesis (for a review see reference [24]) and methanol steam-reforming [25,26], the in¯uence of oxygen on the catalyst surface and on the reaction mechanism is not yet well understood [27]. Alejo et al. [28] showed that the activity of Cu/ZnO-based catalysts is very sensitive towards changes in the redox properties of the reactant mixture under methanol dry partial oxidation conditions. Similarly, Reitz et al. [29] showed that, under small oxygen conversions, the

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

copper in the catalyst was in the oxidised state and that the main reaction pathway was towards carbon dioxide and water production. SchloÈgl and co-workers [30±32] investigated the chemical and structural changes of unsupported copper under methanol dry partial oxidation conditions using in situ X-ray absorption spectroscopy and also found that the overall surface oxidation state was a function of the O2/ CH3OH ratio in the feed. The active form of copper was shown to be small nano-crystals in which the oxygen atoms were intercalated within the large internal surface of the defect structure. From the examined literature, it is clear that the catalyst oxidation state plays a major role in the mechanism of methanol reforming. The present study represents a systematic investigation of the in¯uence of the reaction conditions (temperature and composition of the feed) on the state of the catalyst by postreaction X-ray photoelectron spectroscopy (XPS) and TPD. Since production of H2 by methanol reforming in the presence of both O2 and H2O (methanol wet partial oxidation) appears to be a most promising technical process, special attention has been paid to the in¯uence of water on the catalyst surface. 2. Experimental 2.1. Experimental set-up All experiments were carried out in a UHV system equipped for XPS, TPD and high-pressure sample treatment. XPS was performed in a VG Escalab 220i XL apparatus, equipped with magnetic lenses, using monochromatic Al Ka radiation focused in a 0:3  0:8 mm2 spot. The electron analyser was used in the constant pass energy mode with a pass energy of 20 eV. This gave an FWHM of 0.6 eV for the Ag 3d5/2 peak of clean Ag. The pressure in the analysis chamber during the measurements was always better than 10 9 mbar. TPD spectra were collected using a Balzers QMS 200 Prisma quadrupole mass spectrometer (QMS) capable of simultaneously detecting up to 16 masses. The QMS head was inserted in a differentially pumped nozzle terminating in a hole having a diameter of 1 mm. The nozzle was brought to 0.5 mm from the sample surface prior to the measurement. During the TPD run, the temperature of the sample

61

was raised linearly at a rate of 1 K/s from RT up to 770 K. The CuO/ZnO/Al2O3 catalyst was provided by Johnson Matthey and was crushed and sieved to particle sizes of between 90 and 120 mm prior to use. The BET speci®c surface area was 50 m2/g. Raman microscopy showed that the catalyst contained crystalline graphite whose contribution was also clearly visible in the XPS C 1s spectra. In a typical run 0.070 g of the powder were pressed on a porous glass disk using a weight designed to provide a reproducible pressure of 6 atm. The glass disk was ®xed in a round-shaped stainless steel sample holder equipped with knife edges on the upper and lower side. Heating was accomplished by a heater in thermal contact with the glass disk and the temperature was measured by a thermocouple inserted into the glass disk. Reversible tightening of the high-pressure reactor (HPR) was achieved by knife edge-Cu gasket sealing between the sample holder and two vertically movable stainless steel funnels. The gas stream in the HPR ¯owed perpendicularly through the catalyst bed, entering from the side that would ultimately be subjected to XPS and TPD analysis. The intensity of the TPD spectra measured after exposing the sample holder without catalyst to various reaction conditions was <1% of the intensity measured in the corresponding experiments with the catalyst. This demonstrates that the measured TPD signal originates from molecules desorbing from the catalyst and not from the surrounding material. The gaseous feed for the HPR was provided by a gas distribution system equipped with mass-¯ow controllers. Methanol was supplied in two different ways according to the weight-hourly space velocity (WHSV) needed. In the methanol decomposition and dry partial oxidation experiments, the methanol WHSV was 2 h 1. In this case, an Ar stream (purity 99.999%, fAr ˆ 10 mlN/min) was saturated with methanol vapour by ¯owing it through thermostated (303 K) liquid methanol (Merk, analytical grade, 300 ppm H2O), before being mixed with the appropriate O2 ¯ow (purity 99.995%). In the wet partial oxidation experiments, the methanol WHSV was chosen to be 8 h 1 in order to use the same O2 ¯ows as in the experiments without water. The results of the wet partial oxidation experiments will be presented in terms of the O2/CH3OHexc ratio, where CH3OHexc

62

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

is de®ned by CH3 OHexc ˆ nCH3 OH nH2 O. In this case a water/methanol solution (3:4 molar ratio) was fed by a precision double-piston pump (fliq ˆ 0:89 ml/h). Evaporation of the liquid took place smoothly due to the progressive temperature increase of the tubing in which the solution ¯owed. Mixing with an Ar stream (fAr ˆ 10 mlN/min) and with the appropriate O2 ¯ow took place before the HPR inlet. The composition of both the feed and the product mixture was routinely analysed by leaking a portion of the gas through a heated capillary into a separate UHV chamber equipped with a QMS (VG MONITORR 100D). 2.2. Experimental procedure After introduction of the catalyst in the UHV system, the following mild reduction procedure was applied. The sample was heated in vacuum at a rate of 10 K/min up to 510 K, before tightening the HPR and ®lling it with Ar (fAr ˆ 10 mlN/min) up to a pressure of 1.5 bar. Then a steady ¯ow was initiated and the H2 content in the ¯owing gas mixture was increased at a rate of 10% every 5 min until pure H2 (purity 99.9999%) ¯owed through the HPR (®nal fH2 ˆ 10 mlN/min). The treatment under pure H2 lasted for 40 min. After this time the HPR was evacuated, opened towards the UHV system and the sample was transferred to the analysis chamber. After XPS analysis the sample was transferred back to the HPR, where it was heated up under vacuum to the chosen reaction temperature at a rate of 10 K/min. Then the HPR was tightened again and ®lled with Ar up to a pressure of 1.5 bar. The feed was then switched to the chosen reaction mixture and reaction conditions were maintained for 16 h. After this time the feed was switched to pure Ar and the sample was allowed to cool. When a temperature of 373 K was reached, the Ar ¯ow was stopped, the HPR was evacuated and opened towards the UHV system. Post-reaction XPS analysis, a TPD experiment and post-desorption XPS analysis were performed. Each treatment was performed on a fresh catalyst sample. The quanti®cation of the XPS spectra was carried out using the cross-sections calculated by Sco®eld [33]. No charging took place when Cu(0) or Cu(I) were the predominant copper species; when a considerable amount of Cu(II) was present, charging

could be observed (<2 eV). In this case charge correction was carried out using the C 1s peak of graphite (284.4 eV). After charge correction, the Zn 2p3/2 peak was always centred at 1022.4 eV, indicating that both the graphite C 1s component and the Zn 2p3/2 peak might be used as reference for this system. 3. Results 3.1. Cu oxidation state and distribution The oxidation state of Cu in Cu/ZnO-based materials can be easily determined by considering the position of both the Cu 2p3/2 and Cu LMM XPS peaks. Cu(0) and Cu(I) present very similar Cu 2p3/2 peak positions (BE close to 932.8 eV), however they have distinct Cu LMM signals (KE 918.7 and 916.6 eV, respectively). Cu(II) shows a broader Cu 2p3/2 peak centred at about 933.7 eV and a characteristic broad shake-up signal between 940 and 945 eV. Its Cu LMM peak is positioned at 917.7 eV [34]. The predominant oxidation state of Cu in the catalyst and the elemental distribution within the region probed by XPS after various treatments are summarised in Table 1. 3.1.1. Reduction The XPS spectral features of the catalyst after reduction were consistent with the presence of metallic copper, ZnO and Al2O3. No evidence of oxidised copper was present either in the Cu 2p or in the Cu LMM region, indicating a complete reduction of the oxidised copper phases after the application of the reductive treatment [35,36]. The Cu 2p3/2/Cu 3s intensity ratio measured after catalyst reduction was consistent with the value obtained for a clean copper sample, indicating that no surface segregation of Cu or ZnO occurred in the catalyst under these conditions. 3.1.2. Methanol decomposition After methanol decomposition in the temperature range of 470±590 K, no signi®cant changes were observed in the Cu and Zn spectral features, indicating that copper remained in the metallic state and that signi®cant reduction of ZnO to metallic Zn did not take place. The Cu 2p3/2/Zn 2p3/2 (! in Fig. 1) and Cu 3s/Al 2s (not shown) atomic ratios, as well as the Cu

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

63

Table 1 Predominating Cu oxidation state and observed distribution changes with respect to the freshly reduced catalyst after exposure of the catalyst to various reaction conditionsa Reaction

T (K)

O2/CH3OH

Predominating Cu oxidation state

Elemental distribution changes

Methanol decomposition

470±590

±

Cu(0)

±

Methanol dry partial oxidation

510 510 470±510 530±550

0±0.25 0.42±0.80 0.42 0.42

Cu(0) Cu(I) Cu(I) Cu(0)

± ± ± Decrease of Cu surface concentration

Methanol wet partial oxidation

510 510 510 470±485 510 530±550

0±0.25 0.42 0.50 0.42 0.42 0.42

Cu(0) Cu(I) Cu(I), Cu(II) Cu(I), Cu(II) Cu(I) Cu(0)

± ± Cu surface segregation Cu surface segregation ± ±

a

In the case of methanol wet partial oxidation, the O2/CH3OH column contains the corresponding O2/CH3OHexc values (see Section 2).

3.1.3. Methanol dry partial oxidation After methanol dry partial oxidation both the oxidation state and the concentration of Cu in the region

probed by XPS changed according to the reaction conditions. Fig. 3 shows how the oxidation state of Cu changed with the amount of O2 in the feed at 510 K. By increasing the O2/CH3OH ratio from 0 to 0.5, a transition from metallic copper to Cu(I) was induced. After exposure of the catalyst to a stoichiometric mixture of CH3OH and O2 (O2 =CH3 OH ˆ 0:5) only Cu(I) could be detected. Formation of Cu(II) was not observed up to O2 =CH3 OH ˆ 0:8. After all treatments performed at 510 K, the Cu 2p3/2/Zn 2p3/2 atomic ratio and the Cu 2p3/2/Cu 3s intensity ratio underwent only

Fig. 1. Cu 2p3/2/Zn 2p3/2 atomic ratio vs. T, after exposure of the catalyst to various reaction conditions. The horizontal line represents the value obtained for the freshly reduced catalyst.

Fig. 2. Cu 2p3/2/Cu 3s intensity ratio vs. T, after exposure of the catalyst to various reaction conditions. The horizontal line represents the value obtained after reduction.

2p3/2/Cu 3s intensity ratio (! in Fig. 2) were unchanged within the experimental uncertainty with respect to the values measured after reduction. This indicates that no changes in the morphology/distribution of the Cu, ZnO and Al2O3 phases occurred in the surface and near-surface regions during the methanol decomposition treatment.

64

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

Fig. 3. XPS Cu LMM region after methanol dry partial oxidation at 510 K with various O2 contents in the feed.

small changes with respect to the values measured for the freshly reduced catalyst (see ^ in Fig. 4). This implies that, under these conditions, no changes in the distribution of the Cu, ZnO and Al2O3 phases occurred, for both Cu(0)- and Cu(I)-rich surfaces. The temperature at which the dry partial oxidation was carried out strongly in¯uenced the oxidation state (Fig. 5) and the concentration of copper in the surface and near-surface regions (see ^ in Figs. 1 and 2). At 470 K, the Cu was predominantly Cu(I), and only a

Fig. 4. Cu 2p3/2/Zn 2p3/2 atomic ratio vs. O2/CH3OH ratio in the feed after exposure of the catalyst to dry and wet methanol partial oxidation conditions at 510 K. In the case of wet partial oxidation the abscissa represents the O2/CH3OHexc ratio (see Section 2). The horizontal line represents the value measured for the freshly reduced catalyst.

Fig. 5. XPS Cu LMM region after methanol dry partial oxidation at various temperatures. The O2/CH3OH ratio was 0.42 for all experiments.

small amount of metallic copper (10%) was detected. After increasing the reaction temperature, a transition from Cu(I) to Cu(0) occurred. At 530 K, metallic copper was the most abundant Cu species and only 20% of Cu(I) was detected. Figs. 1 and 2 show that the change in Cu oxidation state was accompanied by a variation of the total Cu concentration in the surface and near-surface regions. At temperatures between 470 and 510 K, when Cu(I) was the predominant Cu species, the distribution of the Cu, ZnO and Al2O3 phases in the region probed by XPS was unchanged with respect to the freshly reduced catalyst. This is shown by the fact that the Cu 2p3/2/Zn 2p3/2 atomic ratio and the Cu 2p3/2/Cu 3s intensity ratio were almost unmodi®ed by the reaction treatment. Only at temperatures higher than 510 K, when Cu(0) predominates, both the Cu 2p3/2/Zn 2p3/2 atomic ratio and the Cu 2p3/2/Cu 3s intensity ratio became signi®cantly lower than the values corresponding to the reduced catalyst. As discussed below, these changes indicate a decrease of the total Cu concentration in the region probed by XPS. 3.1.4. Methanol wet partial oxidation The dependency of the Cu oxidation state on the O2/ CH3OHexc ratio after exposure of the catalyst to methanol wet partial oxidation conditions is shown in Fig. 6. As in the case of methanol dry partial oxidation, a transition from metallic Cu to Cu(I)

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

65

Fig. 6. XPS Cu LMM region after methanol wet partial oxidation with various O2/CH3OHexc ratios. The temperature was 510 K during all treatments.

Fig. 7. XPS Cu LMM region after methanol wet partial oxidation at various temperatures. The O2/CH3OHexc ratio was 0.42 for all treatments.

was observed upon increasing the O2/CH3OHexc ratio from 0 to 0.5. However, in contrast to the dry partial oxidation case, a signi®cant amount of Cu(II) (15%) was detected after wet partial oxidation with O2 =CH3 OHexc ˆ 0:5. The presence of water also affected the Cu concentration measured by XPS after exposure to high O2/CH3OH ratios. The symbol & in Fig. 4 shows that, after methanol wet partial oxidation with O2 =CH3 OHexc ˆ 0:5, the Cu 2p3/2/Zn 2p3/2 atomic ratio strongly increased with respect to the value corresponding to the freshly reduced catalyst. This suggests that surface segregation of Cu occurs under these conditions. The presence of water seems to have only a marginal effect at lower O2/CH3OHexc values. Fig. 7 shows the dependency of the Cu oxidation state on the temperature during the exposure of the catalyst to methanol wet partial oxidation conditions. The observed trend was similar to the one obtained after methanol dry partial oxidation. A transition from oxidised to metallic Cu occurred upon increasing the temperature from 470 to 530 K. However, in the presence of water, at temperatures lower than 485 K, signi®cant amounts of Cu(II) (up to 60%) were detected. As shown in Figs. 1 and 2 (&) the presence of Cu(II) was associated with an increase of both the Cu 2p3/2/Zn 2p3/2 atomic ratio and of the Cu 2p3/2/Cu 3s intensity ratio. This indicates that surface segregation

of Cu(II) species occurs under these conditions, giving rise to partial coverage of ZnO by a thin Cu(II)containing layer. The methanol conversion measured during dry partial oxidation with O2 =CH3 OH ˆ 0:42 changed from 70% at 550 K to 45% at 470 K and did not show any abrupt change that could indicate loss of catalytic activity associated with the transition from metallic Cu to Cu(I). However, a very low methanol conversion was observed during wet partial oxidation with O2 =CH3 OHexc ˆ 0:42 at 470 K, indicating that a Cu(II)-rich catalyst surface is inactive. 3.2. Adsorbed species The XPS C 1s spectra obtained after exposure of the catalyst to various conditions are presented in Fig. 8. The corresponding TPD spectra are shown in Fig. 9. The TPD features obtained after exposure of the catalyst to reaction conditions are quite broad, as expected for a highly porous powder sample. As a matter of fact, it has been demonstrated that the readsorption of desorbing molecules within the pore system during the TPD experiment is responsible for extensive broadening of the desorption features and for a shift of their position by up to a few hundreds of Kelvin [37]. The absorbed species identi®ed on the basis of the XPS and TPD data are summarised in Table 2.

66

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

Fig. 8. XPS C1s spectra after exposure of the catalyst to various reaction conditions. The spectra were normalised using the intensity of the graphite component.

3.2.1. Reduction After catalyst reduction, the XPS C 1s spectrum shows a sharp peak at 284.4 eV that is assigned to graphite [38]. A weak shoulder at 285.5 eV, probably due to adsorbed CO, is present [39]. 3.2.2. Methanol decomposition After exposure to methanol decomposition conditions at 510 K, a new intense signal at 287.2 eV appeared. A BE of 287.2 eV is reported in the literature for molecularly adsorbed methanol on various Cu and ZnO single-crystalline surfaces. Adsorbed methoxy is reported to give an XPS C 1s peak in the range 285.8±286.8 eV, depending on the surface and on the

coverage [23,35]. The presence of molecularly adsorbed methanol on the catalyst surface after exposure to reaction conditions is not very likely, since this species is reported to desorb below room temperature from all of the Cu and ZnO surfaces studied so far [15,16,35]. Based on these considerations, we attribute the peak at 287.2 eV to adsorbed methoxy groups. Bowker and co-workers [40] found no evidence for the adsorption of methanol on a vacuum annealed Cu2O powder (whose surface consists of a Cu metal ®lm) upon methanol exposure at RT. On the other hand, adsorbed methoxy groups are easily formed on Zn…1 0 1 0† upon methanol dosage, and they are still present after annealing at 523 K [16]. Therefore, it is very likely that methoxy groups adsorbed on ZnO are responsible for the C 1s XPS component at 287.2 eV. The presence of adsorbed methoxy species is supported by the TPD spectra recorded after exposure of the catalyst to methanol decomposition conditions (Fig. 9a). The detected TPD features can be associated to the following reactions of adsorbed methoxy and formate species (the subscript a indicates adsorbed species): CH3 Oa ‡ OHa ! CH3 OH " ‡ Oa CH3 Oa ‡ Oa ! HCOOa ‡ 2Ha HCOOa ! CO2 " ‡

1 2 H2

…low T†

(4)

…medium T†

(5)

"

(6)

HCOOa ! CO " ‡ OHa CH3 Oa ‡ Oa ! CO " ‡ H2 " ‡ OHa

(7) …high T† (8)

The desorption behaviour of the various species can be rationalised keeping in mind the porous structure of

Table 2 Distinctive TPD features and XPS C 1s BE of the adsorbed species identi®ed on the catalyst surface after various treatments Reaction

Adsorbed species

C 1s BE (eV)

Distinctive TPD features m/e

T (K) 440 520 650

Methanol decomposition

CH3O

287.2

31 44 28

Methanol dry partial oxidation ‡ methanol wet partial oxidation

CH3O CO3 HCOO OH

287.2 289.4 290.0 ±

31 ± 44 ‡ 2 18

Methanol wet partial oxidation

H2O

±

18

440±510 ± 600 >700 500

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

67

Fig. 9. TPD spectra measured after exposure of the catalyst to the following reaction conditions: (a) methanol decomposition at 510 K; (b) methanol dry partial oxidation with O2 =CH3 OH ˆ 0:42 at 530 K; (c) methanol wet partial oxidation with O2 =CH3 OHexc ˆ 0:42 at 470 K; (d) steam reforming at 510 K. The contribution deriving from the fragmentation of CO2 has been subtracted from all m=e ˆ 28 spectra.

the catalyst. In a porous solid, because of diffusion limitations within the pore system, molecules can adsorb and desorb several times before leaving the sample and being detected by the QMS. At temperatures lower than 450 K, the main desorption product is methanol, formed via reaction (4). The methanol molecules that were not able to leave the pore system before the temperature reaches 500 K can re-form methoxy groups upon adsorption and form formate

via reaction (5). Formate decomposes according to reaction (6), forming CO2 and H2. The H2 that is formed partly dissolves in the bulk of ZnO and desorbs slowly giving rise to a continuously increasing m=e ˆ 2 signal. Uptake of hydrogen by ZnO has been reported previously in the literature (see for instance [41]). At temperatures higher than 600 K reaction of adsorbed methoxy species to form CO and H2 (reaction (8)), and possibly formate decomposition according to

68

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

reaction (7) become predominating. Since CO is the main product of the methanol decomposition reaction, desorption of adsorbed CO formed during the exposure of the catalyst to reaction conditions might also contribute to the m=e ˆ 28 signal. However the XPS C 1s spectrum recorded after methanol decomposition does not show a higher intensity at 285.5 eV compared to the spectrum of the freshly reduced catalyst. This suggests that no additional adsorbed CO is present after exposure of the catalyst to methanol decomposition conditions and that decomposition of adsorbed methoxy groups is the main source of the CO detected in the corresponding TPD experiment. 3.2.3. Methanol dry partial oxidation After exposure of the catalyst to methanol dry partial oxidation conditions resulting in a Cu(0)-rich surface (O2 =CH3 OH ˆ 0:42, T ˆ 530 K), an additional peak at 290.0 eV with a shoulder at 289.4 eV appears in the C 1s XPS spectrum. We assign the signal at 290.0 eV to adsorbed formate groups. This attribution agrees with a BE in the range 288.5± 293.6 eV reported in the literature for formate species adsorbed on various ZnO and Cu surfaces [15,16,33]. The shoulder at 289.4 eV is assigned to ZnCO3, in agreement with the BE reported in the literature for this compound [39]. The presence of adsorbed formate is con®rmed by the simultaneous desorption of CO2 and H2 around 600 K observed in the corresponding TPD experiments (see Fig. 9b). As in the case of methanol decomposition, dissolution of H2 in the bulk of ZnO is responsible for the smaller intensity of the m=e ˆ 2 desorption peaks compared to that predicted by reaction (6). Decomposition of adsorbed formate on ZnO powder has been reported in the range between 580 and 630 K [15]. On the other hand, decomposition of adsorbed formate on vacuumannealed Cu2O (whose surface consists of a metallic Cu ®lm) was observed at 485 K [40]. This suggests that the formate species present on the catalyst after exposure to methanol dry partial oxidation conditions is probably adsorbed on ZnO. Decomposition of ZnCO3 also contributes to the desorption of CO2, according to the reaction CO3a ! CO2 " ‡ Oa

(9)

The presence of gas-phase O2 is also associated with a high temperature H2O desorption signal, probably due

to disproportionation of OH groups adsorbed on ZnO, according to the reaction 2OHa ! H2 O " ‡ Oa

(10)

Desorption of water formed by disproportionation of very stable OH surface groups on ZnO at temperatures above 670 K has been reported in the literature [42]. 3.2.4. Methanol wet partial oxidation Exposure of the catalyst to methanol wet partial oxidation conditions resulting in a Cu(II)-rich surface (O2 =CH3 OHexc ˆ 0:42, T ˆ 470 K) gives rise to a lower intensity of the XPS C 1s signal associated with adsorbed methoxy and formate species. This indicates that a smaller amount of these adsorbed species forms on the Cu(II)-rich surface compared to the Cu(0)-rich surface obtained after methanol dry partial oxidation. This is con®rmed by the lower intensity of the m=e ˆ 31 and m=e ˆ 44 desorption peaks in the corresponding TPD experiment (Fig. 9c). The C 1s intensity observed in the 285.5±287 eV region is attributed to adsorbed CHx species, in accordance with [43]. The water TPD spectrum shows an intense peak at 500 K and the same high temperature component attributed to disproportionation of surface OH groups on ZnO. Since a high intensity of the feature at 500 K is associated with the presence of water in the feed, we assign it to the desorption of molecularly adsorbed water. 3.2.5. Methanol steam reforming After methanol steam reforming the XPS C 1s spectrum is similar to the one recorded after methanol wet partial oxidation, with the remarkable difference that, in the steam reforming case, the signal due to ZnCO3 predominates over the one of adsorbed formate. This supports the idea that the formation of surface formate is strongly favoured by the presence of O2 in the feed. 4. Discussion After methanol decomposition at all temperatures studied, the catalyst remained in a thoroughly reduced state and no indications of morphological changes in the surface and near-surface regions were obtained by post-reaction XPS. This indicates that thermal sintering of the copper crystallites is not occurring in the

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

present conditions. After exposure of the catalyst to methanol dry partial oxidation conditions with O2 =CH3 OH ˆ 0:42, the copper in the catalyst was prevalently Cu(I) for T  510 K, whereas it was almost exclusively Cu(0) for T  530 K. The observed change of the Cu oxidation state with the temperature can be rationalised by assuming a different temperature dependence of the two concurring reactions given below 4Cu ‡ O2 ! 2Cu2 O

(11)

3Cu2 O ‡ CH3 OH ! 6Cu ‡ 2H2 O ‡ CO2

(12)

The reduction of cuprous oxide by methanol (reaction (12)) is ef®cient only at high temperatures, whereas oxidation of Cu by oxygen (reaction (11)) is already fast in the low temperature region. This hypothesis was con®rmed by differential thermo-gravimetry (DTG) experiments performed under ¯ow of O2/He and CH3OH/He mixtures. Reduction of the calcined catalyst by a CH3OH/He mixture containing 13% CH3OH occurred, as a sharp transition, at 520 K, whereas reoxidation of the reduced catalyst by a O2/He mixture containing 30% O2 was already complete below 470 K (in both cases the heating rate was 10 K/min, and the He ¯ow 50 ml/min). These results clearly show that, in the temperature range in which post-reaction XPS analysis indicated Cu(I) being the predominant Cu-species, the rate at which the reduced catalyst is oxidised by O2 is already signi®cant, whereas the rate of reduction of oxidised Cu by methanol is still low. Besides in¯uencing the oxidation state of copper, the temperature at which the methanol dry partial oxidation treatment was carried out had a pronounced effect on the intensity of the Cu XPS peaks. As shown in Figs. 1 and 2, for T  530 K, an increase of the temperature at which the methanol dry partial oxidation treatment was carried out, caused a decrease of both the Cu 2p3/2/Zn 2p3/2 and the Cu 2p3/2/Cu 3s XPS ratios with respect to the values measured after reduction. For T  510 K both ratios remained almost unchanged within the experimental uncertainty. Both observed trends can be explained by a decrease of the Cu concentration in the near-surface region of the catalyst with increasing reaction temperature. The depth of the region probed by different photo-electrons depends, for a certain material, on their kinetic energy. From the calculations of Tanuma et al. [44], a value equal to 1.1 nm can be estimated for the

69

inelastic mean free path of both Cu 2p3/2 and Zn 2p3/2 photo-electrons. Due to their higher kinetic energy, the inelastic mean free path of Cu 3s photoelectrons is larger, about 2.1 nm, implying that these electrons carry information from a region extending deeper within the sample. Therefore, the decrease of the Cu 2p3/2/Cu 3s ratio observed after methanol dry partial oxidation at T  530 K indicates that the surface region of the catalyst becomes Cu-depleted with respect to the freshly reduced catalyst. This is con®rmed by the concomitant decrease of the Cu 2p3/2/Zn 2p3/2 ratio at temperatures higher than 530 K. The mechanism resulting in the observed decrease of the Cu surface concentration is not entirely clear. Growth of the Cu crystallites on the catalyst surface would reduce the amount of Cu detected by XPS with a simultaneous increase of the Zn surface concentration. This would explain the observed decrease of the Cu 2p3/2/Zn 2p3/2 ratio. An increase of the Cu crystallite size by 40% has been observed by GuÈnter et al. [45] in a Cu/ZnO catalyst at 523 K, after oxidation in a CH3OH/H2O/O2 stream, followed by reduction under CH3OH/H2O ¯ow. However, the change of the Cu 2p3/ 2/Zn 2p3/2 ratio observed after methanol dry partial oxidation was accompanied by a decrease of the Cu 2p3/2/Cu 3s ratio below the value measured for bulk metallic Cu. This cannot be explained by growth of the Cu crystallites alone, since the Cu 2p3/2/Cu 3s ratio of bulk Cu represents a limit value corresponding to an in®nitely thick Cu layer. The most plausible mechanism that can account for lower Cu 2p3/2/Cu 3s ratios is coverage of the Cu crystallites by another material. Under methanol dry partial oxidation conditions above 510 K the Cu crystallites on the catalyst surface might be encapsulated by a layer of ZnO. Migration of ZnOx onto the Cu surface has been observed by Fujitani and Nakamura [46] in a Cu/ZnO catalyst exposed to methanol synthesis conditions at T  650 K. The reason why such a migration would occur at lower temperatures under methanol dry partial oxidation conditions is not clear and needs further investigation. As shown in Fig. 7, the above considerations about the in¯uence of temperature on the redox behaviour of copper during exposure to methanol dry partial oxidation conditions are still relevant when water is present in the feed. In contrast to the ``dry'' case, in the presence of water, formation of Cu(II) occurred after treatment at T  485 K. The fact that the formation of

70

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

Cu(II) was only observed when water was present in the feed suggests that the interaction of water with the catalyst plays a major role in the oxidation of Cu to its highest oxidation state. A similar effect was observed by Morterra et al. [47] for Cu/ZrO2 catalysts. In that work, water was shown to cause an increase of the amount of Cu(II) already at room temperature. The formation of Cu(II) was accompanied by an increase of the Cu surface concentration, indicating that surface segregation of Cu(II) occurs (see & in Figs. 1 and 2). The ef®ciency of water in increasing the dispersion of supported Cu(II)-containing phases has been previously reported in the literature. Morterra et al. [47] showed that an increase of the dispersion of Cu(II) species on zirconia could be obtained by the action of water a room temperature. Surface segregation of CuO was also observed by Chadwick et al. [48] in nanocrystalline Cu(II)-doped ZnO particles prepared in aqueous solution. We propose that the strong interaction of water with Cu(II) species makes their formation and spreading on the support surface energetically favourable. By comparing the desorption features in Fig. 9a±d, it is evident that they are strongly in¯uenced by the applied reaction conditions, suggesting a change of the type and relative amount of surface species. However one needs to be careful in relating the information obtained by post-reaction UHV TPD experiments to the actual population of surface intermediates under reaction conditions. Because of the pressure-gap existing between analysis and reaction conditions it is likely that many of the adsorbed species on the catalyst surface desorb or further react in the lapse of time between the quenching of the reaction and the TPD experiment. It is expected that the adsorbed surface species detected by the methods applied here are either the most stable present under reaction conditions or the products of further transformation of more weakly bound surface species. In other words, the applied approach makes the detection of labile reaction intermediates unlikely. The only adsorbed species that could be unambiguously identi®ed by post-reaction C 1s XPS and TPD data were methoxy and formate groups. Comparable amounts of adsorbed methoxy groups were detected by XPS after methanol decomposition and dry partial oxidation at 530 K. The formation of adsorbed methoxy species appears to be hindered by the presence of water in the feed. This

is probably due to competitive molecular adsorption of water and to higher ef®ciency of reaction (4) due to the presence of a higher amount of surface hydroxy groups formed by water dissociative chemisorption. A similar effect was observed on a iron±molybdenum oxide by Holstein and Machiels [49]. The presence of O2 in the feed is also associated with the presence of adsorbed formate. This indicates the participation of surface atomic oxygen, formed by dissociative adsorption of gas phase O2, in the formation of surface formate by reaction (5). 5. Conclusions It was demonstrated that the oxidation state of Cu in a commercial Cu/ZnO/Al2O3 catalyst after methanol reforming is a function of the feed composition and of the reaction temperature. For a slightly sub-stoichiometric O2/CH3OH ratio, Cu(I) is predominant at T  510 K, whereas at higher temperatures the Cu is in the metallic state. The catalyst shows a substantial activity towards the production of hydrogen when Cu(0) or Cu(I) are detected by post-reaction XPS analysis. With the same oxygen content in the feed, in the presence of water, Cu is oxidised to Cu(II) at T  485 K. The resulting Cu(II)-containing phase tended to spread on the catalyst surface, that itself became inactive in the production of H2 from methanol. When oxygen was present in the feed at T  530 K the Cu surface concentration measured by XPS decreased, possibly due to partial encapsulation of the surface Cu crystallites by ZnO. This process was suppressed by the presence of water in the feed. Adsorbed methoxy and formate groups were identi®ed by post-reaction XPS and TPD experiments. The presence of O2 in the feed strongly enhanced the formation of surface formate. The presence of water in the reactant mixture brought about a decrease of the amount of adsorbed methoxy groups, probably due to competitive molecular adsorption of water and to higher concentration of surface hydroxy groups. Acknowledgements Thanks are due to Johnson Matthey plc. for supplying the methanol reforming catalyst. The authors are

F. Raimondi et al. / Applied Surface Science 189 (2002) 59±71

grateful to J. Hobley, T. Lippert, E. Newson and B. Schnyder for fruitful discussion. References [1] [2] [3] [4] [5]

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]

W. DoÈnitz, Bunsenmagazin 1 (2001) 3. Y. Borodko, G. Somorjai, Appl. Catal. A 186 (1999) 355. A. Bill, Ph.D. Thesis, EPF, Lausanne, 1998. F. Peltier, P. Chaumette, J. Saussey, M. Bettahar, J. Lavalley, J. Mol. Catal. A 132 (1998) 91. B. HoÈhlein, in: Minutes of the Second IEA Advanced Fuel Cell Workshop on Fuel Processing for Modular (<100 kW) Fuel Cell Power Packages, 1997, Wislikofen, Switzerland. K. Geissler, E. Newson, F. Vogel, T. Truong, P. Hottinger, A. Wokaun, Phys. Chem. Chem. Phys. 3 (2001) 289. N. Takezawa, N. Iwasa, Catal. Today 36 (1997) 45. H.-F. Oetjen, V.M. Schmidt, U. Stimming, F. Trila, J. Electrochem. Soc. 143 (1996) 3838. B.A. Sexton, A.E. Hughes, N.A. Avery, Appl. Surf. Sci. 22/23 (1985) 404. R. Ryberg, J. Chem. Phys. 82 (1985) 567. Q. Dai, A.J. Gellman, J. Phys. Chem. 97 (1993) 10783. A.K. Chen, R. Masel, Surf. Sci. 343 (1995) 17. I.E. Wachs, R.J. Madix, J. Catal. 53 (1978) 208. R. Zhang, A. Ludviksson, C.T. Campbell, Catal. Lett. 25 (1994) 277. J.M. Vohs, M.A. Barteau, Surf. Sci. 176 (1986) 91. P.M. Jones, J.A. May, J.B. Reitz, E.I. Salomon, J. Am. Chem. Soc. 120 (1998) 1506. A.F. Carley, P.R. Davies, G.G. Mariotti, S. Read, Surf. Sci. 364 (1996) L525. M. Bowker, S. Poulston, R.A. Bennett, A.H. Jones, Catal. Lett. 43 (1997) 267. I.E. Wachs, R. Madix, Surf. Sci. 84 (1979) 375. P.R. Davies, G.G. Mariotti, Catal. Lett. 43 (1997) 261. G.J. Millar, C.H. Rochester, K.C. Waugh, J. Chem. Soc., Faraday Trans. 88 (15) (1992) 2257. S.G. Neophitides, A.J. Marchi, G.F. Froment, Appl. Catal. A 86 (1992) 45. P. Petersen, W. Krasser, Appl. Surf. Sci. 103 (1996) 91. J. Skrzypek, J. Sloczynski, S. Ledakowicz, Methanol Synthesis, Polish Scienti®c Publishers, Warszawa, 1994.

71

[25] B. Peppley, J. Amphlett, L. Kearns, R. Mann, Appl. Catal. A 179 (1999) 21. [26] B. Peppley, J. Amphlett, L. Kearns, R. Mann, Appl. Catal. A 179 (1999) 31. [27] T. Huang, S. Chren, Appl. Catal. 40 (1988) 43. [28] L. Alejo, R. Lago, M. Pena, J. Fierro, Stud. Surf. Sci. Catal. 110 (1997) 623. [29] T. Reitz, M. Ahmed, R. Kumar, H. Kung, Stud. Surf. Sci. Catal. 130 (2000) 3645. [30] I. BoÈttger, T. Schedel-Niedrig, O. Timpe, R. Gottschall, M. HaÈvecker, T. Ressler, R. SchloÈgl, Chem. Eur. J. 6 (2000) 1870. [31] M. HaÈvecker, A. Knop-Gericke, T. Schedel-Niedrig, R. SchloÈgl, Angew. Chem. Int. Ed. 37 (1998) 1939. [32] H. Werner, D. Herein, G. Schulz, U. Wild, R. SchloÈgl, Catal. Lett. 49 (1997) 109. [33] J.H. Sco®eld, J. Electron Spectrosc. Relat. Phenom. 8 (1976) 129. [34] K.R. Harikumar, C.N.R. Rao, Catal. Lett. 47 (1997) 265. [35] B. Peplinski, W.E.S. Unger, I. Grohmann, Appl. Surf. Sci. 62 (1992) 115. [36] G. Moretti, G. Fierro, M. Lo Jacono, P. Porta, Surf. Interf. Anal. 14 (1989) 325. [37] R.K. Herz, J.B. Kiela, S.P. Marin, J. Catal. 73 (1982) 66. [38] C. Kozlowski, P.M.D. Sherwood, J. Chem. Soc., Faraday Trans. I 80 (1984) 2099. [39] K.R. Harikumar, C.N.R. Rao, Appl. Surf. Sci. 125 (1998) 245. [40] S. Poulston, E. Rowbotham, P. Stone, P. Parlett, M. Bowker, Catal. Lett. 52 (1998) 63. [41] J. Yoshihara, C.T. Campbell, Surf. Sci. 407 (1998) 256. [42] R. Zhang, A. Ludviksson, C.T. Campbell, Surf. Sci. 289 (1993) 1. [43] S. Badrinarayanan, A.B. Mandale, S.R. Sainkar, J. Mater. Res. 11 (1996) 1605. [44] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interf. Anal. 17 (1991) 911. [45] M.M. GuÈnter, T. Ressler, R.E. Jentoft, B. Bems, J. Catal. 203 (2001) 133. [46] T. Fujitani, J. Nakamura, Appl. Catal. A 191 (2000) 111. [47] C. Morterra, E. Giamello, G. Cerrato, G. Centi, S. Perathoner, J. Catal. 179 (1998) 111. [48] A.V. Chadwick, A. Harsch, N.V. Russell, K.F. Tse, A.R. Witham, A. Wilson, Radiat. Eff. Defects Solids 137 (1995) 51. [49] W.L. Holstein, C.J. Machiels, J. Catal. 162 (1996) 118.