Interaction of NH 3 with oxygen-pretreated Ni at 600 K

Transcription

1 SURFACE AND INTERFACE ANALYSIS Surf. Interface Anal. 2002; 34: Published online in Wiley InterScience ( DOI: /sia.1360 Interaction of NH 3 with oxygen-pretreated Ni at 600 K B. Lescop, 1 G. Fanjoux, 1 A. Galtayries 2 and A. Le Nadan 1 1 Laboratoire des Collisions Electroniques et Atomiques, UFR Sciences et Techniques, 6 av. V. Le Gorgeu, Brest Cedex, France 2 Laboratoire de Physico-Chimie des Surfaces, CNRS, Ecole Nationale Supérieure de Chimie de Paris, 11 rue Pierre et Marie Curie, Paris, France Received 16 July 2001; Revised 30 November 2001; Accepted 27 December 2001 The aim of the present study was to investigate the reaction between ammonia and pre-adsorbed oxygen on Ni at 600 K using metastable impact electron spectroscopy (MIES) and UPS techniques. For a given oxygen-pretreated nickel surface, the changes in the MIES spectra along with ammonia adsorption can be decomposed into three different steps: a first domain without significant changes, a second domain corresponding to drastic changes in the intensity of a signal at 15.8 ev; and a third domain where no more change is detected. The first domain may correspond to the ammonia chemisorption on the surface, whereas the changes in the second domain would be attributed to surface dehydroxylation. The existence of NH x groups on the surface has been detected by UPS; they result from ammonia chemisorption. The changes in the MIES spectra have also been studied versus the initial oxygen uptake on Ni. It emerged that the first domain is larger with increasing initial oxygen exposure: this may be explained by the poorer surface reactivity of NiO, contrary to the oxygen-adsorbed phase. Copyright 2002 John Wiley & Sons, Ltd. KEYWORDS: ammonia; Ni; oxygen interaction; MIES; UPS; kinetics INTRODUCTION The interaction of ammonia with metal surfaces has been studied extensively. 1 3 However, only a few studies have been performed on the adsorption of ammonia on oxygenpretreated metal surfaces. 2 Open questions remain as far as adsorption mechanisms are concerned for this type of system. It has been shown recently by XPS that, on Ni(111), the oxygen-adsorbed phase formed during the first stages of oxygen interaction presents reactivity towards NH 3 adsorption at room temperature, contrary to both clean Ni(111) and NiO thin films. 4 Nickel surface pretreatment with oxygen greatly enhances NH 3 decomposition. 5 At high temperatures (>500 K), NH 3 was adsorbed and decomposed into NH x adspecies (x D 0, 1, 2). The adsorption and dissociation reactions take place concomitantly. 6 NH 3 (g) C O(adsorbed)! NH 2 (adsorbed) C OH (adsorbed) NH 2 (adsorbed)! NH (adsorbed) C H(adsorbed) NH (adsorbed)! N(adsorbed) C H(adsorbed) Ł Correspondence to: B. Lescop, Laboratoire des Collisions Electroniques et Atomiques, UFR Sciences et Techniques, 6 av. V. Le Gorgeu, Brest Cedex, France. benoit.lescop@univ-brest.fr According to the results of Sprunger et al., 7 the hydroxyl groups may then combine either with adsorbed hydrogen OH (adsorbed) C H(adsorbed)! H 2 O or follow the disproportionation reaction OH (adsorbed) C OH (adsorbed)! H 2 O 4 C O(adsorbed) 5 Within the framework of a fundamental approach, our work was aimed at investigating the NH 3 adsorption and the reaction mechanisms on oxygen-pretreated Ni by an in situ technique at 600 K. Besides thermal desorption studies, to our knowledge no spectroscopic data are available about the mechanisms of adsorption under such temperature conditions. The interaction of O 2 and NH 3 with the Ni surface was followed by metastable impact electron spectroscopy (MIES) in combination with UPS (He(I)). The former technique is non-destructive and based on the interaction of metastable helium atoms (He Ł (2 3 S D 19.8 ev) with the surface. Its main asset is that it probes the surface outermost layer in situ with high sensitivity. 8 EXPERIMENTAL The experimental set-up has been described already elsewhere, 9 so only a few details are given here. The apparatus consists of a source for metastable atoms and photons in an ultrahigh vacuum (UHV) chamber whose base pressure is 5 ð Torr. Briefly, a helium cold-cathode gas discharge Copyright 2002 John Wiley & Sons, Ltd.

2 Interaction of NH 3 with O 2 -pretreated Ni at 600 K 561 provides both ultraviolet photons (He(I)) and metastable He Ł atoms (2 1,3 S) with thermal kinetic energies. 10 The tripletto-singlet ratio has been measured as 7 : 1 for this type of MIES source. 10 Metastable atoms and photons within the beam are separated by means of a time-of-flight method using a mechanical chopper. The angle of incidence of the probe beam is 45 with respect to the surface normal and the electrons are collected in the direction perpendicular to the surface. The MIES and UPS measurements were performed using a 127 analyser with a 6 ev fixed energy leading to a resolution of 250 mev. Biasing the sample (6 V negative on the sample) permits direct measurement of the surface work function from the high-binding-energy cut-off of the spectra. The polycrystalline Ni sample was first mechanically polished before its introduction into the analysis chamber; it was cleaned by repeated cycles of Ar C bombardment (3 kev; 2 ð 10 6 Torr) and heating in UHV (650 K) with cycles of oxygen and hydrogen atmosphere (10 6 Torr). According to Morgner et al., 11 MIES is sensitive enough to check for surface cleanliness, we thus considered the lack of change in the MIES spectrum after several cleaning cycles as a good indicator to end the cleaning process. Exposures to oxygen (99.995% purity) or ammonia (99.995% purity) were performed by admitting the gas through two sapphire leak valves at a constant pressure of typically Torr. Oxygen exposures varied between 5 and 500 L (1Langmuir D 1 ð 10 6 TorrÐs). Ammonia exposure was with a continuous flow of 1.4 ð 10 6 Torr. Between two series of ammonia interaction on the oxygen-pretreated sample, the surface was sputter-cleaned by Ar C bombardment. All measurements were performed at 600 K. RESULTS AND DISCUSSION Interaction of ammonia with oxygen-pretreated Ni surface: MIES study Figure 1(a) presents the MIES spectra for the clean Ni surface and for the surface after oxygen pretreatment and during ammonia interaction. The spectrum corresponding to the clean Ni surface is structureless despite the existence of the well-known Ni 3d band near the Fermi level. 12 The lack of signals in this spectrum can be explained by the fact that the interaction is dominated by the resonant ionization C Auger neutralization (RI C AN) process, 8 which dominates when the work function is larger than the He Ł ionization energy (3.5 ev). The electrons emitted from this process reflect the self-convolution of the local surface density of states (SDOS). This also explains the absence of signal up to the metastable energy. Figure 1 illustrates that once oxygen has been adsorbed it causes a marked reduction of the signal located at 16 ev and a slight increase of the work function (C0.3 ev). The spectrum is still dominated by the RI C AN process, which is in agreement with spectra previously published by Morgner et al. 11,13 In our series of experiments, we also studied the surface work function by exposing Ni to L of oxygen at 600 K (not shown). The surface work function reached a maximum of C0.6eVfor<10 L and decreased continuously down to 0.2 ev. From the interpretation of the mechanisms currently accepted for Ni oxidation, which involve nucleation and Figure 1. (a) The MIES spectra for a pre-oxidized Ni surface exposed to ammonia deposition. Clean Ni and pre-oxidized spectra are indicated; NH 3 exposure is 100 L per spectrum. (b) Difference between the final spectrum and the clean Ni spectrum. growth of oxide islands within the oxygen-adsorbed phase (see Ref. 14 and references therein), we attribute the maximum work function value to the oxygen chemisorption, whereas the negative value is attributed to NiO formation. Note that above 500 K adsorbed oxygen may also diffuse into the bulk 14 and thus reduce the work function value. For 200 L of oxygen its value is intermediate between both limits, which suggests that the surface is covered by both an oxygen-adsorbed phase and NiO islands. This is consistent with an XPS study of the interaction of O 2 with Ni(111) at 650 K. 4 It is also noteworthy that above 500 K the non-polar NiO(100) starts to be formed irreversibly, which leads to a non-hydroxylated oxidized surface Figure 1(a) presents the NH 3 interaction on the Ni surface pretreated by 200 L of oxygen at 600 K. Ammonia exposure was up to 3000 L. During ammonia adsorption, each MIES spectrum in Fig. 1(a) was taken after a dose of 100 L of NH 3. The spectrum corresponding to the first dose is very similar to the previous one in the absence of ammonia. However, after subtraction (not shown), the result exhibited a weak structure at ¾11 ev indicative of the start of adsorption on the surface. With increasing doses of NH 3 we noticed changes both in the work function values and in the signal at 15.8 ev. This signal has a preferentially 3d character even if the spectrum corresponds to the self-convolution of the SDOS. Figures 2(a)

3 562 B. Lescop et al. surface. The lack of structure at ¾16 ev may result from remaining adspecies on the surface; proven evidence is a surface work function value smaller than that for the clean Ni surface. However, missing structures cannot be attributed irrevocably to the presence of adsorbed species orbitals, because the MIES spectra reproduce the self-convolution of the SDOS. Figure 2. Changes as a function of ammonia exposure of: (a) the 15.2 ev signal intensity (MIES); (b) the 18 work function variations compared with Ni. The first value corresponds to the oxygen pretreated surface. and 2(b) illustrate the variations in signal intensity as a function of NH 3 exposure and the changes in the relative work function, respectively. Both figures highlight three different domains: (1) Up to 850 L of NH 3, no drastic modification in the shape of the MIES spectrum is noticed but the work function slightly increases (<0.1eV). (2) Above 900 L a marked rise of the 15.8 ev signal intensity is concomitant with a decrease in the surface work function. In this domain, the MIES spectra tend to become closer to the clean Ni spectrum. (3) Above 1750 L, the 15.8 ev signal intensity reaches a plateau, whereas a slow-down is noticed in the decrease of the surface work function value. With ammonia deposition there is no longer any change in the spectra. The subtraction between the initial Ni and the final spectrum (Fig. 1(b)) highlights the reduction of the signal at 15.8 ev, and thus the presence of adsorbed species. In the first domain, the established surface reactivity of the oxygen-adsorbed phase present on the pretreated Ni surface suggests that, at 600 K, NH 3 has adsorbed and is immediately decomposed into NH x adspecies (x D 0, 1, 2). A mixture of NH 2, NH and N adspecies may be present on the surface (Eqns (1) (3)). In the second domain, by analogy with the results obtained after hydrogen interaction with an oxygenpretreated Ni surface 18 by MIES, we attribute the changes in the spectra to oxygen removal from the surface. The removal of adsorbed oxygen must be related to an NH x O(adsorbed) interaction leading to hydroxyl group formation. 4 At 600 K these groups are then desorbed (Eqns (4) (5)). 18,19 In this scheme, the weak structure (at ¾11 ev) shown in Fig. 1(a) after ammonia adsorption and the work function increase can be attributed to the hydroxyl groups. These structures disappear in the third domain. At 600 K, the remaining adsorbed oxygen can also diffuse in the substrate. For the third domain, Fig. 1(b) illustrates the difference between the MIES spectra in this area and the clean Ni Interaction of ammonia with oxygen-pretreated Ni surface: UPS study The RI C AN process leading to self-convolution of the SDOS makes it difficult to identify the adsorbed species orbitals. The UPS data are used to observe directly the adsorbed species orbitals. Figure 3(a) shows the UPS spectra of NH 3 adsorption on oxygen-pretreated Ni. The clean Ni UPS spectrum (solid line) obviously exhibits the Ni 3d band at the Fermi level. 20 Upon an oxygen uptake of 200 L at 600 K (dotted line) the features of the O 2p adsorbed oxygen contribution are at ¾15 ev 21 and the intensity of the Ni 3d band located near the Fermi level decreases. The shoulder at 19 ev is also attributed to the formation of NiO islands. 21 Both the weakness of this structure and the work function increase confirm that the surface is also composed of an oxygen-adsorbed phase. Figure 3(a) displays only the final ammonia exposure (2400 L) (bold dashed line). The Ni 3d band intensity has increased but the O 2p oxygen feature has been reduced. However, the 3d band did not reach the clean Ni value, which gives evidence for the existence of remaining adsorbed species. A signal from 17 to 6 ev is clearly superimposed on the Ni spectrum. Given the high temperature, no hydroxyls are present on the surface. 18,19 The plateau observed in Fig. 3 is composed of ammonia decomposition products. The spectrum subtraction from Ni after NH 3 (Fig. 3(b)) exhibits broad structures located at ¾10, 12 and 14 ev (binding energies of ¾12, 9 and 7 ev). The structures can be related to the presence of NH x groups. Structures corresponding to the orbitals of NH compounds are observed at 13.7 and 12 ev (7.5 and 9.2 ev binding energies). 12,20 The third features at ¾10 ev can be attributed to NH Such a feature has been determined by calculations and attributed to NH 2.No structure is observed at 16 ev, which indicates the absence of atomic nitrogen on the surface. 20 In a similar way to what had been done with the MIES spectra, the 3d band intensity from the UPS spectra was measured versus NH 3 exposure (not shown). We noticed a continuous rise in intensity but this trend does not allow one to determine the different domains, as was done with MIES. This can be explained by a different probing depth: MIES only probes the first monolayer, whereas up to three or four monolayers are analysed in a UPS spectrum. However, the UPS spectra also show a decrease in the oxygen coverage. Interaction of ammonia with different oxygen-pretreated Ni surfaces Similar to the experiment performed with the Ni surface after an oxygen pretreatment of 200 L, we studied the ammonia interaction after different oxygen exposures by MIES. Figure 4 depicts the three domains described above and experimentally determined after oxygen doses of L.

4 Interaction of NH 3 with O 2 -pretreated Ni at 600 K 563 the reactions corresponding to the three domains are, up to 500 L of O 2, linearly dependent on the oxygen uptake. The first reaction (restricted to the first domain) is longer as more oxygen is pre-exposed. This is in good agreement with a decrease in the NH 3 chemisorption process, whose reactivity is correlated to the increase in the oxygenadsorbed phase coverage. The same conclusion had been drawn from an XPS study of the NH 3 interaction, at room temperature, on different oxygen-pretreated Ni(111) surfaces. 4 Accordingly, the surface dehydroxylation, for which we have defined the second domain, decreases with greater oxygen exposure. Once again, the more reactive the surface is towards ammonia chemisorption, the more important the surface hydroxylation is. Consequently, higher rates of dehydroxylation occur on the more hydroxylated surfaces. In the last domain, the oxygen has to be absent and the surface sites are still occupied by NH x (x D 1, 2) adspecies, so further ammonia adsorption becomes impossible. These first results obtained on an Ni polycrystalline sample need confirmation, particularly as a function of temperature of adsorption and orientation of the crystal face. CONCLUSION Figure 3. The UPS spectra (a) and difference spectra (b) for an oxygen pretreated Ni surface exposed to ammonia: clean Ni (solid line); oxygen pretreated Ni (dotted line); 2400 L of NH 3 (bold dashed line). The reaction between ammonia and pre-adsorbed oxygen on Ni was investigated at 600 K using MIES and UPS. The changes of the MIES spectra along with ammonia uptake allowed the distinction of three domains: one related to ammonia chemisorption on the surface; one attributed to surface dehydroxylation; and one where surface saturation had occurred. The UPS data exhibited the existence of NH x groups on the surface. The reactivity of the surface is proportional to the oxygen-adsorbed phase coverage. The UPS technique enabled us to identify only the adsorbed species orbitals, whereas MIES probed the outermost layer of the NH 3 /O/Ni system. REFERENCES Figure 4. Ammonia exposure as a function of oxygen exposure of the initial surface. Each point delimits one domain (see text). The ammonia exposure corresponding to the limit of the first and second domains was plotted versus the oxygen exposure of the surface. The compiled results indicate that 1. Lambert RM, Bridge ME. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, vol. 3B, King DA, Woodruff WP (eds). Elsevier: Amsterdam, 1984; Thornburg DM, Madix RJ. Surf. Sci. 1989; 220: Chrysostomou D, Flowers J, Zaera F. Surf. Sci. 1999; 439: Galtayries A, Laksono E, Siffre JM, Argile C, Marcus P. Surf. Interface Anal. 2000; 30: Madey TE, Benndorf C. Surf. Sci. 1985; 152/153: Bassignana IC, Wagemann K, Küppers J, Ertl G. Surf. Sci. 1986; 175: Sprunger PT, Okawa Y, Besenbacher, Stensgaard I, Tanaka K. Surf. Sci. 1995; 344: Harada Y, Masuda S, Ozaki H. Chem. Rev. 1997; 97: Fanjoux G, Fornander-Billault H, Lescop B, LeNadan A. J. Electron Spectrosc. Relat. Phenom. 2001; 119: Maus-Friedrichs W, Wehrhahn M, Diekhoff S, Kempter V. Surf. Sci. 1990; 237: Morgner H, Tackenberg M. Surf. Sci. 1994; 301: Jacobi K, Jensen E, Rhodin T, Merrill R. Surf. Sci. 1981; 108: Kubiak R, Morgner H, Rakhovskaya O. Surf. Sci. 1994; 321: Brundle CR, Broughton JQ. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, vol.3a, King DA, Woodruff WP (eds). Elsevier: Amsterdam, 1990; Besenbacher F, Nørskov JK. Prog. Surf. Sci. 1993; 44: 5.

5 564 B. Lescop et al. 16. Rohr F, Wirth K, Libuda J, Cappus D, Bäumer M, Freund HJ. Surf. Sci. 1994; 315: L Kitakatsu N, Maurice V, Hinnen C, Marcus P. Surf. Sci. 1998; 407: Fanjoux G, Lescop B, LeNadan A. Surf. Interface Anal. 2002; Villarubia JS, Ho W. Surf. Sci. 1984; 144: Benndorf C, Nöbl C, Madey TE. Surf. Sci. 1984; 138: McKay JM, Henrich VE. Phys.Rev.Lett.1984; 53: Seabury CW, Rhodin TN, Purtell RJ, Merrill RP. Surf. Sci. 1980; 93: 117.