pubs.acs.org/Langmuir © 2010 American Chemical Society
Transient Oxygen States in Catalysis: Ammonia Oxidation at Ag(111)†,‡ Albert F. Carley, Philip R. Davies, M. Wyn Roberts,* and Kevin K. Thomas Cardiff Institute of Catalysis, School of Chemistry, Cardiff University, Cardiff, CF10 3AT Received March 10, 2010. Revised Manuscript Received April 15, 2010 Although the reactive sticking probability of oxygen at Ag(111) is of the order of 10-6 at 295 K, ammonia oxidation is a facile process at low temperatures. A combination of quantitative analysis of photoelectron spectra together with high resolution electron energy loss spectroscopy provides kinetic and spectroscopic evidence for an ammonia-dioxygen complex, stable at 100 K, as the key intermediate. The reactive oxygen O2(s) is a transient dioxygen precursor of the unreactive peroxo state O2δ-(a). It is present as a complex when ammonia and dioxygen are coadsorbed at low temperature (100 K) with evidence from both O(1s) and energy loss spectra. Hydroxyl and amide/imide species are formed, followed by dehydroxylation and “oxide” formation at 260 K. This is a further example (zinc was the first) of how an sp-metal, where dioxygen bond cleavage is slow, provides an alternative pathway via a transient dioxygen state to catalytic oxidation through precursor assisted dioxygen bond cleavage. Whether it is a general characteristic of spmetals remains to be established. Comparisons are made with the homogeneously catalyzed Gif reaction, the selective oxidation of hydrocarbons by dioxygen.
1. Introduction Cryogenic studies of molecular processes at metal surfaces have provided evidence for the facile nature of bond breaking and bond making and, in particular, the role of oxygen transient states in controlling reaction pathways in oxidation catalysis.1-4 Our earlier studies used Mg(0001), Cu(110), and Cu(111) as model systems in which the final chemisorbed oxygen state designated O2-(a) and associated with surface reconstruction is unreactive. These showed that reactive oxygen transients Oδ-(s) arising from bond cleavage of nitric oxide, nitrous oxide, and dioxygen have sufficient surface lifetimes under dynamic conditions to be able to control reaction pathways.1,2 It was a new concept which was shown to have wide relevance to the mechanism of surface reactions, including hydrocarbon oxidation,3,5 and relied on real-time photoelectron spectroscopic (XPS) studies. The concept was subsequently shown to be sustained at the atom resolved level by scanning tunneling microscopy (STM) with in situ XPS (ref 5a and reviewed in ref 5b). Disordered mobile oxygen states were shown to be very reactive, oxidizing ammonia to nitrogen adatoms, in contrast to the unreactive “final” ordered states. Recently, Friend et al.6 have also reported that disordered oxygen states present at low temperature at Au(111) are active in the oxidation of propene to give partial oxidation products, which provides a new view on † Part of the Molecular Surface Chemistry and Its Applications special issue. ‡ It is to celebrate Gabor's seminal contributions to the molecular understanding of catalysis over the last 40 years and his 75th birthday that we dedicate this paper. *To whom correspondence should be addressed. E-mail: robertsmw@ cf.ac.uk.
(1) (a) Au, C. T.; Roberts, M. W. Nature 1986, 319, 206. (b) Au, C. T.; Roberts, M. W. J. Chem. Soc., Faraday Trans. 1 1987, 83, 2047. (c) Boronin, A; Pashuski, A; Roberts, M. W. Catal. Lett. 1992, 16, 345. (2) Carley, A. F.; Roberts, M. W. J. Chem. Soc. Chem. Commun. 1987, 355. (3) Au, C. T.; Li, X.-c.; Roberts, M. W. J. Catal. 1987, 106, 538. (4) (a) Roberts, M. W. Chem. Soc. Rev. 1989, 18, 451. (b) Roberts, M. W. Chem. Soc. Rev 1996, 6, 437. (c) Roberts, M. W. Surf. Sci. 1994, 299/300, 769. (5) (a) Carley, A. F.; Davies, P. R.; Roberts, M. W. Chem. Commun. 1998, 1793. (b) Carley, A. F.; Davies, P. R.; Roberts, M. W. Philos. Trans. R. Soc., A 2005, 363, 829. (6) Min, B. K.; Deng, X; Liu, X; Friend, C. M.; Alemozafar, A. R. ChemCatChem. 2009, 1, 116.
Langmuir 2010, 26(21), 16221–16225
catalytic oxidation at gold surfaces. When dioxygen bond cleavage is slow, as at Zn(0001), it was shown7 in a dynamic study that, for ammonia-rich oxygen-ammonia mixtures, the kinetics have an inverse rate dependence on temperature, with the rate of oxygen cleavage enhanced by a factor of about 300 at low temperatures. An oxygen-ammonia complex was proposed as providing a low energy route to oxygen cleavage and the formation of hydroxyl, amide, and “oxide” species. The background, with details as to how the transient concept evolved, is discussed in a recent monograph.8 To explore the implications of the transient concept further, we chose to study Ag(111), since dioxygen bond cleavage is slow at this surface, with a reactive sticking probability of about 10-6 at ambient temperature;9,10 there is therefore a kinetic window in which to observe the effects of molecular oxygen transients.11 Furthermore, since silver has an important role in industrial oxidation catalysis, we hoped these studies might also give an insight into their technological relevance.
2. Experimental Section The Ag(111) surface was first polished to a mirror finish and then cleaned in situ in the ultrahigh vacuum (UHV) spectrometer by means of successive cycles of annealing at 600 °C and sputtering by argon-ion bombardment at the annealing temperature. The spectrometer had both X-ray photoelectron spectroscopy (XPS) and high resolution electron energy loss spectroscopic (HREELS) facilities, with concentrations of surface species determined from the intensities of O(1s), N(1s), and Ag(3d) spectra. The clean Ag(111) surface was exposed to spec-pure oxygen and ammonia in the pressure range 10-6-10-8 Torr, and exposures are reported in Langmuirs (1 L 10-6 Torr s). (7) (a) Carley, A. F.; Roberts, M. W.; Song, Y. J Chem. Soc., Chem. Commun. 1988, 267. (b) Carley, A. F.; Roberts, M. W.; Song, Y. J. Chem. Soc., Faraday Trans. 1990, 86, 2701. (8) Davies, P. R.; Roberts, M. W. Atom Resolved Surface Reactions: Nanocatalysis; RSC Publishing: Cambridge, 2008; ISBN: 978-0-85404-269-2. (9) Campbell, C. T. Surf. Sci. 1985, 157, 143. (10) Carley, A. F.; Davies, P. R.; Roberts, M. W.; Thomas, K. K. Surf. Sci. Lett. 1990, 238, L467. (11) Carley, A. F.; Davies, P. R.; Roberts, M. W.; Thomas, K. K.; Yan, S. Chem. Commun. 1998, 35.
Published on Web 05/06/2010
DOI: 10.1021/la100953m
16221
Article
Carley et al.
3. Results 3.1. Oxygen and Ammonia Adsorption. Both the HREEL and XP spectra (Figure 1) establish that a weakly adsorbed dioxygen state is present at 120 K but has desorbed at 210 K. There are two loss peaks at 220 and 700 cm-1 (Figure 1) assigned to υ(Ag-O2) and υ(O-O), respectively, and a single O(1s) peak at 532 eV. All these spectral features are absent after warming the adsorbed layer to 210 K. The sticking probability for the formation of the O2δ-(a) peroxo-state is very low and reported previously as ∼10-4 at 120 K and ∼10-6 at 150 K.9 Table 1 shows the O2δ-(a) concentrations measured as a function of oxygen exposure at 80, 120, and 150 K, with a maximum concentration calculated from the O(1s) intensity of 1.6 1014 cm-2 after an exposure of 1000 L at 80 K. The peroxo state is inactive in ammonia oxidation at low temperatures as shown by the O(1s), N(1s), and HREEL spectra observed when the peroxo state was exposed to NH3 with no evidence for hydroxyl or imide/ amide species at 210 K (Figure 2d). 3.2. Activation of Oxygen at Ag(111)-NH3(a). Ammonia adsorbed in a molecular state at 110 K to a concentration of 2 1014 cm-2, calculated from the N(1s) intensity at 400 eV, exhibited
Figure 1. (a) HREEL and (b) XP spectra for oxygen (2500 L) adsorbed at Ag(111) at 120 K; the peroxo O2δ-(a) state is characterized by loss peaks at 220 cm-1 (υAg-O) and 700 cm-1 (υO-O) and an O(1s) peak at 532 eV. The peroxo species desorbs below 210 K.
characteristic loss peaks at 250 cm-1 (υAg-N), 1080 cm-1 (δs(NH3)), and 3390 cm-1 (υNH) (Figure 3). When exposed to oxygen (20 L) at the same temperature, two peaks were resolved in the O(1s) spectrum with binding energies of 530.3 and 532 eV, assigned to hydroxyl and molecular oxygen states, respectively (Figure 4c). With further oxygen exposure (120 L), there was an increase in the intensity of the 530.3 eV component, and on warming to 220 K it was the major component (Figure 4d). Both components are characterized by a high sticking probability as compared with the peroxo state. There developed simultaneously a shoulder to lower binding energy of the main N(1s) intensity at 400 eV (Figure 4c), but the overall N(1s) intensity was unchanged. At 220 K, the main N(1s) intensity was at 399 eV, assigned to NH2(a), and the total intensity decreased due to the desorption of unreacted ammonia. The peak at 530.3 eV we assign to surface hydroxyls. On warming to 260 K (Figure 4e), asymmetry developed, indicating dehydroxylation and “oxide” formation. Figure 5c shows the corresponding HREEL spectra with strong loss peaks at 250, 300, 700, 1120, 1490, 1640, and 3390 cm-1 at 110 K. On warming to 210 K, there emerged a new loss peak at 3610 cm-1; the latter is assigned to υOH associated with hydroxyl species. The 1490 and 1640 cm-1 peaks we attribute to molecular oxygen states which we discuss later; these loss peaks are also present when ND3 is used.10 There is a substantial decrease in the intensities of the 1120 and 3390 cm-1 loss peaks, and the 300 cm-1 peak is assigned to υAg-OH present at both 110 and 210 K (Figure 5d). On warming to 260 K (Figure 5e), the only loss peak present is associated with Ag(111)-O and Ag(111)-OH, the chemisorbed oxygen and hydroxyl species, with vibrations at ∼300 cm-1. There is a clear analogy with what we reported earlier10 for the activation of water by oxygen at low temperatures at Ag(111). 3.3. Coadsorption of Oxygen and Ammonia. The exposure (500 L) of an oxygen-ammonia (100:1) mixture to a clean Ag(111) surface at 100 K (Figure 6b) resulted in an O(1s) peak with two components at binding energies of 530.3 and 532 eV, of approximately equal intensities. Further exposure (total of 1100 L) resulted in an increase in both components (Figure 6c). Under these dynamic coadsorption conditions, there is, in contrast to the interaction of oxygen with preadsorbed ammonia (Figure 4), no discrete evidence in the N(1s) spectrum for molecularly adsorbed ammonia, characterized by a binding energy of 400 eV; oxydehydrogenation is fast and efficient. The N(1s) feature at 100 K is asymmetric toward lower binding energy (spectra b and c), characteristic of NHx (x = 1 or 2) species, and this is in keeping with the hydroxyl state present with an O(1s) binding energy of 530.3 eV (Figure 6). The HREEL spectrum at this exposure had loss peaks at 250, 700, 1120, 1460, 1640, and 3390 cm-1; however, on warming to 210 K, peaks were only present at 250, 1120, and 3390 cm-1 together with the emergence of a feature at 3640 cm-1, assigned to surface hydroxyls (υOH). Although there was evidence for NH3(a) in the HREEL spectrum at 210 K, it was below the level of detection in the N(1s) spectrum. The O(1s) intensity at 530.3 eV (Figures 4 and 6) is evidence for hydroxyl species being present at 100 K, and the corresponding
Table 1. Surface Oxygen Concentrations (cm-2) as a Function of Exposure of Ag(111) to Dioxygen at Sample Temperatures of 80, 120, and 150 K exposure/L
oxygen concentration at 80 K/cm-2
20 100 1000
0.5 1014 0.8 1014 1.6 1014
16222 DOI: 10.1021/la100953m
exposure/L
oxygen concentration at 120 K/cm-2
exposure/L
oxygen concentration at 150 K/cm-2
500 1500
0.5 1014 1.4 1014
500 1500
not observed 0.2 1014
Langmuir 2010, 26(21), 16221–16225
Carley et al.
Article
Figure 2. XP and HREEL spectra for NH3 adsorption at Ag(111) surface pre-exposed to O2 at 100 K. (a) Clean surface, (b) after exposure to 5000 L O2 at 100 K, (c) ∼1 L NH3 at 100 K, and (d) after warming to 210 K. Note the absence of any surface species after warming to 210 K.
Figure 3. HREEL and XP spectra for ammonia adsorbed at Ag-1
-1
(111) at 110 K, with loss peaks at 250 cm (υAg-NH3), 1120 cm (δs(HNH)), and 3390 cm-1 (δs(NH)) and an N(1s) peak at 400 eV. Ammonia desorbs below 260 K.
HREEL spectrum has no discernible loss peak at 3640 cm-1; it is, however, present on warming to 210 K. This was also observed in Langmuir 2010, 26(21), 16221–16225
Figure 4. XP spectra N(1s) and O(1s) for (b) ammonia adsorbed at 110 K followed by (c) oxygen (20 L) and further exposure to O2(g) (120 L) and warming to (d) 220 K and (e) 260 K .
our earlier studies10 of the oxygen-water system at Ag(111) and attributed to a change in the orientation of the OH species, with the linear form being preferred at 210 K. DOI: 10.1021/la100953m
16223
Article
Carley et al.
about 1 1014 cm-2 and after 1100 L to be 1.6 1014 cm-2 (Figure 1 and Table 1). We conclude that there is a second contribution to the intensity at 532 eV which arises from a molecular oxygen state present at a concentration of 3.3 1014 cm-2 after an exposure of 500 L and 4.6 1014 cm-2 after an exposure of 1100 L. Associated with this state is the strong loss peak at 1490 cm-1 which is not associated with the peroxo state (loss peak at 700 cm-1, Figure 1). The 1490 cm-1 loss is only observed when ammonia and oxygen are both present at low temperatures, and it has also been observed in earlier studies11 of oxygen coadsorbed with ammonia (NH3 and ND3) and also at Cu(111).15 An ammonia-dioxygen surface complex, stable at 100 K, is implicated, where the oxygen is a transient O2(s), a precursor of the peroxo state. Under these high surface coverages, the dissociation reaction pathway is inhibited and the complex is trapped as a metastable “cryogenic complex”. The small loss peak observed at 1640 cm-1 is close to the υ(O-O) value of the dioxygen state reported by Otto et al.12 for dioxygen adsorbed at a silver film at 20 K, with no electrons in the antibonding Π* orbital, with a loss peak υ(O-O) observed at 1600 cm-1. The following scheme summarizes the steps involved in the oxidation of ammonia: O2 ðgÞ f O2 ðsÞ
ð1Þ
O2 ðsÞ f O2 δ - ðaÞ
ð2Þ
O2 ðsÞ þ NH3 ðaÞ f ðO2 3 3 3 NH3 Þ
ð3Þ
ðO2 ðsÞ 3 3 3 NH3 Þ f OðsÞ þ OHðaÞ þ NH2 ðaÞ
ð4Þ
OðsÞ þ NH2 ðaÞ f OHðaÞ þ NHðaÞ
ð5Þ
1 OHðaÞ f OðaÞ þ H2 ðgÞ 2
ð6Þ
Figure 5. HREEL spectra corresponding to XP spectra (c-e) in Figure 4. Losses observed at 250, 300, 700, 1120, 1490, 1640, and 3390 cm-1 at 110 K; 1120, 3390, and 3610 cm-1 (υOH) at 210 K; and 300 cm-1 at 260 K.
4. Discussion A key observation from the data described above is that the peroxo state that forms when oxygen is preadsorbed at the Ag(111) surface at low temperatures is unreactive toward ammonia whereas oxygen adsorbed in the presence of ammonia is reactive. The formation of the peroxo state has a low sticking probability and is characterized by an XP binding energy of 532 eV and a vibrational band at 700 cm-1 (Figure 1) and is unaffected by subsequent ammonia adsorption (Figure 2). In contrast, when ammonia is preadsorbed at 110 K and then exposed to oxygen, the resulting peak in the O(1s) spectrum is made up of two components with binding energies of 532 and 530.3 eV. Furthermore, by comparison with the slow rate of formation of the peroxo state (Figure 1 and Table 1), it is a kinetically fast process (Figure 4). Exposure of a clean surface to an oxygen-ammonia (100:1) mixture (Figure 6) is also kinetically fast at 100 K and results in an O(1s) peak composed of two components at 530.3 and 532 eV. In both cases, the N(1s) peak at 100 K indicates that ammonia has been oxidized, with NHx(a) (x = 1 or 2) being present with, at 210 K, the characteristic υOH loss peak at 3640 cm-1. What then is the oxygen state that contributes to the intensity of the 532 eV peak? The concentration of the peroxo state O2δ-(a) after an exposure to oxygen of 500 L would be expected to be 16224 DOI: 10.1021/la100953m
(1) (2) (3)
Accommodation of dioxygen; a transient state The peroxide route; kinetically slow Complex formation at low temperatures involving the dioxygen transient (4,5) Complex dissociation and stepwise oxy-dehydrogenation of ammonia thermodynamically driven and resulting in NHx (x = 1 or 2) and OH formation. It is a further example of precursor assisted oxygen bond cleavage. (6) Dehydroxylation at above 220 K resulting in chemisorbed oxygen; facile route to surface “oxide”
5. Summary Our early transient studies1-4 used the probe-molecule approach to explore whether “hot” oxygen adatoms were present under dynamic conditions as a result of the exothermicity associated with bond cleavage of molecules such as O2, NO, and N2O. We established that surface lifetimes of these transient oxygen-atom states were sufficiently long to be able to control catalytic oxidation (12) Pettenkoffer, C.; Pockrand, I.; Otto, A. Surf. Sci. 1983, 135, 52.
Langmuir 2010, 26(21), 16221–16225
Carley et al.
Article
Figure 6. XP and HREEL spectra for exposure of Ag(111) to an O2-NH3 (100:1) mixture (500 and 1100 L) at 100 K with loss peaks at 300, 700, 1120, 1460, 1640, and 3390 cm-1; at 210 K with loss peaks at 250, 1120, 3390, and 3640 cm-1. O(1s) and N(1s) spectra: (a) clean surface, (b,c) after exposures of (b) 500 L and (c) 1100 L, and (d) after warming to 210 K.
pathways at magnesium, copper, and aluminum surfaces.1,5,13,16 At a Zn(0001) surface, on the other hand, where oxygen bond cleavage is slow,7 a transient dioxygen state was shown to complex with ammonia to provide a low energy pathway to oxygen bond cleavage. The rate of reaction was 300 times faster than that in the absence of ammonia. A pyridine-oxygen complex was similarly effective.14 Ag(111) offered an opportunity through the availability of a kinetic window to explore this concept further through a combination of XPS and HREELS at cryogenic temperatures. A dioxygen-ammonia complex, with the oxygen being a transient precursor of the stable peroxo state, was shown through analysis of the O(1s) and N(1s) spectra supported by HREELS to provide a low energy kinetic pathway to dioxygen bond cleavage, and hydroxyl and NHx formation. To confirm the mechanism, atom-resolved evidence from STM for the formation of the dioxygen-ammonia complex at cryogenic temperatures is a further objective. At Cu(110), it was direct evidence from STM for the role of disordered mobile oxygen states that confirmed the earlier XPS evidence.4,5,8,17 There is also a strong similarity between the oxidation reactions at low temperatures at zinc, silver, and magnesium surfaces5b,8 (13) Afsin, B.; Davies, P. R.; Pashuski, A.; Roberts, M. W.; Vincent, D. Surf. Sci. 1993, 284, 189. (14) Carley, A. F.; Roberts, M. W.; Yan, S. Catal. Lett. 1988, 1, 265. (15) Davies, P. R.; Roberts, M. W.; Shukla, N.; Vincent, D. J. Surf. Sci. 1995, 325, 50. (16) Carley, A. F.; Roberts, M. W. Chem. Commun. 1987, 355. (17) Carley, A. F.; Davies, P. R.; Kulkarni, G. U.; Roberts, M. W. Catal. Lett. 1999, 58, 33.
Langmuir 2010, 26(21), 16221–16225
and also with the homogeneous Gif reaction for the oxidation of hydrocarbons. Zinc was one of the first metals to be effective in the Gif reaction where molecular oxygen is the oxidant and pyridine is present as a solvent.18 At Zn(0001) surfaces, it has been shown that, in the presence of pyridine, oxygen is activated as O2δ-, resulting in an enhancement of the rate of zinc oxidation.14 The model proposed for the latter case involves pyridine acting as a Lewis base in a three-centered mechanism;14,19 a similar mechanism involving an ammonia-dioxygen complex that is a precursor of the peroxo state would account for the present results. Catalysis is a dynamic process, and it was with this in mind that we explored whether transient oxygen states were present in the chemisorption of oxygen at metal surfaces with established methods in surface science.1-4,8 Although the studies at copper, zinc, magnesium, and silver surfaces have been at low pressure and temperatures, there is a view that, provided the reactions are not kinetically controlled, they could conceivably be considered as mimicking reactions at higher temperatures and pressures. Acknowledgment. We are grateful to EPSRC and ICI for their support over many years of our studies in surface chemistry and catalysis. (18) Balavoine, G.; Barton, D. H. R.; Geletii, Y.; Hill, D. R. In The activation of dioxygen and homogeneous oxidation; Barton, D. H. R., Martell, A. E., Sawayer, D. T., Eds.; Plenum Press: New York, 1993; p225, ISBN 10: 0306445913. (19) Roberts, M. W. Chem. Soc. Rev. 1989, 18, 451.
DOI: 10.1021/la100953m
16225
