Reactivity of a Hydroxylated Alumina Surface in the Presence of NO

Feb 6, 2008 - Laboratoire de Re´actiVite´ de Surface, UMR CNRS 7609, UniVersite´ ... 2 × 105 Pa, RT) with a hydroxylated alumina surface was inves...
1 downloads 0 Views 367KB Size
2964

J. Phys. Chem. C 2008, 112, 2964-2971

Reactivity of a Hydroxylated Alumina Surface in the Presence of NO Diluted in N2: A PM-IRRAS in Situ Investigation A. Delebecque,*,†,‡ C. Thomas,† C.-M. Pradier,*,† C. Methivier,† E. Coffre,‡ H. Paoli,‡ and M. Carre‡ Laboratoire de Re´ actiVite´ de Surface, UMR CNRS 7609, UniVersite´ Pierre et Marie Curie-Paris 6, 4 Place Jussieu, 75005 Paris, France, and Centre de Recherche Claude Delorme, Air Liquide, 1 Chemin de la Porte des Loges Les-Loges en Josas 78350 Jouy-en-Josas, France ReceiVed: July 5, 2007; In Final Form: NoVember 12, 2007

The interaction of NO (900 ppm NO/N2, 2 × 105 Pa, RT) with a hydroxylated alumina surface was investigated by in situ polarization modulation infrared reflexion absorption spectroscopy (PM-IRRAS). The resulting spectra reveal two distinct periods, namely 0-1 day and 1-13 days of exposure. During the first period, NO exposure primarily leads to the formation of nitrites (1523, 1470, 1330, 1235, and 1130 cm-1), HONO (1235 cm-1), and perhaps also some nitrates (1650-1620 cm-1) and hyponitrites (1340 cm-1). After one day of exposure, the transformation of nitrites into nitrates is demonstrated by the decrease in intensity of the band at 1235 cm-1, attributed to the ν1 stretching vibration of bridging bidentate nitrites and also by the presence of an isosbestic point at 1260 cm-1. Various nitrate species including monodentate, bidentate, bridged, and water-solvated nitrates (1600-1280 cm-1) are then formed on the surface. The stability of the formed ad-NOx species during N2 purges, NO/N2, O2 or H2O/O2 exposures was also investigated in detail in this study. These experiments revealed an equilibrium at the surface between water-solvated and oxide-coordinated nitrates.

Introduction When properly alloyed, aluminum exhibits improved mechanical properties which makes it an ideal candidate for numerous industrial applications including the storage of gases at high pressure (150 × 105 Pa).1 In the case of the storage of reactive gas mixtures such as nitrogen monoxide (NO) in nitrogen (N2), the gas-phase reactive molecules may interact with the inner surfaces of storage cylinders leading to the undesired and uncontrolled formation of adsorbed species that may also desorb. Since the quantities of the newly formed species may not be negligible with respect to the nominal concentration of the mixture, the risk of being out of specifications is high, if the preparation of the cylinder surface is not correct. The characterization of the species formed under NO/ N2 exposure on the inner surfaces of cylinders is therefore of the utmost interest to develop passivation procedures that could guarantee gas storage at the nominal concentration. Nevertheless, understanding the gas surface interaction is difficult partly because the inner surface of the aluminum alloy cylinder has a complex composition and also due to the difficulty of characterizing the gas-cylinder interface. Aluminum exposed to ambient air at room temperature is known to be rapidly covered with a thin layer (18 MΩ) for 2 min and dried under nitrogen for a few minutes. Gas Distribution Panel and Methods. The gases (Air Liquide) had the following purity classes: O2 (2 ppm H2O), NO (891 ( 12 ppm) in N2 (O2 < 10 ppb, H2O ) 0.2 ppm) and N2. Residual traces of H2O in N2 and O2 or of O2 in N2 were removed by a molecular sieve or by a deoxo filter (Altech), respectively. The gas flow was injected perpendicularly to the sample. First, the sample was flushed with N2 at a flow rate of 8.33 × 10-6 m3/s at room temperature for 30 min. The sample was then annealed under 3.33 × 10-6 m3/s of N2 at 443 K for 1 h, before cooling to 298 K. N2 was then introduced with a total pressure of 2 × 105 Pa (2 bar) for 16 h. The pressure was then decreased to 1 × 105 Pa and the NO/N2 mixture was introduced in the analysis chamber with a flow rate of 1.7 × 10-7 m3/s for 2 h to purge the chamber and then stored with a total pressure of 2 × 105 Pa for 13 days. In Situ PM-IRRAS Analysis. The stainless steel analysis chamber was equipped with two ZnSe windows enabling the beam to pass through the chamber. The sample in the analysis chamber was placed in the external beam of a Fourier transform infrared Spectrometer (Nicolet 5700) at grazing incidence, and the reflected light was focused onto a liquid-nitrogen-cooled MCT detector. A ZnSe polarizer and a ZnSe photoelastic modulator, allowing modulation of the polarization of the incident beam between parallel and perpendicular polarizations (HINDS Instruments, PEM 90, modulation frequency 36 kHz), were placed prior to the sample. The sum and difference interferograms were generated by a two channel electronic device. Theses interferograms undergo Fourier transformation and their ratio is proportional to the differential reflectivity ∆R/R. The PM-IRRAS setup and the experimental procedure have been described in detail elsewhere.23 All reported spectra were recorded with a resolution of 8 cm-1 by addition of 128 scans. This polarization modulation enables us to perform analyses of the surface sample in situ under controlled atmosphere, without purging the atmosphere or requiring reference spectra. The curve-fitting of the PM-IRRAS spectra was accomplished using the Origin 7.1 software (Origin Lab Corporation). XPS. XPS spectra were collected on a SPECS (Phoibos MCD 150) X-ray photoelectron spectrometer, using a Mg KR (hν ) 1253.6 eV) X-ray source having a 150 W (12 mA, 12.5 kV) electron beam power and a 7 × 2 mm spot size. The emission of photoelectrons from the sample were analyzed at a takeoff angle (θ) of 90° under ultrahigh vacuum conditions (10-8 Pa). High-resolution spectra were collected at a pass energy of 10

Figure 2. PM-IRRAS spectra of an Al sample exposed to ambient atmosphere after (a) polishing, (b) exposure to boiling water for 2 min, and (c) heating at 443 K and cooling to room temperature.

eV for C1s, O1s, Al2p, and N1s core XPS levels. No charge compensation was applied during acquisition. After collection, the binding energies were calibrated with respect to either the C-C/C-H components of the C1s peak at a binding energy of 284.8 eV24 or the Al0 component of the Al2p peak at a binding energy of 73 eV. The peak areas of the spectra were determined after subtraction of a Shirley background. The atomic ratio calculations were performed after normalization using Scofield factors.25 All spectra processing were carried out using the Casa XPS software package and Origin 7.1 (Origin Lab Corporation). The spectral decomposition was performed by using Gaussian functions after background subtraction. Results and Discussion Surface Characterization of the Samples Prior to NO/N2 Exposure. The PM-IRRAS spectrum of a sample exposed 1 day to ambient atmosphere shows bands of weak intensity at about 950, 1460, 1600, and 2930 cm-1, and a broad contribution in the 3600-2500 cm-1 region (Figure 2, spectrum a). After exposure to boiling water for 2 min, the spectrum of the hydroxylated sample exhibits an intense band at 1072 cm-1 and three bands of weaker intensities at about 1400, 1640, and 3400 cm-1 with a shoulder at around 3250 cm-1 (Figure 2, spectrum b). After annealing at 443 K, the broad band peaking at 3400 cm-1 and that at 1640 cm-1 decrease significantly and new maxima are revealed at 3666, 3300, and 3250 cm-1. In addition, the bands at 1600 and 1460 cm-1 are clearly reappearing (Figure 2, spectrum c). The band at 950 cm-1 is attributed to the Al-O-Al stretching vibration of the amorphous aluminum oxide layer.26 A limited contamination by carbonates, (νsym(OCO) at 1460 and νasym(OCO) at 1600 cm-1) and hydrocarbons (ν(CH) at 2930 cm-1) is also observed on the sample. The broad feature observed in the 3600-2500 cm-1 region is attributed to the ν(OH) vibrations of hydroxyl groups and adsorbed water on alumina. The presence of adsorbed water is corroborated by the existence of a band at 1640 cm-1 (δ(H2O)).27,28 The intense band at 1072 cm-1 on the hydroxylated sample is characteristic of the Al-OH hydroxyl bending mode δ(OH) of Al oxyhydroxide.4,26-30 The broad asymmetric band between 3700 and 2500 cm-1, is composed of an absorption contribution with a maximum at about 3450 cm-1, due to the stretching vibration of the interlamellar water in the oxide structure,28,30 and a shoulder at lower frequency, 3250 cm-1,4,28,30 which is attributed to the ν(OH) vibration of bulk hydroxyls in the pseudo-boehmite layer. The rather low frequency of this ν(OH)

Reactivity of a Hydroxylated Alumina Surface

J. Phys. Chem. C, Vol. 112, No. 8, 2008 2967

Figure 3. XPS spectra of a non-hydroxylated (Al, -O-) and a hydroxylated (Al-OH, -) samples. The C1s core level (d) has been decomposed into five contributions, as recommended by Alexander et al.32

vibration band, compared to that reported earlier on Al2O3,28 indicates significant hydrogen bonding of the hydroxyls in the oxide layer. The contribution at about 3450 cm-1 has an associated bending vibration δ(H2O) at around 1630-1640 cm-1,27,28 as shown on spectrum b of Figure 2. When annealing at 443 K (Figure 2, spectrum c), the decrease of the bands peaking at 3400 and 1640 cm-1 indicates a partial removal of the adsorbed water. In agreement with this, a band appears at 3666 cm-1 upon annealing, which can be ascribed to surface OH groups not involved in H-bonding.27 On Al2O3, Raybaud et al.31 estimated that the stretching vibration of isolated surface OH groups appeared between 3800 and 3660 cm-1 indeed. The XPS spectra of nonhydroxylated (air-passivated) and hydroxylated samples shown in Figure 3 were calibrated against the lowest binding energy contribution attributed to metallic Al (Al0) at 73 eV for the non-hydroxylated sample or to oxidized Al (Al3+) at 75.6 eV for the hydroxylated sample. The Al2p core level of the non-hydroxylated sample exhibits two peaks at 73 and 75.6 eV, whereas only one peak is observed at 75.6 eV for the hydroxylated sample (Figure 3a). As only oxidized aluminum (Al3+) is observed on the Al2p core level of the hydroxylated sample, the formed oxide layer must have a thickness greater than the XPS analysis depth (i.e., 8-10 nm). The C1s and the O1s core levels are also shown in Figure 3 (b, c) for the hydroxylated and non-hydroxylated samples, and the corresponding atomic percentages are listed in Table 1. Figure 3 and Table 1 clearly show that the method used in this work to increase the OH coverage of the sample also leads to a drastic decrease of the carbon contamination (Figure 3c), as the C1s contribution decreases from 37 at. % on the nonhydroxylated sample to 5 at. % on the hydroxylated one (Table 1). The estimation of the amount of oxygen involved in the contaminating carbon (O1scont., Table 1) was done according to the decomposition of the C1s peak into 5 contributions (Figure 3d) at 284.8, 285.1, 286.4, 287.6, and 289.0 eV (( 0.2 eV) related to C-C/C-H, C-C(dO)OX, C(-O)2, C-O, and C-C(dO)OX, respectively, as reported previously.32 The values of the O1s contamination (involved in C-O or CdO bonds) are estimated to be 9.1 and 2.4 at. % for the nonhydroxylated

TABLE 1: XPS Analyses of a Nonhydroxylated and a Hydroxylated Alumina Samples atomic % nonhydroxylated hydroxylated binding fwhm/ sample sample energy/eV eV C1s C-C/C-H C-C(dO)OX C-C(dO)OX C-O C(-O)2 O1s O1sCont. Al2p Al0 Al3+

37 28.7 2.3 2.3 3.0 0.8 37 9.1 26 6.2 19.8

5 2.8 0.8 0.8 0.4 0.2 64 2.4 31 0 31

284.8 285.2 289.0 286.4 287.6

1.5 1.9 1.5 1.5 1.5

531.7

2.9

71.2 74.2

1.8 2.1

and the hydroxylated samples, respectively. The ratios of the oxygen bonded to Al3+, expressed as the total amount of oxygen (O1s) minus that involved in carbon contamination (O1scont.), over the amount of Al3+ are therefore 1.4 and 2.0 for the nonhydroxylated and the hydroxylated samples, respectively. These ratios are consistent with the expected stoichiometry of alumina (Al2O3) and boehmite (AlOOH). In summary, PM-IRRAS and XPS analyses show that exposure of the Al sample to boiling water for 2 min leads to a growth of the oxide layer, to the formation of a pseudoboehmite (AlOOH) layer having a thickness greater than 8-10 nm and to a significant decrease in the amount of contaminating carbon. The annealing treatment removes part of the adsorbed water and allows to obtain a reproducible surface. Exposure of the Hydroxylated Alumina Sample to the NO/ N2 Mixture. After annealing, the sample was cooled to room temperature under a flow of N2 and stored under a 2 × 105 Pa pressure for 5 days. Under these conditions, no significant deviation of the IR baseline could be observed. N2 was then removed and the NO/N2 mixture was introduced into the analysis chamber. Figure 4 shows the spectral evolution of the hydroxylated alumina sample exposed to the NO/N2 mixture (900 ppm) at RT under a 2 × 105 Pa total pressure for 13 days. This evolution can be divided into two periods, namely 0-1 day and 1-13 days of exposure.

2968 J. Phys. Chem. C, Vol. 112, No. 8, 2008

Delebecque et al.

Figure 4. PM-IRRAS spectra of a hydroxylated sample exposed to a NO/N2 mixture (891 ( 12 ppm at 2 × 105 Pa). Spectra were ratioed to a reference spectrum under N2. Spectrum after (a) 30 min of exposure, (b) 3 h, (c) 1 day, (d) 1.5 days, (e) 2 days, (f) 4 days, (g) 8 days, and (h) 13 days.

During the first day of exposure to NO/N2, (insert of Figure 4), several absorption bands are observed between 1700 and 1000 cm-1. This broad continuum of bands presents maxima at 1235, 1340, 1523 and 1640 cm-1 which increase continually up to 1 day of exposure to NO/N2 (Figure 4 a, b and c). In the OH stretching region, the exposure of the hydroxylated sample to the NO/N2 mixture results in a slight increase of the broad band at 3600-2500 cm-1 and a slight decrease of that peaking at 3700 cm-1. The band observed at 1235 cm-1 is clearly attributed to the ν3 asymmetric stretch vibration of bridged nitrites.6-8 As mentioned by Bo¨rensen et al.,8 the free nitrite ion (point group C2V) has three fundamental vibrations: ν1 (symmetric stretch), ν2 (bending vibration) and ν3 (asymmetric stretch). The two last vibrations are infrared active, and the typical associated frequencies are 806 (ν2) and 1286-1260 cm-1 (ν3).12 When coordinated to a metal cation, the ν1 symmetric stretch becomes IR active and three main configurations may be observed: linear nitrites (Al-O-N-O), bridged nitrites ((AlO)2N) or nitro compounds (Al-N(O)O-Al). IR frequencies of these different configurations are reported in Figure 1. As shown by Westerberg et al.,6 the ν3 and ν1 vibrations of bridged nitrites adsorbed on aluminum oxide, are observed at 1320 and 1230 cm-1, respectively. The presence of other bands at 1340, 1470, and 1200-1000 and at 1523 and 1130 cm-1 may be ascribed to hyponitrite (Al+NO-), to linear nitrites and to nitro compounds, respectively.6 Frequencies of bands at around 1620-1650 cm-1 that are too high to be attributed to nitrites are most probably due to the ν3as stretching vibration of surface nitrates7 or to the bending vibration δ(H2O) of adsorbed water. The evolution of the ν(OH) absorption band reveals an increase in the H-bonded OH groups and a decrease in the free OH groups and may be explained by the formation of the adNOx species at the same site or in the vicinity of surface hydroxyls, as previously mentioned by Hadjiivanov et al.7 in the case NO adsorption on activated γ-Al2O3. The observed evolution may also indicate the formation of HONO interacting with the surface OH groups. The ν1(O-N-O) vibration of HONO is also reported at 1231 cm-1 and cannot be easily discriminated from that of bridged nitrites.8 These results show that during the first period, NO exposure leads to the formation of mainly nitrites (1523, 1470, 1330, 1235, and 1130 cm-1) with HONO (1235 cm-1), a few nitrates (1650-1620 cm-1), and hyponitrites (1340 cm-1). These results

Figure 5. Evolution of the ad-NOx species followed by PM-IRRAS: (A) spectra after 13 days of exposure to NO/N2, (B) N2 purge (spectra were ratioed with respect to the spectrum obtained after 13 days of exposure to NO; (C) NO/N2 exposure: 900 ppm, 2 × 105 Pa, 20 h of storage, spectra were ratioed with the spectra obtained after the N2 purge; (D) N2 purge, spectra were ratioed with the spectra obtained after 20 h of storage in NO/N2; (E) H2O-free O2 exposure: 2 × 105 Pa, 2 h storage, spectra were ratioed with the spectra obtained after the N2 purge; (F) O2 (2 ppm H2O) exposure: 2 × 105 Pa, 2 h storage, spectra were ratioed with the spectra obtained after the H2O-free O2 exposure.

are consistent with those obtained upon NO adsorption on alumina powders at RT.7,13,16 From 1 to 13 days of exposure to NO/N2 (Figure 4, spectra c-h), a large change in the IR signal is observed in the 17001000 cm-1 absorption region. The broad band between 1700 and 1260 cm-1 grows significantly with the appearance of two very strong contributions peaking at 1340 and 1420 cm-1. A shoulder can also be seen at 1520 cm-1 and a broad band of a weaker intensity is observed at 1620 cm-1. The band at 1235 cm-1 exhibits a completely different behavior. This latter band decreases for times of exposure longer than 1 day, whereas the contributions at 1340 and 1420 cm-1 start growing. Another interesting point is the presence of an isosbestic point appearing at 1260 cm-1 in between the 1235 and 1341 cm-1 contributions. In the OH stretching region, the evolution is similar to that observed during the first day of exposure. In the 2200-2000 cm-1 region, no contribution was observed up to 13 days of exposure to NO/N2. Study of the Stability of the ad-NOx Formed Species. The changes observed between one to 13 days of exposure are complex since part of the bands decreases, whereas others increase. From Figure 1, the presence of two strong bands within the 1600-1280 cm-1 absorption region may be due to nitrate or to nitro/nitrite compounds.7,33 To help the identification, the stability of the formed ad-NOx species was investigated under various gaseous atmospheres consisting of N2, NO/N2 and O2. After exposure of the hydroxylated sample for 13 days to NO/N2 (Figure 5A), the sequence of experiments consisted of (B) a purge under N2 (3.3 × 10-6 m3/s, 1 h), (C) NO/N2 exposure (900 ppm, 2 × 105 Pa, 19 h), (D) a purge under N2 (3.3 × 10-6 m3/s, 1 h), (E) exposure to H2O-free O2 (2 × 105 Pa, 2.5 h), and (F) exposure to O2 containing 2 ppm of H2O(2 × 105 Pa, 2.5 h). Under N2 (Figure 5B), the intensities of the groups of bands between 1440 and 1330 cm-1 and 1700-1600 cm-1 decrease.

Reactivity of a Hydroxylated Alumina Surface Simultaneously, the bands peaking at 1310 and 1525 cm-1 increase. It is noteworthy that two isosbestic points at 1320 cm-1 and 1480 cm-1 are observed, indicating the transformation of some species to others. The spectral changes observed during a subsequent NO/N2 exposure (Figure 5 (C)) are opposite to those shown in Figure 5B. The equilibrium is almost reached after 1 h under a N2 purge (Figure 5B), whereas it is still not reached after 19 h in the case of a repeated exposure to NO/N2 (Figure 5C). Under a subsequent N2 purge (Figure 5D), the observed spectral changes are similar to those observed during the first N2 purge (Figure 5B). It must be noted that the formed ad-NOx species are unaffected upon exposure to H2O-free O2 (Figure 5E). On the contrary, when trace amounts of water (2 ppm) are present in O2, bands at 1440-1330 and 1700-1600 cm-1 increase and bands peaking at 1525 and 1310 cm-1 decrease (Figure 5F). Even if these changes are comparable to those found upon NO/ N2 exposure, their kinetics are clearly different, the equilibrium being reached after about 1 h of exposure to H2O-O2 instead of more than 19 h during NO/N2 storage as shown in Figure 5. The two isosbestic points at 1320 cm-1 and 1480 cm-1 are also present upon H2O-O2 exposure. In the OH stretching region (Figure 5), a decrease of the broad band at around 3600-3200 cm-1 and a slight increase of a band at 3740-3700 cm-1 upon dry nitrogen purges (Figure 5B,D) are observed. Additional exposure to NO/N2 leads to a slow increase of the broad band at 3600-3200 cm-1 and a slight decrease of that around 3740-3700 cm-1. No change appears in the OH stretching region upon H2O-free O2 exposure (Figure 5E). In contrast, the presence of a small amount of water in O2 leads to a quick increase of the bands between 3600 and 3200 cm-1 (Figure 5F). The changes observed under N2, NO/N2, and O2 clearly highlight the distinction between the groups of bands in the 1600-1470 and 1310-1280 cm-1 regions, from those in the 1440-1400 and 1360-1330 cm-1 regions. Since no change was observed during H2O-free O2 exposure and because nitrates do not react with O2,7,33 the bands appearing from 1 to 13 days of exposure to NO exposure are attributed to the formation of various types of nitrates (oxide-coordinated and water-solvated nitrates). It is known from the literature that nitrate ions have four fundamental vibrations: ν1 (symmetric stretch, 1050 cm-1),12 ν2 (bending out of plane, 831 cm-1), ν3 (asymmetric stretch, 1405 cm-1), and ν4 (bending in plane, 692 cm-1), with ν2, ν3, and ν4 being IR active. Nitrate adsorption induces a loss of degeneration of the ν3 stretch vibration. On Al2O3, Westerberg et al. estimated ∆ν3 to be 247 and 362 cm-1 for monodentate nitrates and bridged nitrates, respectively.6 The first group of bands (1600-1470 and 1310-1280 cm-1) has a ∆ν3 between 160 and 320 cm-1, which is a fingerprint of oxide-coordinated nitrates.6 In contrast, the second group of bands (1440-1400 and 1360-1330 cm-1) exhibits a much smaller ∆ν3, ranging from 40 to 110 cm-1. These values are close to those reported for nitrate salts (such as NaNO3: ∆ν3 ) 69 cm-1)15 or, more likely, of solvated nitrates (OH‚‚‚HNO3: ∆ν3 ) 50-115 or 112 cm-1).8,19 This latter assignment is also supported by the experiment in which H2O-O2 was contacted with a NOx saturated sample (Figure 5F). Thus, the first group of bands (1600-1470 and 1310-1280 cm-1) may be attributed to oxidecoordinated nitrates including bridged, bidentate and monodentate nitrates, whereas the second group of bands (1440-1400 and 1360-1330 cm-1) may be assigned to water-solvated

J. Phys. Chem. C, Vol. 112, No. 8, 2008 2969 nitrates. The spectral changes shown in Figure 5F and particularly the presence of the two isosbestic points at 1320 and 1480 cm-1, which has never been reported previously to our knowledge, clearly illustrate the transformation of oxidecoordinated into water-solvated nitrates. This transformation may be explained by the presence of an equilibrium between nitrates and adsorbed water, as also suggested by Bo¨rensen et al.8

{(AlO)+ (NO3 )} + H2O S {AlOOH‚‚‚HNO3}

(1)

The introduction of water shifts the equilibrium to the right and leads to the formation of {AlOOH‚‚‚HNO3} on the surface. The evolutions observed in Figure 5B,D during N2 exposure clearly reveal the reversibility of the equilibrium (1). The changes observed in Figure 5C during NO/N2 exposure are similar to those observed during H2O-O2 exposure. It must be reminded that the analysis of the NO/N2 mixture showed the presence of trace amounts of water below 0.2 ppm. This may explain first that part of the observed evolution is likely due to the small amount of water present in NO/N2 mixture and second that the different kinetics observed between Figure 5C,F are due to a difference in water concentration. A limited adsorption of NO on nonreacted OH groups may also be considered. NO adsorption may then lead to the release of traces of water that may react with adsorbed nitrates. It cannot be excluded formally, however, that part of the spectral changes reported in Figure 5F may also be due to the transformation of bidentate to monodentate nitrates, as recently reported by Szanyi et al.19 and Angelini et al.34 Decomposition of Spectra Obtained during NO Exposure. It seemed of interest to further analyze the successive steps of nitrates formation upon the first NO exposure (Figures 4 and 5A). A curve fitting-procedure using Gaussian curves was used to decompose all of the overlapping bands in the 1700-1200 cm-1 region. The initial iteration was performed using an initial fwhm (full width at half-maximum) of 30 cm-1 (and with a maximum of 40 cm-1 constrained). After optimization of the band position and the fwhm by taking into account spectra recorded after various times of exposure to NO/N2, these parameters were kept constant for all spectra. The final fit consists in 18 bands between 1200 and 1700 cm-1. As an example, Figure 6A shows the spectral decomposition of the PM-IRRAS spectra recorded after 13 days of exposure to NO/ N2. By considering five absorption regions included between 1600 and 1200 cm-1, as already discussed in the above sections, the calculated contributions could be reasonably gathered as follows: 1600-1470, 1440-1400, 1360-1330, 1310-1280, and 1240-1200 cm-1. The integrated areas of these different ensembles were then plotted as a function of the time of exposure (Figure 6B). This figure shows the predominance of bands attributed to water-solvated nitrates (1440-1400 and 1360-1330 cm-1) as the time of exposure increases. Figure 6B also clearly shows the singular evolution of the band peaking at 1235 cm-1, attributed to the formation of nitrites. The integrated area of this band reaches a maximum after 1 day of exposure to NO/N2 and then vanishes after about 7 days. Such a typical behavior indicates that nitrites behave as intermediate species. This phenomenon has been reported by Bo¨rensen et al.8 when studying NO2 adsorption on alumina8 but not for NO adsorption on alumina (as we may know). From this, the presence of traces of O2 or NO2 cannot be excluded formally.

2970 J. Phys. Chem. C, Vol. 112, No. 8, 2008

Delebecque et al.

Figure 6. (a) PM-IRRAS spectrum recorded after 13 days of exposure to NO/N2. Gaussian functions were used to fit the ad-NOx IR bands. (b) PM-IRRAS band areas as a function of time of exposure.

Differences must be pointed out between our work and that of Bo¨rensen et al.8 These authors proposed a mechanism including the adsorption of NO2 on surface OH groups, characterized by the presence of a band at 1690 cm-1, followed by the disproportionation of NO2 molecules resulting in the formation of oxide-coordinated nitrates and nitrites plus water. These authors attributed the decrease in the nitrite contribution to a recombination of the latter species with adsorbed water to form adsorbed nitrous acid (HONO), which was easily desorbed under flowing conditions. In the present study, however, the band at 1690 cm-1, attributed to NO2 adsorption on aluminum OH groups,8 is not observed as clearly as in the study of Bo¨rensen et al.8 Second, the decrease in the band peaking at 1235 cm-1 in the present study is correlated to the presence of an isosbestic point at 1260 cm-1, which was not observed in the study of Bo¨rensen et al.8 It seems thus difficult to attribute the decrease in the 1235 cm-1 band to the release of HONO in the gas phase, since the isosbestic point clearly indicates the transformation of the formed nitrites species into other species. To summarize, from one to 13 days of exposure to NO/N2, the IR spectral changes indicate unambiguously (i) the conversion of nitrites into nitrates, (ii) the formation of at least two different groups of nitrates attributed to oxide-coordinated and water-coordinated nitrates and (iii) the predominance of the bands attributed to water-coordinated nitrates with increasing times of exposure. XPS Analysis of the ad-NOx Species. Prior to XPS analyses, PM-IRRAS measurements were carried out to confirm that adNOx species were not affected by exposure to ambient air (not

Figure 7. N1s core levels of a hydrated alumina sample before (a) and after (b) NO/N2 exposure.

shown). The XPS analyses of the NO-reacted sample and a freshly hydroxylated sample are shown in Figure 7. The binding energy was calibrated against the Al3+ component of the Al2p peak at a binding energy of 75.6 eV. Figure 7 shows that the N1s signal was detected only on the sample exposed to the NO/N2 mixture. A difficulty encountered in the interpretation of the N1s binding energy is its susceptibility to radiation damage. Exposure of the sample to X-rays leads to a vanishing of the signal. Beard et al.35 previously mentioned this phenomenon in the case of cellulose nitrates. Although the N1s signal was recorded before the C1s, O1s, and Al2p core levels, the N1s signal was no longer observed after a few minutes of analysis. In the first 2 min of analysis, Figure 7B shows that two N1s peaks are observed at binding energies of 408 and 405 eV. From previous studies,36-39 the more intense N1s peak at 408 eV can be attributed to nitrogen from NO3-, whereas that at 405 eV

Reactivity of a Hydroxylated Alumina Surface can be attributed to nitrogen from NO2-. The presence of nitrites may be assigned to the reduction of nitrates upon X-ray exposure. These results thus support the attribution of the main IR bands observed between 1600 and 1200 cm-1 to nitrates. Conclusion The PM-IRRAS technique has been used to study in situ the interaction between a NO/N2 mixture (900 ppm NO in N2, 2 × 105 Pa) and a hydrated alumina surface at room temperature up to 13 days. The resulting spectra reveal a two steps mechanism. Up to 1 day of exposure to NO, the species formed on the hydrated alumina surface are attributed mainly to nitrite species with the growth of an absorption band at 1235 cm-1. For longer NO contact times, the transformation of nitrites into nitrates is demonstrated unambiguously by the decrease in intensity of the band at 1235 cm-1 and the increase of a broad contribution at 1700-1300 cm-1, together with the presence of an isosbestic point at 1260 cm-1. The adsorbed species formed after 13 days of exposure to NO are attributed to various nitrate species including monodentate, bidentate, bridged and water-solvated nitrates. The formation of nitrates is further supported by XPS data. The exposure of the NOx-saturated hydrated alumina surface to various gaseous atmospheres also led us to conclude that water-solvated nitrates were in equilibrium with the oxidecoordinated ones, depending on the presence of water in the gas phase. The existence of two isosbestic points at 1480 and 1320 cm-1, which are reported for the first time, clearly supports such a proposal. This study thus also allows to delimit ranges of wavenumbers for which the symmetric and asymmetric ν3 stretching vibrations of the water-solvated nitrates are included between 1480 and 1320 cm-1, whereas the symmetric or asymmetric ν3 stretching vibrations of the oxide-coordinated nitrates are observed at frequencies higher than 1480 cm-1 or lower than 1320 cm-1, respectively. Acknowledgment. We gratefully acknowledge the Association Nationale de la Recherche Technique (Grant 2005/C05058) and Air Liquide for supporting this research. We thank Dr. T. Jacksier and Dr. S. Moreau from Air Liquide for fruitful discussions. References and Notes (1) Reboul, M. Techniques de l’inge´ nieur 2005, COR 325, 1. (2) Safrany, J. S. Techniques de l’inge´ nieur 2001, M 1 630, 1. (3) Vedder, W.; Vermilyea, D. A. Trans. Faraday Soc. 1969, 65, 561. (4) Alwitt, R. S. Oxides and oxide films; Diggle, J. W., Ed.; Dekker: New York, 1976; Vol. 4. (5) Van den Brand, J.; Van Gils, S.; Beentjes, P. C. J.; Terryn, H.; de Wit, J. H. W. Appl. Surf. Sci. 2004, 235, 465.

J. Phys. Chem. C, Vol. 112, No. 8, 2008 2971 (6) Westerberg, B.; Fridell, E. J. Mol. Catal. A: Chem. 2001, 165, 249. (7) Venkov, T.; Hadjiivanov, K.; Klissurski, D. Phys. Chem. Chem. Phys. 2002, 4, 2443. (8) Bo¨rensen, C.; Kirchner, U.; Scheer, V.; Vogt, R.; Zellner, R. J. Phys. Chem. A 2000, 104, 5036. (9) Centi, G.; Perathoner, S.; Biglino, D.; Giamello, E. J. Catal. 1995, 151, 75. (10) Schauermann, S.; Johanek, V.; Laurin, M.; Libuda, J.; Freund, H.J. Chem. Phys. Lett. 2003, 381, 298. (11) Prinetto, F.; Ghiotti, G.; Nova, I.; Lietti, L.; Tronconi, E.; Forzatti, P. J. Phys. Chem. B 2001, 105, 12732. (12) Hadjiivanov, K. Catal. ReV., Sci. Eng. 2000, 42, 71. (13) Hoost, T. E.; Otto, K.; Laframboise, K. A. J. Catal. 1995, 155, 301. (14) Szanyi, J.; Kwak, J. H.; Kim, D. H.; Burton, D. S.; Peden, H. F. C. J. Phys. Chem. B 2005, 109, 27. (15) Goodman, A. L.; Miller, T. M.; Grassian, V. H. J. Vac. Sci. Technol. A 1998, 16, 2585. (16) Pozdnyakov, D. V.; Filimonov, V. N. AdV. Mol. Relax. Proc. 1973, 5, 55. (17) Miller, T. M.; Grassian, V. H. Geophys. Res. Lett. 1998, 25, 3835. (18) Pozdnyakov, D. V.; Filimonov, V. N. Kinet. Katal. 1973, 14, 760. (19) Szanyi, J.; Kwak, J. H.; Chimentao, R. J.; Peden, C. H. F. J. Phys. Chem. C 2007, 111, 2661. (20) Knoezinger, H.; Ratnasamy, P. Catal. ReV. - Sci. Eng. 1978, 17, 31. (21) Ozensoy, E.; Peden, C. H. F.; Szanyi, J. J. Phys. Chem. B 2005, 109, 15977. (22) Ozensoy, E.; Peden, C. H. F.; Szanyi, J. J. Phys. Chem. B 2006, 110, 8025. (23) Buffeteau, T., Ph.D. dissertation, Universite´ de Bordeaux I, France, 1988. (24) Barr, T. L.; Seal, S.; Chen, L. M.; Kao, C. C. Thin Solid Films 1994, 253, 277. (25) Duc, T. M. Techniques de l’inge´ nieur 1998, Traite´ d’analyse et caracte´ risation; ETI Sciences & Techniques: Paris; P 2 625, 1. (26) Van Gils, S.; Melendres, C. A.; Terryn, H. Surf. Int. Anal. 2003, 35, 387. (27) Morterra, C.; Emanuel, C.; Cerrato, G.; Magnacca, G. J. Chem. Soc. Faraday Trans 1992, 88, 339. (28) Van den Brand, J.; Blajiev, O.; Beentjes, P. C. J.; Terryn, H.; deWit, J. H. W. Langmuir 2004, 20, 6308. (29) Kiss, A. B.; Keresztury, G.; Farkas, L. Spectrochim. Acta, Part A 1980, 36A, 653. (30) Melendres, C. A.; Van Gils, S.; Terryn, H. Electrochem. Commun. 2001, 3, 737. (31) Raybaud, P.; Digne, M.; Iftimie, R.; Wellens, W.; Euzen, P.; Toulhoat, H. J. Catal. 2001, 201, 236. (32) Alexander, M. R.; Thompson, G. E.; Beamson, G. Surf. Int. Anal. 2000, 29, 468. (33) Hadjiivanov, K. Cat. Lett. 2000, 68, 157. (34) Angelini, M. M.; Garrard, R. J.; Rosen, S. J.; Hinrichs, R. Z. J. Phys. Chem. A 2007, 111, 3326. (35) Beard, B. C. Appl. Surf. Sci. 1990, 45, 221. (36) Swartz, W. E., Jr.; Youssefi, M. J. Elec. Spec. Relat. Phenom. 1976, 8, 61. (37) Matsuta, H.; Hirokawa, K. Appl. Surf. Sci. 1988, 35, 10. (38) Roberts, M. W.; Smart, R. S. C. Surf. Sci. 1980, 100, 590. (39) Tabata, K.; Kamada, M.; Choso, T.; Munakata, H. Appl. Surf. Sci. 1998, 125, 93.