Applied Surface Science 258 (2011) 1541–1550 Contents lists available at SciVerse ScienceDirect Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc Electronic state of ruthenium deposited onto oxide supports: An XPS study taking into account the final state effects Yurii V. Larichev a,b , Boris L. Moroz a,b,∗ , Valerii I. Bukhtiyarov a,b a b G.K. Boreskov Institute of Catalysis, Siberian Branch of the Russian Academy of Sciences, 5 Avenue Akademika Lavrentieva, 630090 Novosibirsk, Russia Novosibirsk State University, 2, Pirogova Street, 630090 Novosibirsk, Russia a r t i c l e i n f o Article history: Received 16 February 2011 Received in revised form 23 August 2011 Accepted 29 September 2011 Available online 6 October 2011 Keywords: Ruthenium Oxide supports Electronic state XPS Differential charging a b s t r a c t The electronic state of ruthenium in the supported Ru/EOx (EOx = MgO, Al2 O3 or SiO2 ) catalysts prepared by with the use of Ru(OH)Cl3 or Ru(acac)3 (acac = acetylacetonate) and reduced with H2 at 723 K is characterized by X-ray photoelectron spectroscopy (XPS) in the Ru 3d, Cl 2p and O 1s regions. The influence of the final state effects (the differential charging and variation of the relaxation energy) on the binding energy (BE) of Ru 3d5/2 core level measured for supported Ru nanoparticles is estimated by comparison of the Fermi levels and the modified Auger parameters determined for the Ru/EOx samples with the corresponding characteristics of the bulk Ru metal. It is found that the negative shift of the Ru 3d5/2 peak which is observed in the spectrum of ruthenium deposited onto MgO (BE = 279.5–279.7 eV) with respect to that of Ru black (BE = 280.2 eV) or ruthenium supported on ␥-Al2 O3 and SiO2 (BE = 280.4 eV) is caused not by the transfer of electron density from basic sites of MgO, as considered earlier, but by the differential charging of the supported Ru particles compared with the support surface. Correction for the differential charging value reveals that the initial state energies of ruthenium in the Ru/EOx systems are almost identical (BE = 280.5 ± 0.1 eV) irrespectively of acid–base properties of the support, the mean size of supported Ru crystallites (within the range of 2–10 nm) and the surface Cl content. The results obtained suggest that the difference in ammonia synthesis activity between the Ru catalysts supported on MgO and on the acidic supports is accounted for by not different electronic state of ruthenium on the surface of these oxides but by some other reasons. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Supported catalysts containing metallic ruthenium are utilized for many reactions including the Fischer–Tropsch synthesis [1,2] and synthesis of ammonia [3–6]. The electronic state of ruthenium is among important factors affecting the catalytic activity and selectivity; it can be modified through the metal-support interaction. Thereby catalytic properties of the supported ruthenium may depend on donor/acceptor properties of the support surface. As an example, the correlation between the chemical nature of the oxide support (EOx ) and the activity of supported Ru catalysts for ammonia synthesis is often mentioned: the stronger the support basicity, the higher the catalyst activity [7–11]. Indeed, the catalysts prepared by depositing ruthenium on MgO or CaO show considerable activity to ammonia synthesis even at 573–673 K and under the pressure of a N2 /H2 mixture close to ambient. The substitution of ∗ Corresponding author at: G.K. Boreskov Institute of Catalysis, Siberian Branch of the Russian Academy of Sciences, 5 Avenue Akademika Lavrentieva, 630090 Novosibirsk, Russia. Tel.: +7 3833269521; fax: +7 3833308056. E-mail address: moroz@catalysis.ru (B.L. Moroz). 0169-4332/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2011.09.127 acidic oxides (Al2 O3 and, in particular, SiO2 ) for MgO leads to a sharp decrease in the catalytic activity of the supported ruthenium. X-ray photoelectron spectroscopy was used to characterize the electronic state of ruthenium on the acidic and basic supports in order to identify the reason for the observed differences in catalytic behavior of Ru/EOx systems [7,8,12–17]. The experimentally determined binding energy (BE) of the Ru 3d5/2 core level was ranged between 279.5 and 280.0 eV for ruthenium supported on MgO or CaO [7,8,12], but between 280.2 and 280.6 eV for ruthenium supported on Al2 O3 or SiO2 [13–17]. For comparison, the reported BE(Ru 3d5/2 ) values for bulk Ru metal vary from 279.9 to 280.2 eV depending on the instrument calibration [18]. The fact that the BE(Ru 3d5/2 ) values are lower for the Ru/MgO and Ru/CaO catalysts than for bulk Ru metal gave Aika et al. an impetus to explain the promoting effect of basic supports on the activity of ruthenium catalysts for ammonia synthesis by their ability to donate electrons to surface Ru atoms [7,19,20]. The shift of the Ru 3d core level spectrum of supported ruthenium towards lower BEs with respect to that of the bulk metal indicates the excess electron density, which can be transferred from the d orbitals of Ru atoms to the antibonding orbitals of N2 molecules upon their adsorption on the surface of Ru crystallites. This transfer weakens the N–N bonds 1542 Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 and therefore facilitates the dissociation of dinitrogen, which is the rate-determining step of ammonia synthesis. The said hypothesis was accepted by other researchers (see, e.g., Refs [21,22]), but the mechanism of the electron density transfer from basic sites of the support to supported Ru particles still remains unclear. Examination of Ru/EOx catalysts with alternative research techniques such as infrared spectroscopy of adsorbed CO did not reveal any dependence of the electronic state of supported ruthenium on acid-base properties of the support surface [23]. Eventually, it should be kept in mind that the negative shift of the core level spectra from supported metal particles compared to the bulk metal can reflect not only the presence of a negative charge on metal particles due to their interaction with the support, but may also be caused by the so-called final state effects appearing as a result of photoelectron emission [24,25]. Among these are the relaxation effect representing a change in the relaxation energy due to reorganization of electrons of the solid that provides the screening of the photoelectron holes remaining after electron emission, and the differential charging effect, i.e., appearance of a potential difference between the surface of supported metal particles and the support surface. The latter phenomenon originates from that the supported metallic nanocrystallites due to their internal conductivity can provide easier compensation of the surface positive charge accumulated as a result of photoelectron emission than the dielectric support. Neither relaxation nor differential charging effects in the heterogeneous systems like low-loaded supported metal catalysts can be eliminated using the internal reference method, since the positions of XPS peaks in their spectra are usually referenced to the selected peak of an element contained in the dielectric support. Our earlier studies [26,27] revealed that namely the differential charging effect is responsible for variations of the binding energy of Ag 3d5/2 core level measured for supported silver in the Ag/␣-Al2 O3 catalysts for ethylene epoxidation. Later, it has been assumed [28,29] that differential charging may also account for shifting the Ru 3d core level spectrum of the catalyst prepared by reduction of deposited ruthenium chloride on the MgO surface with respect to the spectrum of the bulk metal. The present paper reports the results of comparative XPS study on the chemical state of ruthenium deposited on oxides with different acid–base properties (MgO, ␥-Al2 O3 , and SiO2 ). The samples under study were prepared using various Ru precursors (ruthenium hydroxychloride, Ru(OH)Cl3 , or ruthenium(III) acetylacetonate, Ru(acac)3 ) and differed by the mean size of the metal particles, as was confirmed by X-ray diffraction (XRD) and transmission electron microscopy (TEM) analysis. When inspecting the XPS data obtained, we took into consideration a probable influence of the final state effects such as differential charging of the metallic Ru crystallites supported on a dielectric support and the relaxation effect on observed BE values. It was revealed that differential charging is considerable if the metallic ruthenium is supported on MgO but negligible when ␥Al2 O3 and SiO2 are used as the supports. Taking this into account, we found that the energies of the initial state of supported Ru on all the examined supports are approximately identical that differs from previous conclusions. 2. Experimental 2.1. Materials “Ruthenium hydroxychloride” (Ru(OH)Cl3 , 44.8 wt% Ru), aqueous formaldehyde (40 wt%) and acetone were of analytical grade and used as received without further purification. Ruthenium(III) acetylacetonate, Ru(acac)3 was synthesized as described elsewhere [30]. Table 1 Textural properties of supports. Support S (m2 /g) Vpore (cm3 /g) Rpore (nm)a MgO ␥-Al2 O3 SiO2 228 220 160 0.80 0.50 0.95 4 and 20–30 7.5 30 As determined by N2 adsorption at 77 K using a Micromeritics ASAP-2400 instrument (before measuring, all samples were heated in vacuum at 423 K for 4 h). a Corresponding to the maxima of the pore-size distribution. Magnesia prepared in G.K. Boreskov Institute of Catalysis by precipitation from an aqueous solution of Mg(NO3 )2 , ␥-alumina supplied by Ryazan Oil Refinery Company (Russia) and Silochrom2 silica purchased from Siberian Catalyst Co. (Novosibirsk) were used as the supports in the form of 0.2–0.5 mm granules. Textural properties of the supports are given in Table 1. Prior the impregnation with Ru compound, MgO and ␥-Al2 O3 were evacuated at 523 K for 3 h, while SiO2 was calcined in air at the same temperature for 3 h. 2.2. Sample preparation and characterization Ruthenium deposition was carried out by incipient-wetness impregnation of the support with an acetone solution of Ru(OH)Cl3 or Ru(acac)3 , followed by drying in an airflow at room temperature. Then the samples were evacuated at 293 K for 2 h and at 323 K for another 2 h. Due to the low solubility of Ru(OH)Cl3 and Ru(acac)3 in acetone, the impregnation procedure was twice repeated. After the last portion of a Ru precursor was deposited, the samples were evacuated at 323 K for 6 h, heated in flowing hydrogen to 723 K for 2.5 h and reduced under these conditions for another 6 h. After reduction, they were cooled in a H2 flow to room temperature, transferred from the reactor to a glass ampoule without atmospheric exposure and stored in the ampoule filled with argon. As a reference sample, ruthenium black was synthesized by reaction of Ru(OH)Cl3 with alkaline formaldehyde solution at 353 K, washed with distilled water, dried and reduced again by heating in flowing hydrogen at 623 K for 5 h [31]. The Ru content of the Ru/EOx samples was measured using Xray fluorescence technique on a VRA-30 instrument equipped with a W-anode. TEM (JEOL JEM-2010 microscope with a lattice resolution of 0.14 nm operating at 200 kV) and XRD (HZG-4 diffractometer with Cu K␣ radiation and a graphite monochromator) analyses were performed to determine the size of Ru crystallites. Diffractograms were collected at a scanning rate of 0.60 K/min with a 0.05◦ step size in the 2 range between 30◦ and 90◦ . Prior to XRD analysis, a sample was quickly ground in air and wrapped in an organic polymer film to prevent it partly from contacting air during the XRD pattern acquiring. 2.3. Catalytic tests The catalytic activity of the Ru/EOx samples for ammonia synthesis was tested under continuous flow conditions at 300–400 ◦ C under 101 kPa using a stoichiometric (1:3) N2 /H2 mixture at the flow rate of 170 mL/min. The reaction feed was deoxidized (to the level <0.1 ppm O2 ) and dried (<0.5 ppm H2 O) by passing successively through the columns filled with active alumina, a reduced Ni-Cr catalyst and molecular sieves NaA and NaX. A catalyst sample (2.0–2.4 g) was transferred to a glass plug-flow reactor under synthesis gas (1:3 N2 /H2 ) atmosphere (without contact with air) and heated to 300 ◦ C. As soon as the outlet NH3 concentration ceased to change noticeably with time (pseudo-stationary state), the reaction temperature was risen by 50 ◦ C, and the reaction again was carried out until reaching the pseudo-stationary state of the catalyst at the Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 2.4. X-ray photoelectron spectroscopy XPS spectra were measured with a VG ESCALAB HP spectrometer (Vacuum Generators, Great Britain) using a non-monochromated Al K␣ radiation (Eh = 1486.6 eV) operated at 200 W as the excitation source. The BE scale was pre-calibrated against the position of the peaks of the Au 4f7/2 (BE = 84.0 eV) and Cu 2p3/2 (BE = 932.6 eV) core levels from polycrystalline foils of gold and copper, respectively. The charging effect on the peak positions in the spectra of Ru/EOx was corrected using peaks Mg 2s (BE = 88.1 eV), Al 2p (BE = 74.6 eV) or Si 2p (BE = 104.0 eV) as the internal references at EOx = MgO, Al2 O3 or SiO2 , respectively. The correction factor for charge induced shift of XPS peaks was the difference between the experimentally measured and tabulated BE of the Mg 2s, Al 2p or Si 2p core levels. For measuring the spectrum, a sample was pressed into a Ni grid immediately after unsealing the storage ampoule and transferred to a preparation chamber of spectrometer (the samples were exposed to air for no more than 5 min during this procedure). In the preparation chamber, the sample was re-reduced with H2 under static conditions at 623 K and 101 kPa for 1 h, then outgassed to 10−5 Pa, cooled to room temperature and transferred into the analyzer. The BEs and the areas of XPS peaks were determined after Shirley background subtraction and analysis of the line shapes by curve fitting of the experimental using the mixture of a Gaussian function and a Lorentzian function at their varied relationship. The standard deviation for the BEs was 0.1 eV. The atomic ratios of the elements at the sample surface were calculated by formula: nx/y = Ix /ASFx Iy /ASFy (1) where nx/y is the ratio of atomic concentrations of elements X and Y in the area of analysis; Ix and Iy are the integrated intensities (areas) of the XPS peaks of elements X and Y, respectively; ASFx and ASFy are the atomic sensitivity factors of elements X and Y, respectively, taken from Refs. [18,32]. ** ** Intensity, a.u. given temperature. For determination of the outlet NH3 concentration, the gas flow after reactor was passed through an aqueous H2 SO4 solution of a known concentration and the time interval needed to neutralize a certain amount of the acid at constant flow rate, temperature, and pressure was measured. The stationary outlet NH3 concentration was calculated by averaging results of 6–10 measurements at identical conditions (the standard deviation from the mean value were within the range 0.002–0.008 vol%). The detection limit of the ammonia concentration was 0.005 vol%. 1543 ** ** * ** ** * ** * * ** Ru(Cl)/MgO MgO 40 60 80 100 120 2Θ Fig. 1. XRD patterns of MgO and a sample prepared by impregnation of MgO with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K (* and ** label reflections of the Ru and MgO phases, respectively). measuring diameters (di ) of Ru particles seen in TEM micrographs were used for determination of particle size distributions, no less than 400 crystals being included in the distribution for each sample. Fig. 2 demonstrates Ru particle size distributions of all the samples under study. Linear mean (dl ) and mass mean (dm ) diameters of Ru particles calculated from the TEM histograms are presented in Table 2. The value of dm is close to the crystallite size (d0 0 1 ) determined by XRD for the same sample. The latter suggests that Ru particles contained in all the samples under consideration do not tend to twinning and are monocrystalline. From TEM data (Fig. 2), the samples prepared by depositing Ru(OH)Cl3 onto MgO, ␥-Al2 O3 and SiO2 with large enough surface area (SBET = 160–220 m2 /g) and reduced with dihydrogen at 723 K contain mainly Ru metal particles less than 5 nm in diameter. In addition, few larger Ru particles (di = 5–20 nm) are seen in the micrographs of Ru(Cl)/MgO and Ru(Cl)/SiO2 samples but no in the micrographs of Ru(Cl)/␥-Al2 O3 sample. When Ru(acac)3 is used instead of Ru(OH)Cl3 as the precursor, the dispersion of Ru is much higher on MgO than on ␥-Al2 O3 (di = 3.9 and 11.0 nm for the Ru(AA)/MgO and Ru(AA)/␥-Al2 O samples, respectively). It is most likely due to the fact that Ru(acac)3 , unlike chloride and hydroxychloride complexes of Ru(III, IV), interacts stronger with the MgO surface than with the alumina surface [33]. 3. Results and discussion 3.2. Catalytic activity in ammonia synthesis 3.1. Ru particle size Table 3 shows results of testing the catalytic activity of the Ru/EOx samples for ammonia synthesis at atmospheric pressure. With the samples prepared by depositing ruthenium onto ␥-Al2 O3 and SiO2 , the NH3 concentration in the reactor outlet gas mixture is below the detection limit (<0.005 vol%) over the whole temperature range examined (300–400 ◦ C). At the same time, the formation of NH3 over the Ru(AA)/MgO sample is detected at 300 ◦ C. At 350 and 400 ◦ C, the NH3 outlet concentration reaches 5% and 39% of the corresponding NH3 equilibrium yields, respectively, i.e. under the given conditions, the ammonia synthesis activity of the catalyst containing metallic Ru particles on the surface of basic MgO support is higher at least by an order of magnitude than the activity of the Ru/␥-Al2 O3 and Ru/SiO2 samples prepared in the same way. This result agrees with the literature data [7,8,10,21,35] indicating that the Ru/EOx samples prepared in this work can be used as objects for studying the support effect which is discussed in the literature (see Section 1). Table 2 lists the samples under study with their designations and Ru weight contents. XRD patterns of the Ru/EOx samples display intense diffraction peaks from the relevant support (EOx = MgO, Al2 O3 or SiO2 ) along with weak, broadened reflections related to Ru metal (see Fig. 1 as an example). At the same time, no peaks that may relate to any of the known ruthenium oxides (RuO2 or RuO4 ) are observed. The sizes of Ru crystallites in the Ru/EOx samples and Ru black calculated from the width of Ru(0 0 1) reflections are given in Table 2. The lattice constants of supported ruthenium in all the Ru/EOx samples are identical within the experimental error to those of Ru black (a = 0.270(4) and 0.2707(2) nm, c = 0.428(2) and 0.4280(3) nm for the supported and bulk ruthenium, respectively). TEM micrographs of the Ru/EOx samples taken with a medium magnification demonstrate spherical or near-spherical Ru particles which are evenly distributed over the support surface. Results of 1544 Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 Table 2 Nomenclature and main characteristics of Ru/EOx samples (supports, Ru precursors, Ru content and mean particle diameter). Sample designation Support Precursor Ru content (wt%) Ru(Cl)/MgO Ru(AA)/MgO Ru(Cl)/␥-Al2 O3 Ru(AA)/␥-Al2 O3 Ru(Cl)/SiO2 Ru black MgO MgO ␥-Al2 O3 ␥-Al2 O3 SiO2 Ru(OH)Cl3 Ru(acac)3 Ru(OH)Cl3 Ru(acac)3 Ru(OH)Cl3 Ru(OH)Cl3 5.0 5.0 4.1 4.8 5.0 n.d.   Mean particle diameter (nm) dl dm d0 0 1 3.5 2.2 2.0 8.6 3.4 33.0 10.8 3.9 3.0 11.0 7.9 80.0 10.0 ≤3.0 ≤3.0 10.0 6.2 45.0 Abbreviations: acac = acetylacetonate; n.d. = not determined; dl = di /N, dm = di4 / di3 , where dl , the diameter of i-particle of Ru measured by TEM; N, the total number of measured particles; d0 0 1 , the size of Ru crystallites related to diffraction peak Ru(0 0 1) at 38◦ (in 2 scale) and calculated by the Scherrer equation. Table 3 Results of testing the catalytic activity of Ru/EOx samples in ammonia synthesis. Sample Reaction rate × 104 (mol min−1 g (Ru)−1 ) at 350 ◦ C Stationary NH3 concentration in the reactor outlet gas (vol%) at the given temperature (◦ C)a Ru(AA)/MgO Ru(AA)/␥-Al2 O3 Ru(Cl)/␥-Al2 O3 Ru(Cl)/SiO2 300 350 400 0.006 <0.005 <0.005 <0.005 0.042 <0.005 <0.005 <0.005 0.17 <0.005 <0.005 <0.005 0.33 <0.04 <0.04 <0.04 Reaction conditions: 101 kPa of 1:3 N2 /H2 gas mixture, total flow rate of 170 mL/min, catalyst weight of 2.0–2.4 g (ca. 1.0 mg-at Ru). Nomenclature and characteristics of Ru/EOx samples are presented in Table 2. a The equilibrium NH3 concentrations are 2.17, 0.86 and 0.44 vol% at 300, 350 and 400 ◦ C, respectively [34]. 3.3. Chemical composition of the sample surface and electronic state of Ru deposited onto various oxides (according to XPS analysis) The survey XPS spectra of the Ru/EOx samples contain the core level peaks which are characteristic of the support (the Mg 2s, Al Ru(Cl)/MgO 300 Ru(Cl)/γ -Al2O3 300 200 200 100 100 0 0 0 5 10 15 20 25 30 35 0 40 400 Number of Particles 2p, Si 2p and O 1s peaks), ruthenium (the Ru 3d peak), adsorbed carbonaceous species (CHx ) (the C 1s peak) and chlorine impurity (the Cl 2p peak with BE ranging from 198.3 to 199.0 eV that is characteristic of chloride ions [32]). While this impurity may modify the ruthenium surface electronically, XPS was used to determine the chemical state and the surface Cl content in the samples after 5 10 15 60 25 30 35 40 Ru(AA)/γ -Al2O3 Ru(AA)/MgO 300 20 40 200 20 100 0 0 0 5 10 15 20 25 30 35 0 40 5 10 15 20 25 30 35 40 35 40 200 Ru black Ru(Cl)/SiO2 50 100 0 0 0 20 40 60 80 100 120 140160 0 5 10 15 20 25 30 Particle Diameter, nm Fig. 2. Ruthenium particle size distributions of Ru black and Ru/EOx samples (EOx = MgO, ␥-Al2 O3 , SiO2 ). Mean diameters of Ru particles calculated from the histograms are given in Table 2. Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 198.4 1545 198.9 (A) (B) Ru(Cl)/MgO Intensity, a.u. Ru(Cl)/γ-Al2O3 MgO 192 196 200 204 γ-Al2O3 192 196 200 204 Binding energy, eV Fig. 3. Cl 2p core level spectra: A – MgO support and the sample prepared by impregnation of MgO with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K; B – ␥-Al2 O3 support and the sample prepared by impregnation of ␥-Al2 O3 with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K. The experimental spectra represented by thick lines are fitted as a sum of two components, Cl 2p3/2 (dashed shaded peak) and Cl 2p1/2 (dashed open peak), which appear as a result of a spin-orbital splitting. they were treated with H2 at high temperature. As an example, Fig. 3 shows the Cl 2p core level spectra from the Ru/EOx samples prepared by depositing Ru(OH)Cl3 onto MgO and ␥-Al2 O3 and treated with H2 at 723 K. The Cl 2p spectra from the supports taken under identical conditions are also shown for comparison. For all the samples, the Cl 2p peak has a characteristic asymmetric shape and can be decomposed into two components, Cl 2p3/2 and Cl 2p1/2 , which appear as a result of a spin-orbital splitting. Table 4 presents BEs corresponding to the most intense Cl 2p3/2 peak as well as the Cl/E atomic ratios calculated for the EOx and Ru/EOx samples from the integrated intensities of Cl 2p, Mg 2s, Al 2p and Si 2p peaks. The BEs of the Cl 2p3/2 core level experimentally determined for the Ru(Cl)/MgO (BE = 198.4–198.5 eV) and Ru(Cl)/␥-Al2 O3 (198.8–199.0 eV) samples differ from the value found for Ru(OH)Cl3 (BE(Cl 2p3/2 ) = 198.2 eV) but coincide (within the measurement accuracy) with the values determined for the Table 4 The Cl 2p3/2 binding energies and Cl/E atomic ratios obtained by XPS for the EOx and Ru/EOx samples (EOx = MgO, Al2 O3 , SiO2 ). Sample BE (Cl 2p3/2 ) (eV) Cl/Mg(Al, Si) atomic ratioa MgO Ru(Cl)/MgO Ru(AA)/MgO ␥-Al2 O3 Ru(Cl)/␥-Al2 O3 Ru(AA)/␥-Al2 O3 SiO2 Ru(Cl)/SiO2 Ru(OH)Cl3 198.4 198.5 198.4 198.9 198.8 199.0 n.d. n.d. 198.2 0.003 0.033 0.003 0.042 0.043 0.024 <0.001 <0.001 Nomenclature of Ru/EOx samples is presented in Table 2. Abbreviations: BE = binding energy; n.d. = not determined because of very small peak intensity. a Calculated from the integrated intensities of the Cl 2p, Mg 2s, Al 2p and Si 2p peaks using the corresponding atomic sensitivity factors taken from Refs [18,32]. supports containing no ruthenium (BE(Cl 2p3/2 ) = 198.4 ± 0.1 and 198.9 ± 0.1 eV for MgO and ␥-Al2 O3 , respectively). Hence, it is reasonable to suppose that Cl− ions are preferably localized on the surface of the supports in the Ru/MgO and Ru/␥-Al2 O3 samples and bound to Mg2+ or Al3+ cations. As seen from Table 4, a considerable dependence of the surface Cl content in a Ru/EOx sample on whether Cl-containing or Cl-free ruthenium complex is used for its preparation is only observed in the case of EOx = MgO. Upon treatment with H2 at 723 K, the residual surface Cl content in the Ru/MgO sample prepared using Ru(OH)Cl3 is higher by an order of magnitude than in the sample prepared from Ru(acac)3 and in the support. On the contrary, the surface Cl content in the Ru(Cl)/␥Al2 O3 sample after reduction with H2 is approximately equal to the Cl content in the initial support and differs inconsiderably from that in the Ru(AA)/␥-Al2 O3 sample. In the spectra of the initial SiO2 and Ru(Cl)/SiO2 samples, the intensities of the Cl 2p peak are at the noise level (Cl/Si < 0.001), i.e. Cl− ions introduced with Ru(OH)Cl3 are completely removed from Ru/SiO2 when it is reduced with H2 at 723 K. The data obtained are in agreement with the conclusion by Murata and Aika [36] that the tendency of Ru/EOx systems to hold chlorine corresponds mainly to the basicity of the support (MgO > Al2 O3 ≫ SiO2 ) and to the stability of the surface chloride (MgCl2 ≫ AlCl3 > SiCl4 ). This is also evidence, even though not a direct, that the Cl− ions are retained on the support surface rather than on the Ru metal particles. Fig. 4 shows the Ru 3d core level spectra measured from the Ru/EOx samples. The Ru 3d spectrum from ruthenium black taken under identical conditions is also shown for comparison. The spectra consist of the Ru 3d5/2 and Ru 3d3/2 peaks appearing as a result of a spin-orbital splitting. The Ru 3d3/2 peak is overlapped with the C 1s peak at ∼285 eV derived from CHx impurities present on the sample surface. For this reason, BEs of the Ru 3d5/2 core level are only used in the following discussion. The Ru 3d peaks obtained from the Ru/EOx samples are broad with full width at half 1546 Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 279.5 280.4 Ru(Cl)/γ-Al2O3 Ru(Cl)/MgO Intensity, a. u. 279.7 280.2 280.2 Ru(AA)/MgO Ru(AA)/γ-Al2O3 280.4 Ru 275 280 285 Ru(Cl)/SiO2 290 275 280 285 290 Binding Energy, eV Fig. 4. Ru 3d core level spectra of Ru black and Ru/EOx samples (EOx = MgO, ␥-Al2 O3 , SiO2 ). For sample nomenclature see Table 2. The experimental spectra represented by thick lines are shown after a Shirley background has been subtracted. They are fitted as a sum of two components, Ru 3d5/2 (dashed shaded peak) and Ru 3d3/2 (dashed open peak), which appear as a result of a spin-orbital splitting. The Ru 3d3/2 peaks are overlapped with the C 1s peaks (solid open peak) derived from CHx impurities originally present on the sample surface or slowly accumulated in the spectrometer. maximum ca. 3 eV but are fairly symmetric. Hence, it is reasonable to assume the presence of ruthenium in a single valence state, while the observed broadening of the Ru 3d peaks is accounted for by the charging of the sample surface due to photoelectron emission and/or by inhomogeneous size distribution of the supported Ru particles. Fig. 5A demonstrates that the intensity of the Ru 3d core level spectrum from the Ru(Cl)/MgO sample strongly depends on the degree of sample grinding before the spectrum registration. Grinding the sample in an agate mortar to powder with particles of several microns in size causes a decrease in the spectrum intensity by an order of magnitude and corresponding decrease in the determined Ru/Mg atomic ratio, while BE of the Ru 3d5/2 peak remains unchanged. TEM examination of the grinded Ru(Cl)/MgO sample did not reveal any changes in the diameters of Ru particles as compared with those in the as-prepared sample. Hence, sintering of metal particles (for example, due to local overheating during grinding) cannot be the reason for low intensity of the Ru 3d core level spectrum measured from the grinded sample. It is most likely that the observed effect is accounted for by low depth of Ru penetration inside the support granules. Upon grinding the sample, interior of the granules becomes exposed and, therefore, a decrease in the intensity of the Ru 3d spectrum, which is observed for the grinded Ru(Cl)/MgO sample as compared with that for the non-grinded specimen of the same sample, indicates the preferable ruthenium localization on the external surface of the support granules (the “egg shell” type of distribution of the supported component). A similar, but less considerable though, decrease in the intensity of the Ru 3d spectrum also is observed as a result of grinding the Ru(Cl)/␥-Al2 O3 (Fig. 5B) and Ru(Cl)/SiO2 samples. In these cases, the Ru/Al(Si) atomic ratio for the grinded sample is smaller than that for the non-grinded sample by only a factor of 1.2–1.7. This indicates much more uniform ruthenium distribution between the external and internal surfaces of support granules in the Ru(Cl)/␥-Al2 O3 and Ru(Cl)/SiO2 samples as compared with the Ru(Cl)/MgO sample. The BE values of the Ru 3d5/2 core level experimentally determined in this paper and presented in Fig. 4 coincide with the BEs reported by earlier workers [7,8,12–17] for the analogous Ru/EOx systems. When ruthenium is supported on the ␥-Al2 O3 or SiO2 surface, the BE (Ru 3d5/2 ) values are either equal to or little higher than the value observed for Ru black (280.2 eV). On the contrary, the Ru 3d5/2 peak from the Ru(Cl)/MgO and Ru(AA)/MgO samples is shifted to lower BEs by 0.4 and 0.7 eV, respectively, compared to that of Ru black. In interpreting the data obtained, the fact that the experimentally observed BE of the Ru 3d5/2 core level (BE(Ru 3d5/2 )exp ) may characterize not only the initial state of Ru atoms in the supported metal particles was taken into consideration. This value also may be contributed by such final state effects caused by photoelectron emission as a change in the relaxation energy and differential charging [24,25]. Taking this into account, the Ru 3d5/2 orbital energy of the initial state of atoms in the supported Ru particles (E0 (Ru 3d5/2 )) can be determined by correcting the value of BE(Ru 3d5/2 )exp for the values of the final state effects: E0 (Ru 3d5/2 ) = BE(Ru 3d5/2 )exp + Edif + ER (2) where Edif is the differential charging value; ER is the variation of the relaxation energy for the supported Ru particles with respect to the bulk metal. The variation of the relaxation energy, ER , can be estimated from the change in the value of the modified Auger parameter, ˛, for supported ruthenium in the Ru/EOx samples compared to the bulk metal [37,38]. This parameter represents the sum of a core level BE of an element and a kinetic energy of the most intense Auger peak of the same element. Thus, in the case under consideration ER = 0.5 × ˛ = 0.5 × [BE(Ru 3d5/2 )exp + Ekin (Ru MNN)] (3) where BE(Ru 3d5/2 )exp and Ekin (Ru MNN) are the differences between the Ru 3d5/2 binding energies and between the Ru MNN Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 (A) Ru(Cl)/MgO 279.5 (B) 1547 280.4 Ru(Cl)/γ-Al2O3 Ru/Mg=0.026 Ru/Al=0.043 Intensity, a.u. grinded grinded Ru/Mg=0.217 Ru/Al=0.052 as prepared as prepared 276 280 284 288 292 276 280 284 288 292 Binding Energy, eV Fig. 5. Ru 3d core level spectra of the samples prepared by impregnation of MgO (A) and ␥-Al2 O3 (B) with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K. The samples were taken for the spectra measuring both as prepared and after grinding in an agate mortar. Atomic ratios of Ru/Mg and Ru/Al calculated from the integrated intensities of the Ru 3d, Mg 2s and Al 2p peaks are indicated within the figure. Auger kinetic energies, respectively, measured for supported and bulk Ru metal. The values of ˛ and, hence, ER are independent (in the first approximation) on the charging, because increasing the core level binding energy with an increase in the charging potential is compensated for by a decrease in the Auger kinetic energy by the same value. The differential charging value, Edif , is determined from the shift of Fermi level in the valence band spectrum of supported metal particles in respect relative to the Fermi level of bulk metal, which is always in electrochemical equilibrium with the Fermi level of the spectrometer [26,39,40]. Fig. 6 shows the Ru MNN Auger spectra measured for the Ru black and Ru/EOx samples in order to determine the Ekin (Ru MNN) values. The ˛ values calculated by Eq. (3) from the experimental Ru 3d5/2 binding and Ru MNN Auger kinetic energies for the Ru/EOx 275.2 Intensity, a.u. 275.2 Ru(Cl)/MgO 274.5 Ru(AA)/MgO Edif = ϕsup − ϕRu 274.3 Ru(Cl)/γ-Al2O3 274.5 Ru(Cl)/SiO2 Ru black 264 270 276 samples coincide with or to be only 0.2 eV higher than that found for Ru black (˛ = 554.7 eV). The value of ˛ ≤ 0.2 eV corresponds to a very small value of ER (ER ≤ 0.1 eV) and indicates that the relaxation gives essentially no contribution to the shift of the Ru 3d spectra of the Ru/EOx samples with respect to metallic Ru. The valence band spectra measured with the Ru/EOx samples (the examples are shown in Fig. 7) contain the contribution of the support in addition to the signals from the valence Ru 4d/5s orbitals. To detect the differential charging effect, one should pay main attention to the BE region of −1 ± 5 eV where the contribution from ruthenium predominates. Comparison of the valence band spectra of the Ru(Cl)/MgO or Ru(AA)/MgO samples with the spectrum measured with the Ru black sample (Fig. 7A) reveals that the spectra of the heterogeneous systems in the region close to the Fermi level are shifted by 1.0 and 0.8 eV, respectively, towards lower BEs with respect to the spectrum of bulk Ru metal. Since the mean diameter of Ru particles in the studied samples is higher than 2 nm (see Table 2), any changes of the valence band spectrum caused by the small sizes of supported metal crystallites are hardly observed [41], and, therefore, the shift of the spectrum can be attributed exclusively to differential charging of the supported Ru particles compared to the support microglobules. The value of this shift represents the Edif value [39,40] 282 288 Kinetic Energy, eV Fig. 6. Auger Ru MNN spectra (excitation energy Al K␣ , 1486.6 eV) of Ru black and Ru/EOx (EOx = MgO, ␥-Al2 O3 , SiO2 ) samples. For sample nomenclature see Table 2. (4) where ϕsup and ϕRu are the charging potentials induced by photoelectron emission at the support and ruthenium surfaces, respectively. Correspondingly, the Ru 3d core level spectra measured for the Ru(Cl)/MgO and Ru(AA)/MgO samples and shown in Fig. 4 should be shifted to higher BEs by the Edif values, i.e. by 1.0 and 0.8 eV, respectively. On the contrary, Fig. 7B demonstrates that the Fermi level of supported ruthenium in the Ru/␥-Al2 O3 and Ru/SiO2 samples prepared from Ru(OH)Cl3 coincides, with an accuracy of 0.1 eV, with the Fermi level of the bulk metal. Therefore, it is reasonable to conclude that photoelectron emission does not cause differential 1548 Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 (B) (A) Ru(Cl)/γ-Al2O3 +1.0 eV +0.1 eV Intensity, a.u. Ru(Cl)/MgO +0.8 eV Ru(AA)/MgO Ru(Cl)/SiO2 +0.1 eV -3 0 3 6 9 -3 0 3 6 9 Binding Energy, eV Fig. 7. Valence band spectra: A – samples prepared by impregnation of MgO with solutions of Ru(OH)Cl3 (upper) and Ru(acac)3 (bottom); B – samples prepared by impregnation of ␥-Al2 O3 and SiO2 with a Ru(OH)Cl3 solution. All the samples were reduced with H2 at 723 K. The dashed line below the spectrum of each sample shows the valence band spectrum of Ru black. The digit over an arrow indicates a distance to shift the spectrum of the sample up to the coincidence with the spectrum of Ru black at BE = 0 eV. charging in these samples. The differential charging effect does occur in the Ru/␥-Al2 O3 sample prepared from Ru(acac)3 , while its value (Edif = +0.4 eV) is far inferior than that for the both of Ru/MgO samples under consideration. Table 5 summarizes all the parameters (BE(Ru 3d5/2 )exp , Ekin (Ru MNN), ˛, ER and Edif ) found for the Ru black and Ru/EOx samples by means of Ru 3d core level, valence band, and Ru MNN Auger spectra, as well as the 3d5/2 orbital energies of Ru atoms in the Ru/EOx samples calculated by Eq. (2) from these data. One can see that the experimentally observed BEs of the Ru 3d5/2 core level for the Ru/␥Al2 O3 and Ru/SiO2 samples (BE(Ru3d5/2 )exp = 280.2–280.4 eV) are close to the energy of the initial state of supported ruthenium in these samples (E0 (Ru 3d5/2 ) = 280.5–280.6 eV) due to minor influence of the final state effects (the differential charging and variation in the relaxation energy). On the contrary, the considerable differential charging of supported Ru particles compared with the support surface (Edif = +0.8 or +1.0 eV) appears as a result of photoelectron emission in case of MgO is used as the support. After the correction for the differential charging (Table 5), the Ru 3d5/2 binding energies of surface Ru atoms appears practically identical and equal to 280.5–280.6 eV for the both Ru/MgO samples under consideration irrespectively of the mean size of Ru crystallites (at least within the range of 2–10 nm) and surface Cl content. Hence, the initial state energy of ruthenium supported on MgO is similar to that found for Ru nanoparticles supported on Al2 O3 or SiO2 and by 0.3 ± 0.1 eV higher than that in the bulk metal. This fact indicates that the supported Ru particles located on the MgO surface are positively charged similarly to the Ru crystallites supported on the Al2 O3 or SiO2 surface. Note that the low electron density is characteristic of oxide-supported metal particles of several nanometers in size and commonly attributed to their interaction with functional groups on the support surface [42]. Thus, the negative shift of the Ru 3d core level spectrum which is observed for ruthenium deposited onto MgO with respect to ruthenium supported on acidic supports (Al2 O3 , SiO2 ) is caused not by the different electronic state of ruthenium on these oxides, as was supposed before [7,8,20], but by the differential charging effect. Fig. 8 shows the O 1s core level spectra of Ru(Cl)/MgO and Ru(Cl)/␥-Al2 O3 samples in comparison with the spectra of MgO and ␥-Al2 O3 , respectively. The spectra of MgO and Ru(Cl)/MgO (Fig. 8A) contain an asymmetric peak which can be decomposed into two features related to different surface states of oxygen. According to the literature data [43], the main feature at BE ≈ 530 eV is derived from O2− ions of MgO lattice, whereas the less intense feature at BE ≈ 532 eV can be assigned to OH groups on the MgO surface. In the spectrum of Ru(Cl)/MgO sample, the feature at 532 eV is shifted by 0.6 eV to lower BEs with respect to the corresponding feature in the spectrum of MgO. This shift indicates an increase in the negative charge on oxygen atoms of the surface OH groups in the Ru(Cl)/MgO sample as compared to MgO containing no supported ruthenium. The fact that the negative shift of the O 1s peak at 532 eV is observed along with an increase in the initial state energy of Ru atoms (compared to that of ruthenium black) suggests the electron transfer from the supported Ru particles to OH groups on the MgO surface. Unfortunately, it is impossible to extract the contributions of different oxygen states to the total spectrum in the case of alumina or silica is used as the support, since the O 1s peak derived from oxygen atoms of ␥-Al2 O3 (Fig. 8B) and SiO2 is broader and more symmetric than that derived from MgO. Furthermore, as mentioned above, ruthenium is distributed rather uniformly between the external and internal surfaces of the support granules in the Ru(Cl)/␥-Al2 O3 and Ru(Cl)/SiO2 samples, but localized predominantly on the external surface of the support in the Ru(Cl)/MgO samples. As a result, the content of Ru atoms is comparable to the content of OH groups on the MgO surface (the Ru/O atomic ratio is equal to 1.6, as calculated from the integrated intensities of the Ru 3d5/2 peak and the O 1s feature at ≈532 eV) but considerably lower on the surface of ␥Al2 O3 or SiO2 . Therefore, the interaction between Ru atoms and OH groups of alumina or silica may remain unnoticed if the peak shift in the O 1s spectrum is used as an only argument for its presence. Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 1549 Table 5 Experimental Ru 3d5/2 binding and Ru MNN Auger kinetic energies for the Ru/EOx (EOx = MgO, ␥-Al2 O3 , SiO2 ) and Ru black samples, values of final state effects, and the initial state energies of supported and bulk ruthenium calculated by Eq. (2). Sample BE (Ru 3d5/2 )exp BE Edif Ekin (Ru MNN) ˛ ER Ru black Ru(Cl)/MgO Ru(AA)/MgO Ru(Cl)/␥-Al2 O3 Ru(AA)/␥-Al2 O3 Ru(Cl)/SiO2 280.2 279.5 279.7 280.4 280.2 280.4 0 −0.7 −0.5 +0.2 0 +0.2 0 +1.0 +0.8 +0.1 +0.4 +0.1 274.5 275.2 275.2 274.5 274.5 274.3 554.7 554.7 554.9 554.9 554.7 554.7 0 0 +0.1 +0.1 0 0 E0 (Ru 3d5/2 ) 280.2 280.5 280.6 280.6 280.6 280.5 Nomenclature of Ru/EOx samples is presented in Table 2. Designations: BE(Ru 3d5/2 )exp and Ekin (Ru MNN), the experimental Ru 3d5/2 binding and kinetic Ru MNN energies, respectively; BE, the difference between the Ru 3d5/2 binding energies measured for supported and bulk Ru metal; Edif , the differential charging value (the difference between the charging potentials at the support surface and at the surface of supported Ru particles) determined from the shift of the valence band spectrum of supported Ru with respect to that of bulk Ru metal (see Fig. 7); ˛, the modified Auger parameter defined as ˛ = BE(Ru 3d5/2 )exp + Ekin (Ru MNN); ER , the variation of relaxation energy defined as Erelax = 0.5˛, where ˛, the Auger parameter shift between supported and bulk Ru metal; E0 (Ru 3d5/2 ), the 3d5/2 orbital energy of the initial state of Ru atoms (the Ru 3d5/2 binding energy corrected for the final state effects) defined as E0 (Ru 3d5/2 ) = BE(Ru 3d5/2 )exp + Edif + Erelax . All values in the table are given in eV. 531.5 529.8 Intensity, a.u. 531.7 Ru(Cl)/MgO Ru(Cl)/γ-Al2O3 532.3 γ-Al2O3 MgO 525 528 531 534 537 525 528 531 534 537 Binding Energy, eV Fig. 8. O 1s core level spectra: A – MgO support and the sample prepared by impregnation of MgO with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K; B – ␥-Al2 O3 support and the sample prepared by impregnation of ␥-Al2 O3 with a Ru(OH)Cl3 solution followed by reduction with H2 at 723 K. The shaded and open peaks represent the components used for spectra decomposition. 4. Conclusion The data presented in this paper lead to conclude that the initial state energies of Ru atoms in the metal crystallites supported on MgO and acidic supports (␥-Al2 O3 and SiO2 ) are almost identical. Metallic ruthenium deposited on all these oxides is characterized by a deficit of electron density compared to the bulk metal, probably, due to the interaction between Ru atoms and OH groups on the support surface. The mean diameter of supported Ru crystallites and the surface Cl content (at least within the range of variations in our experiment) affect inconsiderably the energy of the initial state of the surface Ru atoms. The negative shift of the Ru 3d core level spectrum which is observed for ruthenium deposited onto MgO with respect to ruthenium supported on acidic supports (␥Al2 O3 , SiO2 ) is caused not by the transfer of electron density from basic sites of MgO, as suggested earlier, but by the differential charging of the supported ruthenium compared with the support surface, which is the final state effect. This phenomenon originates from a higher internal conductivity of the supported ruthenium in comparison to the conductivity of MgO which is a genuine dielectric. The differential charging effect is absent or negligible in the Ru/␥-Al2 O3 and Ru/SiO2 systems, probably, due to much lower electrical resistivity of ␥-Al2 O3 and SiO2 in comparison with MgO. The above statement also is supported by the fact that no differential charging was detected with the samples containing ruthenium metal on the surface of highly conductive carbon supports [44]. The results obtained suggest that the difference in ammonia synthesis activity between the Ru catalysts supported on MgO and on the acidic supports (Al2 O3 , SiO2 ) is accounted for by not different electronic state of ruthenium on the surface of these oxides but by other reasons. Further studies are required in order to clarify these reasons. Acknowledgements The work was supported by the Russian Academy of Sciences (integral project no. 10.4). The authors are grateful to their colleagues from the G.K. Boreskov Institute of Catalysis, Novosibirsk, 1550 Y.V. Larichev et al. / Applied Surface Science 258 (2011) 1541–1550 namely to Prof. A.S. Ivanova and Mrs. N.V. Karasyuk for the preparation of MgO, Prof. E.M. Moroz for assistance in the XRD studies, Dr. V.I. Zaikovskii for TEM measurements, Mrs. I.L. Kraevskaya for X-ray fluorescence analysis, and Dr. I.P. Prosvirin for technical assistance in the XPS experiments. We also thank Prof. V.B. Shur, Dr. S.M. Yunusov and Dr. E.S. Kalyuzhnaya from the A.N. Nesmeyanov Institute of Oranoelement Compounds, Moscow, for testing the catalytic activity of the Ru/EOx samples towards ammonia synthesis. References [1] P. Moggi, G. Predieri, F. Di Silvestri, A. Ferretti, Appl. Catal. A 182 (1999) 287. [2] K. Okabe, I. Takahara, M. Inaba, K. Murata, Y. Yoshimura, J. Jpn. Petrol. Inst. 50 (2007) 65. [3] T.A. Czuppon, S.A. Knez, R.V. Schneider, G. Worobets, Chem. Eng. 100 (1993) 1. [4] Z. Kowalczyk, S. Jodzis,.W. Rarog, J. Zielinski, J. Pielaszek, Appl. Catal. A 173 (1998) 153. [5] K. Aika, Y. Niwa, Stud. Surf. Sci. Catal. 121 (1999) 327. [6] H. Bielawa, O. Hinrichsen, A. Birkner, M. Muhler, Angew. Chem. Int. Ed. 40 (2001) 1061. [7] K. Aika, A. Ohya, A. Ozaki, Y. Inoe, Y. Yasumori, J. Catal. 92 (1985) 305. [8] K. Aika, T. Takano, S. Murata, J. Catal. 136 (1992) 126. [9] M.S. Khaja, R.K.S. Rama, R.P. Rama, Indian J. Chem. A 32 (1993) 383. [10] S.M. Yunusov, B.L. Moroz, A.S. Ivanova, V.A. Likholobov, V.B. Shur, J. Mol. Catal. A 132 (1998) 263. [11] P. Moggi, G. Predieri, A. Malone, Catal. Lett. 79 (2002) 7. [12] M. Muhler, F. Rosowski, O. Hinrichsen, A. Hornung, G. Ertl, Stud. Surf. Sci. Catal. 101 (1996) 317. [13] D.K. Chakrabarty, A. Joshi, S. Unnikrishnan, P.D. Prabhawalkar, React. Kinet. Catal. Lett. 26 (1984) 143. [14] K. Aika, K. Shimazaki, Y. Hattori, A. Ohya, S. Ohshima, K. Shirota, A. Ozaki, J. Catal. 92 (1985) 296. [15] P.C.H. Mitchell, C.E. Scott, J. Bonnelle, J.G. Grimblot, J. Catal. 107 (1987) 482. [16] M.G. Cattania, F. Parmigiani, V. Ragani, Surf. Sci. 211/212 (1989) 1097. [17] S.B. Halligudi, M.M. Taqui Khan, B.L. Moroz, A.L. Chuvilin, I.P. Prosvirin, V.A. Likholobov, React. Kinet. Catal. Lett. 44 (1991) 139. [18] M.P. Seah, D. Briggs (Eds.), Practical Surface Analysis by Auger and Xray Photoelectron Spectroscopy, second ed., Wiley & Sons, Chichester, UK, 1992. [19] K. Aika, K. Tamaru, in: A. Nielsen (Ed.), Ammonia: Catalysis and Manufacture, Springer, Berlin, 1995, p. 104. [20] K. Aika, J. Kubota, Y. Kadowaki, Y. Niwa, Y. Izumi, Appl. Surf. Sci. 121/122 (1997) 488. [21] O. Hinrichsen, F. Rosowski, A. Hornung, M. Muhler, G. Ertl, J. Catal. 163 (1997) 33. [22] S. Dahl, A. Logadottir, C.J.H. Jacobsen, J.K. Norskov, Appl. Catal. A 222 (2001) 19. [23] K.I. Hadjiivanov, G.K. Vayssilov, Adv. Catal. 47 (2002) 307. [24] G. Hohlneicher, H. Pulm, H.-J. Freund, J. Electron Spectrosc. Relat. Phenom. 37 (1985) 209. [25] V.I. Bukhtiyarov, M.G. Slin’ko, Russ. Chem. Rev. 70 (2001) 147. [26] V.I. Bukhtiyarov, I.P. Prosvirin, R.I. Kvon, J. Electron Spectrosc. Relat. Phenom. 77 (1996) 7. [27] V.I. Bukhtiyarov, I.P. Prosvirin, R.I. Kvon, S.N. Goncharova, B.S. Balzhinimaev, J. Chem. Soc. Faraday Trans. 93 (1997) 2323. [28] Y.V. Larichev, B.L. Moroz, I.P. Prosvirin, V.I. Bukhtiyarov, V.A. Likholobov, Chem. Sustainable Dev 11 (2003) 155. [29] Y.V. Larichev, B.L. Moroz, V.I. Zaikovskii, S.M. Yunusov, E.S. Kalyuzhnaya, V.B. Shur, V.I. Bukhtiyarov, J. Phys. Chem. C 111 (2007) 9427. [30] I.K. Igumenov, P.P. Semyannikov, S.V. Trubin, N.B. Morozova, N.V. Gelfond, A.V. Mischenko, J.A. Norman, Surf. Coat. Technol. 201 (2007) 9099. [31] M. Kobayashi, T. Shirasaki, J. Catal. 28 (1973) 289. [32] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer Corp, Eden Prairie (MN, USA), 1992. [33] G. Neri, L. Mercadante, A. Donato, M. Visco, S. Galvagno, Catal. Lett. 29 (1994) 379. [34] L.D. Kuznetsov, L.M. Dmitrenko, P.D. Rabina, Y.A. Sokolinsky, Ammonia Synthesis, Khimia, Moscow, 1982 (in Russian). [35] F. Rosowski, A. Hornung, O. Henrichsen, D. Herein, M. Muhler, G. Ertl, Appl. Catal. A 151 (1997) 443. [36] S. Murata, K. Aika, Appl. Catal. A 82 (1992) 1. [37] C.D. Wagner, C.D. Gale, R.H. Raymond, Anal. Chem. 51 (1979) 466. [38] R.T. Lewis, M.A. Kelly, J. Electron Spectrosc. Relat. Phenom. 20 (1980) 105. [39] G.K. Werthiem, Z. Phys. B 66 (1987) 53. [40] T.L. Barr, E. Hoppe, T. Dugall, P. Shah, S. Seal, J. Electron Spectrosc. Relat. Phenom. 98/99 (1999) 95. [41] R.C. Bartzold, Surf. Sci. 106 (1981) 243. [42] W.M.H. Sachtler, A.Yu. Stakheev, Catal. Today 12 (1992) 283. [43] V.I. Nefedov, D. Gati, B.F. Dzhurinskii, N.P. Serguhin, Y.V. Salyn, Russ. J. Inorg Chem. 20 (1975) 2307. [44] Y.V. Larichev, D.A. Shlyapin, P.G. Tsyrul’nikov, V.I. Bukhtiyarov, Catal. Lett. 120 (2008) 204.