J. Phys. Chem. B 1997, 101, 9161-9164
9161
Theoretical Calculations and XPS Studies of the Adsorption of NO on a Single Crystal of LiNbO3 Kenji Tabata,* Masahiko Kamada, Tetsuo Choso, and Hiroaki Munakata Research Institute of InnoVatiVe Technology for the Earth (RITE), Kizugawadai, Kizu-cho, Soraku-gun, Kyoto 619-02, Japan ReceiVed: April 1, 1997X
Density functional theory (DFT) was applied to study nitrogen monoxide adsorption and reactivity on LiNbO3 surface, and the results were compared with the experimental ones for a single-crystal sample by X-ray photoelectron spectroscopy (XPS). The total density of state of the cluster model of LiNbO3 (Li15Nb4O22H8) well represented the experimental result. This agreement supports the accuracy of the procedure of this DFT computation. An unrestricted local density functional (LDF) geometry optimization of NO on two surface cluster models, Li2Nb2O10H6 and Li2Nb2O9H6, which had one or two vacant coordination sites of oxygen atoms around niobium, were converged well as NO3- and NO2-, respectively, by adding one electron. These results agreed well with the assignments of observed peak of the spectrum for the N 1s level after the adsorption of NO at room temperature on every facial cut sample, x-, y-, and z-cut, with XPS. The coordinatively unsaturated surface lattice oxygen atoms reacted with NO, and these reacted oxygen atoms were introduced into the adsorbates and produced NO3- and NO2- as shown in both the results of the computed simulation and the spectrum for the N 1s level of the exposed sample at room temperature. This explanation for the reactivity of coordinatively unsaturated surface lattice oxygen was confirmed also from the change of the spectrum for the O 1s level of every facial cut sample before and after NO exposure at room temperature.
Introduction The oxygen species and its reactivity on a heterogeneous catalyst have been reviewed.1 It is pointed out in the review that an oxygen species is an important factor to control selectivity. It is generally assumed that in, for example, allylic oxidation, an intermediate formed on the surface is oxidized by a specific type of lattice oxygen of the catalyst rather than an adsorbed oxygen species to form the reactive products such as acrolein. On an oxide surface, a coordinatively unsaturated oxygen atom is generally thought to play an important role in a catalytic reaction.2 The interaction of this type of oxygen at low coordinate site with a simple molecule such as NO has been examined.3 Oxygen exchange between N18O and Mg16O through an NO2-type intermediate involving lattice oxygen, which may be a coordinatively unsaturated oxygen atom, was suggested from the observed temperature-programmed desorption (TPD) spectrum. We examined previously the adsorbed states of NO on a z-cut single crystal of LiNbO3 with X-ray photoelectron spectroscopy (XPS). NO reacted with a sample at room temperature and produced the broad peak on the photoline of the C 1s level around 406 eV.4 This broad peak was assigned as the overlap of each spectrum for NO2- and NO3-. However, the exact process on the reaction of NO with the surface oxygen of LiNbO3 still remains unclear. On the other hand, a perovskite-type mixed oxide is convenient to study the role of oxygen species in a reaction because an adsorbed oxygen and a lattice one can be easily discerned with XPS.5 In this work, the adsorbed states of NO on a LiNbO3 with ilumenite structure have been studied theoretically with the density functional theory (DFT), and experimentally on a single crystal of each facial cut sample of x, y, and z by XPS. The objective * To whom correspondence should be addressed. E-mail:
[email protected]. Tel: +81 774 75 2305. Fax: +81 774 75 2318. X Abstract published in AdVance ACS Abstracts, October 15, 1997.
S1089-5647(97)01140-1 CCC: $14.00
of this study is to clarify the reactivity of coordinatively unsaturated oxygen species to NO at room temperature. Experimental Section A single crystal of LiNbO3 was obtained from Nippon Kessho Kogaku Co., Ltd. The size of a sample was 13 mm × 13 mm square, and the thickness was 1 mm. This plate was polished to optical finish by the manufacturer. The samples were x-cut, y-cut, and z-cut (each axis normal to the plane). XPS experiments were performed with an angle-resolved Shimadzu ESCA-KM spectrometer which was equipped with a concentric hemispherical analyzer. Mg KR (1253.6 eV) X-ray source was used for the excitation. The spectrometer worked under a base pressure of 7 × 10-8 Pa in a chamber. NO exposure was carried out in a preparation chamber. The sample was indirectly heated by a resistivity heater. Binding energy (BE) was calibrated with respect to the C 1s value of a contaminated carbon as 285.0 eV. Photoelectron was detected at an angle of 15° to the surface of a sample. The treatment of acquired spectrum was carried out with the software Vision produced by Kratos Analytical. The atomic ratio was calculatd from each photoline area with its sensitivity factor by Wagner.6 NO gas was UHP grade (>99.5%) obtained from Ueno Gas Co., Ltd. NO exposure was carried out in a preparation chamber of XPS. The sample was exposed to NO under the pressure of 6.6 × 103 Pa for 1 h at room temperature. After evacuation, XPS photoline was observed under the pressure of 10-6 Pa. DFT Calculation Method Cluster electronic energies and energy gradients were computed mostly at the local density functional (LDF) level of theory with DMol using the VWN local correction functionals derived in the Vosko-Wilk-Nusair parametrization of Ceperley and Alder’s electron gas Monte Carlo data.7-11 The atomic orbital basis set used throughout was the standard DMol double © 1997 American Chemical Society
9162 J. Phys. Chem. B, Vol. 101, No. 45, 1997
Tabata et al.
Figure 1. Cluster model, Li15Nb4O22H8 (CL-1).
numerical + polarization (DNP) basis obtained from exact numerical LDF calculations on the constituent atoms and the cationic states of these same atoms. Additionally, only the molecular orbital (MO) coefficients corresponding to valence orbital were included in the self-consistent field (SCF) procedure. Thus the MO coefficients corresponding to the core levels such as Nb 3d, O 1s, and N 1s were fixed at their atomic values in the molecular SCF procedure. DMol has been enhanced by implementation of Becke’s 1988 version of a gradient corrected functional (B88) and the Lee-Yang-Parr (1988) correlation functional (LYP).12,13 A nonlocal geometry optimization with the B88 and Lee-Yang-Parr functional was performed on the system of NO adsorbed on the cluster model. Insight II14 and the modules developed by the Catalysis & Sorption Project15 were used for the graphics display and analysis.
Figure 2. Comparison of the computed total DOS with the experimental spectrum of XPS.
Results and Discussion Cluster Model Preparation and Geometry Optimization. The crystal system of LiNbO3 is rhombohedral with lattice constants aH ) 5.14829 ( 0.00002 Å, cH ) 13.8631 ( 0.0004 Å, and space group R3c (C3c6) at 23 °C.16 The X-ray crystal structure of LiNbO3 was built,17and a part of the crystal surface [110] was extracted and modeled to get the cluster model, Li15Nb4O22H8 (CL-1), as shown in Figure 1. Each of four Nb atoms is coordinated by six oxygen atoms, and there are two of a bridging oxygen atom connecting two Nb atoms. The oxygen dangling bonds underneath the crystal [110] surface were terminated by adding hydrogen atoms in the direction of connecting niobium atom. An unrestricted LDF energy calculation was performed on this cluster model, and then total density of state (TDOS) was computed with total charge of -1 and compared with the experimental XPS spectrum of a z-cut single crystal of LiNbO3 in Figure 2. The peak P1 of the computed TDOS graph was aligned with the valence band (VB) peak A of the experimental graph. The peaks P2 and P3 corresponded to the peaks B (O 2s) and C (Nb 4p). The peaks P4 and P5 seem to correspond to the peaks of Li 1s and Nb 4s, respectively.18 However, these peaks on the experimental spectrum at that region (D) are not clear. The DFT-computed electronic structure of the cluster model, CL-1, well represented the experimental result. This means that the procedure of this DFT computation can be supported for its accuracy. In order to facilitate the DFT geometry optimization, and to reduce the number of atoms in a cluster model, the smaller size cluster model, Li2Nb2O10H6 (CL-2), was extracted from the cluster model CL-1. To model NO adsorption and reaction site, one oxygen atom with dangling bond located at central part of
Figure 3. (A) Cluster model, Li2Nb2O10H6 (CL-2), and (B) the geometrically optimized structure of NO adsorption.
the cluster was removed to produce a vacant coordination site on Nb1 atom. The dangling bonds of two oxygen atoms which were located above the surface plane and far from the cluster model center were terminated by adding hydrogen atoms. The prepared cluster model, CL-2, is shown in Figure 3A. Nitrogen monoxide was placed near the vacant site on the Nb1 atom of the first surface cluster model, CL-2. An unrestricted LDF geometry optimization was performed only relaxing the Cartesian coordinates of atoms, N1, ON1 O1, O2, and O3, but subject to a geometrical constrain on all other atoms by fixing Cartesian coordinates of those atoms. The geometry optimization on this system had some difficulty with SCF convergence, but by adding one electron to this system, which changed the SCF active electron number from odd to even, the geometry optimization was well achieved. The optimized structure is
Adsorption of NO on a Single Crystal of LiNbO3
J. Phys. Chem. B, Vol. 101, No. 45, 1997 9163
TABLE 1: Geometrically Optimized Interatomic Distances of the Cluster Model Li2Nb2O10H6 (CL-2) before and after NO Adsorption geometrically optimized interatomic distances
Nb1-O1 Nb2-O2 Nb2-O3 O2-O1 Li1-O1 Li1-O3 Li2-O3 Nb-O (lattice) Nb1-N1 Nb1-ON1 N1-ON1 N1-O1 N1-O2
before NO adsorption (Å)
after NO adsorption (Å)
1.76 1.79 1.99 2.63 2.22 2.02 1.79 2.07
2.18 2.06 1.87 2.79 1.91 2.09 1.78 2.07 2.00 2.18 1.34 1.38 1.33
TABLE 2: Interatomic Distances (Å) in LiNbO3 Which Were Observed from the X-ray Diffraction Experiment at 24 °C19 Nb-O
1.889 ( 3 2.112 ( 4
Li-O
2.068 ( 11 2.238 ( 23
O-O
2.719 ( 4 2.801 ( 1 2.840 ( 1 2.879 ( 4 3.042 ( 2 3.362 ( 4
shown in Figure 3B. The optimized structure in Figure 3B showed that two oxygen atoms with dangling bonds, O1 and O2, coordinated to Nb1 and Nb2 atoms formed bonds with the nitrogen atom N1, representing a N1-O1 bond distance of 1.38 Å and a N1-O2 bond distance of 1.33 Å. The atoms of both N1 and ON1, derived from NO substrate, coordinated to Nb1 atom, showed a Nb1-N1 distance of 2.00 Å and a Nb1-ON1 distance of 2.18 Å. The Li1-O1 distance of 2.22 Å converged as 1.91 Å after the geometry optimization. From these results, it was suggested that NO adsorbed as NO3- on the crystal, and a coordinatively unsaturated surface lattice oxygen atom such as O1 or O2 reacted with NO and was involved in NO3- as shown in Figure 3B. Since the distances of N1-Nb1 and O1Li1 were 2.00 and 1.91 Å, respectively, it was suggested that the electronic states of adsorbed NO3- were affected by both Nb and Li ions. Each result of geometrically optimized interatomic distance before and after NO exposure is tabulated in Table 1. The result of experimantal interatomic distance with the X-ray diffraction by Abrahams et al. is also shown in Table 2 for comparison.19 The second surface cluster model, Li2Nb2O9H6 (CL-3), was prepared by removal of one more dangling bond oxygen atom from Nb2 in the cluster model, CL-2, as shown in Figure 4A. NO was located in the space of neighboring vacant sites of Nb1 and Nb2, where the nitrogen atom N1 of NO was placed at the vacant site of Nb1 and the oxygen atom ON1 was at the vacant site of Nb2. An unrestricted LDF geometry optimization was carried out on the NO adsorbed cluster model, only relaxing the Cartesian coordinates of atoms N1, ON1, O1, and O3, but was subjected to a geometrical constrain on other atoms by fixing Cartesian coordinates of these atoms. The geometry optimization with one electron addition to this system, was well converged. The optimized geometry of the NO adsorbed cluster model is shown in Figure 4(B). Since the optimized structure showed the distances of N1-O1 and N1-ON1 bonds were 1.44 and 1.34 Å, respectively, it was suggested that NO adsorbed as NO2- on the cluster model of CL-3, and a coordinatively unsaturated lattice oxygen atom O1 reacted and was involved in NO2- as shown in Figure 4B. The distances of N1-Nb1 and O1-Li 1 were 2.09 and 1.89 Å, respectively. The adsorbed NO2- seemed to be affected by both Nb and Li ions.
Figure 4. (A) Cluster model, Li2Nb2O9H6 (CL-3), and (B) the geometrically optimized structure of NO adsorption.
Figure 5. N 1s level spectra for a preheated z-cut sample without NO exposure, and samples with NO exposure samples, x-, y-, and z-cut at room temperature under the pressure of 6.6 × 103 Pa for 1 h.
XPS Experiment. The sample of a single crystal of LiNbO3 was preheated at 400 °C for 1 h in a preparation chamber of XPS, and then exposed to NO at room temperature. The spectrum of the N 1s level for the preheated z-cut sample without NO exposure is shown in Figure 5. There is no clear peak in this region after the heating at 400 °C for 1 h. Both samples of x-cut and y-cut also have no clear peak in this region. The broad peak near 406 eV appeared after NO exposure under the pressure of 6.6 × 103 Pa for 1 h at room temperature in every facial cut sample as shown in Figure 5. These peak positions are almost the same. XPS studies of the adsorption of NO and NO2 on various transition metals and metal oxides have been reported.20-22 In these reports, the peaks of surface species such as NO3- and NO2- were appeared at around 407 and 404 eV, respectively. The values of the peak position and the full width at half-maximum (fwhm) of a z-cut sample on the broad peak in Figure 5 are 405.9 and 3.2 eV, respectively. Our experimental values of the peak position and the fwhm of LiNO3 on the chemical reagent, obtained from Wako Pure Chemical Industries Ltd., were 408.1 and 1.68 eV, respectively. By comparison these values and the reported values 20-22 with our observed broad peaks in Figure 5, the broad peak should be ascribed to overlap of each spectrum for NO2- and NO3- on
9164 J. Phys. Chem. B, Vol. 101, No. 45, 1997
Tabata et al. the involvement into NO3- and NO2- at room temperature. On the other hand, the small increase of the shoulder peak, as shown in Figure 6, is supposed to come from the constituent oxygen in the adsorbates, NO3- and NO2-. Conclusions
Figure 6. O 1s level spectra for a preheated z-cut sample with and without NO exposure at room temperature under the pressure of 6.6 × 103 Pa for 1 h.
the crystal. Since the value of the peak position of every sample is smaller than that of LiNO3, we conceive that this shift explains the produced NO2- and NO3- adsorb on niobium ions as suggested from the theoretically optimized structures in Figures 3 and 4. The effects of adsorption between a terminal oxygen and a lithium ion must be also included in this shift. We consider the reason the addition of one electron is needed to converge the geometrical optimization. This is not clear, however, we considered as follows. It may come from the added hydrogen atoms on the terminal sites, which are needed to neutralize. Another reason may come from the surface composition. The atomic ratio of Li/Nb was calculated from each peak intensity of Li 1s and Nb 3d. The ratios of x-cut, y-cut, and z-cut after the preheating at 400 °C for 1 h are 0.24, 0.36, and 0.33, respectively. Such a loss of lithium on the eliminating process has been reported.18,23 Schirmer et al. suggested that the hole in LiNbO3 was associated with a defect, and a likely defect candidate was a Li vacancy.24 The presence of this Li vacancy may be postulated by the addition of one electron, and this localized electron may be needed to start the reaction with NO, and to converge the geomerical optimization. The surface seemed to be partly reduced from the loss of lithium by the preheating for 1 h in a chamber of XPS. On this surface, NO adsorbed as NO2- and NO3- at room temperature. This result of XPS experiment agrees well with those of the theoretically optimized structure of NO adsorption on a LiNbO3 in Figures 3 and 4. This means that the reduction by the preheating at 400 °C for 1 h is weak, and the surface partly keep the state of the crystal of LiNbO3. Figure 6 shows the spectra for the O 1s level of the preheated sample with and without NO exposure. These spectra have a peak near 530 eV, and a shoulder peak on a higher binding energy side. The peak position of the O 1s photoline for the lattice oxygen of LiNbO3 was reported as 530 eV.18 Our experimental result of the peak position of the chemical reagent LiNO3 for the O 1s level is 534.0 eV, and this peak position is included in the shoulder peak of the spectra in Figure 6. From these results, we can say that the peak of the lattice oxygen of LiNbO3 decreased after the exposure to NO at room temperature; however, the shoulder peak near 534 eV increased in comparison with that of the sample which was merely pretreated at 400 °C as shown in Figure 6. This change in Figure 6 can be explained well by comparing with the results of the theoretical simulation. The optimized adsorbed structure for the surface cluster models of CL-2 and CL-3 are NO3- and NO2-, respectively, as shown in Figures 3B and 4B. For producing the NO3- and NO2-, the coordinatively unsaturated lattice oxygen atoms were reacted with NO and introduced into the adsorbates at room temperature. From these theoretical simulations, the decrease of the lattice oxygen at 530 eV in Figure 6 is supposed to mean the reaction of the coordinatively unsaturated lattice oxygen with NO and
Density functional theory (DFT) was applied to study nitrogen monoxide adsorption and reactivity on LiNbO3 surface, and the results were compared with the experimental results on a single crystal with XPS. The total density of state of the cluster model of LiNbO3 (Li15Nb4O22H8) well represented the experimental one. This agreement supports the accuracy of the procedure of this DFT computation. An unrestricted local density functional (LDF) geometry optimization of NO on two surface cluster models, Li2Nb2O10H6 and Li2Nb2O9H6, which had one or two vacant coordination sites of oxygen atoms around niobium, were converged well as NO3- and NO2-, respectively, by adding one electron. These results agreed well with the assignments of observed peak of the spectrum for the N 1s level after the adsorption of NO at room temperature on every facial cut sample, x-, y-, and z-cut, with XPS. The coordinatively unsaturated surface lattice oxygen atoms reacted with NO, and these reacted oxygen atoms were involved into the adsorbates, and produced NO3- and NO2- as shown in both the results of geometrically optimized structure and the spectrum for the N 1s level for the exposed sample at room temperature. This explanation on the reactivity of coordinatively unsaturated surface lattice oxygen was convinced also from the change of the spectrum for the O 1s level of every facial cut sample before and after NO exposure at room temperature. Acknowledgment. This work has been supported by the New Energy and Industrial Technology Development Organization (NEDO). References and Notes (1) Che, M.; Tench, A. J. AdV. Catal. 1983, 32, 1. (2) Henrich, V. E.; Cox, P. A. Appl. Surf. Sci. 1993, 72, 277. (3) Yanagisawa, Y. Appl. Surf. Sci. 1995, 89, 251. (4) Tabata, K.; Kamada, M.; Choso, T.; Munakata, H. Appl. Surf. Sci., submitted for publication. (5) Tabata, K.; Kohiki, S. J. Mater. Sci. 1987, 22, 1882. (6) Wagner, C. D. Anal. Chem. 1977, 49, 1282. (7) Delley, B. J. Chem. Phys. 1990, 92, 508. (8) Delley, B. J. Chem. Phys. 1991, 94, 7245. (9) D Mol, BiosymTechnologies, San Diego, 1992, version 2.2.0; 1993, version 2.3.0. (10) Vosko, S. J.; Wilk, L.; Nuair, M. Can. J. Phys. 1980, 58, 1200. (11) Ceperley, D. M.; Alder, B. J. Phys. ReV. Lett. 1980, 45, 566. (12) Becke, A. D. J. Chem. Phys. 1988, 88, 2547. (13) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 786. (14) Insight II, Biosym Technologies, San Diego, CA, 1993, version 2.3.0. (15) Catalysis & Sorption Project modules, Biosym Technologies, San Diego, CA, 1993, version 4.0. (16) Abrahams, S. C.; Reddy, J. M.; Bermstein, J. L. J. Phys. Chem. Solids 1966, 27, 997. (17) Megaw, H. D. Acta Crystallogr. 1968, A24, 583. (18) Courths, R.; Steiner, P.; Hochst, H.; Hu¨fner, S. Appl. Phys. 1980, 21, 345. (19) Abrahams, S. C.; Reddy, J. M.; Bernstein, J. L. J. Phys. Chem. Solids 1966, 27, 997. (20) Swartz Jr., W. E.; Youssefi, M. J. Electron Spectrosc. 1976, 8, 61. (21) Matsuta, H.; Hirokawa, K. Appl. Surf. Sci. 1988, 35, 10. (22) Roberts, M. W.; Smart, R. St. C. Surf. Sci. 1980, 100, 590. (23) Cha´b V.; Kuba´tov´, J. Appl. Phys. 1980, 21, 345. (24) Scirmer, O. F.; Linde von der, D. Appl. Phys. Lett. 1978, 33, 35.
