Atomic Structure and Catalytic Activity of W-Modified Ni 2 P Surface Alloy by Photoelectron Diffraction and Spectroscopy

The surface alloying is the important topics in Ni2P study which is expected for next generation hydrodesulfurization and hydrodenitrogenation catalyst. The atomic structure and catalytic properties of a single crystalline Ni2P surface modified with W (W-Ni2P) was investigated by photoelectron diffraction (PED) and spectroscopy. PED is an element and site selective surface structure analysis method that enables observation of three-dimensional atomic configurations of the surface local structure. The selective replacement of W to the Ni site in the Ni2P crystal was clarified by PED. Chemical reactivities for NO molecules of the W-Ni2P and the clean Ni2P surfaces were compared. On the clean Ni2P surface, the NO adsorption did not occur, whereas W-Ni2P surface showed remarkable activity for NO adsorption. [DOI: 10.1380/ejssnt.2014.53]


I. INTRODUCTION
Removal of the sulfide and nitride contents in natural petroleum feedstock is an urgent issue in the refining industry.The surface properties of transitional metal phosphides (TMPs) have attracted much attention as a new catalyst group for hydrodesulfurization (HDS) and hydrodenitrogenation (HDN) reactions [1].Among the TMPs studied so far, Ni 2 P has been found to have the highest activity for HDS and HDN reactions [1].Current studies in Ni 2 P have been revealing the catalytic behavior of bimetallic Ni 2 P modified with Mo, Co and Fe to explore other phosphide catalysts that can be controlled by alloying [2].However, these bimetallic phosphides have not shown enhanced activity until now.Here, we focused on a W modification effect, since early work showed that Ni 2 P and WP have the first and the second highest HDS and HDN activities in TMPs [1].Local atomic structure analysis is necessary for clarifying the active phase of a Ni 2 P surface alloy at an atomic level.
Surface science studies using a single crystalline sample have been conducted to elucidate the relation between the surface atomic structure and catalytic properties.Previous scanning tunneling microscope (STM) and low energy electron diffraction (LEED) studies have observed a long range periodicity of terminated P atoms on a single crystalline Ni 2 P surface [3,4].However, the former analyses did not give a three-dimensional atomic configuration with element and site selectivity.
A photoelectron from a localized core-level is an excellent element selective probe for surface structure analysis [5].Forward focusing peaks (FFPs) appearing in the photoelectron intensity angular distribution (PIAD) indicate the directions of surrounding atoms seen from the photo- * Corresponding author: m-hirosuke@ms.naist.jpelectron emitter atom [6].The atomic distance between the emitter and scatterer atoms can be deduced from the circular-dichroism shift of FFP position [6][7][8] as well as from the opening angles of diffraction rings (DRs) appearing around the FFP [8].Kato et al. succeeded in determining the B dopant site in a superconductive diamond crystal, and showed an ability to analyze dilute species less than two percent [9].
In this study, we have characterized a W-modified Ni 2 P surface (W-Ni 2 P) by PED and x-ray photoelectron spectroscopy (XPS).Element specific PIADs clarified the atomic site of W in the Ni 2 P crystal.Furthermore, the surface reactivity of both the W-modified and clean Ni 2 P surfaces were evaluated by NO adsorption.

II. EXPERIMENTAL
A single crystalline Ni 2 P(1010) surface was used as the substrate [10].This surface has a stoichiometric composition of Ni/P=2 which is same as that of bulk, and was suitable for studying the bulk chemical property.The sample size was 5×10×1 mm 3 .We made an atomically clean surface following the method reported in the previous study [4].The surface was mechanically polished to a mirror finish and was cleaned by Ar + sputtering at 2.5 keV (current: 1 µA, time: 30 min, Ar pressure: 1×10 −3 Pa) and annealing at 350 • C by direct current injection repeatedly after transferring into an ultra-high vacuum below 3×10 −8 Pa.The annealing temperature was monitored by a K-type thermocouple.The surface structure was determined by reflection high-energy electron diffraction (RHEED).A sharp (1×1) structure coming from the clean surface was confirmed.The chemical composition of the sample surface was observed by XPS.
The W-Ni 2 P surface was prepared by depositing W on a clean Ni 2 P surface under H 2 atmosphere with a pressure of 3×10 −3 Pa.W filament with ϕ=0.2 mm and a length PIAD from the sample at a specific kinetic energy was most efficiently measured using a two-dimensional display-type spherical mirror analyzer (DIANA) which was installed at the circularly-polarized soft x-ray beamline BL25SU at SPring-8 in Japan [11].A 2π-steradian PIAD was obtained by scanning of the sample azimuth for 360 • .A pair of PIADs using σ + and σ − helicity light were averaged and the total acquisition time was 2.5 hr.All data were measured at room temperature.The energy window width of DIANA was set to 1 eV for spectroscopy measurements, while it was set to 30 eV for PED measurements, to achieve the best energy and angular resolution, respectively.All PIADs were obtained with a photoelectron kinetic energy of 600 eV, and displayed in azimuth equidistant projection.NO adsorption method is widely used for the study of active surface site for HDS catalyst [12].Figure 3(a) shows the wide area XPS spectra taken before and after NO dosing for both surfaces.NO did not adsorb on the clean surface, while the significant increase of N 1s and O 1s core-level peak intensities were confirmed on the W-Ni 2 P surface.The intensity ratio of O 1s (7.9 counts) versus N 1s (2.9 counts) based on the spectra shown in Fig. 3(a) was estimated considering cross section for O 1s (0.18) and N 1s (0.11) excitations.The intensity of O was 1.7 times as large than that of N, implying that NO gas was dissociated on the surface and a part of N was desorbed.Figure 3(b) shows the detailed XPS spectra for W 4f , Ni 3p and P 2p.In the XPS taken before NO dosing, Ni 3p and P 2p core-level peaks were simple Gaussian dispersions, while W 4f had a shoulder structure shifted by 2 eV to higher binding energy due to spin orbit interaction of 4f 7/2 and 4f 5/2 .After NO dosing, new shoulder structures appeared in the W 4f and P 2p spectra shifted by 2 and 3.5 eV to higher binding energy, respectively.This indicates that the dissociated O oxidized W and P. W-Ni 2 P surface showed a possibility of high catalytic properties which was not observed on the Ni 2 P surface.

IV. CONCLUSION
In conclusion, we have characterized the clean and Wmodified Ni 2 P(1010) surfaces by PED.Element specific PIADs clarified the replacement of W onto the Ni site, while maintaining the Ni 2 P structure.Newly observed W-Ni 2 P on the Ni 2 P substrate showed a remarkable activity for NO adsorption which was not observed on the clean Ni 2 P surface.Therefore, this surface is expected to have a new catalytic properties.PED combined with XPS was an useful analysis tool for clarifying the complex alloyed structure and the catalytic properties.
FIG. 1: (a) Cystalline structure of Ni2P.Red and yellow spheres indicate Ni and P atoms, respectively.(b) The set of atomic arrangement in Ni3P2 and Ni3P layers.Yellow dotted circles indicate the out plane positions of P atoms.(c) and (d) Set of 2π-steradian Ni 3p and P 2p photoelectron intensity angular distributions (PIADs) from the clean Ni2P(1010) surface, respectively.All PIADs were obtained with the kinetic energy of 600 eV and displayed in azimuth equidistant projection.

Figure 1 (
Figure 1(a) shows the crystalline structure of Ni 2 P. Ni 2 P belongs to the space group P 62m of hexagonal symmetry with a=b=0.586nm and c=0.338 nm.In the bulk, two kinds of stoichiometric layers, namely Ni 3 P and Ni 3 P 2 layers, stack alternatively along the [0001] direction.The blue colored rectangle in Fig. 1(a) indicates the {1010} plane which was used in this study.The (1010) surface has a mirror symmetric structure with respect to the {0001} plane.Ni has two kinds of atomic sites which are surrounded by tetrahedral and pyramidal P structures in Ni 3 P 2 and Ni 3 P layers, respectively.Since these tetrahedral and pyramidal Ni sites rotate by 120 • in each other, six kinds of photoelectron emitter sites (Ni1-6) are expected as shown in Fig. 1(b).P has three kinds of emitter sites: P1 and P2 sites in Ni 3 P 2 layer and P3 site in Ni 3 P layer, respectively.Figures 1(c) and 1(d) show the set of 2π-steradian Ni 3p and P 2p PIADs from the clean Ni 2 P(1010) surface.The different structures appeared in each PIAD owing to the different surrounding atomic configuration for each atomic site.The left and right horizontal directions correspond to the [0001] and [0001], respectively.Mirror symmetric structure with respect to the {0001} plane was observed.A Kikuchi-band-like feature, which originated from the bulk atomic arrangement, was observed along the projection of {0001} plane.Each colored dots along the {0001} plane corresponds to the direction shown in Fig. 1(b).In the P 2p PIAD shown in Fig. 1(d), the additional Kikuchi-band-like features at {1100} planes appeared.DR, which corresponds to the bond between the P3 atom and the first neighboring Ni5 atom in Ni 3 P layer, was observed around the [1120] direction shown with dotted line.The opening angle of DR (44 • ) corresponded to the inter atomic distance of 0.24 nm.Figures 2(a), 2(b) and 2(c) are 2π-steradian Ni 3p, P 2p and W 4f PIADs from the W-Ni 2 P surface.The FFP positions, diffraction patterns and their circular dichroism from Ni 3p and P 2p PIADs were basically similar before and after the W deposition.This indicates that the Ni 2 P structure was not changed by the W deposition.The W 4f PIAD was similar to that of Ni 3p but not to that of P 2p.The positions of characteristic low intensity areas were marked with the sky blue dotted circles in Figs.2(a), 2(b) and 2(c).Note that these positions are different for Ni 3p and P 2p, while those of Ni 3p and W 4f coincided.The signal intensity profiles along the {0001} plane for Figs.2(a), 2(b) and 2(c) are displayed in Fig. 2(d).The W and Ni profiles showed similar features, while that of P did not have intensity around the [0110] and [1100] directions.Thus, the substitution of W to the Ni site in the Ni 2 P crystal was clarified.The destination of substituted Ni atom is still an open question.The chemical reactivity difference between W-Ni 2 P and clean Ni 2 P surfaces was studied by NO adsorption.This