Chinese Journal of Chemical Physics  2019, Vol. 32 Issue (6): 753-759

#### The article information

Rui Wang, Ding Ding, Wei Wei, Yi Cui

Near Ambient Pressure Adsorption of Nickel Carbonyl Contaminated CO on Cu(111) Surface

Chinese Journal of Chemical Physics, 2019, 32(6): 753-759

http://dx.doi.org/10.1063/1674-0068/cjcp1904066

### Article history

Accepted on: April 30, 2019
Near Ambient Pressure Adsorption of Nickel Carbonyl Contaminated CO on Cu(111) Surface
Rui Wanga,b , Ding Dinga , Wei Weia , Yi Cuia
Dated: Received on April 1, 2019; Accepted on April 30, 2019
a. Vacuum Interconnected Worksation, Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Science, Suzhou 215123, China;
b. Nano Science and Technology Institute, University of Science and Technology of China, Hefei 230026, China
Abstract: Formation of volatile nickel carbonyls with CO in catalytic reaction is one of the mechanisms of catalyst deactivation. CO is one of the most popular probe molecules to study the surface properties in model catalysis. Under ultra-high vacuum (UHV) conditions, the problem of nickel carbonyl impurity almost does not exist in the case that a high purity of CO is used directly. While in the near ambient pressure (NAP) range, nickel carbonyl is easily found on the surface by passing through the Ni containing tubes. Here, the NAP techniques such as NAP-X-ray photoelectron spectroscopy and NAP-scanning tunneling microscopy are used to study the adsorption of nickel carbonyl contaminated CO gas on Cu(111) surface in UHV and NAP conditions. By controlling the pressure of contaminated CO, the Ni-Cu bimetallic catalyst can form on Cu(111) surface. Furthermore, we investigate the process of CO adsorption and dissociation on the formed Ni-Cu bi-metal surface, and several high-pressure phases of CO structures are reported. This work contributes to understanding the interaction of nickel carbonyl with Cu(111) at room temperature, and reminds the consideration of CO molecules contaminated by nickel carbonyl especially in the NAP range study.
Key words: Nickel carbonyl    Carbon monoxide    Cu(111)    Near ambient pressure X-ray photoelectron spectroscopy    Near ambient pressure scanning tunneling microscopy
Ⅰ. INTRODUCTION

Nickel carbonyl, Ni(CO)$_4$, is mainly used as an intermediate in the Mond process for nickel refining [1], and also used for vapor-plating in the metallurgical process as a catalyst for synthesis of acrylic monomers in the plastics industry [2, 3]. The molecule has a tetrahedral configuration with one nickel atom in the center and four carbonyl ligands attached to it via carbon atoms. Extensive studies on the formation of Ni(CO)$_4$ has been reported [4-6], for example, the kinetic data for the formation of nickel carbonyl were obtained by passing purified carbon monoxide through a fixed bed of freshly reduced nickel powder at temperatures from 25 ℃ to 150 ℃ and pressures from 1 atm to 4 atm [6]. The Ni(CO)$_4$ vapor partially decomposes at temperatures as low as 298-303 K and decomposes quickly in air, with a half-life of about only 40 s [7]. Derouane et al. have studied the preparation of Ni catalysts by decomposing Ni(CO)$_4$ on Al$_2$O$_3$ fibers at 473 K [8]. It is a good way to prepare uniform Ni-containing bimetallic catalyst by dissociating nickel carbonyl completely [9]. For instance, Ni-Cu alloy is one of the most representative bimetallic catalysts for studying the relationship between structure and activity [10-14].

In many catalytic reactions, including hydrogenation, dehydrogenation, hydrogenolysis, isomerization, and Fischer-Tropsch synthesis, Cu-Ni bimetallic catalysts have been studied [15-18]. For many CO evolved catalytic reactions on Ni-based catalysts, the controllable formation and dissociation of nickel carbonyl becomes important in order to keep the catalysts alive. In model catalysis study, CO is usually used as a probe molecular to determine the morphological and/or electronic structure of Ni-Cu catalysts [19-21]. While there is the famous "pressure gap" between model catalysis and real catalysis, some catalytic reaction mechanisms may not be the same in the real catalytic process. Thanks to the near ambient pressure techniques' development in surface science in the last decade, many new insights have been shed light on the understanding of the real catalysis [22].

Here, the near ambient pressure X-ray photoelectron spectroscopy (NAP-XPS) and near ambient pressure scanning tunneling microscope (NAP-STM) are used to re-visit the CO adsorption and activation on Cu(111) surface. While the CO gas is not purified through the liquid nitrogen cold trap, the introduced CO is actually contaminated with traced nickel carbonyl. In the ultra-high vacuum (UHV) range of adsorption, it's not surprising there is no obvious change on Cu(111) surface in the case of dosing the contaminated CO at room temperature and at pressures up to 10$^{-6}$ mbar range, while there appear some CO adsorption structures on the surface when increasing the CO partial pressure to the mbar range. XPS spectrum indicates there are no other pollution like iron carbonyl, chromium carbonyl, except the nickel deposit and nickel carbonyl on the surface. The interaction between CO and the formed copper-nickel bimetallic catalyst is further investigated by in-situ NAP-XPS and in-situ NAP-STM. The process of copper-nickel alloying which derives from the dissociation of nickel carbonyl is observed by in-situ NAP-STM. Detailed STM pictures show that differently periodic CO absorption structures appear on the Ni/Cu interface with elevated pressures. In addition, we investigate that when X-ray participates in the reaction, the XPS spectra signals of carbide and oxide state of Ni are stronger, showing the X-ray can promote the progress of CO dissociation.

Ⅱ. EXPERIMENTS

The experiments were carried out in a combined UHV system, including NAP-XPS (Specs) and NAP-STM (Specs) along with a preparation chamber. The base pressures are in the low 10$^{-10}$ mbar range for all the three chambers. The preparation chamber was equipped with standard cleaning ion source and evaporation sources including an E-beam evaporator for Ni. The NAP-XPS can operate at a pressure range of UHV to 20 mbar and temperature range of 90 K (LN$_2$) to 1500 K (laser heating), while samples in the NAP-STM can be heated to 770 K at UHV or 500 K under NAP condition.

Cu(111) metal crystal (HF-Kejing, China) was cleaned by cycles of Ar$^+$ sputtering at 1 keV and UHV annealing at 770-800 K. The clean and ordered surface was confirmed by XPS and STM. CO gas was used as received from MESSER company (CO purity of 99.999%) without further purification in both NAP-XPS and NAP-STM experiments at various pressures and temperatures. Since the CO gas was demonstrated to be contaminated by nickel carbonyl through the gas tubes in the experiment, the CO gas was mentioned as contaminated CO in below. Ar$^+$ etched tungsten tip was used for NAP-STM and the images were processed with SPECS software.

Ⅲ. RESULTS AND DISCUSSION A. Nickel carbonyl decomposition on Cu(111) surface at room temperature

Comparing most studies carried out in ultrahigh vacuum, surface science study under realistic ambient conditions became possible in the last few years, yet challenges remain under certain conditions. It has been reported that CO adsorption on Cu(111) at room temperature in the pressure range of 0.1 torr to 100 torr will cause the surface to be rough, and even derive the surface to decompose into nanoclusters [22]. In order to explore CO adsorption on Ni-Cu bimetal surface, we tried to perform the CO adsorption on Cu(111) surface firstly. Seen in FIG. 1, under vacuum condition (10$^{-6}$ mbar), there appears no change in Ni 2p, C 1s and O 1s spectra compared to clean Cu(111). While the pressure of CO is increased up to 0.2 mbar, there appears a strong peak for Ni 2p at 852 eV, indicating the formation of metal nickel deposition on Cu(111) surface (FIG. 1(b)). According to the C 1s and O 1s spectra, one can also obviously find that there appeared C and O signal upon CO's pressure increasing to 0.2 mbar. This means that the adsorption and deposition of nickel carbonyl can only be observed under near ambient pressure conditions. While this is not the real "pressure gap" normally mentioned, this means the CO gas is contaminated when passing through the tubes, at near ambient pressure range, the traced nickel carbonyl get enriched on the Cu(111) surface. Several shake-up peaks (941.0, 944.1, 946.2, 960.6, and 966.0 eV) in Cu 2p spectra are visible at CO pressure of 1 mbar, which has been reported to be CO satellite peak [23]. According to the report in Ref.[24], weak features of Cu$_2$O also showed shake-up peaks in Cu 2p spectrum at the above mentioned positions. However, after evacuating the contaminated CO, the satellite peaks disappear (FIG. 1(a)), suggesting there is no copper oxidation and the shake-up is mainly attributed to interface charge transfer or electronic modification between CO and Ni/Cu interface.

 FIG. 1 In-situ XPS spectra of Cu 2p (a), Ni 2p (b), C 1s (c) and O 1s (d) at 1$\times$10$^{-6}$, 0.2, 1, 2.5, 5 mbar nickel carbonyl contaminated CO exposed at room temperature on Cu(111). (e) Ni/Cu ratio obtained for the Cu(111) surface while varying the contaminated CO pressures from 10$^{-6}$ mbar to 5 mbar at room temperature. (f-h) NAP-STM images of the formed Ni-Cu surface alloy by introducing 0.3 mbar contaminated CO onto Cu(111). Scanning parameters including sample bias $V_ \rm{s}$, tunneling current $I_ \rm{t}$, and scanning temperatures $T$ are (f) $V_ \rm{s}$ = +1.358 V, $I_ \rm{t}$ = 0.12 nA, $T$ = 300 K; (g) $V_ \rm{s}$ = +0.0189 V, $I_ \rm{t}$ = 0.49 nA, $T$ = 300 K; (h) $V_ \rm{s}$ = +0.0104 V, $I_ \rm{t}$ = 0.33 nA, $T$ = 300 K.

Both the C 1s and O 1s peaks of CO gas phase move slightly toward higher binding energy when the pressure of CO increases from 10$^{-6}$ mbar to 5 mbar, this is due to the impact of the interactions between photoelectrons and gaseous molecules [25], which is quite common in the NAP-XPS measurements. C 1s peak at 285.4 eV can be ascribed to the adsorption of CO on Ni/Cu surface, which is consistent with the peak position of CO adsorption on nickel single crystal (FIG. 1(c)) [26, 27], C 1s peak at 283 eV may be attributed to CH$_x$ from combination of C species with the background of hydrogen. With the CO pressure increasing, the adsorption peak of nickel carbonyl is weakened by atmospheric scattering. The phenomenon in our case is becoming serious starting from 1 mbar, accompanying the appearance of gas signals of CO at 291 eV and 537 eV for C 1s and O 1s respectively. After the evacuation of CO gas, except for adsorbed CO species, a strong peak at 284 eV evolves which can be assigned to NiC$_x$ and/or carbon laydown on the surface [28-30]. In addition, the O 1s components at 529.3 eV and 531 eV can be assigned to Ni oxide and residual CO molecules on Ni/Cu surface [31, 32], respectively (FIG. 1(d)). These results give strong evidence that CO adsorbed on Ni/Cu surface can be further dissociated into atomic C and O, at least partially, to form NiC$_x$ and NiO$_x$ species. From Ni/Cu ratio against CO pressure in FIG. 1(e), one can see that it reaches a relatively constant value of about 0.11 as soon as Ni is detected by XPS at 0.2 mbar, indicating an easy dissociation of nickel carbonyl on Cu(111) and the formed Ni layer on Cu(111) has a self-limited deposition.

Furthermore, we explored the process by NAP-STM. As seen in FIG. 1(f), deep stripes with distance of 40 Å shows up at CO pressure of 0.33 mbar. Zoom in scan reveals different-oriented sub-structures of these stipes with a periodicity of $\sim$5 Å (FIG. 1(g)). An atomic resolution image is shown in FIG. 1(h) recalling a reconstructed Ni(1$\times$2) missing-row structure. Similar surface reconstruction of the (1$\times$2) phase has been found on Cu(100) and Au(110) exposed to H$_2$ and CO [33, 34], respectively.

Note that in NAP-STM experiments, tunneling current may have an effect on the adsorption, diffusion and dissociation of gas molecules. So we conducted the controllable experiment in which the clean Cu(111) surface was exposed to a certain amount of CO first, then the STM tip was approached without interference factors of tunneling current coefficient. The results show no difference (not shown here).

B. Ni films formation on the Cu(111) surface by MBE method

In order to determine the thickness of the nickel films by dissociation of nickel carbonyl on Cu(111), layered nickel films were prepared using MBE method and characterized by STM and XPS as reference. With increasing amount of Ni deposited, multi-layered Ni films show up before bilayer Ni nano islands dominate the Cu(111) surface (FIG. 2 (a) and (b)). So a quasi-layer by layer growth mode of Ni/Cu(111) by MBE was confirmed by STM. Then the thickness of Ni films made from dissociation of nickel carbonyl could be deduced by comparing the corresponding Ni/Cu ratio with MBE method. For sample shown in FIG. 2(a), the XPS of Ni/Cu ratio is calculated as 0.1, so the deposited Ni film made by dissociation of nickel carbonyl shown in FIG. 1 (f)-(h) is estimated to be approximately 2-3 ML. Such a thickness of Ni film deposited on Cu(111) could also be ascribed as a surface alloy [9].

 FIG. 2 STM images of 2 ML (a) and 4 ML (b) Ni films grown on the Cu(111) surface at 373 K. (c, d) XPS of Cu 2p and Ni 2p of 2 ML and 4 ML Ni/Cu alloy as shown in (a, b). Scanning parameters are (a) $V_ \rm{s}$ = +1.583 V, $I_ \rm{t}$ = 0.2 nA, $T$ = 300 K; (b) $V_ \rm{s}$ = +1.239 V, $I_ \rm{t}$ = 0.14 nA, $T$ = 300 K.
C. Adsorption of nickel carbonyl on Cu(111) step

Coordinately unsaturated defects like steps are usually considered as highly active reaction sites. FIG. 3 shows the step changes of Cu(111) while the CO atmosphere varies from UHV to 10 mbar. Compared to the clean Cu(111) step at UHV as shown in FIG. 3(a), there appears a protruding chain along the step (FIG. 3(b)) with an apparent height of 0.5 Å (FIG. 3(c)) at 1 mbar. By increasing the CO pressure to 3 mbar, we found a new terrace (FIG. 3(d)) and ordered CO adsorption structure appears on terraces which will be discussed in detail later. As we further increased the pressure of CO to 10 mbar, the newly formed terrace expands with an apparent height of $\sim$1.5 Å (FIG. 3(f)). Combining with the results obtained from previous XPS experiment, we consider that nickel carbonyl is firstly adsorbed at Cu(111) steps, then dissociates into nickel and CO, so the Ni-Cu bimetal alloy can form attaching to the Cu(111) steps.

 FIG. 3 Cu(111) step changes induced by contaminated CO. (a) A clean Cu(111), (b) in the atmosphere of 1 mbar contaminated CO. (c) Height profile of across the bright chain in (b). (d) A new terrace formed between steps with CO pressure enhancing to 3 mbar. (e) High resolution STM image showing the new terrace with the CO pressure increasing to 10 mbar. (f) The height profile of the new terrace displayed in (e). Scanning parameters are (a) $V_ \rm{s}$ = +1.488 V, $I_ \rm{t}$ = 0.32 nA, $T$ = 300 K; (b) $V_ \rm{s}$ = +1.165 V, $I_ \rm{t}$ = 0.26 nA, $T$ = 300 K; (d) $V_ \rm{s}$ = +2.433 V, $I_ \rm{t}$ = 0.15 nA, $T$ = 300 K; (e) $V_ \rm{s}$ = +0.0311 V, $I_ \rm{t}$ = 0.33 nA, $T$ = 300 K.
D. CO adsorption on Ni/Cu interface at elevated pressure and at room temperature

The adsorption structure of CO on the surface of nickel single crystal has long been studied in the low pressure range (UHV to 10$^{-6}$ mbar) by traditional surface science tools like LEED [35], STM, HREELS, etc., significantly different phases may evolve at higher pressures which represents the real catalysis in most cases. FIG. 4 demonstrates various CO adsorption structures on Cu-Ni bimetallic surface prepared from nickel carbonyl dissociation at room temperature. As FIG. 4(a) shows, an ordered CO adsorption structure appears at 1 mbar CO pressure with a period of 5.6 Å (red rhombic unit cell, FIG. 4(a)). When CO pressure is increased to 5 mbar, one can see that the CO adsorption structure transforms into a new one with a period of 8.2 Å (green rhombic unit cell, FIG. 4(b)). When the CO pressure is increased to 10 mbar, the adsorption structure (black rhombic unit cell, FIG. 4(c)) does not change, but rotates 19.1$^{\circ}$ compared to the one in FIG. 4(b). Though the Cu(111) surface undergoes a roughening process into nanoclusters at relatively higher pressure of CO as proved by Eren et al. [22], CO hardly adsorbs on metallic Cu surface, especially under UHV conditions. Temperature programmed desorption (TPD) experiments also confirm that CO desorbs from Cu surfaces at 250 K and above [36, 37]. So the CO adsorption structures shown above give another evidence of exposed Ni sites on the top surface of the Ni-Cu surface.

 FIG. 4 In-situ STM of the adsorption and dissociation of CO on Ni/Cu surface. CO structure formed on terraces at 1 mbar (a), 5 mbar (b), 10 mbar (c) respectively, and after evacuation of CO gas (d-f), high resolution STM images showing one hexagonal hole (e), and the high-pressure CO phase (f) from (d). (g) Height profile of the hole in (e). Model of CO adsorption on Ni/Cu interface (h). Scanning parameters are (a) $V_ \rm{s}$ = +1.317 V, $I_ \rm{t}$ = 0.29 nA, $T$ = 300 K; (b) $V_ \rm{s}$ = +0.0064 V, $I_ \rm{t}$ = 0.410 nA, $T$ = 300 K; (c) $V_ \rm{s}$ = +0.0424 V, $I_ \rm{t}$ = 0.44 nA, $T$ = 300 K. (d) $V_ \rm{s}$ = +1.030V, $I_ \rm{t}$ = 0.14 nA, $T$ = 300 K; (e) $V_ \rm{s}$ = +0.0397 V, $I_ \rm{t}$ = 0.43 nA, $T$ = 300 K; (f) $V_ \rm{s}$ = +0.0397 V, $I_ \rm{t}$ = 0.44 nA, $T$ = 300 K.

After evacuation of gas phase CO, one kind of cubic CO adsorption structure dominates the surface with some randomly distributed hexagonal holes co-existing. Such a cubic CO adsorption structure is similar to c(4$\times$2)-CO on Pt(111) surface [38]. As for those hexagonal holes, in the previous analysis of Cu(111) step changes, Ni(CO)$_4$ is considered to adsorb on the steps and dissociate into Ni and CO. With additional activation of CO by Ni atoms, the dissociated oxygen atoms might oxidize the Ni into a hexagonal phase due to the six-fold symmetry of nickel oxide. After Ni-Cu surface reaches a full coverage over Cu(111), due to the competitive effect of CO adsorption on Ni, further deposition of nickel from nickel carbonyl might be prohibited, this kind of self-limited Ni film deposition is correlated well with in-situ NAP-XPS experiment results discussed earlier.

E. CO dissociation on Ni/Cu interface induced by X-ray

Based on the above results and discussion, it is known that pressure is a key factor influencing the adsorption/dissociation of CO on Ni/Cu surface. To find out whether X-ray is involved in the reaction or not, two sets of comparable experiments with or without X-ray radiation were performed. For experiment "without X-ray", the sample was exposed to X-ray only when measuring and the acquisition time for C 1s and O 1s spectrum was about 3 min. After 1 mbar contaminated CO exposure to Cu(111) surface at room temperature for 30 min, with X-ray on during this process shown as the red curve in FIG. 5(a), only weak features of CO adsorption on Ni-Cu surface (285.6 eV) remain, while the signal of NiC$_x$ (284 eV) is strong, indicating CO dissociation and NiC$_x$ formation. On the contrary, in the C 1s region of the one without X-ray irradiation, stronger signal of CO adsorption on Ni/Cu surface (285.6 eV) and weaker signal of NiC$_x$ (284 eV) are detected. Also in FIG. 5(b), the O 1s signal of NiO$_x$ (529.3 and 530 eV) is more intense, indicating a X-ray accelerated dissociation of CO on Ni/Cu surface.

 FIG. 5 XPS of C 1s (a) and O 1s (b) spectra obtained after 1 mbar CO exposure to Cu(111) surface at room temperature for 30 min. The exposure of CO was without X-ray radiation (black curves) and with radiation (red curves). XPS of C 1s (c) and O 1s (d) spectra obtained in-situ in 1 mbar CO exposure to Cu(111) surface at room temperature after the indicated increasing of X-ray exposure time.

To further evaluate the kinetic process of the X-ray induced nickel carbonyl contaminated CO dissociation on Cu(111) surface, NAP-XPS was measured in-situ in 1 mbar CO atmosphere for prolonged time. FIG. 5 (a) and (b) show the XPS spectra of C 1s and O 1s collected every 30 min, respectively. Signal of CO adsorbed on the Ni-Cu surface (285.6 eV) gradually decreases and signal of NiC$_x$ increases (284 eV) with the X-ray radiation on. After the evacuation of CO, only a trace amount of adsorbed CO is remained on the Ni-Cu surface, while C 1s signal of NiC$_x$ and O 1s signal of NiO$_x$ is stronger (FIG. 5 (c) and (d)). Hence the dynamic progress of the adsorption and dissociation process of CO at Ni-Cu surface was followed by NAP-XPS, in which X-ray is identified to accelerate the dissociation of CO.

Ⅳ. CONCLUSION

The adsorption of CO on Cu(111) and Ni-Cu bimetal surfaces was investigated systematically by NAP-XPS and NAP-STM. The Ni-Cu bimetal surface could be prepared by introducing NAP range of nickel carbonyl contaminated CO gas onto Cu(111) at room temperature, and by this preparation method the nickel deposition on Cu(111) surface has a self-limiting increasing coverage up to about 2 ML. Higher exposure of the contaminated CO gas on the surface could form several high pressure superstructure on the bimetal surface. Moreover, the X-ray radiation had been demonstrated to accelerate the dissociation of both nickel carbonyl and CO molecule on the Cu(111) surface. The results and conclusion here could also be regarded as a reference, reminding that the explanation in surface science study especially in the NAP range should be carefully proposed. Multiple techniques used on the same project, especially different techniques carried on the same sample are very necessary to achieve a comprehensive and correct conclusion.

Ⅴ. ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (No.91845109) and Key Laboratory of Surface Physics and Chemistry Discipline Development Fund (XKFZ201711).

Reference
 [1] L. Mond, C. Langer, and F. Quincke, J. Chem. Soc. Trans. 57, 749(1890). DOI:10.1039/CT8905700749 [2] R. S. H. Yang, and E. J. Rauckman, Toxicology 47, 15(1987). DOI:10.1016/0300-483X(87)90158-2 [3] K. Kester, E. Zag, and J. Falconer, Appl. Catal. 22, 311(1986). DOI:10.1016/S0166-9834(00)82638-X [4] G. Greiner, and D. Menzel, J. Catal. 77, 382(1982). DOI:10.1016/0021-9517(82)90180-4 [5] M. M. Windsor, and A. A. Blanchard, J. Am. Chem. Soc. 55, 1877(1933). DOI:10.1021/ja01332a013 [6] W. M. Goldberger, and D. F. Othmer, Ind. Eng. Chem. Process Des. Dev. 2, 202(1963). [7] D. H. Stedman, D. A. Hikade, R. Pearson, and E. D. Yalvac, Science 208, 1029(1980). DOI:10.1126/science.208.4447.1029 [8] E. G. Derouane, J. B. Nagy, and J. C. Védrine, J. Catal. 46, 434(1977). DOI:10.1016/0021-9517(77)90231-7 [9] P. F. A. Alkemade, H. Fortuin, R. Balkenende, and F. H. P. M. Habraken, Surf. Sci. 225, 307(1990). DOI:10.1016/0039-6028(90)90452-E [10] J. H. Sinfelt, Acc. Chem. Res. 10, 15(1997). [11] I. Alstrup, U. E. Petersen, and J. R. Rostrup-Nielsen, J. Catal. 191, 401(2000). DOI:10.1006/jcat.1999.2812 [12] B. Seemala, C. M. Cai, R. Kumar, C. E. Wyman, and P. Christopher, ACS Sustainable. Chem. Eng. 6, 2152(2017). [13] E. T. Saw, U. Oemar, X. R. Tan, Y. Du, A. Borgna, K. Hidajat, and S. Kawi, J. Catal. 314, 32(2014). DOI:10.1016/j.jcat.2014.03.015 [14] A. R. Naghash, T. H. Etsell, and S. Xu, Chem. Mater. 18, 2480(2006). DOI:10.1021/cm051910o [15] J. H. Sinfelt, J. Catal. 29, 308(1973). DOI:10.1016/0021-9517(73)90234-0 [16] H. H. Brongersma, and M. J. Sparnaay, Surf. Sci. 71, 657(1978). DOI:10.1016/0039-6028(78)90453-3 [17] M. Araki, and V. Ponec, J. Catal. 44, 439(1976). DOI:10.1016/0021-9517(76)90421-8 [18] J. A. Dalmon, and G. A. Martin, J. Catal. 66, 214(1980). DOI:10.1016/0021-9517(80)90023-8 [19] E. Asedegbega-Nieto, A. Guerrero-Ruíz, and I. Rodríguez-Ramos, Thermochim. Acta 434, 113(2005). DOI:10.1016/j.tca.2005.01.026 [20] D. T. Ling, and W. E. Spicer, Surf. Sci. 94, 403(1980). DOI:10.1016/0039-6028(80)90015-1 [21] Y. Yao, and D. W. Goodman, Phys. Chem. Chem. Phys. 16, 3823(2014). DOI:10.1039/c3cp54997f [22] B. Eren, D. Zherebetskyy, and L. L. Patera, Science 351, 475(2016). DOI:10.1126/science.aad8868 [23] X. Zhang, and S. Ptasinska, Phys. Chem. Chem. Phys. 8, 1632(2016). [24] G. Panzner, B. Egert, and H. P. Schmidt, Surf. Sci. 151, 400(1985). DOI:10.1016/0039-6028(85)90383-8 [25] F. F. Tao, and L. Nguyen, Phys. Chem. Chem. Phys. 20, 9812(2018). DOI:10.1039/C7CP08429C [26] T. Fleisch, G. L. Ott, W. N. Delgass, and N. Winograd, Surf. Sci. 81, 1(1979). DOI:10.1016/0039-6028(79)90501-6 [27] C. R. Brundle, and A. F. Carley, Faraday Discuss. Chem. Soc. 60, 51(1975). DOI:10.1039/dc9756000051 [28] A. Furlan, J. Lu, L. Hultman, U. Jansson, and M. Magnuson, J. Phys.: Condens. Matter. 26, 415501(2014). DOI:10.1088/0953-8984/26/41/415501 [29] N. Mahata, A. F. Cunha, J. J. M. Órfão, and J. L. Figueiredo, ChemCatChem. 2, 330(2010). DOI:10.1002/cctc.200900299 [30] I. Czekaj, F. Loviat, F. Raimondi, J. Wambach, S. Biollaz, and A. Wokaun, Appl. Catal. A-Gen. 329, 68(2007). DOI:10.1016/j.apcata.2007.06.027 [31] R. Ebrahim, A. Zomorrodian, N. Wu, and A. Ignatiev, Thin Solid Films 539, 337(2013). DOI:10.1016/j.tsf.2013.04.142 [32] A. N. Mansour, Surf. Sci. Spectra 3, 231(1994). DOI:10.1116/1.1247751 [33] R. T. Vang, E. Laegsgaard, and F. Besenbacher, Phys. Chem. Chem. Phys. 9, 3460(2007). DOI:10.1039/B703328C [34] Y. Jugnet, F. J. C. S. Aires, C. Deranlot, L. Piccolo, and J. C. Bertolini, Surf. Sci. 521, 639(2002). DOI:10.1016/S0039-6028(02)02295-1 [35] J. Engbæk, O. Lytken, J. H. Nielsen, and I. Chorkendorff, Surf. Sci. 602, 733(2008). DOI:10.1016/j.susc.2007.12.008 [36] K. Christmann, O. Schober, and G. Ertl, J. Chem. Phys. 60, 4719(1974). DOI:10.1063/1.1680972 [37] H. Koschel, G. Held, and H. P. Steinruck, Surf. Sci. 453, 201(2000). DOI:10.1016/S0039-6028(00)00349-6 [38] X. Xu, and D. W. Goodman, J. Phys. Chem. 97, 683(1993). DOI:10.1021/j100105a025 [39] H. J. Yang, T. Minato, M. Kawai, and Y. Kim, J. Phys. Chem. C 117, 16429(2013). DOI:10.1021/jp404231t

a. 中国科学院苏州纳米技术与纳米仿生研究所，纳米真空互联实验站，苏州 215123;
b. 中国科学技术大学纳米科学技术学院，合肥 230026