Yu-qing Song,Xiao-ping Wang

a.Department of Physics,University of Science and Technology of China,Hefei 230026,China

b.Hefei National Lab oratory for Physical Sciences at the Microscale,University of Science and Technology of China,Hefei 230026,China

c.Synergetic Innovation Center of Quantum Information&Quantum Physics,University of Science and Technology of China,Hefei 230026,China

Layer Dependence of Graphene for Oxidation Resistance of Cu Surface

Yu-qing Songa?,Xiao-ping Wanga,b,c

a.Department of Physics,University of Science and Technology of China,Hefei 230026,China

b.Hefei National Lab oratory for Physical Sciences at the Microscale,University of Science and Technology of China,Hefei 230026,China

c.Synergetic Innovation Center of Quantum Information&Quantum Physics,University of Science and Technology of China,Hefei 230026,China

(Dated:Received on October 9,2016;Accepted on November 4,2016)

We studied the oxidation resistance of graphene-coated Cu surface and its layer dependence by directly growing monolayer graphene with different multilayer structures coexisted,diminishing the influence induced by residue and transfer technology.It is found that the Cu surface coated with the monolayer graphene demonstrate tremendous difference in oxidation pattern and oxidation rate,compared to that coated with the bilayer graphene,which is considered to be originated from the strain-induced linear oxidation channel in monolayer graphene and the intersection of easily-oxidized directions in each layer of bilayer graphene, respectively.We reveal that the defects on the graphene basal plane but not the boundaries are the main oxidation channel for Cu surface under graphene protection.Our finding indicates that compared to putting forth efforts to improve the quality of monolayer graphene by reducing defects,depositing multilayer graphene directly on metal is a simple and effective way to enhance the oxidation resistance of graphene-coated metals.

Graphene,Chemical vapor deposition,Oxidation resistance,Defect

I.INTRODUCTION

Graphene,a two-dimensional single-atom thick sheet of carbon,has attracted increasing interest due to its unique properties[1–3]and potential applications[4–6].Aside from its excellent electrical property[1]and strength[3],graphene has promising use as an ultrathin protection layer on metal corrosion because of its chemical inertness[7],impermeable to standard gases [8],and exceptionally transparency[2,9].At the same time,the chemical vapor deposition(CVD)has been applied to prepare high-quality graphene on various metals directly,including Cu[10],Ni[11],Pt[12],Ir[13], making the CVD graphene be an ideal optical thin layer for protecting metal surfaces.Chen et al.reported that CVD graphene improves the oxidation resistance of Cu and Cu/Ni[14].However,a number of small oxides, which are shown as bright white spots,are still formed on graphene surface.This is due to the fact that the real CVD graphene is imperfect and wrinkles,point defect as well as graphene boundaries is inevitable during the graphene growth.As to the oxidation mechanism, most previous investigations reveal that the oxygen diffusion starts from the graphene grain boundaries[15–17],which are always considered as the primary factors for the inferior quality of the CVD graphene.However,recent studies demonstrate that the wrinkles of graphene rather than the Cu grain boundaries or the graphene domain boundaries to facilitate the oxygen diffusion and oxidation of Cu surface[18,19].As a result,grain boundaries or wrinkles,which are the dominant factors for oxygen diffusion and oxidation process are still an open question.

It is intuitively that the oxidation resistance of Cu surface can be improved by using multiple graphene layers coating.In this regards,Prasai et al.[20]and Roy et al.[17]reported the ways to increase the degree of protection by transferring multiple graphene layers onto target surfaces.However,it is worth mentioning that the complicated and skilled transfer techniques inevitably resulted in the degradation of materials due to the formation of wrinkles and cracks,and the removal of adhesive residue is not complete[21,22].All of these can bring a lot of external factors for the study on the oxidation resistance of graphene coated Cu surface.

In order to eliminate the problems mentioned above, we investigate the layer dependence of the oxidation resistance by directly CVD growing graphene with different multilayer structure coexisted on the Cu surface. At the same time,we carefully adjust the trace O2contained in Ar/O2mixture to control the oxidation rate, which enable us to improve the controllability and to observe the initial state of oxidation.Optical microscope,Raman spectroscopy,X-ray photoelectron spectroscopy(XPS)and X-ray excited Auger electron spec-troscopy(XAES)were used to gain more insight into the graphene layer dependence of oxidation resistance of Cu surface.We found that the Cu surface coated by the monolayer graphene presents much difference in oxidation pattern and oxidation rate from that coated by the bilayer graphene.The different behavior is considered to be originated from the strain-induced linear oxidation channel in monolayer graphene and the intersection of easily-oxidized directions in each layer of bilayer graphene,respectively.The results can be further verified by the oxidation experiment performed on the different temperatures and the corresponding Arrhenius plots analysis.Our results imply that it is the defects on the graphene basal plane but not the boundaries are the main oxidation channel for Cu surface under graphene protection.Moreover,instead of transfer,direct growth of multilayer graphene on the Cu surface is an effective and practicable method to protect the Cu surface from oxidation.

II.EXPERIMENTS

The monolayer graphene with coexisting multi-layer structure was directly grown onto the Cu foil using traditional CVD method.A piece of 2 cm×2 cm Cu foil (Alfa Aesar,item No.46365)was inserted into a small quartz tube with a diameter of 2.5 cm.After the tube was placed in the CVD furnace with a set of slides, the system was heated to 1045°C in a 270 sccm Ar flow;then graphene could be directly formed on thefoil at 1045°C with a flowing gas mixture of CH4/Ar (406 ppm CH4,15 sccm),H2(8 sccm,99.999%),and Ar(247 sccm,99.999%)at ambient pressure.In order to make a direct comparison between the area with and without graphene coverage in oxidation resistance, graphene growth time was limited in 4 h,so that some of the substrate could not be covered.After growth, the sample was cooled rapidly to room temperature by simply pushing the furnace away from it.

In addition to the controllability of the experiment, the appropriate oxidation temperature should be chosen carefully.Extra-high temperature will lead the oxidation process too fast to observe the initial oxidation state.On the contrary,extra-low temperature will make the oxidation too slow to perform the experiment in the practicable time.Accordingly,we set the oxidation temperature between 500 and 650°C.In the oxidation process,the graphene-coated Cu foil was put into the tube furnace with constant flowing gas.The gas is mixture of Ar(500 sccm,99.999%)and O2/Ar (100 ppm O2,70 sccm).

The process of oxidization on the surface of graphenecoated Cu surface was characterized by optical microscopy(Bx51,OLYMPUS),XPS,and XAES(ESCALAB 250 X-ray photoelectron spectrometer with a monochromatic Al Kα radiation(hν=1486.6 eV)),as well as Raman spectra obtained by a LABRAM-HR Raman spectrometer with an excitation wavelength of 633 nm generated by an Ar+laser.

III.RESULTS AND DISCUSSIONS

The monolayer graphene with coexisted multi-layer structure was successfully grown onto the Cu foil by the CVD method.As shown in Fig.S1 in the supplementary materials,no defect-activated peak D at～1350 cm-1can be observed in both spectra of monolayer and multilayer samples,indicating the high quality of the asgrown graphene.

As mentioned in the experiment section,we first carried out the oxidation process at 550°C.For the sake of comparison,we chose the same area containing bare Cu surface,single layer graphene-coated Cu surface(Cu SLG)and multiple layer graphene-coated Cu surface (Cu MLG)to study the oxidation resistance at different oxidation time.As seen in Fig.1(a)-(f),with the increase of oxidation time,the monolayer graphene was gradually oxidized away,while the multilayer graphene pieces can be preserved.When the oxidation time reaches 50 min,all monolayer graphene disappears and only multilayer graphene part is left.Figure 1(g)and (h)show the Raman characterization corresponding to the area covered by SLG and MLG under different oxidation time.Compared with the Raman spectrum of SLG pristine sample(Fig.S1 in the supplementary materials),it is found that the defect-activated peak at～1350 cm-1appears in the spectrum when oxidation time reached 20 min,implying the oxidation induced disorders and/or defects existed in the SLG[23].As the oxidation time increases to 30 min,the 2D peak disappears completely and just a weak G peak is left, implying the SLG has suffered serious damage.With further oxidation,all peaks of graphene are vanished. On the contrary,as to the MLG,it can be found from Fig.1(h)that,even after 50 min oxidation,both G and 2D peaks are unchanged and their intensity ratios are kept about I2D/IG≈0.4.Moreover,no D peak can be observed in the spectra.These results indicate that MLG can be preserved with high quality under oxidation and can be used as a kind of good protection layer for Cu surface against oxidation.

To further know how the MLG resists the oxidation, we perform the oxidation with extend time for two kinds of MLG samples.One is bilayer graphene plate(sample I),and the other is the bilayer basal plate with multilayer graphene in the center(sample II).Figure 2 shows the optical images of the morphology changes for two kinds of samples when the oxidation time prolongs from 50 min to 340 min.For sample I,we can find the whole plate becomes in pieces continually with the oxidation time(Fig.2(a)-(c)).The Raman spectra shown in Fig.2(d)indicate that the bilayer graphene can still remain after 50 min oxidation with its Raman characteristics of I2D/IG≈1(Fig.2(d),black line).How-ever,the characteristic peaks of graphene completely disappear(Fig.2(d),red line)after 340 min oxidation, implying that the bilayer graphene has been fully destroyed.As compared to sample I,the sample II shows different oxidation behavior.As shown in Fig.2(e)-(g), the pieces appear dominantly in the margin of the sample,while only sparse pieces exist in the middle region. Raman spectra taken from the middle area(Fig.2(h)) show that the multilayer graphene can be preserved after 340 min oxidation.This comparative investigation further confirms the result that the oxidation resistance of graphene increases with its layer thickness.

Based on the results of Fig.1 and Fig.2,we can find that there exists apparent optical contrast for graphene between the area with and without oxidation.Therefore,in order to quantitatively explore the layer dependence of oxidation resistance,the percentage of oxidized area can be analyzed based on image treatment,details are shown in Fig.S3 of the supplementary materials. Figure 3(a)shows the percentage of oxidized area of monolayer and bilayer coated Cu surface with oxidation time,and the corresponding fitting lines are defined asoxidation rate.The fitting results show the oxidation rate of Cu MLG is 0.0025%/s,an order of magnitude slower than that of Cu SLG-0.035%/s(Fig.3(a)),exhibiting the higher oxidation resistance.

FIG.1(a-f)Optical images of degradation of the graphene/Cu surface under oxidation at 550°C with different time. (a)0 min,(b)10 min,(c)20 min,(d)30 min,(e)40 min,(f)50 min.Scale bar:100μm.(g)and(h)Raman spectra for SLG and MLG at corresponding oxidation time.The inset in Fig.1(g)is enlarged G peak of graphene.

FIG.2 Optical images and Raman spectra of two kinds of MLG.(a-d)Sample I of bilayer graphene.(e-h)Sample II with bilayer basal and more layers in the middle.Scale bar:10μm.

FIG.3(a)Percentage of oxidation area with oxidation time for SLG(black dot)and MLG(blue dot)on the Cu surface. Red and green lines are the fitting to deduce the oxidation rate.(b)The dependence of oxidation rate on the oxidation temperature for SLG(black dot)and MLG(red dot).The fitting results are based on Arrhenius equation.

FIG.4(a)XPS Cu2p of Cu,Cu SLG and Cu MLG before and after oxidation.Fitting results of Cu2p3/2peak after oxidation:(b)Cu,(c)Cu SLG,(d)Cu MLG.(Cu or Cu2O,～932.3 eV,blue line;CuO,～933.9 eV,red line;satellite peaks,～940.0-946.0 eV,representation of Cu2+,orange line.).For Cu foil,the peak corresponding to CuO and the satellite peaks corresponding to Cu2+can be more clearly as the layer of graphene coatings increasing.(e)AES Cu LMM of Cu,Cu SLG and Cu MLG before and after oxidation.Fitting results of characteristic peak of Cu after oxidation:Cu(f),Cu SLG(g), Cu MLG(h).(Cu2O:～916.4 eV,blue line;CuO:～917.2 eV,red line;Cu:～918.3 eV,green line.).S1:Cu2O,S2:CuO, S3:Cu,(S1+S2)/S3:degree of oxidation.The oxidation degree is 3.0,1.9,0.3 to Cu,Cu SLG and Cu MLG,implying that oxidation resistance of Cu foil can be dramatically improved under graphene protection,and MLG acts as more effective diffusion barrier than SLG,protecting the underlying Cu foil from oxidation.

Since it has been found that graphene coatings last longer in the oxygen atmosphere as its layer increasing, it is quite necessary to study the oxidation resistance of Cu surface under the protection of different layers of graphene.XPS and XAES were performed on these samples in order to provide an analysis of the metal composition after heat treatment(550°C,30 min).As shown in Fig.4(a),XPS of Cu MLG,Cu SLG,and bareCu surface(Cu)before and after oxidation all show two main peaks at binding energies of 932.3 and 952.0 eV, which correspond to Cu2p3/2and Cu2p1/2[23,24].For Cu and Cu SLG,satellite peaks can be observed at higher binding energies after oxidation(Fig.4(b)and (c)).This satellite feature observed in the samples is an indication of materials having a partially filled d9 shell configuration in the ground state,such as copper dihalides,metallic nickel,or CuO[24,25].In addition,the peak at 933.9 eV corresponding to CuO can be seen in these samples.All above indicate that Cu and Cu SLG has been oxidized and there is deeper oxidation on Cu surface than that on Cu SLG due to the higher intensity of CuO peak.For Cu MLG,however, no satellite peaks can be found in the XPS after oxidation(Fig.4(d)),implying the MLG coatings act as more effective diffusion barrier than SLG,protecting the underlying Cu surface from oxidation.

Considering the characteristic peaks of XPS for Cu and Cu2O are very close to each other,the Cu2p3/2peak at 932.3 eV is mainly due to Cu or Cu2O or both,depending on the experimental conditions,so that XAES is essential to distinguish them[23],as shown in Fig.4(e)-(h).Notably,Cu foil shows broader peak located at 916.4 eV corresponding to Cu2O(Fig.4(e)), indicating that uncoated Cu surface is easy to be oxidized even in the ambient condition.While the Cu surface with graphene coatings,no matter monolayer or multilayer,show narrow peaks located at 918.3 eV corresponding to Cu,exhibiting good oxidation resistance in the ambient atmosphere.After oxidation,peaks of Cu2O and CuO appear with different intensity in the XAES of the three samples.We use the intensity ratio of sum of copper oxides(S1+S2,S1:Cu2O,S2:CuO) to Cu(S3)to show the degree of oxidation,and the results also demonstrate the protection effect of graphene which can be enhanced by the layer increasing.

In order to study the source of the difference in oxidation resistance,more detailed observation on the Cu SLG and Cu MLG has been made.First,it is essential to know the initial position of oxidation.As the oxidation time increasing,the bare Cu surface is the first to be oxidized(Fig.5(a)),and the position of oxidation will be gradually revealed.It is particularly worth noting that oxidation does not begin at the grain boundaries,but in the graphene plane,because no oxidation line is observed at the boundary areas of graphene domains as shown in Fig.5(b).It is different from the results reported in literatures that oxidation begins at grain boundaries[15–17].For Cu SLG,there are linear oxidation stripes,which aligned in parallel,as seen in the Fig.5(c),and the area of Cu MLG can be easy to distinguish simultaneously.For Cu MLG,oxidation particles are significant different.Very uniform granule and occasional linear stripes are shown in Fig.5(d).Accordingly,it can be inferred that there exist different oxidation channels in Cu SLG and Cu MLG,respectively,which may be the possible reason for the distinct oxidation rates.

Due to the distinctly different coefficients of thermal expansion(CTE)between graphene[26]and the Cu foil [27],the substrate-induced compressive strain may exist in the graphene film directly grown on Cu substrates during the CVD process.Thus,winkles perpendicular to strain(Fig.6(a))is induced.At the same time, wrinkles can be etched more easily than in other areas because epoxy chains tend to align along the zigzag direction[28,29],which is nearly perpendicular to the strain.Etching is realized by the transformation of the epoxy chain to the carbonyl pairs,while the epoxy pair acting as an intermediate species[31].Once cracks are formed along the wrinkles perpendicular to the strain, the Cu surface under the wrinkles is oxidized by oxidizing elements which diffuse through the cracks.So the parallel oxidation stripes come out.To MLG,however, the wrinkles in each layer cannot completely coincide because of the larger size of the upper graphene single crystal,which crosses more copper grains and has to be under their influence.In order to better describe the progress,we take bilayer graphene coatings as an example(Fig.6(b)).After the cracks form in the first layer, oxygen diffuse through and contact to the second layer, but probably not to the sites on the wrinkles where can be easily etched,more likely to be at the barrier sites. Unless oxygen just contact to the sites belonging to the wrinkles,continued etching can occur and it will further oxidize the underneath Cu surface.That is to say, oxidation can happen on the sites where the intersection of the wrinkles of the two stacked layers for bilayer graphene.Occasionally,wrinkles of each layer can be coincident or partially coincident,inducing the emergence of linear oxidation stripes.More layers graphene is similar.That is why there were very uniform granule and occasional linear stripes after oxidation for Cu MLG.It should be pointed out that Raman results show that graphene gradually disappear with the increase of oxidation time,accordingly,oxygen not only reacts with Cu surface at the cracks but also oxidize adjacent copper atoms and graphene,as mentioned in the previous report[30,31].As a result,diffusion channels of oxygen get broader,and more copper and graphene can be oxidized.Eventually,the Cu surface will be of complete oxidation as graphene is totally oxidized away.

Based on the above analysis,diffusion channels of MLG are much fewer than that of SLG,which give rise to the slow oxidation rate of Cu MLG.To demonstrate the dominant contribution of the diffusion channel,we carried out oxidation experiments to Cu SLG and Cu MLG at different temperatures of 500,550,600, 650°C.The corresponding oxidation rate is shown in Fig.3(b),which show a significant linear relationship with 1/T for both samples when the longitudinal coordinates are expressed in logarithmic,implying that the oxidation process is related to the activation of temperature.Through the fitting results of Arrhenius equation:

FIG.5(a)Phenomenon of Cu SLG at 10 min oxidation.Scale bar:50μm.(b)More obvious oxidation on graphene plane than that on grain boundary at 20 min oxidation.Oxidation does not begin at the grain boundaries,but in the graphene plane.Scale bar:50μm.(c)Parallel linear stripes on Cu SLG,the hexagonal in the middle is MLG.Scale bar:10μm.(d) Uniform oxidation granule on Cu MLG and occasional linear stripes.Scale bar:10μm.

FIG.6(a)Oxidation sites of Cu SLG:wrinkles.(b)Oxidation sites of Cu MLG:intersection or coincident part of wrinkles of each layer.

where Eais the activation energy and A is preexponential factor,the amount of reaction sites can be obtained.In essence,it is the reaction of oxygen and copper no matter for Cu SLG or Cu MLG,leading to the value of Eabeing approximately the same,1.04 and 1.07 eV for Cu SLG and Cu MLG,respectively.Remarkably,the pre-exponential factor of Cu SLG(890.8 s-1) was 5.5 times as much as that of Cu MLG(161.2 s-1), indicating the number of reaction sites in Cu SLG is larger.The sites allowing oxygen to diffuse are the sites where reaction can occur.Therefore,the fitting results demonstrate our suggestion that the amount of oxygen diffusion channel make a major contribution to the oxidation resistance of graphene-coated Cu surface.

IV.CONCLUSION

We studied the oxidation resistance of graphenecoated Cu surface and its layer dependence by directly growing monolayer graphene coexistent with multilayer structure,diminishing the influence of residue and transfer technology.We reveal that oxidation begins at defects in graphene plane but not grain boundaries. Aiming at the problem of dramatic different oxidation rates of Cu SLG and Cu MLG,we give the opinion that the oxygen diffusion channels are wrinkles induced by strain for Cu SLG,and those of Cu MLG are the intersection of the wrinkles in each stacked layer.The fitting results of experimental data based on Arrhenius formula prove our point of view.As a consequence, compared to putting forth efforts to improve the quality of monolayer graphene by reducing defects,depositing multilayer graphene directly on metal is a simple and effective way to enhance the oxidation resistance of graphene-coated metal.

Supplementary materials:To make sure that the foil surface was not completely covered,we made it oxidized in air for about 1 min at 200°C by a heating platform,as seen in Fig.1(a).Photograph(Fig.S1(a)), optical microscopy(Fig.S1(b))and SEM(Fig.S1(c)) of graphene show that the domain size is～2 mm. Some multilayer graphene can be visible by optical microscopy after being transfer to SiO2/Si substrate (Fig.S1(d)),implying that the graphene was single layer with multi-layer structure of coexistence.Corresponding Raman spectra showed in the Fig.S1(e)-(f)also prove this structure,the intensity ratio of the 2D(I2D) to G band(IG)is about 3 and the FWHM of 2D peak is 29 in Fig.1(e),exhibiting the characteristics of single layer graphene(SLG),but that in Fig.1(f)is 0.5 and56,respectively,corresponding to multilayer graphene (MLG)[23].In addition,no defect-activated D peak at～1350 cm-1can be seen in both spectra,implying the high quality of as-grown graphene.

[1]K.I.Bolotin,K.J.Sikes,Z.Jiang,M.Klima,G.Fudenberg,J.Hone,P.Kim,and H.L.Stormer,Solid State Commun.146,351(2008).

[2]R.R.Nair,P.Blake,A.N.Grigorenko,K.S.Novoselov, T.J.Booth,T.Stauber,N.M.R.Peres,and A.K. Geim,Science 320,1308(2008).

[3]C.Lee,X.Wei,J.W.Kysar,and J.Hone,Science 321, 385(2008).

[4]M.Evaldsson,I.V.Zozoulenko,H.Xu,and T.Heinzel, Phys.Rev.B 78,161407(2008).

[5]F.Xia,T.Mueller,R.Golizadeh-Mojarad,M.Freitag, Y.M.Lin,J.Tsang,V.Perebeinos,and P.Avouris, Nano Lett.9,1039(2009).

[6]F.Wang,Y.Zhang,C.Tian,C.Girit,A.Zettl,M. Crommie,and Y.R.Shen,Science 320,206(2008).

[7]L.Liu,S.Ryu,M.R.Tomasik,E.Stolyarova,N.Jung, M.S.Hybertsen,M.L.Steigerwald,L.E.Brus,and G. W.Flynn,Nano Lett.8,1965(2008).

[8]J.S.Bunch,S.S.Verbridge,J.S.Alden,A.M.van der Zande,J.M.Parpia,H.G.Craighead,and P.L. McEuen,Nano Lett.8,2458(2008).

[9]S.Bae,H.Kim,Y.Lee,X.Xu,J.S.Park,Y.Zheng, J.Balakrishnan,T.Lei,H.Ri Kim,Y.I.Song,Y.J. Kim,K.S.Kim,B.Ozyilmaz,J.H.Ahn,B.H.Hong, and S.Iijima,Nat.Nano 5,574(2010).

[10]H.Zhou,W.J.Yu,L.Liu,R.Cheng,Y.Chen,X. Huang,Y.Liu,Y.Wang,Y.Huang,and X.Duan,Nat. Commun.4,2096(2013).

[11]K.S.Kim,Y.Zhao,H.Jang,S.Y.Lee,J.M.Kim,K. S.Kim,J.H.Ahn,P.Kim,J.Y.Choi,and B.H.Hong, Nature 457,706(2009).

[12]T.Fujita,W.Kobayashi,and C.Oshima,Surf.Interface Anal.37,120(2005).

[13]J.Coraux,A.T.N’Diaye,C.Busse,and T.Michely, Nano Lett.8,565(2008).

[14]S.Chen,L.Brown,M.Levendorf,W.Cai,S.Y.Ju,J. Edgeworth,X.Li,C.W.Magnuson,A.Velamakanni, R.D.Piner,J.Kang,J.Park,and R.S.Ruof f,ACS Nano 5,1321(2011).

[15]D.L.Duong,G.H.Han,S.M.Lee,F.Gunes,E.S. Kim,S.T.Kim,H.Kim,Q.H.Ta,K.P.So,S.J. Yoon,S.J.Chae,Y.W.Jo,M.H.Park,S.H.Chae, S.C.Lim,J.Y.Choi,and Y.H.Lee,Nature 490,235 (2012).

[16]B.Wang,Y.S.Puzyrev,and S.T.Pantelides,Polyhedron 64,158(2013).

[17]S.S.Roy and M.S.Arnold,Adv.Func.Mater.23,3638 (2013).

[18]Y.H.Zhang,B.Wang,H.R.Zhang,Z.Y.Chen,Y.Q. Zhang,B.Wang,Y.P.Sui,X.L.Li,X.M.Xie,G.H. Yu,Z.Jin,and X.Y.Liu,Carbon 70,81(2014).

[19]L.Ma,J.Wang,and F.Ding,Angew Chem.Int.Ed. Engl.51,1161(2012).

[20]D.Prasai,J.C.Tuberquia,R.R.Harl,G.K.Jennings, and K.I.Bolotin,ACS Nano 6,1102(2012).

[21]X.Li,Y.Zhu,W.Cai,M.Borysiak,B.Han,D.Chen, R.D.Piner,L.Colombo,and R.S.Ruof f,Nano Lett. 9,4359(2009).

[22]Y.C.Lin,C.C.Lu,C.H.Yeh,C.Jin,K.Suenaga, and P.W.Chiu,Nano Lett.12,414(2012).

[23]J.Ghijsen,L.H.Tjeng,J.van Elp,H.Eskes,J.Westerink,G.A.Sawatzky,and M.T.Czyzyk,Phys.Rev. B 38,11322(1988).

[24]C.Malitesta,T.Rotunno,L.Sabbatini,and P.G.Zambonin,J.Chem.Soc.Faraday Trans.86,3607(1990).

[25]C.C.Chusuei,M.A.Brookshier,and D.W.Goodman, Langmuir 15,2806(1999).

[26]N.Mounet and N.Marzari,Phys.Rev.B 71,205214 (2005).

[27]G.K.White and S.J.Collocott,J.Phys.Chem.Ref. Data 13,1251(1984).

[28]J.L.Li,K.N.Kudin,M.J.McAllister,R.K. Prud’homme,I.A.Aksay,and R.Car,Phys.Rev.Lett. 96,176101(2006).

[29]Z.Li,W.Zhang,Y.Luo,J.Yang,and J.G.Hou,J. Am.Chem.Soc.131,6320(2009).

[30]E.Starodub,N.C.Bartelt,and K.F.McCarty,J.Phys. Chem.C 114,5134(2010).

[31]V.Joh′anek,G.W.Cushing,J.K.Navin,and I.Harrison,Surf.Sci.644,165(2016).

?Author to whom correspondence should be addressed.E-mail: yqsong@mail.ustc.edu.cn