Graphene/Substrate Charge Transfer Characterized by Inver Photoelectron Spectroscopy
Lingmei Kong,†Cameron Bjelkevig,‡,§Sneha Gaddam,‡Mi Zhou,‡Young Hee Lee,|
Gang Hee Han,|Hae Kyung Jeong,⊥Ning Wu,†Zhengzheng Zhang,†Jie Xiao,†P.A.Dowben,†and Jeffry A.Kelber*,‡
Department of Physics and Astronomy,Nebraska Center for Nanostructures and Materials,Theodore Jorgenn Hall,855North 16th Street,Uni V ersity of Nebraska s Lincoln,Lincoln,
Nebraska 68588-0111,United States,Department of Chemistry and Center for Electronic Materials Processing and Integration.Uni V ersity of North Texas,1155Union Circle #305070,Denton,Texas 76203-5017,United States,Department of Physics,Department of Energy Science,and Center for Nanotubes and Nanostructured Composites,Sungkyunkwan Ad V anced Institute of Nanotechnology,Sungkyunkwan Uni V ersity,Suwon,440-746Korea (ROK),and Department of Physics,Daegu Uni V ersity,Gyeongsan,712-714Korea (ROK)Recei V ed:September 9,2010;Re V id Manuscript Recei V ed:October 26,2010
Wave vector-resolved inver photoelectron spectroscopy (IPES)measurements demonstrate that ther凯撒豪庭
e is a large variation of interfacial charge transfer for graphene grown by chemical vapor deposition (CVD)on a range of dielectric or metallic substrates.Monolayer graphene grown by CVD on monolayer BN(0001)/Ru(0001)exhibits strong charge transfer from the substrate to graphene of 0.07(1)e -per carbon atom,as manifested by filling of the π*band and displacement of the Fermi level.IPES measurements of CVD single layer graphene on Ru indicate a substrate-to-graphene charge transfer from the substrate of 0.06(1)e -per carbon atom,in agreement with reported angle-resolved photoemission results.The IPES spectra of CVD single layer graphene on Ni(poly)and on Cu(poly)indicate 0.03(1)e -per carbon atom charge transfer from Ni and Cu substrates.Single layer graphene has also been grown by free radical-assisted CVD on MgO(111),resulting in a layer of graphene and an oxidized carbon interfacial layer between the graphene and the substrate.IPES measurements indicate that 0.02(1)e -per carbon atom charge is transferred from graphene to the MgO substrate.Additionally,IPES and photoemission data indicate that single layer graphene/MgO(111)exhibits a band gap.The data demonstrate that IPES is an effective method for preci measurement of substrate/graphene charge transfer and related electronic interactions,in part becau of the extreme surface nsitivity of the technique,and suggest new strategies for extrinsic doping of graphene for controlled mobilities for device applications.
1.Introduction
Graphene,due to extremely high room temperature electron/hole mobilities 1-3and polarizabilities,4,5is of great interest for nanoelectronic and spintronic device applications.Interfacial interactions of graphene with adjacent layers are therefore of practical importance,as both adsorbate-induced charge transfer and interaction with dielectric substrates can result in signifi-cantly enhanced or reduced mobilities.Of interest for device applications,adsorbate-induced hole or charge transfer generally results in reduced mobilities but has resulted in enhanced Hall mobility,6while proximity to high-k dielectric substrates sometimes yields incread electron or hole mobility at room temperature,apparently through screening of carriers from charged impurities.7Graphene/substrate interactions may also yield a band gap in the graphene band structure,8and this is of critical concern in the development of graphene-bad logic devices with true “off”states.
Recently,wave (k )-vector resolved inver photoelectron spectroscopy (IPES)has demonstrated the prence of substan-tial BN-to-graphene charge transfer 9for graphene grown by
CVD on BN(0001)/Ru(0001).An order-of-magnitude enhance-ment of graphene room temperature mobility relative to graphene/SiO 2has recently been reported 10for graphene sheets physically transferred to BN substrates,consistent with predicted results for extrinsic doping,6but also possibly
due to decread phonon interactions with hexagonal BN(0001)layers.The facts,combined with monolayer nsitivity,11demonstrate the potential for IPES to delineate electronic interactions between graphene and substrates in a layer-by-layer fashion and to provide a quantitative basis for predicting the effects of such materials interactions on graphene electronic properties.
We report here the results of IPES measurements of electronic charge transfer between graphene and various substrates,including BN/Ru(0001),Ru(0001),Ni(poly),Cu(poly),and MgO(111),as well as previously reported 12IPES data of graphene on SiC(0001).Notably,the IPES and photoemission data indicate the prence of a band gap for graphene/MgO(111),suggesting that graphene on this substrate may be of particular utility for logic device applications,with the band gap allowing a true “off”state.IPES data further indicate that significant interfacial charge transfer and band-filling do not appreciably alter the fundamental conduction band electronic structure of graphene and suggest practical routes toward the formation of extrinsically doped graphene layers with controlled electron mobilities for device applications.
*To whom correspondence should be addresd.E-mail:Kelber@unt.edu.†
University of Nebraska s Lincoln.‡
University of North Texas.§
Prent address:Intel Corp.,4100Sara Rd.SE,Rio Rancho,NM 87124.|
Sungkyunkwan University.⊥
Daegu University.
J.Phys.Chem.C 2010,114,21618–21624
2161810.1021/jp108616h 2010American Chemical Society
Published on Web 11/17/2010
2.Experimental Methods
The fabrication of the monolayer graphene/BN/Ru(0001)sample,and characterization by scanning tunneling microscopy/spectroscopy (STM/STS),low energy electron diffraction (LEED),Raman spectroscopy,photoemission spectroscopy (PES),and IPES measurements,has been previously reported.9The data are included here for direct comparison with results for graphene on other substr
变色龙作者
ates.Graphene on Ru(0001)was prepared in a manner similar to that described in the literature 13by CVD of C 2H 4at 1×10-8Torr,2min of exposure at 700K on Ru(0001),followed by annealing in ultrahigh vacuum (UHV)to 1000K,in a UHV system described previously.14Graphene layers on polycrystalline substrates s Ni(poly)and Cu(poly)s were prepared by CVD and characterized as described previously.15,16
A graphene film was also grown by CVD using thermally dissociated C 2H 4(free radical assisted CVD;FRA-CVD)on a MgO(111)single crystal substrate in a system equipped for LEED and X-ray excited photoelectron spectroscopy (XPS),which has been described previously.17,18A more detailed account of the preparation and characterization of graphene films on this surface,and the associated interfacial chemistry,will be published elwhere.Briefly,a MgO(111)substrate 1cm in diameter and 0.5mm thick was cleaned at room temperature by exposure to atomic oxygen from a commercially available thermal catalytic cracker (Oxford Applied Instruments).A flux of dissociated C 2H 4was supplied from the same source,which has also been described previously.19,20XPS spectra were acquired at a constant pass energy (44eV)using Mg K R radiation and analyzed using commercial software (ESCA-TOOLS)according to standard methods.
IPES data were acquired as described previously,21,22in the isochromat mode using a Geiger -Mu ¨l二话不说的意思
ler detector t to detect
photons at 9.7eV.The IPES measurements were limited by an
instrumental resolution of ∼400meV.The IPES spectra were acquired along the surface normal.Thus,the relative positions of the π*and σ*bands in the IPES data reported here are at the Γpoint of the Brilloin zone,at the σ*minimum.9,12The angle-integrated PES and wave vector-resolved IPES were undertaken to study the molecular orbital placement of both occupied and unoccupied orbitals of graphene on MgO but here have not been corrected for final state effects arising from the insulating nature of the substrate,which may lead to an overestimation of the true band gap.In addition to such effects,the angle-integrated nature of the PES measurements and wave vector-resolved nature of the IPES measurements obviate the u of the data to determine energy splitting between specific valence and conduction band features,such as the π-π*splitting.PES data were acquired using a He I source UV source (21.2eV)in the same vacuum system and with the analyzer aligned with the surface normal.
3.Results
3.1.Characterization of Single Layer Graphene on Ru(0001).A cleaned and ordered Ru(0001)single
crystal was expod to 1.2L of C 2H 4at 700K,followed by annealing to 1000K in UHV.A sharp,bifurcated LEED pattern was obrved (Figure 1a),due to the Ru and graphene lattice mismatch.Assuming a Ru lattice constant of 2.7Å,the outer diffraction spots (Figure 1b)indicate a lattice constant of 2.5(1)Ås graphene.The thickness of the graphene overlayer was estimated from decreas in intensity of a Ru Auger feature (Figure 1c,arrow),which does not overlap with the C(KVV)Auger feature.Using a calculated 23inelastic mean free path of 10.5Å,a total average thickness of the graphene layer after读后感写法
Figure 1.Characterization of single layer graphene on Ru(0001).(a)LEED (beam energy )70eV)after 1.2L C 2H 4exposure to Ru(0001)at 700K,followed by annealing to 1000K in UHV.(b)Line scan of LEED pattern,showing bifurcated spots.(c)Auger spectra before/after ethylene exposure and annealing.Changes in peak-to-peak height for marked Auger feature (arrow)were ud to calculate changes in intensity and therefore average thickness of carbon overlayer.
Graphene/Substrate Charge Transfer Characterized by IPES J.Phys.Chem.C,Vol.114,No.49,2010
21619
annealing is estimated at 1.3((0.2)monolayers.The uncertainty in this figure derives from the fact that
the experimental error in the reproducibility of absolute Auger electron intensities in this system,conrvatively estimated at 10%.Exposure of this sample to ambient resulted in no discernible evidence of oxidation as determined by Auger or changes to the LEED pattern.
3.2.Characterization of Single Layer Graphene on MgO(111).A graphene film was grown on an MgO(111)single crystal by first cleaning the surface in atomic O at room temperature and then exposure of the (disordered)MgO surface to a flux of dissociated C 2H 4at ∼600K (5×10-7Torr,25min).No LEED pattern was obrved before or after the exposure to dissociated ethylene.From the C(1s)XPS spectrum (Figure 2a),a total average carbon thickness of 3Åwas determined.The sample was re-expod to ambient and placed on a different sample holder to permit more efficient heating.Subquent annealing to 1000K in UHV produced the obrved change in C(1s)XPS (Figure 2a).(Becau of the significant and variable sample charging,the peak maxima of the two carbon spectra have been aligned to compare changes in peak shape.)The total average thickness of the carbon layer after annealing to 1000K was 2.5Å,indicating that exposure to ambient and subquent annealing in UHV had resulted in no significant increa in surface carbon.The annealing process did,however,result in a significant broadening of the C(1s)spectrum (Figure 2a)toward higher binding energy,indicating the prence of surface carbon in multiple oxidati
on states.After annealing,the expected 6-fold LEED image for a graphene film was obrved (Figure 2b).Becau atomically clean MgO(111)(1×1)yields a 3-fold LEED pattern at this beam energy (∼75eV),24the prence of a 6-fold LEED pattern indicates formation of a graphene surface layer.This conclusion is corroborated by the fact that decomposition of the XPS spectrum of the annealed sample into two regions s one centered near 292eV binding energy (uncorrected for charging)with the same fwhm as the spectrum prior to annealing and the cond including the higher binding energy tail s indicates that the thickness of the lower binding energy component is 1.5Å,corresponding to a monolayer of graphene.The XPS data in Figure 2,however,indicate that carbon is prent on the MgO surface in multiple oxidation states.Subquent exposure of the annealed sample to ambient resulted in no significant changes to the LEED pattern (Figure 2b).The LEED and XPS data together therefore indicate the formation of an ordered graphene layer in the
prence of an interfacial layer that includes carbon in higher oxidation states.
3.3.Inver Photoemission Measurements of Interfacial Charge Transfer.The position of a specific conduction band feature in the IPES spectrum,relative to the Fermi level,provides quantitative information on charge transfer between substrate and graphene layers.The IPES data are displayed in Figure 3for single layer graphene on BN(0001)/Ru(0001)(Figure 3a),on Ru(0001)(Figure 3b),on C
u(poly)(Figure 3c),on Ni(poly)(Figure 3d),and compared to the previously reported 12data for multilayer graphene grown on SiC(0001)by thermal evaporation (Figure 3e)and to our results for graphene/MgO(111)(Figure 3f).Although charge transfer between the single layer graphene and the SiC substrate is known to depend
Figure 2.Graphene formation on MgO(111)(a)C(1s)XPS spectra after exposure of a clean,disordered MgO surface to dissociated ethylene at 5×10-7Torr,25min at ∼600K (open circles),and after subquent exposure to ambient and annealing in UHV to 1000K (solid line).The two spectra have been manually aligned and have not been corrected for sample charging.(b)Corresponding LEED pattern at 75eV beam energy after annealing to 1000K.
Figure 3.IPES spectra of graphene on various substrates as labeled (background subtracted):(a)BN(0001)/Ru(0001),(b)Ru(0001),(c)Cu foil,(d)Ni foil,(e)SiC (adapted from ref 12),and (f)MgO(111).Features are labeled as (1)scattering from the π*,(2)σ*and low lying π*,and (3)σ*(Γ1+).
不做你的爱人
21620J.Phys.Chem.C,Vol.114,No.49,2010Kong et
al.
on the Si(n type)vs C(p type)nature of the substrate surface termination,25,26such interface effects(including band gap formation)are rapidly screened upon multilayer graphene formation.8Becau the data of Forbeaux et al.(Figure3e)were acquired for multilayer graphene,12we take the corresponding IPES spectrum as a standard for a graphene layer with a weakly interacting substrate,with negligible interfacial charge transfer. The spectra in Figure3are plotted as a function of energy from the Fermi level,E-E F.The data show the main features in the vicinity of the Fermi level and are plotted relative to the features identified for multilayer graphene on SiC,12en also with single crystal graphite.27Becau of imperfections in some of the graphene overlayers,the spectral features are similar but not identical due to scattering contributions.In addition,the features obrved in IPES shift progressively clor to the Fermi level as the substrate is varied from MgO(Figure3f)to SiC (Figure3e),to Ni(Figure3d)or Cu(Figure3c),to Ru(Figure 3b),and then to BN/Ru(Figure3a).This is evidence of band-filling.28,29The IPES spectrum displays only the unoc-cupied density of states,but becau the IPES spectra are extremely surface nsitive,the degree of charge transfer between the substrate to or from graphene can be quantitatively estimated from the binding energy positions of known features (e.g.,the mainσ*feature)in the spectrum relative
to the Fermi level.At a charge neutral condition,the Fermi energy of free-standing graphene coincides with the conical point(the Dirac point),and the density of states is dominated near the Fermi level by the Dirac cone.26The introduction of extra electrons to graphenefills any empty states near the Fermi level E F,with states clost to the Fermi levelfilledfirst,thus decreasing the energy difference of conduction band features relative to E F.In contrast,the subtraction of electrons from graphene increas this energy difference.It can be en in Figure3that corre-sponding features in the IPES spectra of graphene on Ni,Cu, Ru,and BN shift progressively clor to the Fermi level, indicating increasing charge transfer from the substrates to graphene.The bands of graphene on MgO,however,shift away from the Fermi level and indicate charge transfer from graphene to the MgO substrate.The amount of charge transfer can be calculated from the shift of the Fermi level relative to the conduction band edge.A larger shift indicates more charge transfer.The amount of charge transferred is proportional to the shift of the Fermi level for small amounts of charge transfer, becau of the density of states being dominated by the unique band structure near the Fermi level.The amount of charge transfer roughly equals the shift of the Fermi level×0.02e-/ carbon atom.1For larger shifts,a density of states correction is required.Quantitative measurements of interfacial charge transfer derived from the IPES spectra(Figure3)are listed in Table1.
Results of the analys of the IPES spectra in Figure3are in agreement with results obtained by other experimental methods or by theory.The IPES-deduced charge transfer from BN(0001)/ Ru(0001)to graphene(Figure1a)indicates a substrate-to-graphene charge transfer of-0.07(1)e-/carbon atom(the negative sign indicating substrate-to-graphene charge transfer and the number in parenthes indicating the uncertainty in the last digit),in qualitative agreement with DFT calculations and consistent with a pronounced red shift of the2D Raman feature for this system(Table1).9This value is less than the previously reported value9due to the assumption(Figure3and Table1) of a charge neutral condition for multilayer graphene on SiC, not assumed previously.The interlayer charge transfer results obtained from analysis of the IPES spectra of single layer graphene/Ru(0001)(Figure1b),-0.06(1)e-/carbon atom,are in good agreement with the result of-0.05e-/carbon atom obtained by angle-resolved PES.13,30The large charge transfers determined for graphene/BN/Ru(0001)and for graphene/ Ru(0001)would indicate complete or partialfilling of the grapheneπ*band,eliminating or obscuring the features in the IPES spectrum.Consistent with this,theπ*feature is not obrved for graphene/BN/Ru(Figure3a)9and is at least significantly obscured for single layer graphene/Ru(0001) (Figure3b).The result of-0.03(1)e-/carbon atom for graphene on Ni(poly)(Figure1c)is in agreement with photoemission studies indicating weak substrate-to-graphene charge transfer due to Ni(3d)/graphene(π)mixing31an
d in good agreement with DFT results32indicating a0.35eV downward shift of graphene band features for this reason.A similar substrate-to-graphene charge transfer is obrved for graphene/Cu(poly)(Figure1d). Using the multilayer graphene/SiC IPES spectrum(Figure3e) as the“standard”for zero interfacial charge transfer,a slight [+0.02(1)]e-/carbon atom charge transfer is obrved from graphene to the MgO(111)substrate(Figure3f).The latter measurements cannot be taken as accurate as thefinal state effects have not been completely excluded and cannot be taken fully into account from this data.Suchfinal state effects will result in the occupied states and unoccupied states appearing to be placed farther from the Fermi level than may be in fact true in the charge neutral ground state.33
The effects of the MgO substrate on the electronic structure of graphene deposited by FRA-CVD can be determined in further detail by combining angle integrated PES and wave vector-resolved IPES spectra(Figure4).The data are plotted relative to a common Fermi level(E F).The assignment of the
TABLE1:Comparison of2D Peak Positions and Substrate/Graphene Charge Transfer
sample a G peak position(cm-1)2D peak position(cm-1)
IPES-determined charge
transfer(e-per carbon atom)b remarks/charge transfer
SLG/BN(0001)/Ru(0001)1540924009-0.07(1)previously reported value,-0.129 SLG/Ru(0001)no spectrum43no spectrum43-0.06(1)ARXPS value,-0.0513,30
DLG/Ru(0001)159943267843042
SLG/Cu(poly)159044267844-0.03(1)
SLG/Ni(poly)158045,46270945,46-0.03(1)-0.02
SLG/SiC1591.547271047-0.0227
MLG/SiC15904827204808
SLG/MgO+0.02(1)
鑫是什么意思a SLG,single layer graphene;DLG,double layer graphene;and MLG,multilayer graphene.
b A negative charge transfer number refers to charge donation from the substrate to graphene(n type doping);a positive charge transfer number refers to charge transfer from graphene to the substrate(p
金川梨花type).The amount of charge transfer equals the shift of the Fermi level×0.02e-/carbon atom.The number in parenthes is the uncertainty in thefinal digit.
Graphene/Substrate Charge Transfer Characterized by IPES J.Phys.Chem.C,Vol.114,No.49,201021621
valence band features in the PES spectrum is in good agreement with results on,for example,Ni(111),34while the conduction band features are in accord with assignments for graphene/SiC(0001).12The inversion of relative π/σand π*/σ*ordering (Figure 4)reflects the angle-integrated nature of the PES spectrum as oppod to the k -resolved nature of the IPES measurements.Significant sample charging was encountered for graphene/MgO(111),and in this ca,the placement of the Fermi level is approximated from the position of a copper contact to the sample.The resulting placement of the Fermi level near the top of the valence band edge (Figure 4),however,is consistent with obrved (Figure 3f)slightly p type doping of graphene by MgO.A band gap is obrved for graphene/MgO(111)(Figure 4).The magnitude of the band gap,∼1eV,is independent of the preci placement of the Fermi level.The prence of a substantial band gap is consistent with charging encountered for the monolayer nsitive-IPES and LEED measurements.A previous brief report 35also suggested the formation of an insulating or miconducting graphene fil金六
m resulting from carbon evaporation onto MgO(111).4.Discussion
The above data demonstrate the utility of IPES for asssing interfacial charge transfer between graphene and dielectric or metallic substrates.The data also suggest new device ap-plications for certain graphene/dielectric heterostructures.The results for graphene/BN/Ru indicate substantial charge transfer to graphene,esntially filling the π*band.9Adsorbate-induced charge transfer has been correlated 6with order-of-magnitude increas in electron mobility.Recent transport measurements on graphene sheets physically transferred to BN substrates 10indicate just such a factor of 10increa in room temperature mobility relative to graphene/SiO 2.Other factors,such as substrate phonon effects,may have impacted the above result.However,the correlation between obrved charge transfer in graphene/BN heterojunctions 9and predicted 6and obrved 10effects upon electron transport are certainly motivation for further study of such systems for device applications.
This is the first report of the structural (LEED),chemical (XPS),and electronic (PES,IPES)characterization of graphene layer formation on MgO(111).The results,showing the prence of a band gap,are consistent with a brief abstract published previously for graphene deposition by MBE on this surface 35but indicate that the formation of a graphene layer (Figure 2)is
associated with the formation of an oxidized carbon interfacial
layer.Indeed,the higher binding energy tail of the XPS C(1s)spectrum obtained after annealing to 1000K (Figure 2a)is similar to the XPS C(1s)spectrum for graphene oxide.36Band gaps varying from ∼0.237to 1.7eV 38for oxidized graphene flakes have been reported.However,the XPS data (Figure 2a)indicate two layers of carbon,while the surface nsitive LEED (Figure 2b)and PES/IPES measurements (Figure 4)indicate a graphenelike surface layer,with intact πand π*features.One possibility is that the sample consists of an oxidized carbon interfacial layer,hexagonally ordered,interacting with a graphene overlayer.The similar hexagonal symmetry of both layers would lead to A -B symmetry breaking in the graphene layer,and similar arguments have been advanced for formation of a 0.26eV band gap for single layer graphene on SiC(0001).3The evolution of the oxidized carbon region upon annealing (Figure 2a)further suggests that this oxidized carbon interfacial layer reflects a fundamental aspect of carbon/MgO(111)surface chemistry at elevated temperatures.The possible formation of BN(0001)and then graphene/BN heterojunctions on MgO(111)are suggested by the data,as a route toward the formation of extrinsically doped graphene layers on a high-k dielectric films and for moderating the effects of carbon/MgO interfacial interactions for device applications.The fact that MgO(111)can be grown on Si(100)39is further motiv
ation for examining this system from the point of view of future device applications.Finally,the recent report of graphitic nanoflake formation on MgO powders under CVD conditions 40suggests that graphene growth on other MgO orientations or on amorphous substrates might be possible.
The effects of charge transfer on the change of the energies of graphene Raman spectral features 41-48is of practical interest in monitoring doping interactions involving graphene layers.The reported energy of Raman “G”(ring breathing mode)and “2D”(a two phonon mode)features for graphene on various substrates are therefore compared in Table 1.As shown in Table 1,the 2D position for graphene/BN(0001)/Ru is red-shifted by ∼300cm -1from the 2D energy for graphene on other substrates,and this also corresponds to a large charge transfer to graphene of 0.07(1)e -/carbon atom.The energy of the corresponding G feature s 1540cm -1s is only slightly red-shifted relative to the other G mode energies (Table 1)or to the corresponding energy in HOPG.9The other graphene Raman spectra exhibit 2D features in the region of 2670-2710cm -1,and G feature energies in the range of 1580-1600cm -1,corresponding to negligible interlayer charge transfer ((0.02e -/carbon atom).Importantly,although the monolayer graphene/Ru(0001)system exhibits significant substrate-to-graphene charge transfer [-0.0.6(1)e -/carbon atom;Figure 3],no Raman spectrum is obrved for single layer graphene on Ru(0001).43This is consistent with a large charge transfer,as the strong int
eractions required for large charge transfers on metal substrates can also lead to overdamping of the vibrational mode intensities of graphene in the Raman spectra.The data in Table 1therefore indicate that the energies of graphene Raman features for graphene on different substrates are at best very qualitative guides to interlayer charge transfer.Our own measure of the Raman shifts,however,indicates that there is a likely mode stiffening of the G band with charge abstraction from the graphene layer,even if slight,as indicated in Figure 5.This is in consistent with the previous report for phonon stiffening in carbon nanotubes.49
As a final point,it is worth noting key differences between IPES and near-edge X-ray absorption fine structure spectroscopy (NEXAFS),which is also an important probe of conduction band
Figure 4.Combined PES and IPES of graphene on MgO substrates (background subtracted).For assignment of the features in the valence and conduction bands,e refs 34and 12,respectively.
21622J.Phys.Chem.C,Vol.114,No.49,2010Kong et
al.
