nature ARTICLES chemistry PUBLISHED ONLINE:12 AUGUST 2012|DOI:10.1038/NCHEM.1421 Understanding and controlling the substrate effect on graphene electron-transfer chemistry via reactivity imprint lithography Qing Hua Wang',Zhong Jin,Ki Kang Kim2,Andrew J.Hilmer,Geraldine L.C.Paulus',Chih-Jen Shih, Moon-Ho Ham3,Javier D.Sanchez-Yamagishi4,Kenji Watanabe5,Takashi Taniguchis,Jing Kong?, Pablo Jarillo-Herrero4 and Michael S.Strano1* Graphene has exceptional electronic,optical,mechanical and thermal properties,which provide it with great potential for use in electronic,optoelectronic and sensing applications.The chemical functionalization of graphene has been investigated with a view to controlling its electronic properties and interactions with other materials.Covalent modification of graphene by organic diazonium salts has been used to achieve these goals,but because graphene comprises only a single atomic layer,it is strongly influenced by the underlying substrate.Here,we show a stark difference in the rate of electron-transfer reactions with organic diazonium salts for monolayer graphene supported on a variety of substrates. Reactions proceed rapidly for graphene supported on SiO2 and Al2O(sapphire),but negligibly on alkyl-terminated and hexagonal boron nitride(hBN)surfaces,as shown by Raman spectroscopy.We also develop a model of reactivity based on substrate-induced electron-hole puddles in graphene,and achieve spatial patterning of chemical reactions in graphene by patterning the substrate. raphene is a two-dimensional,atomically thin lattice of because the SAMs prevent dipolar contaminants from adsorbing on sp2-bonded carbon atoms and has exceptional electronic, the substrate,prevent charge injection from the graphene to the mechanical and thermal properties2.Modifying the basic dielectric interface and screen the effect of charged impurities electronic,chemical and structural properties of graphene is impor- within the substrate26.27.29 tant for incorporating graphene into a variety of applications includ- In this Article,we demonstrate that the substrate on which gra- ing electronic devices,biosensors and composite materials3.The phene rests strongly influences the chemical reactions on the top chemical functionalization of graphene is critical for enabling surface of the graphene.We also demonstrate spatial control of these applications and has been explored for both covalents and the chemical reactivity of graphene with micrometre-scale resol- non-covalent-8 schemes.Functionalizing graphene with aryl diazo ution to achieve wafer-scale patterning of chemical reactions on gra- nium salts49-16 results in the opening of a bandgap10.13.17-19 and phene.A previous report has shown differences in reactivity for shifting of the Fermi level,both of which are desirable in the fab- small mechanically exfoliated flakes of graphene on SiO,and hexa- rication of electronic devices.In addition,the functional groups on methyldisilazane (HMDS)-treated SiO,(ref.30).In the present the diazonium moiety can be tailored by organic chemistry to allow work,chemical vapour deposition(CVD)-grown graphene is depos- various chemical characteristics to be coupled to the graphene ited on a variety of substrates and covalently functionalized with aryl Graphene is strongly influenced by the underlying substrate. diazonium salts.Using Raman spectroscopic mapping,we find that SiO,-covered silicon substrates are compatible with device fabrica- the substrate-induced electron-hole charge fluctuations in graphene tion,but they have rough surfaces and contain charged impurities. greatly influence the chemical reactivity.Graphene on SiO,and These lead to electron-hole charge fluctuations (or 'puddles')in Al2O(sapphire)substrates is highly reactive,but graphene on an the graphene,which scatter charge carriers and inhibit electronic alkyl-terminated monolayer and hBN is much less reactive.We device performance2021.Graphene devices suspended over gaps also develop a new lithographic patterning technique,reactivity exhibit the highest carrier mobilities2,but are not robust for prac- imprint lithography(RIL),where the underlying substrate is chemi- tical use.Recently,single-crystal hexagonal boron nitride(hBN)2425 cally patterned to achieve spatial control of the graphene chemical and self-assembled monolayers(SAMs)of hydrophobic molecules reactivity.This method allows chemical reactions on graphene to grafted on SiO,substrates26-29 have been explored as alternative sub- be spatially patterned over large areas without the use of disruptive strates for graphene electronics.Graphene on hBN,which is atom- materials such as photoresists or chemical etchants.Here,RIL is ically smooth,chemically inert and electrically insulating,has used to spatially control the conjugation of enhanced green fluor- significantly smaller electron-hole charge fluctuations and higher escent protein(EGFP)on graphene,directly from solution,demon- mobilities2425.Graphene devices on SAM-covered substrates also strating the advantages of the technique for producing structures for exhibit lower charge inhomogeneity and performance hysteresis26.27 sensor and microarray applications. 'Department of Chemical Engineering,Massachusetts Institute of Technology,Cambridge,Massachusetts 02139,USA,Department of Electrical Engineering and Computer Science,Massachusetts Institute of Technology,Cambridge,Massachusetts 02139,USA,School of Materials Science and Engineering,Gwangju Institute of Science and Technology,Gwangju 500-712,South Korea,Department of Physics,Massachusetts Institute of Technology, Cambridge,Massachusetts 02139,USA,Advanced Materials Laboratory,National Institute for Materials Science,1-1 Namiki,Tsukuba 305-0044,Japan. "e-mail:strano@mit.edu NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
Understanding and controlling the substrate effect on graphene electron-transfer chemistry via reactivity imprint lithography Qing Hua Wang1 , Zhong Jin1 , Ki Kang Kim2, Andrew J. Hilmer1 , Geraldine L. C. Paulus1 , Chih-Jen Shih1 , Moon-Ho Ham3, Javier D. Sanchez-Yamagishi4, Kenji Watanabe5, Takashi Taniguchi5, Jing Kong2, Pablo Jarillo-Herrero4 and Michael S. Strano1 * Graphene has exceptional electronic, optical, mechanical and thermal properties, which provide it with great potential for use in electronic, optoelectronic and sensing applications. The chemical functionalization of graphene has been investigated with a view to controlling its electronic properties and interactions with other materials. Covalent modification of graphene by organic diazonium salts has been used to achieve these goals, but because graphene comprises only a single atomic layer, it is strongly influenced by the underlying substrate. Here, we show a stark difference in the rate of electron-transfer reactions with organic diazonium salts for monolayer graphene supported on a variety of substrates. Reactions proceed rapidly for graphene supported on SiO2 and Al2O3 (sapphire), but negligibly on alkyl-terminated and hexagonal boron nitride (hBN) surfaces, as shown by Raman spectroscopy. We also develop a model of reactivity based on substrate-induced electron–hole puddles in graphene, and achieve spatial patterning of chemical reactions in graphene by patterning the substrate. Graphene is a two-dimensional, atomically thin lattice of sp2 -bonded carbon atoms and has exceptional electronic, mechanical and thermal properties1,2. Modifying the basic electronic, chemical and structural properties of graphene is important for incorporating graphene into a variety of applications including electronic devices, biosensors and composite materials3 . The chemical functionalization of graphene is critical for enabling these applications and has been explored for both covalent4,5 and non-covalent6–8 schemes. Functionalizing graphene with aryl diazonium salts4,9–16 results in the opening of a bandgap10,13,17–19 and shifting of the Fermi level10, both of which are desirable in the fabrication of electronic devices. In addition, the functional groups on the diazonium moiety can be tailored by organic chemistry to allow various chemical characteristics to be coupled to the graphene9 . Graphene is strongly influenced by the underlying substrate. SiO2-covered silicon substrates are compatible with device fabrication, but they have rough surfaces and contain charged impurities. These lead to electron–hole charge fluctuations (or ‘puddles’) in the graphene, which scatter charge carriers and inhibit electronic device performance20,21. Graphene devices suspended over gaps exhibit the highest carrier mobilities22,23, but are not robust for practical use. Recently, single-crystal hexagonal boron nitride (hBN)24,25 and self-assembled monolayers (SAMs) of hydrophobic molecules grafted on SiO2 substrates26–29 have been explored as alternative substrates for graphene electronics. Graphene on hBN, which is atomically smooth, chemically inert and electrically insulating, has significantly smaller electron–hole charge fluctuations and higher mobilities24,25. Graphene devices on SAM-covered substrates also exhibit lower charge inhomogeneity and performance hysteresis26,27 because the SAMs prevent dipolar contaminants from adsorbing on the substrate, prevent charge injection from the graphene to the dielectric interface and screen the effect of charged impurities within the substrate26,27,29. In this Article, we demonstrate that the substrate on which graphene rests strongly influences the chemical reactions on the top surface of the graphene. We also demonstrate spatial control of the chemical reactivity of graphene with micrometre-scale resolution to achieve wafer-scale patterning of chemical reactions on graphene. A previous report has shown differences in reactivity for small mechanically exfoliated flakes of graphene on SiO2 and hexamethyldisilazane (HMDS)-treated SiO2 (ref. 30). In the present work, chemical vapour deposition (CVD)-grown graphene is deposited on a variety of substrates and covalently functionalized with aryl diazonium salts. Using Raman spectroscopic mapping, we find that the substrate-induced electron–hole charge fluctuations in graphene greatly influence the chemical reactivity. Graphene on SiO2 and Al2O3 (sapphire) substrates is highly reactive, but graphene on an alkyl-terminated monolayer and hBN is much less reactive. We also develop a new lithographic patterning technique, reactivity imprint lithography (RIL), where the underlying substrate is chemically patterned to achieve spatial control of the graphene chemical reactivity. This method allows chemical reactions on graphene to be spatially patterned over large areas without the use of disruptive materials such as photoresists or chemical etchants. Here, RIL is used to spatially control the conjugation of enhanced green fluorescent protein (EGFP) on graphene, directly from solution, demonstrating the advantages of the technique for producing structures for sensor and microarray applications. 1 Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA, 2 Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA, 3 School of Materials Science and Engineering, Gwangju Institute of Science and Technology, Gwangju 500-712, South Korea, 4 Department of Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA, 5 Advanced Materials Laboratory, National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan. *e-mail: strano@mit.edu ARTICLES PUBLISHED ONLINE: 12 AUGUST 2012 | DOI: 10.1038/NCHEM.1421 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 1 © 2012 Macmillan Publishers Limited. All rights reserved
ARTICLES NATURE CHEMISTRY DOL:10.1038/NCHEM.1421 a b20 2D Graphene 0 GD N2*BF4 (AL2O.functionalized AlOa.pristine A1203 0 BN hBN.functionalized 1.4 Si2,plasma-Siplasma-cleaned 10 hBN.pristine cleaned SiO,piranha-cleaned 1.2 ■SiO2.as-received ■AlO3(sapphire) 1.0 ■OTS D OTS,functionalized 0.8 OTS,pristine OTS 0.6 0 0 C1- 0.4 as-received SiO2,functionalized 0.2 SiO,,pristine 0.01 0- 0 40 0 80 100 120 1.2001.4001.600 2.4002.6002.800 Contact angle(deg) Raman shift(cm-1) Figure 1 Chemical reactivity of graphene supported on different substrates.a,Reaction scheme of covalent chemical functionalization of graphene by 4-nitrobenzenediazonium tetrafluoroborate.b,Representative Raman spectra of CVD-grown graphene deposited on different substrate materials before and after diazonium functionalization,normalized to the G peak height.These substrates are,from bottom to top,300-nm-thick SiO,on silicon,SiO. functionalized by an OTS SAM,single-crystal hBN flakes deposited on SiO,and single-crystal o-Al,O(c-face sapphire).The SiO,substrate here was plasma-cleaned.c,Change in intensity ratio of Raman D and G peaks (p/lG)after diazonium functionalization (difference between functionalized and unfunctionalized ratios)plotted as a function of water contact-angle of the substrate before graphene deposition.The dashed line is an exponential fit of the data.Raman spectra were taken with a laser excitation wavelength of 633 nm. Results and discussion Ip/IG ratios.After diazonium functionalization,prominent D Chemical reactivity of graphene on different substrates.Large- peaks and small D'peaks appear on the SiO2 and Al2O3 substrates, area monolayer graphene grown by CVD on copper foils31 was cut indicating the significant formation of sp3 bonds.On the OTS into smaller pieces and transferred onto several different substrates and hBN substrates,very small D peaks appear,indicating using a poly(methyl methacrylate)(PMMA)-mediated transfer sparse covalent functionalization.For all substrates,the G and 2D method32.Graphene grown by CVD on copper foils is peaks are shifted up in position,and the 2D peak intensity predominantly monolayer,but polycrystalline(Supplementary Figs is decreased. S1,S2)3334.Covalent functionalization via an electron-transfer The correlation of chemical reactivity with the hydrophobicity of reaction with 4-nitrobenzenediazonium (4-NBD)tetrafluoroborate the underlying substrate is shown in Fig.Ic.In addition to the results in nitrobenzene groups being covalently attached to the oxygen-plasma-cleaned bare SiO2,we studied SiO,cleaned by graphene lattice(Fig.la).Figure 1b(right)shows the substrates used piranha solution (3:1 solution of sulfuric acid and 30%hydrogen in this work:300-nm-thick SiO,on a silicon wafer;a SAM of peroxide),which also produces a hydrophilic surface,and a octadecyltrichlorosilane (OTS)on 300 nm SiOz;a mechanically sample that was used as received.The hBN flakes were typically exfoliated flake of 90-nm-thick single-crystal hBN deposited on under 100 um in diameter and were too small for macroscopic 300 nm SiO,;and a single-crystal wafer of a-AlO,(polished contact-angle measurements.In general,the contact angle of the sapphire,c-plane).The SiO,substrate was cleaned by oxygen plasma substrate appears to be inversely correlated with graphene chemical to generate a hydrophilic surface terminated with-OH groups. reactivity.Low contact angles indicate hydrophilicity due to polar Figure 1b presents representative Raman spectra of graphene on chemical groups at the surface,which can induce electron-hole each substrate before and after diazonium functionalization.The puddles in graphene,whereas high contact angles indicate nonpolar primary peaks are the G peak near 1,580 cm,the D peak near surfaces.Further analysis of Raman spectra was conducted to clarify 1,300-1,350 cm'and the 2D peak near 2,600-2,700 cm (refs the role of the substrate in changing the chemical reactivity 35,36).The G and 2D peaks provide information about the level of graphene. of doping,strain and layer number5-3,and the D peak is activated by lattice defects3 including physical damage38.40 and the formation Analysis of Raman spectroscopic maps.Two-dimensional Raman of sphybridization by covalent chemistrys..The integrated inten- maps,with 121 spectra each and for points spaced 1 um apart,were sity ratio of the D and G peaks(Ip/IG)is a measure of the concen- taken in the same 10 um x 10 um sample areas before and after tration of covalent defect sites,and has been used by other diazonium functionalization.The Raman mapping enables a researchers to characterize the degree of covalent functionaliza- statistical analysis of many spectra and accounts for spatial tion10.Under our reaction conditions,physical damage is not heterogeneity in the graphene properties across the samples. incurred by the graphene lattice,so the increase in the D peak can Regions of uniform monolayer graphene were chosen to be attributed directly to the formation of covalent bonds as a avoid bilayer or multilayer islands,wrinkles and edges (see result of diazonium functionalization.In the spectra for pristine gra- Supplementary Fig.S1 for optical microscope images and phene in Fig.1b,which are normalized to the G peak height,the D additional Raman spectra of the initial graphene).The average peak is very small on all substrates and differences are seen in the peak parameters from fitting the peaks to Lorentzian functions NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
Results and discussion Chemical reactivity of graphene on different substrates. Largearea monolayer graphene grown by CVD on copper foils31 was cut into smaller pieces and transferred onto several different substrates using a poly(methyl methacrylate) (PMMA)-mediated transfer method32. Graphene grown by CVD on copper foils is predominantly monolayer, but polycrystalline (Supplementary Figs S1,S2)33,34. Covalent functionalization via an electron-transfer reaction with 4-nitrobenzenediazonium (4-NBD) tetrafluoroborate results in nitrobenzene groups being covalently attached to the graphene lattice (Fig. 1a). Figure 1b (right) shows the substrates used in this work: 300-nm-thick SiO2 on a silicon wafer; a SAM of octadecyltrichlorosilane (OTS) on 300 nm SiO2; a mechanically exfoliated flake of 90-nm-thick single-crystal hBN deposited on 300 nm SiO2; and a single-crystal wafer of a-Al2O3 (polished sapphire, c-plane). The SiO2 substrate was cleaned by oxygen plasma to generate a hydrophilic surface terminated with –OH groups. Figure 1b presents representative Raman spectra of graphene on each substrate before and after diazonium functionalization. The primary peaks are the G peak near 1,580 cm21 , the D peak near 1,300–1,350 cm21 and the 2D peak near 2,600–2,700 cm21 (refs 35,36). The G and 2D peaks provide information about the level of doping, strain and layer number35–38, and the D peak is activated by lattice defects39 including physical damage38,40 and the formation of sp3 hybridization by covalent chemistry5,10. The integrated intensity ratio of the D and G peaks (ID/IG) is a measure of the concentration of covalent defect sites, and has been used by other researchers to characterize the degree of covalent functionalization10. Under our reaction conditions, physical damage is not incurred by the graphene lattice, so the increase in the D peak can be attributed directly to the formation of covalent bonds as a result of diazonium functionalization. In the spectra for pristine graphene in Fig. 1b, which are normalized to the G peak height, the D peak is very small on all substrates and differences are seen in the I2D/IG ratios. After diazonium functionalization, prominent D peaks and small D′ peaks appear on the SiO2 and Al2O3 substrates, indicating the significant formation of sp3 bonds. On the OTS and hBN substrates, very small D peaks appear, indicating sparse covalent functionalization. For all substrates, the G and 2D peaks are shifted up in position, and the 2D peak intensity is decreased. The correlation of chemical reactivity with the hydrophobicity of the underlying substrate is shown in Fig. 1c. In addition to the oxygen-plasma-cleaned bare SiO2, we studied SiO2 cleaned by piranha solution (3:1 solution of sulfuric acid and 30% hydrogen peroxide), which also produces a hydrophilic surface, and a sample that was used as received. The hBN flakes were typically under 100 mm in diameter and were too small for macroscopic contact-angle measurements. In general, the contact angle of the substrate appears to be inversely correlated with graphene chemical reactivity. Low contact angles indicate hydrophilicity due to polar chemical groups at the surface, which can induce electron–hole puddles in graphene, whereas high contact angles indicate nonpolar surfaces. Further analysis of Raman spectra was conducted to clarify the role of the substrate in changing the chemical reactivity of graphene. Analysis of Raman spectroscopic maps. Two-dimensional Raman maps, with 121 spectra each and for points spaced 1 mm apart, were taken in the same 10 mm × 10 mm sample areas before and after diazonium functionalization. The Raman mapping enables a statistical analysis of many spectra and accounts for spatial heterogeneity in the graphene properties across the samples. Regions of uniform monolayer graphene were chosen to avoid bilayer or multilayer islands, wrinkles and edges (see Supplementary Fig. S1 for optical microscope images and additional Raman spectra of the initial graphene). The average peak parameters from fitting the peaks to Lorentzian functions Si Si Si Al2O3 SiO2 Graphene hBN 20 15 10 5 0 Normalized intensity (a.u.) 1,200 1,400 1,600 2,400 2,600 2,800 Al2O3, functionalized Al2O3, pristine hBN, pristine hBN, functionalized OTS, pristine OTS, functionalized SiO2, pristine SiO2, functionalized D G D′ 2D D D BN G D′ D SiO2, as-received OTS SiO2, piranha-cleaned Al2O3 (sapphire) SiO2, plasma-cleaned 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 20 40 60 80 100 120 OTS Contact angle (deg) ΔI D/I G SiO2, as-received SiO2, plasmacleaned a c b Raman shift (cm–1) Al2O3 NO2 NO2 BF4 – N2 + Si Cl Cl Cl Figure 1 | Chemical reactivity of graphene supported on different substrates. a, Reaction scheme of covalent chemical functionalization of graphene by 4-nitrobenzenediazonium tetrafluoroborate. b, Representative Raman spectra of CVD-grown graphene deposited on different substrate materials before and after diazonium functionalization, normalized to the G peak height. These substrates are, from bottom to top, 300-nm-thick SiO2 on silicon, SiO2 functionalized by an OTS SAM, single-crystal hBN flakes deposited on SiO2 and single-crystal a-Al2O3 (c-face sapphire). The SiO2 substrate here was plasma-cleaned. c, Change in intensity ratio of Raman D and G peaks (I D/I G) after diazonium functionalization (difference between functionalized and unfunctionalized ratios) plotted as a function of water contact-angle of the substrate before graphene deposition. The dashed line is an exponential fit of the data. Raman spectra were taken with a laser excitation wavelength of 633 nm. ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 2 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2012 Macmillan Publishers Limited. All rights reserved
NATURE CHEMISTRY DOI:10.1038/NCHEM.1421 ARTICLES Table 1 Summary of graphene Raman peak parameters before and after diazonium functionalization. WG (cm) Ia(cm) @2p (cm) Ip(cm) Ip/Is Bp/Is o (cm SiO,,pristine 1,588.6 14.4 2,644.1 33.7 0.11 4.24 7.1×10" SiO,,functionalized 1.591.9 18.1 2,649.8 36.1 1.42 1.64 11×103 OTS,pristine 1,5883 12.7 2,644.8 292 0.12 6.20 7.8×10 OTS,functionalized 1,596.7 12.4 2,6511 33.0 0.25 2.66 1.6×102 hBN,pristine 1584.7 14.5 2,645.6 27.8 0.13 9.88 8.4×10m hBN,functionalized 1,595.6 12.1 2,655.8 30.4 0.27 4.51 1.8x102 Al,O(sapphire),pristine 1595.6 12.5 2,653.7 30.7 0 6.01 0 AlO(sapphire),functionalized 1598.0 16.3 2,657.6 33.6 116 331 8.5×102 Average values for key Raman peak parameter ed for (plas ed).OTS.hBN and AlO The eak positiosof Gand2Dpeaks ( 2D pe ks (Ts and)a ensity ratios (p/l and/l)- sites (is calculated from equation (4). are summarized in Table 1.Histograms of the Ip/IG ratio in Fig.2a G peak position3641.42.This trend line has been shifted upwards show very low initial defect concentrations.After diazonium to accommodate the wider G peak in CVD graphene.Pristine gra- functionalization,the centres of the distributions have increased phene on each of the substrates generally follows the doping trend to ~1.2 for Al,O,and ~1.4 for SiO,,indicating a relatively high line,with hBN closer to the undoped region and Al2O3 closer to degree of covalent functionalization.The histograms are also the more doped region.However,electron and hole doping wider,suggesting an increased spatial inhomogeneity.For hBN cannot be distinguished from this plot,and graphene that is uni- and OTS,the Ip/IG ratio has only slightly increased,to ~0.25, formly electron-or hole-doped cannot be distinguished from gra- indicating much lower reactivity. phene with many electron-and hole-doped charge puddles.After Scatter plots of the Raman peak parameters are shown in Fig.2b-e. diazonium functionalization,is upshifted for all substrates, Data from literature reports of mechanically exfoliated monolayer suggesting increased doping,while IG is also much higher for graphene doped by electrostatic gating are included on these plots SiO2 and Al2O3,suggesting increased disorder3 as comparisons741.In Fig.2b,the full-width at half-maximum The G and 2D peak positions (@G and @D)are plotted against (FWHM)of the G peak(T)is plotted against the position of the each other in Fig.2c together with comparison data to distinguish G peak (G).The dashed trend line indicates that increasing n-or between n-and p-doping trends.The unfunctionalized graphene in p-doping leads to narrowing of the G peak and an increase of the our samples lies in the slightly p-doped region of this plot,with the SiO2,pristine o OTS.pristine △hBN,pristine VAl2Os.pristine Gated SiO2,functionalized ·OTS,functionalized▲hBN,functionalized Al,O,functionalized graphene a 100 Pristine Al2O(sapphire) 6 24 2,665 2.660 50 Functionalized 20 2,655 0 16 .2.650 2,645 40 hBN 2.640 2,635 1,5801,5851,5901,5951,6001.605 1,5801.5851,5901.5951,6001.605 wG (cm-1) WG(cm-1) istine OTS g Functionalized 12 % 10 35 6 n-or p 40 Functionalized doping 25 0 20 0 0.5 1.0 1.5 2.0 2,640 2.650 2.660 1.5801,5851,5901.5951.6001.605 bolle "2D(c-1) "e(cm-1) Figure 2 I Raman spectroscopy peak parameter analysis.Spatial Raman maps were collected for graphene supported on each substrate for the same 10 um x 10 um regions before and after diazonium functionalization,with 121 spectra in each map.a,Histograms of lp/lG ratios before and after functionalization.A low degree of covalent functionalization (small increase in Ip/l)is seen for OTS and hBN,and a much higher degree for SiO(plasma- cleaned)and AlO.b-e,Scatter plots of Raman peak parameters with data points adapted from pristine,mechanically exfoliated graphene doped by electrostatic gating dashed lines added to guide the eye are included to aid comparison37.4.b,G peak full-width at half-maximum(FWHM,I)versus G peak position ()Comparison data from ref.41 are shifted up to fit the higher FWHM of CVD graphene.Before reaction,graphene follows the doping trend,but highly functionalized samples significantly deviate above the curve.c,2D peak position (versus G peak position(,with additional data points adapted from ref.41 for distinguishing n-doped and p-doped exfoliated monolayer graphene,shifted to account for the dependence ofon excitation laser wavelength56.Diazonium-functionalized graphene in our experimental data is p-doped,but deviates left from the trend of pristine,gated graphene.d,2D peak FWHM(versus 2D peak position showing clearly distinguished clusters for each substrate before and after functionalization.Increasing I values before functionalization reflect inhomogeneous broadening due to electron-hole charge fluctuations.e,p/lG intensity ratio versus G peak position (G),with comparison data adapted from ref.37 showing the doping trend.Raman spectra were taken at 633 nm laser excitation wavelength. NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
are summarized in Table 1. Histograms of the ID/IG ratio in Fig. 2a show very low initial defect concentrations. After diazonium functionalization, the centres of the distributions have increased to 1.2 for Al2O3 and 1.4 for SiO2, indicating a relatively high degree of covalent functionalization. The histograms are also wider, suggesting an increased spatial inhomogeneity. For hBN and OTS, the ID/IG ratio has only slightly increased, to 0.25, indicating much lower reactivity. Scatter plots of the Raman peak parameters are shown in Fig. 2b–e. Data from literature reports of mechanically exfoliated monolayer graphene doped by electrostatic gating are included on these plots as comparisons37,41. In Fig. 2b, the full-width at half-maximum (FWHM) of the G peak (GG) is plotted against the position of the G peak (vG). The dashed trend line indicates that increasing n- or p-doping leads to narrowing of the G peak and an increase of the G peak position36,41,42. This trend line has been shifted upwards to accommodate the wider G peak in CVD graphene. Pristine graphene on each of the substrates generally follows the doping trend line, with hBN closer to the undoped region and Al2O3 closer to the more doped region. However, electron and hole doping cannot be distinguished from this plot, and graphene that is uniformly electron- or hole-doped cannot be distinguished from graphene with many electron- and hole-doped charge puddles. After diazonium functionalization, vG is upshifted for all substrates, suggesting increased doping, while GG is also much higher for SiO2 and Al2O3, suggesting increased disorder43. The G and 2D peak positions (vG and v2D) are plotted against each other in Fig. 2c together with comparison data41 to distinguish between n- and p-doping trends. The unfunctionalized graphene in our samples lies in the slightly p-doped region of this plot, with the Table 1 | Summary of graphene Raman peak parameters before and after diazonium functionalization. vG (cm–1) GG (cm–1) v2D (cm–1) G2D (cm–1) ID/IG I2D/IG s (cm–2) SiO2, pristine 1,588.6 14.4 2,644.1 33.7 0.11 4.24 7.1 × 1011 SiO2, functionalized 1,591.9 18.1 2,649.8 36.1 1.42 1.64 1.1 × 1013 OTS, pristine 1,588.3 12.7 2,644.8 29.2 0.12 6.20 7.8 × 1011 OTS, functionalized 1,596.7 12.4 2,651.1 33.0 0.25 2.66 1.6 × 1012 hBN, pristine 1,584.7 14.5 2,645.6 27.8 0.13 9.88 8.4 × 1011 hBN, functionalized 1,595.6 12.1 2,655.8 30.4 0.27 4.51 1.8 × 1012 Al2O3 (sapphire), pristine 1,595.6 12.5 2,653.7 30.7 0 6.01 0 Al2O3 (sapphire), functionalized 1,598.0 16.3 2,657.6 33.6 1.16 3.31 8.5 × 1012 Average values for key Raman peak parameters are summarized for pristine and functionalized graphene on SiO2 (plasma-cleaned), OTS, hBN and Al2O3 (sapphire) substrates. The parameters shown are the peak positions of G and 2D peaks (vG and v2D) and FWHM values of G and 2D peaks (GG and G2D) and D/G and 2D/G integrated intensity ratios (I D/I G and I2D/I G). The area concentration of defects or reacted sites (s) is calculated from equation (4). a b d c e 40 20 0 40 20 0 40 20 0 100 50 0 0 0.5 1.0 1.5 2.0 Number of spectra Al2O3 (sapphire) hBN OTS SiO2 Pristine Pristine Pristine Pristine Functionalized Functionalized Functionalized Functionalized 24 20 16 12 1,580 1,585 1,590 1,595 1,600 1,605 2,665 2,660 2,655 2,650 2,645 2,640 2,635 1,580 1,585 1,590 1,595 1,600 1,605 p-doping n-doping 45 40 35 30 25 20 2,640 2,650 2,660 12 10 8 6 4 2 0 1,580 1,585 1,590 1,595 1,600 1,605 I D /I G ΓG (cm–1) ω2D (cm–1) Γ2D (cm–1) I2D/I G (cm–1) ωG (cm–1) ωG (cm–1) ω2D (cm–1) ωG (cm–1) Disorder n- or pdoping n- or pdoping SiO2, pristine SiO2, functionalized OTS, pristine OTS, functionalized hBN, pristine hBN, functionalized Al2O3, pristine Al2O3, functionalized Gated graphene Figure 2 | Raman spectroscopy peak parameter analysis. Spatial Raman maps were collected for graphene supported on each substrate for the same 10mm × 10 mm regions before and after diazonium functionalization, with 121 spectra in each map. a, Histograms of I D/I G ratios before and after functionalization. A low degree of covalent functionalization (small increase in I D/I G) is seen for OTS and hBN, and a much higher degree for SiO2 (plasmacleaned) and Al2O3. b–e, Scatter plots of Raman peak parameters with data points adapted from pristine, mechanically exfoliated graphene doped by electrostatic gating; dashed lines added to guide the eye are included to aid comparison37,41. b, G peak full-width at half-maximum (FWHM, GG) versus G peak position (vG). Comparison data from ref. 41 are shifted up to fit the higher FWHM of CVD graphene. Before reaction, graphene follows the doping trend, but highly functionalized samples significantly deviate above the curve. c, 2D peak position (v2D) versus G peak position (vG), with additional data points adapted from ref. 41 for distinguishing n-doped and p-doped exfoliated monolayer graphene, shifted to account for the dependence of v2D on excitation laser wavelength56. Diazonium-functionalized graphene in our experimental data is p-doped, but deviates left from the trend of pristine, gated graphene. d, 2D peak FWHM (G2D) versus 2D peak position (v2D), showing clearly distinguished clusters for each substrate before and after functionalization. Increasing G2D values before functionalization reflect inhomogeneous broadening due to electron–hole charge fluctuations. e, I 2D/I G intensity ratio versus G peak position (vG), with comparison data adapted from ref. 37 showing the doping trend. Raman spectra were taken at 633 nm laser excitation wavelength. NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 ARTICLES NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 3 © 2012 Macmillan Publishers Limited. All rights reserved.
ARTICLES NATURE CHEMISTRY DOL:10.1038/NCHEM.1421 hBN surface being less doped,but graphene on Al2O3 is on the of the transition between the on-OTS and on-SiO2 regions,and is p-doping branch.After functionalization,graphene on all substrates ~0.85 um.The Ip/IG profile for graphene across the edge of a is further along the p-doping branch.However,covalent defects are flake of hBN is plotted and fitted similarly in Fig.3e,with a variance expected to cause deviations from these doping-related Raman of 0.76 pm.These variances are comparable to the 0.71 um diag- trends,which were measured on pristine graphene.The p-doping onal of the pixel size (0.5 um x 0.5 um)and the ~0.9 um laser after reaction has contributions from the covalent bond formation spot size.Therefore the measured resolution of the RIL patterns is itself and from the non-covalent adsorption of the diazonium limited by the optical characterization technique,and the true res- cation and oligomers213.18.Strain effects are ruled out as the olution of the chemical patterning is primarily determined by the cause of these peak position shifts because the graphene samples spatial resolution of the substrate patterning technique and spatial rest conformally on very flat substrates,and mechanical strain size of the electron-hole puddles on a given substrate,which the causes simultaneous downshifting of both the G and 2D peak pos- data indicate as less than 1 um. itions below the values for undoped graphene4,instead of the upshift that is observed here. Patterned attachment of proteins on graphene.Spatial control of The FWHM of the 2D peak(T2D)is plotted against its position surface chemistry is important for biological applications such as (@)in Fig.2d.Because the 2D peak position shifts in opposite microarrays,biosensors and tissue engineering.Many important directions for electron or hole doping (Fig.2c),the presence of macromolecules such as proteins,antibodies or DNA are not electron-hole puddles with spatial extents significantly smaller compatible with conventional lithographic techniques.RIL allows than the Raman laser spot size would result in a broadened 2D these biomolecules to be attached to graphene as the final processing peak.In our Raman system,the laser spot size is ~0.9 um in diam- step in aqueous solution.The patterning of biomolecules on eter,and the sizes of electron-hole puddles have been measured to graphene using RIL is schematically illustrated in Fig.4a.CVD be ~5-10 nm in diameter for graphene on SiO,and ~100 nm for graphene is transferred to an OTS-patterned substrate and graphene on hBN25.We therefore propose that a higher I is cor- functionalized by 4-carboxybenzenediazonium tetrafluoroborate. related with higher amplitudes of charge fluctuations.Graphene on The graphene is then reacted with NN-bis(carboxymethyl)-L- SiO,exhibits the highest I2p values,and graphene on hBN has the lysine hydrate (NTA-NH,)followed by reaction with NiCl,to lowest.This trend is in general agreement with the amplitudes of complex the Ni ions with the NTA structure.Finally,the sample charge fluctuations on SiO,and hBN measured by scanning tunnel- is incubated with a solution of polyhistidine (His)-tagged EGFP to ling spectroscopy25.On OTS,the T2p is slightly higher than on hBN form the graphene-NTA-Ni-His-EGFP complex. and notably lower than on Si.. Attachment of the carboxybenzene group is demonstrated by The integrated area intensity ratio Ip/IG is plotted against G in attenuated total reflectance infrared (ATR-IR)spectra of the pristine Fig.2e,with additional comparison data for gated pristine graphene CVD graphene (blue curve)and functionalized graphene (red adapted from ref.37,and shows that the Ip/IG ratio decreases and curve)in Fig.4b.Vibrations from carboxyl groups are seen at @G increases for increasing n-and p-doping.Graphene on hBN is ~1,730 cm(C=O stretching)and ~3,330 cm(O-H stretch- closest to the undoped region of the plot,followed by OTS,SiO ing).Confocal fluorescence microscopy after incubation in EGFP and finally Al2O,at the more highly doped region.(Although the shows bright green stripes,confirming the spatial patterning of peak intensities on Al,O have not been corrected for optical inter- the protein tethering reaction (Fig.4c).The wider,bright lines ference effects from the different substrate5,the peak positions are correspond to graphene resting on SiO,where the higher con- accurate.)After diazonium functionalization,the data points from centration of diazonium attachment sites results in a high coverage all substrates move further along the doping trend line.Again,we of EGFP.The narrower,dark lines correspond to graphene resting observe that diazonium functionalization increases the p-doping on OTS where the low reactivity results in fewer EGFP.The inset of the graphene. shows the fluorescence intensity profile along the white line. Graphene on the various substrates displays different extents of This tethering scheme is very robust because of the covalent overall p-doping and apparent intensities of electron-hole charge attachment site,and is also chemically reversible due to the metal fluctuations.Graphene on hBN is the least doped,with the lowest ion chelation,in contrast to a previous report of proteins patterned degree of charge fluctuations,followed by OTS.In contrast, on graphene by physisorption graphene on SiO,and Al2O3 are more highly p-doped,and on SiO,the I2p is the highest,indicating the greatest broadening of Reactivity model:the influence of electron puddles.To explain the the 2D peak from electron-hole puddles.After reaction,graphene chemical reactivity of graphene on the different substrates,we use a on all substrates shows increased p-doping.For the substrates model describing the reaction kinetics from electron-transfer theory with a low degree of sp3 hybridization,the p-doping arises from as a function of the Fermi level of graphene and relating the reacted diazonium molecules non-covalently deposited on the graphene. site density to an experimentally measurable Raman Ip/IG ratio. The role of electron-hole puddles in the reactivity of graphene is Owing to the overlap between graphene and the diazonium states, discussed in the following sections. the electron-transfer theory below shows that the reactivity increases for increasingly n-doped graphene and is negligible for Spatial patterning of chemical reactivity.With our RIL technique, p-doped graphene.The schematic in Fig.5a shows how a a substrate with OTS micropatterned46.47 on SiO2 was used to graphene sheet that is overall p-doped but with a high electron- spatially control the chemical reactivity of graphene (Fig.3a).The hole charge fluctuation amplitude can have much higher reactivity patterned surface in the topographic atomic force microscopy due to the locally n-doped puddles. (AFM)image of Fig.3b comprises ~2-um-wide OTS lines and In a first-order electron-transfer reaction model,the density ~7 um wide SiOz gaps.Graphene was transferred onto this of reacted lattice sites o is given by substrate and functionalized by diazonium salts.Figure 3c shows the resulting spatial Raman map of Ip/IG.The narrower regions of Pc(1-exp(-(KE.[D]s/Pc)t)) (1) low functionalization correspond to graphene over OTS-covered areas and the wider stripes of high functionalization the SiO,regions. where pc is the number of carbon atoms per unit area in graphene,[D]s The Ip/IG spatial profile at the edge of a stripe was fit using is the concentration of diazonium ions,ker is the rate constant of elec- an integral Gaussian distribution in Fig.3d (Supplementary tron transfer,and t is the reaction time.The overall reaction rate is Information,Page 8).The variance of this fit indicates the sharpness limited by the electron transfer rate from graphene to diazonium,as NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
hBN surface being less doped, but graphene on Al2O3 is on the p-doping branch. After functionalization, graphene on all substrates is further along the p-doping branch. However, covalent defects are expected to cause deviations from these doping-related Raman trends, which were measured on pristine graphene. The p-doping after reaction has contributions from the covalent bond formation itself and from the non-covalent adsorption of the diazonium cation and oligomers12,13,18. Strain effects are ruled out as the cause of these peak position shifts because the graphene samples rest conformally on very flat substrates, and mechanical strain causes simultaneous downshifting of both the G and 2D peak positions below the values for undoped graphene44, instead of the upshift that is observed here. The FWHM of the 2D peak (G2D) is plotted against its position (v2D) in Fig. 2d. Because the 2D peak position shifts in opposite directions for electron or hole doping (Fig. 2c), the presence of electron–hole puddles with spatial extents significantly smaller than the Raman laser spot size would result in a broadened 2D peak. In our Raman system, the laser spot size is 0.9 mm in diameter, and the sizes of electron–hole puddles have been measured to be 5–10 nm in diameter for graphene on SiO2 and 100 nm for graphene on hBN25. We therefore propose that a higher G2D is correlated with higher amplitudes of charge fluctuations. Graphene on SiO2 exhibits the highest G2D values, and graphene on hBN has the lowest. This trend is in general agreement with the amplitudes of charge fluctuations on SiO2 and hBN measured by scanning tunnelling spectroscopy25. On OTS, the G2D is slightly higher than on hBN and notably lower than on SiO2. The integrated area intensity ratio I2D/IG is plotted against vG in Fig. 2e, with additional comparison data for gated pristine graphene adapted from ref. 37, and shows that the I2D/IG ratio decreases and vG increases for increasing n- and p-doping. Graphene on hBN is closest to the undoped region of the plot, followed by OTS, SiO2 and finally Al2O3 at the more highly doped region. (Although the peak intensities on Al2O3 have not been corrected for optical interference effects from the different substrate45, the peak positions are accurate.) After diazonium functionalization, the data points from all substrates move further along the doping trend line. Again, we observe that diazonium functionalization increases the p-doping of the graphene. Graphene on the various substrates displays different extents of overall p-doping and apparent intensities of electron–hole charge fluctuations. Graphene on hBN is the least doped, with the lowest degree of charge fluctuations, followed by OTS. In contrast, graphene on SiO2 and Al2O3 are more highly p-doped, and on SiO2 the G2D is the highest, indicating the greatest broadening of the 2D peak from electron–hole puddles. After reaction, graphene on all substrates shows increased p-doping. For the substrates with a low degree of sp3 hybridization, the p-doping arises from diazonium molecules non-covalently deposited on the graphene. The role of electron–hole puddles in the reactivity of graphene is discussed in the following sections. Spatial patterning of chemical reactivity. With our RIL technique, a substrate with OTS micropatterned46,47 on SiO2 was used to spatially control the chemical reactivity of graphene (Fig. 3a). The patterned surface in the topographic atomic force microscopy (AFM) image of Fig. 3b comprises 2-mm-wide OTS lines and 7 mm wide SiO2 gaps. Graphene was transferred onto this substrate and functionalized by diazonium salts. Figure 3c shows the resulting spatial Raman map of ID/IG. The narrower regions of low functionalization correspond to graphene over OTS-covered areas and the wider stripes of high functionalization the SiO2 regions. The ID/IG spatial profile at the edge of a stripe was fit using an integral Gaussian distribution in Fig. 3d (Supplementary Information, Page 8). The variance of this fit indicates the sharpness of the transition between the on-OTS and on-SiO2 regions, and is 0.85 mm. The ID/IG profile for graphene across the edge of a flake of hBN is plotted and fitted similarly in Fig. 3e, with a variance of 0.76 mm. These variances are comparable to the 0.71 mm diagonal of the pixel size (0.5 mm × 0.5 mm) and the 0.9 mm laser spot size. Therefore the measured resolution of the RIL patterns is limited by the optical characterization technique, and the true resolution of the chemical patterning is primarily determined by the spatial resolution of the substrate patterning technique and spatial size of the electron–hole puddles on a given substrate, which the data indicate as less than 1 mm. Patterned attachment of proteins on graphene. Spatial control of surface chemistry is important for biological applications such as microarrays, biosensors and tissue engineering. Many important macromolecules such as proteins, antibodies or DNA are not compatible with conventional lithographic techniques. RIL allows these biomolecules to be attached to graphene as the final processing step in aqueous solution. The patterning of biomolecules on graphene using RIL is schematically illustrated in Fig. 4a. CVD graphene is transferred to an OTS-patterned substrate and functionalized by 4-carboxybenzenediazonium tetrafluoroborate. The graphene is then reacted with Na,Na-bis(carboxymethyl)-Llysine hydrate (NTA–NH2) followed by reaction with NiCl2 to complex the Ni2þ ions with the NTA structure. Finally, the sample is incubated with a solution of polyhistidine (His)-tagged EGFP to form the graphene–NTA–Ni–His–EGFP complex. Attachment of the carboxybenzene group is demonstrated by attenuated total reflectance infrared (ATR-IR) spectra of the pristine CVD graphene (blue curve) and functionalized graphene (red curve) in Fig. 4b. Vibrations from carboxyl groups are seen at 1,730 cm21 (C¼O stretching) and 3,330 cm21 (O–H stretching). Confocal fluorescence microscopy after incubation in EGFP shows bright green stripes, confirming the spatial patterning of the protein tethering reaction (Fig. 4c). The wider, bright lines correspond to graphene resting on SiO2 where the higher concentration of diazonium attachment sites results in a high coverage of EGFP. The narrower, dark lines correspond to graphene resting on OTS where the low reactivity results in fewer EGFP. The inset shows the fluorescence intensity profile along the white line. This tethering scheme is very robust because of the covalent attachment site, and is also chemically reversible due to the metal ion chelation, in contrast to a previous report of proteins patterned on graphene by physisorption48. Reactivity model: the influence of electron puddles. To explain the chemical reactivity of graphene on the different substrates, we use a model describing the reaction kinetics from electron-transfer theory as a function of the Fermi level of graphene and relating the reacted site density to an experimentally measurable Raman ID/IG ratio. Owing to the overlap between graphene and the diazonium states, the electron-transfer theory below shows that the reactivity increases for increasingly n-doped graphene and is negligible for p-doped graphene. The schematic in Fig. 5a shows how a graphene sheet that is overall p-doped but with a high electron– hole charge fluctuation amplitude can have much higher reactivity due to the locally n-doped puddles. In a first-order electron-transfer reaction model, the density of reacted lattice sites s is given by s = rC(1 − exp(−(kET [D]S/rC)t)) (1) where rC is the number of carbon atoms per unit area in graphene, [D]S is the concentration of diazonium ions, kET is the rate constant of electron transfer, and t is the reaction time. The overall reaction rate is limited by the electron transfer rate from graphene to diazonium, as ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 4 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY DOI:10.1038/NCHEM.1421 ARTICLES a PDMS stamp OTS ink Graphene Reactivity pattem OTS pattern OTS pattern Sio. Si Height(nm) 5μm d Dat 12 一Fit hBN 0.8 1.0 B 0.8 0.6 un 0.6 0.4 0.4 0.2 0.2- 0.0 0.0 0 3 4 681012 14 Distance (um) Distance(um) Figure 3 Spatial control of reactivity of graphene on patterned substrates.a,Schematic illustration of RIL.The SiO,substrate is patterned by a PDMS stamp inked with OTS.Graphene is transferred over the OTS-patterned substrate and reacted with 4-NBD tetrafluoroborate.b,AFM topographic image of the OTS stripes (narrower raised regions)on SiO2 before graphene deposition.c,Raman spatial map of Ip/lG intensity ratio after diazonium functionalization The narrow,mildly functionalized stripes correspond to the regions over the OTS pattern and the wide,strongly functionalized stripes correspond to the regions over the SiO,gaps.d,Spatial profile of p/l for the stripe pattern (blue curve)along the line A-B in the Raman map (inset),and a fit to an integrated Gaussian function with a variance of 0.85 um.e,A spatial Raman map (lower left inset)was measured for a region of graphene covering both SiO and a flake of hBN(white box in optical image in upper right inset).The lp/lG spatial profile along the line C-D is shown together with the integrated Gaussian fit,which has a variance of 0.76 um. is the case for carbon nanotubes,and depends on the overlap of states of the 4-NBD diazonium salt53 and DOSG(E)is the electronic between graphene and diazonium.Once the diazonium radical forms, density of states of graphene.The electron-transfer frequency v it is highly reactive and can be quenched readily by a variety of and integral prefactor rd are treated as a single fitting parameter substratessas1.Because the rate-limiting step is electron transfer,the eThe distribution of oxidized states of the solvated diazonium Fermi level of graphene determines the influence of the substrate on molecule W (E)is given by graphene reactivity,and the specific interactions of the charged states in graphene with the diazonium radical can be neglected.The rate 1 constant ker is described using Gerischer-Marcus theorys2. Wox(E)= (E-(Eedox+)) √4Tk7 exp (3) 4λkT red(E)DOSG(E)W(E)dE (2) where k is the Boltzmann constant,T is the absolute temperature and A is the energy difference between the standard potential for the redox couple of the diazonium salt and the energy for where EG=-4.66 eV is the Fermi level of undoped graphene, maximum probability of finding a vacant state.This parameter is Eredox=-5.15 eV is the standard potential for the redox couple also known as the reorganization energy and is ~0.7 ev for NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
is the case for carbon nanotubes49, and depends on the overlap of states between graphene and diazonium. Once the diazonium radical forms, it is highly reactive and can be quenched readily by a variety of substrates50,51. Because the rate-limiting step is electron transfer, the Fermi level of graphene determines the influence of the substrate on graphene reactivity, and the specific interactions of the charged states in graphene with the diazonium radical can be neglected. The rate constant kET is described using Gerischer–Marcus theory52: kET = nn EF,G Eredox 1red( ) E DOSG( ) E Wox( ) E dE (2) where EF,G ¼ –4.66 eV is the Fermi level of undoped graphene, Eredox ¼ –5.15 eV is the standard potential for the redox couple of the 4-NBD diazonium salt53 and DOSG(E) is the electronic density of states of graphene. The electron-transfer frequency nn and integral prefactor 1red are treated as a single fitting parameter nn1red. The distribution of oxidized states of the solvated diazonium molecule Wox(E) is given by Wox( )= E 1 4plkT √ exp − E − Eredox + l 2 4lkT (3) where k is the Boltzmann constant, T is the absolute temperature and l is the energy difference between the standard potential for the redox couple of the diazonium salt and the energy for maximum probability of finding a vacant state. This parameter is also known as the reorganization energy and is 0.7 eV for b c 5 μm OTS SiO2 8 0 Height (nm) 1.4 0 2 µm ID / G I a PDMS stamp OTS ink Si SiO2 OTS pattern Graphene OTS pattern Reactivity pattern d e 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 Data Fit 1.0 0.8 0.6 0.4 0.2 0.0 0 2 4 6 8 10 12 14 I D/I G I D/I G A B C D A B C D 5 μm hBN SiO2 Distance (μm) Distance (μm) Figure 3 | Spatial control of reactivity of graphene on patterned substrates. a, Schematic illustration of RIL. The SiO2 substrate is patterned by a PDMS stamp inked with OTS. Graphene is transferred over the OTS-patterned substrate and reacted with 4-NBD tetrafluoroborate. b, AFM topographic image of the OTS stripes (narrower raised regions) on SiO2 before graphene deposition. c, Raman spatial map of I D/I G intensity ratio after diazonium functionalization. The narrow, mildly functionalized stripes correspond to the regions over the OTS pattern and the wide, strongly functionalized stripes correspond to the regions over the SiO2 gaps. d, Spatial profile of I D/I G for the stripe pattern (blue curve) along the line A–B in the Raman map (inset), and a fit to an integrated Gaussian function with a variance of 0.85 mm. e, A spatial Raman map (lower left inset) was measured for a region of graphene covering both SiO2 and a flake of hBN (white box in optical image in upper right inset). The I D/I G spatial profile along the line C–D is shown together with the integrated Gaussian fit, which has a variance of 0.76 mm. NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 ARTICLES NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 5 © 2012 Macmillan Publishers Limited. All rights reserved.
ARTICLES NATURE CHEMISTRY DOL:10.1038/NCHEM.1421 CO2H N、CO2H 02H His-EGFP ●0H NiCk 100.0 99.0 8 Pristine graphene 98.0 万 97.0 100.0 CO,H-diazonium functionalized 30 99.0 20 98.0 10 vO-H 97.0 1020. 30 3,500 3.000 2,500 2.000 1,500 Distance(μm) Wavenumber(cm-1) Figure 4 Patterning of proteins on graphene.a,Schematic illustration of protein-attachment chemistry.The graphene is covalently functionalized with 4-carboxybenzenediazonium tetrafluoroborate,then NTA-NH2.Reaction with NiCl causes Ni ions to complex to the covalently attached structures,and link to polyhistidine (His)-tagged EGFP.(Image of EGFP is taken from the RCSB PDB (www.pdb.org)from ref.57.)b,ATR-IR spectra of pristine CVD graphene (blue curve)and COzH-diazonium functionalized CVD graphene(red curve),showing O-H and C=O vibrations from the carboxyl groups. c Confocal fluorescence microscope image of EGFP attached to graphene resting on a substrate with alternating stripes of bare SiO2 and OTS patterned on graphene.The bright green stripes,indicating a higher concentration of EGFP attachment,corresponds to graphene resting on bare SiO2,and the darker stripes correspond to graphene resting on OTS-patterned regions where very little EGFP was able to attach.Inset:intensity profile of fluorescence along the white line indicated in c. single-walled carbon nanotubes9;it is assumed to be similar for gra- sparse coverage12.Additional STM imaging as well as Raman phene.The model here shows how the electronic density of states spectroscopy would be valuable for clarifying the relation between and Fermi level of graphene directly influence the reaction rate. reacted site concentration and Ip/IG and for elucidating the The density of reacted sites o was quantitatively related to the graphene microstructure. Ip/IG ratio by Lucchese et al.40 The model curve is plotted together with experimental data from several samples of graphene on different substrate materials in 是-c((到 Fig.5b,c.To obtain the average Fermi level,the Raman IG/Ip ratio was used54: +c-()] (4) (5) D =C(Ye-ph +0.07Eg.avg where the distance between defects is Lp=1/o.Around each where Y-ph33 meV is the average energy of electron scattering defect site is a structurally damaged region with radius r=rs and due to phonon emission and C10ev-(ref.54).We have used around that an activated region between r=rs and r=r that is pri- the versus G data (Fig.2c)to determine Eav <0.However, marily responsible for an increase in the D peak.In ref.40,changes the hole-doped data show little agreement with the model in in I/I are caused by ion bombardment damage,but covalent Fig.5b.To account for electron-hole puddles as illustrated in functionalization with diazonium salts results in a slightly different Fig.5a,we note that the reactivity is instead dominated by the behaviour of Ip/IG (ref.10).Accordingly,we used smaller values of sum of the average Fermi leve and the amplitude of the rs=0.07 nm and r =1.0 nm because a covalent attachment site is puddle,which should be proportional to the increase in Tap com- much less disruptive to the lattice than an ion bombardment defect. pared to the case with negligible puddle influence.Specifically,the Parameters Ca and Cs are similar to the values used in ref.40. effective Fermi level of the n-doped puddles E is Combining equations (1)to (4)results in a curve showing the Ip/IG after diazonium functionalization as a function of graphene EE.n EE.avg a(T2D-T2D.o) (6) E with vred as the fitting parameter.The surface concentration of defect sites o for graphene on each substrate was estimated where a is a proportionality constant,and Tis the FWHM of the from equation(4)and is summarized in Table 1.Our estimated 2D peak for graphene with no charge puddles.In Fig.5c,the data reacted site concentration is ~1x 1012 to ~1x 1013 cm2 points were shifted using a=0.08 eV cm and T2D=26 cm which is much lower than the estimate for near-saturation of and the model curve is plotted with red=0.105.(Note that 1x 1015 cm-2 reported elsewhere,but is consistent with a<o for Ep in the p-doped puddles.)After the adjustment in molecularly resolved scanning tunnelling microscopy (STM)of equation (6)to account for the n-doped puddles,the data are diazonium-functionalized graphene showing a much more much better described by the model. NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
single-walled carbon nanotubes49; it is assumed to be similar for graphene. The model here shows how the electronic density of states and Fermi level of graphene directly influence the reaction rate. The density of reacted sites s was quantitatively related to the ID/IG ratio by Lucchese et al.40 ID IG = CA r 2 A − r 2 S r2 A − 2r2 S exp − pr 2 S L2 D − exp − p r 2 A − r 2 S L2 D + CS 1 − exp − pr 2 S L2 D (4) where the distance between defects is LD ¼ 1/ ps. Around each defect site is a structurally damaged region with radius r ¼ rS and around that an activated region between r ¼ rS and r ¼ rA that is primarily responsible for an increase in the D peak. In ref. 40, changes in ID/IG are caused by ion bombardment damage, but covalent functionalization with diazonium salts results in a slightly different behaviour of ID/IG (ref. 10). Accordingly, we used smaller values of rS ¼ 0.07 nm and rA ¼ 1.0 nm because a covalent attachment site is much less disruptive to the lattice than an ion bombardment defect. Parameters CA and CS are similar to the values used in ref. 40. Combining equations (1) to (4) results in a curve showing the ID/IG after diazonium functionalization as a function of graphene EF with nn1red as the fitting parameter. The surface concentration of defect sites s for graphene on each substrate was estimated from equation (4) and is summarized in Table 1. Our estimated reacted site concentration is 1 × 1012 to 1 × 1013 cm22 , which is much lower than the estimate for near-saturation of 1 × 1015 cm22 reported elsewhere4 , but is consistent with molecularly resolved scanning tunnelling microscopy (STM) of diazonium-functionalized graphene showing a much more sparse coverage12. Additional STM imaging as well as Raman spectroscopy would be valuable for clarifying the relation between reacted site concentration and ID/IG and for elucidating the graphene microstructure. The model curve is plotted together with experimental data from several samples of graphene on different substrate materials in Fig. 5b,c. To obtain the average Fermi level, the Raman IG/I2D ratio was used54: IG I2D = C ge−ph + 0.07 EF,avg (5) where ge–ph ¼ 33 meV is the average energy of electron scattering due to phonon emission and C ≈ 10 eV21 (ref. 54). We have used the v2D versus vG data (Fig. 2c) to determine EF,avg , 0. However, the hole-doped data show little agreement with the model in Fig. 5b. To account for electron–hole puddles as illustrated in Fig. 5a, we note that the reactivity is instead dominated by the sum of the average Fermi level EF,avg and the amplitude of the puddle, which should be proportional to the increase in G2D compared to the case with negligible puddle influence. Specifically, the effective Fermi level of the n-doped puddles EF,n is EF,n = EF,avg + a G2D − G2D,0 (6) where a is a proportionality constant, and G2D,0 is the FWHM of the 2D peak for graphene with no charge puddles. In Fig. 5c, the data points were shifted using a ¼ 0.08 eV cm and G2D,0 ¼ 26 cm21 , and the model curve is plotted with nn1red ¼ 0.105. (Note that a , 0 for EF,p in the p-doped puddles.) After the adjustment in equation (6) to account for the n-doped puddles, the data are much better described by the model. BF4 – CO2H N2 + CO2H N HO2C CO2H CO2H H2N O H N Ni N O O O O O O O H N CO2H N CO2H CO2H His-EGFP NiCl2 3,500 3,000 2,500 2,000 1,500 100.0 99.0 98.0 97.0 Normalized transmittance (%) 100.0 99.0 98.0 97.0 Pristine graphene CO2H-diazonium functionalized νO H− νC=O b 0 10 20 30 Distance (μm) 30 20 10 0 Intensity (a.u.) 15 μm c Wavenumber (cm–1) a Figure 4 | Patterning of proteins on graphene. a, Schematic illustration of protein-attachment chemistry. The graphene is covalently functionalized with 4-carboxybenzenediazonium tetrafluoroborate, then NTA–NH2. Reaction with NiCl2 causes Ni2þ ions to complex to the covalently attached structures, and link to polyhistidine (His)-tagged EGFP. (Image of EGFP is taken from the RCSB PDB (www.pdb.org) from ref. 57.) b, ATR–IR spectra of pristine CVD graphene (blue curve) and CO2H-diazonium functionalized CVD graphene (red curve), showing O2H and C¼O vibrations from the carboxyl groups. c, Confocal fluorescence microscope image of EGFP attached to graphene resting on a substrate with alternating stripes of bare SiO2 and OTS patterned on graphene. The bright green stripes, indicating a higher concentration of EGFP attachment, corresponds to graphene resting on bare SiO2, and the darker stripes correspond to graphene resting on OTS-patterned regions where very little EGFP was able to attach. Inset: intensity profile of fluorescence along the white line indicated in c. ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 6 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY DOI:10.1038/NCHEM.1421 ARTICLES a High p-doping,high electron-hole Low p-doping,low electron-hole puddle amplitude puddle amplitude Low reactivity,e.g.hBN High reactivity,e.g.SiOz ■sio2,plasma-cleaned■SiOz,as--received■OTs -Model ■Sio2,piranha-cleaned■Al2O3(sapphire)■hBN b 1.4 c1.4 1.2 12 1.0 1.0 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0.2 0 0.2 0.4 -0.2 0 0.2 0.4 Fermi level(eV) Fermi level(eV) Increased diazonium Decreased diazonium concentration concentration d 3.5 ●100nmSi02 3.0 ■300nmSi02 2.5 Effect of 2.0 higher Ep 1.5 T : 1.0 Effect of lower E 0.5 01 -40 -20 0 20 40 ve(v) Figure 5 Modelling of substrate-influenced reactivity.a,Schematic of the role of electron-hole charge fluctuations in graphene reactivity.Solid curves indicate spatial variation of the local Fermi level in charge puddles,and the dashed lines indicate the average Fermi level.The green curve (left)represents graphene on a substrate that causes it to be mildly p-doped with small charge fluctuations,and the red curve (right)represents higher p-doping and large charge fluctuations.According to electron-transfer theory,n-doped puddles have a higher reactivity towards diazonium functionalization and the p-doped puddles have very low reactivity.b,Experimental data from graphene on various substrates are plotted together with the curve from the electron-transfer model for the initial graphene Fermi level (E)and change in /ratio after diazonium functionalization.The experimental average E values are calculated from the l2p/lG ratios before functionalization.Each experimental point is the average value for a particular sample taken from 121 Raman spectra in a map, and the error bars represent standard deviation.The average doping for all samples is p-type,and does not agree with the model.c,Average E values are offset by considering the FWHM of 2D peaks,which reflects inhomogeneous broadening due to electron-hole charge fluctuations,to reflect the maximum n-doping.d,Resulting/ratio changes measured after electrochemical functionalization experiments at different applied gate voltages for samples on 100nm and 300 nm SiO dielectric layers,showing the effect of Fermi level shifts and field-induced diazonium concentration change on overall reactivity. Considering the Raman spectral analysis and the modelling results An unknown film of organic contamination probably covers as- above,we can summarize the effects of the different substrates on the received SiO,substrates and serves a similar role as the OTS monolayer. chemical reactivity of graphene.Graphene on hBN and OTS has low Although the Al2Os substrates are single crystals in the bulk,their sur- electron-hole fluctuations and hence lower diazonium reactivity, faces are likely to be similar to the amorphous SiO,substrates. whereas graphene on hydrophilic SiO2(plasma-cleaned and piranha- The Fermi level offsets calculated in equation(6)are larger than cleaned)and AlO,has higher charge fluctuations that result in more the electron-hole fluctuations reported earlier for mechanically n-doped reactive regions.The charge fluctuations on SiO,are caused exfoliated single-crystal graphene25.This difference may be by charged impurities in the substrate and polar adsorbates on the explained by grain boundaries and other contaminants in the surface,so adding the OTS monolayer decreases the fluctuations by CVD graphene that can increase the reactivity for a lower Fermi increasing the distance between the graphene and the charged impuri- level shift.Furthermore,the 2D Raman peaks from graphene with ties and by reducing the adsorption of polar adsorbates such as water. high electron-hole fluctuations would also have lower intensities, NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
Considering the Raman spectral analysis and the modelling results above, we can summarize the effects of the different substrates on the chemical reactivity of graphene. Graphene on hBN and OTS has low electron–hole fluctuations and hence lower diazonium reactivity, whereas graphene on hydrophilic SiO2 (plasma-cleaned and piranhacleaned) and Al2O3 has higher charge fluctuations that result in more n-doped reactive regions. The charge fluctuations on SiO2 are caused by charged impurities in the substrate and polar adsorbates on the surface, so adding the OTS monolayer decreases the fluctuations by increasing the distance between the graphene and the charged impurities and by reducing the adsorption of polar adsorbates such as water. An unknown film of organic contamination probably covers asreceived SiO2 substrates and serves a similar role as the OTS monolayer. Although the Al2O3 substrates are single crystals in the bulk, their surfaces are likely to be similar to the amorphous SiO2 substrates. The Fermi level offsets calculated in equation (6) are larger than the electron–hole fluctuations reported earlier for mechanically exfoliated single-crystal graphene25. This difference may be explained by grain boundaries and other contaminants in the CVD graphene that can increase the reactivity for a lower Fermi level shift. Furthermore, the 2D Raman peaks from graphene with high electron–hole fluctuations would also have lower intensities, Δl D/l G 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 –0.2 0.2 Fermi level (eV) 0 0.4 1.4 1.2 1.0 0.8 Δ 0.6 l D/l G 0.4 0.2 0 –0.2 0.2 Fermi level (eV) 0 0.4 Low p-doping, low electron-hole puddle amplitude a b c d Low reactivity, e.g. hBN High reactivity, e.g. SiO2 Fermi level High p-doping, high electron-hole puddle amplitude r Al2O3 SiO (sapphire) 2, piranha-cleaned hBN SiO2, plasma-cleaned SiO2, as-received OTS Model Δl D/l G Increased diazonium concentration Effect of higher EF Effect of lower EF Decreased diazonium concentration 3.0 3.5 2.5 2.0 1.5 1.0 0.5 0 –40 –20 20 40 0 Vg (V) 100 nm SiO2 300 nm SiO2 Figure 5 | Modelling of substrate-influenced reactivity. a, Schematic of the role of electron–hole charge fluctuations in graphene reactivity. Solid curves indicate spatial variation of the local Fermi level in charge puddles, and the dashed lines indicate the average Fermi level. The green curve (left) represents graphene on a substrate that causes it to be mildly p-doped with small charge fluctuations, and the red curve (right) represents higher p-doping and large charge fluctuations. According to electron-transfer theory, n-doped puddles have a higher reactivity towards diazonium functionalization and the p-doped puddles have very low reactivity. b, Experimental data from graphene on various substrates are plotted together with the curve from the electron-transfer model for the initial graphene Fermi level (EF) and change in I D/I G ratio after diazonium functionalization. The experimental average EF values are calculated from the I 2D/I G ratios before functionalization37. Each experimental point is the average value for a particular sample taken from 121 Raman spectra in a map, and the error bars represent standard deviation. The average doping for all samples is p-type, and does not agree with the model. c, Average EF values are offset by considering the FWHM of 2D peaks, which reflects inhomogeneous broadening due to electron–hole charge fluctuations, to reflect the maximum n-doping. d, Resulting I D/I G ratio changes measured after electrochemical functionalization experiments at different applied gate voltages for samples on 100 nm and 300 nm SiO2 dielectric layers, showing the effect of Fermi level shifts and field-induced diazonium concentration change on overall reactivity. NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 ARTICLES NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 7 © 2012 Macmillan Publishers Limited. All rights reserved
ARTICLES NATURE CHEMISTRY DOL:10.1038/NCHEM.1421 causing the Eavs calculated from equation(5)to be further from acetone to dissolve the PMMA,followed by rinsing in isopropanol and drying with neutrality and requiring a larger shift in equation(6)to fit the nitrogen gas. Ip/I.The hydrophobicity of the substrate is an initial predictor Surface preparation of wafer substrates.Silicon wafers with 300 nm SiO,were of the chemical reactivity as shown in Fig.Ic because the surface ultrasonically ceaned in sequential baths of acetone and isopropanol,blown dry energy of the substrate relates to the presence of charged impurities with nitrogen,and subjected to additional surface treatments.Plasma-cleaned and polar surface groups that can induce electron-hole charge fluc- samples:these were exposed to oxygen plasma (AutoGlow Plasma System,Glow Research)for 10-30 min at 200 W power and 0.5 torr.Piranha-cleaned samples: tuations in graphene. these were immersed in piranha solution (3:1 solution of sulfuric acid and 30% To test the implications of our substrate-dependent graphene hydrogen peroxide)for 15 min and rinsed in ultrapure water.(Warning:piranha reactivity results and model,we conducted electrochemical functio- solution is a strong oxidizing agent and should be handled with extreme care.) nalization experiments where the graphene was electrically doped As-received samples:these were subjected to no additional treatment.Sapphire by an applied backgate voltage during reaction (see full details in wafers (a-Al2O3,c-plane,05 mm thick,MTI Corp.)were ultrasonically cleaned in the Supplementary Information,Pages 10-13 and Supplementary acetone and isopropanol and blown dry with nitrogen. Fig.S8).Our model suggests that for sufficiently large shifts in the OTS monolayer on SiO,.Octadecyltrichlorosilane (OTS)(Sigma-Aldrich,90+%) Fermi level,the contribution of the electron-hole puddles is over- SAMs were formed on freshly plasma-cleaned SiO,substrates in OTS solution come by the overall Fermi level,with overall reactivity decreasing (10 mM in toluene)overnight in a closed vial,then rinsed in fresh toluene and blown for overall highly p-doped graphene and increasing for overall dry with nitrogen. highly n-doped graphene.This approach is complicated by the dia Surface patterning of substrates.OTS patterns were formed on freshly plasma zonium cation being either attracted or repelled by the applied gate cleaned SiO,substrates by printing with polydimethylsiloxane(PDMS)stamps. voltage so that the concentration of diazonium at the graphene Master patterns were formed by electron-beam lithography of PMMA resists on surface is significantly increased or decreased.This concentration silicon wafers.PDMS(10:1 mass ratio of base to curing agent,Dow Corning Sylgard 184)was poured into the master patterns,degassed in vacuum for 45 min,and cured effect at the ionic double layer near the graphene surface saturates at 100Cfor 2 hon a hotplate.The stamps were inked by spin-coating 10 mM OTS with electric field,so comparisons can still be made between reac- in anhydrous toluene (3,000 r.p.m.,30 s).then gently brought into contact with the tions at different electric field strengths in each of these limits.In substrates for 60 s. Fig 5d,the resulting Ip/IG ratios are plotted as a function of gate hBN preparation.The hBN flakes used in this study were prepared by mechanical voltage during reaction on two different SiO,layer thicknesses exfoliation of an ultrapure single crystal of hBN on piranha-cleaned SiO./silicon (100 nm and 300 nm)to compare the effect of different electric substrates.The hBN was s grown using a method described previouslyss. field strengths.At positive gate voltage,reactivity is higher due to Diazonium functionalization of graphene.Graphene samples supported on the higher Fermi level,as expected,even though the diazonium con- substrates were immersed in aqueous solutions of 10 mM 4-NBD tetrafluoroborate centration at the graphene surface is lower.At negative gate voltage, and 0.5 wt%sodium dodecyl sulfate (SDS)with c constant stirring at ~35C.Most the increase in diazonium concentration dominates and causes a samples were reacted for 16.5 h to reach full reaction conversion (the sample in Fig.4 significantly increased reactivity.However,for the highest negative was reacted for 1.5 h to improve Ip/IG spatial contrast).After reaction,samples were rinsed in ultrapure water and blown dry with nitrogen.NMR and optical absorbance fields,occurring with the thinner dielectric layer,the diazonium spectroscopies were used to verify diazonium stability (Supplementary Fig.S6). concentration can be seen to saturate and the p-doping of the gra- phene begins to decrease the reactivity as expected.These exper- Raman spectroscopy and mapping.Raman spectroscopy was performed on a iments therefore support the electron-transfer rate model Horiba Jobin Yvon LabRAM HR800 system using a 633 nm excitation laser,x100 objective lens with ~1-um-diameter spot size and a motorized XYZ stage.The G. developed above.Additional exploration of electrochemical functiona- 2D and D peaks were fit to Lorentzian functions. lization will provide further insight into this reaction mechanism. Contact angle.The contact angles of the substrates were measured using a Rame- Conclusions Hart goniometer and 2 ul sessile droplets of ultrapure water.Several droplets were measured in different sample locations and the results were averaged. In summary,the effect of the underlying substrate on the chemical reactivity of graphene has been explored using detailed Raman spec- Atomic force microscopy.AFM imaging was conducted on an Asylum Research troscopy.Graphene on SiO,and Al,O,is more reactive towards MFP-3D system in a.c.(non-contact)mode using silicon probes(Olympus OMCL- covalent functionalization by aryl diazonium salts than graphene AC240TS).Images were processed using the Gwyddion software package on hBN or on an alkyl-terminated monolayer.The reactivity Binding of proteins on graphene.Graphene samples on OTS-patterned SiO contrast is attributed to higher amplitudes of the substrate-induced substrates were immersed in an aqueous solution of 1 wt%SDS and 50 uM 4- electron-hole charge fluctuations for graphene on SiO,and Al,O carboxybenzenediazonium tetrafluoroborate and stirred at 45C for 12 h.They were Micrometre-scale spatial control of the chemical reactivity of gra- then immersed in a phosphate buffered solution(pH 8.3)with 100 uM of (NTA- NH,)at room temperature for 8 h,followed by an aqueous solution of 20 uM NiCl, phene was demonstrated by chemically patterning the substrate at room temperature for 4 h to complex the Ni ions to the NTA structure.They before deposition of graphene.Owing to the versatility and chemical were then immersed in an aqueous solution of 1 uM polyhistidine (His)-tagged tailorability of the RIL technique,it can be used for the modification EGFP at room temperature for 1 h.Between each step above,the sample was rinsed and manipulation of graphene.This chemical patterning technique with water,acetone and isopropanol and blown dry with nitrogen.ATR-IR spectra was also applied to the spatial patterning of protein molecules were obtained using a Thermo Nicolet 4700 spectrometer.Confocal fluorescence on graphene, demonstrating the potential for applications microscopy images were captured using a Zeiss LSM 710 NLO with 633 nm laser excitation in biosensing. Received 30 January 2012;accepted 27 June 2012; Methods published online 12 August 2012 Graphene synthesis and transfer.Copper foil substrates(25 um,99.8%,Alfa Aesar) were annealed under a hydrogen atmosphere(1,000C,30 min,10 s.c.c.m. References hydrogen,~330 mtorr total pressure)followed by graphene synthesis with methane 1.Geim,A.K.Graphene:status and prospects.Science 324,1530-1534(2009). (1,000C,40 min,15 s.c.c.m.methane and 50 s.c.c.m.hydrogen,~1.5 torr total 2. Novoselov,K.S.et al.Electric field effect in atomically thin carbon films.Science pressure).Graphene on copper was coated in PMMA (950PMMA,A4,MicroChem) 306,666-669(2004). by spin-coating (3,000 r.p.m.,1 min),then dried in air (30 min).Graphene on the 3.Rao,C.N.R.,Sood,A.K.,Subrahmanyam,K.S.Govindaraj.A.Graphene: reverse side was removed by reactive ion etching(Plasmatherm RIE,100 W,7 mtorr the new two-dimensional nanomaterial.Angew.Chem.Int.Ed.48, oxygen,5 min).The PMMA-graphene-copper stack was placed on the surface of 7752-7777(2009). copper etchant(6 M HCl and 1 M CuCl,in water).After copper etching(~30 min). 4. Bekyarova,E.et al.Chemical modification of epitaxial graphene:spontaneous the PMMA-graphene layer was scooped out with a clean wafer and floated on grafting of aryl groups.J.Am.Chem.Soc.131,1336-1337(2009). several baths of ultrapure water for rinsing.It was then scooped out with the target 5.Elias,D.C.et al.Control of graphene's properties by reversible hydrogenation: substrate and dried in air overnight before immersion in several baths of clean evidence for graphane.Science 323,610-613(2009) NATURE CHEMISTRY ADVANCE ONLINE PUBLICATION|www.nature.com/naturechemistry 2012 Macmillan Publishers Limited.All rights reserved
causing the EF,avg calculated from equation (5) to be further from neutrality and requiring a larger shift in equation (6) to fit the ID/IG. The hydrophobicity of the substrate is an initial predictor of the chemical reactivity as shown in Fig. 1c because the surface energy of the substrate relates to the presence of charged impurities and polar surface groups that can induce electron–hole charge fluctuations in graphene. To test the implications of our substrate-dependent graphene reactivity results and model, we conducted electrochemical functionalization experiments where the graphene was electrically doped by an applied backgate voltage during reaction (see full details in the Supplementary Information, Pages 10–13 and Supplementary Fig. S8). Our model suggests that for sufficiently large shifts in the Fermi level, the contribution of the electron–hole puddles is overcome by the overall Fermi level, with overall reactivity decreasing for overall highly p-doped graphene and increasing for overall highly n-doped graphene. This approach is complicated by the diazonium cation being either attracted or repelled by the applied gate voltage so that the concentration of diazonium at the graphene surface is significantly increased or decreased. This concentration effect at the ionic double layer near the graphene surface saturates with electric field, so comparisons can still be made between reactions at different electric field strengths in each of these limits. In Fig. 5d, the resulting ID/IG ratios are plotted as a function of gate voltage during reaction on two different SiO2 layer thicknesses (100 nm and 300 nm) to compare the effect of different electric field strengths. At positive gate voltage, reactivity is higher due to the higher Fermi level, as expected, even though the diazonium concentration at the graphene surface is lower. At negative gate voltage, the increase in diazonium concentration dominates and causes a significantly increased reactivity. However, for the highest negative fields, occurring with the thinner dielectric layer, the diazonium concentration can be seen to saturate and the p-doping of the graphene begins to decrease the reactivity as expected. These experiments therefore support the electron-transfer rate model developed above. Additional exploration of electrochemical functionalization will provide further insight into this reaction mechanism. Conclusions In summary, the effect of the underlying substrate on the chemical reactivity of graphene has been explored using detailed Raman spectroscopy. Graphene on SiO2 and Al2O3 is more reactive towards covalent functionalization by aryl diazonium salts than graphene on hBN or on an alkyl-terminated monolayer. The reactivity contrast is attributed to higher amplitudes of the substrate-induced electron–hole charge fluctuations for graphene on SiO2 and Al2O3. Micrometre-scale spatial control of the chemical reactivity of graphene was demonstrated by chemically patterning the substrate before deposition of graphene. Owing to the versatility and chemical tailorability of the RIL technique, it can be used for the modification and manipulation of graphene. This chemical patterning technique was also applied to the spatial patterning of protein molecules on graphene, demonstrating the potential for applications in biosensing. Methods Graphene synthesis and transfer. Copper foil substrates (25 mm, 99.8%, Alfa Aesar) were annealed under a hydrogen atmosphere (1,000 8C, 30 min, 10 s.c.c.m. hydrogen, 330 mtorr total pressure) followed by graphene synthesis with methane (1,000 8C, 40 min, 15 s.c.c.m. methane and 50 s.c.c.m. hydrogen, 1.5 torr total pressure). Graphene on copper was coated in PMMA (950PMMA, A4, MicroChem) by spin-coating (3,000 r.p.m., 1 min), then dried in air (30 min). Graphene on the reverse side was removed by reactive ion etching (Plasmatherm RIE, 100 W, 7 mtorr oxygen, 5 min). The PMMA–graphene–copper stack was placed on the surface of copper etchant (6 M HCl and 1 M CuCl2 in water). After copper etching (30 min), the PMMA–graphene layer was scooped out with a clean wafer and floated on several baths of ultrapure water for rinsing. It was then scooped out with the target substrate and dried in air overnight before immersion in several baths of clean acetone to dissolve the PMMA, followed by rinsing in isopropanol and drying with nitrogen gas. Surface preparation of wafer substrates. Silicon wafers with 300 nm SiO2 were ultrasonically cleaned in sequential baths of acetone and isopropanol, blown dry with nitrogen, and subjected to additional surface treatments. Plasma-cleaned samples: these were exposed to oxygen plasma (AutoGlow Plasma System, Glow Research) for 10–30 min at 200 W power and 0.5 torr. Piranha-cleaned samples: these were immersed in piranha solution (3:1 solution of sulfuric acid and 30% hydrogen peroxide) for 15 min and rinsed in ultrapure water. (Warning: piranha solution is a strong oxidizing agent and should be handled with extreme care.) As-received samples: these were subjected to no additional treatment. Sapphire wafers (a-Al2O3, c-plane, 0.5 mm thick, MTI Corp.) were ultrasonically cleaned in acetone and isopropanol and blown dry with nitrogen. OTS monolayer on SiO2. Octadecyltrichlorosilane (OTS) (Sigma-Aldrich, 90þ%) SAMs were formed on freshly plasma-cleaned SiO2 substrates in OTS solution (10 mM in toluene) overnight in a closed vial, then rinsed in fresh toluene and blown dry with nitrogen. Surface patterning of substrates. OTS patterns were formed on freshly plasmacleaned SiO2 substrates by printing with polydimethylsiloxane (PDMS) stamps. Master patterns were formed by electron-beam lithography of PMMA resists on silicon wafers. PDMS (10:1 mass ratio of base to curing agent, Dow Corning Sylgard 184) was poured into the master patterns, degassed in vacuum for 45 min, and cured at 100 8C for 2 h on a hotplate. The stamps were inked by spin-coating 10 mM OTS in anhydrous toluene (3,000 r.p.m., 30 s), then gently brought into contact with the substrates for 60 s. hBN preparation. The hBN flakes used in this study were prepared by mechanical exfoliation of an ultrapure single crystal of hBN on piranha-cleaned SiO2/silicon substrates. The hBN crystal was grown using a method described previously55. Diazonium functionalization of graphene. Graphene samples supported on substrates were immersed in aqueous solutions of 10 mM 4-NBD tetrafluoroborate and 0.5 wt% sodium dodecyl sulfate (SDS) with constant stirring at 35 8C. Most samples were reacted for 16.5 h to reach full reaction conversion (the sample in Fig. 4 was reacted for 1.5 h to improve ID/IG spatial contrast). After reaction, samples were rinsed in ultrapure water and blown dry with nitrogen. NMR and optical absorbance spectroscopies were used to verify diazonium stability (Supplementary Fig. S6). Raman spectroscopy and mapping. Raman spectroscopy was performed on a Horiba Jobin Yvon LabRAM HR800 system using a 633 nm excitation laser, ×100 objective lens with 1-mm-diameter spot size and a motorized XYZ stage. The G, 2D and D peaks were fit to Lorentzian functions. Contact angle. The contact angles of the substrates were measured using a Rame´– Hart goniometer and 2 ml sessile droplets of ultrapure water. Several droplets were measured in different sample locations and the results were averaged. Atomic force microscopy. AFM imaging was conducted on an Asylum Research MFP-3D system in a.c. (non-contact) mode using silicon probes (Olympus OMCLAC240TS). Images were processed using the Gwyddion software package. Binding of proteins on graphene. Graphene samples on OTS-patterned SiO2 substrates were immersed in an aqueous solution of 1 wt% SDS and 50 mM 4- carboxybenzenediazonium tetrafluoroborate and stirred at 45 8C for 12 h. They were then immersed in a phosphate buffered solution (pH 8.3) with 100 mM of (NTA– NH2) at room temperature for 8 h, followed by an aqueous solution of 20 mM NiCl2 at room temperature for 4 h to complex the Ni2þ ions to the NTA structure. They were then immersed in an aqueous solution of 1 mM polyhistidine (His)-tagged EGFP at room temperature for 1 h. Between each step above, the sample was rinsed with water, acetone and isopropanol and blown dry with nitrogen. ATR-IR spectra were obtained using a Thermo Nicolet 4700 spectrometer. Confocal fluorescence microscopy images were captured using a Zeiss LSM 710 NLO with 633 nm laser excitation. Received 30 January 2012; accepted 27 June 2012; published online 12 August 2012 References 1. Geim, A. K. Graphene: status and prospects. Science 324, 1530–1534 (2009). 2. Novoselov, K. S. et al. Electric field effect in atomically thin carbon films. Science 306, 666–669 (2004). 3. Rao, C. N. R., Sood, A. K., Subrahmanyam, K. S. & Govindaraj, A. Graphene: the new two-dimensional nanomaterial. Angew. Chem. Int. Ed. 48, 7752–7777 (2009). 4. Bekyarova, E. et al. Chemical modification of epitaxial graphene: spontaneous grafting of aryl groups. J. Am. Chem. Soc. 131, 1336–1337 (2009). 5. Elias, D. C. et al. Control of graphene’s properties by reversible hydrogenation: evidence for graphane. Science 323, 610–613 (2009). ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 8 NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry © 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY DOI:10.1038/NCHEM.1421 ARTICLES 6.Wang,Q.H.Hersam,M.C.Nanofabrication of heteromolecular organic 37.Basko,D.M.,Piscanec,S.Ferrari,A.C.Electron-electron interactions nanostructures on epitaxial graphene via room temperature feedback-controlled and doping dependence of the two-phonon Raman intensity in graphene. lithography.Nano Lett.11,589-593 (2010). Ph5.ReB80,165413(2009). 7.Chen,W.Chen,SQi,D.C.Gao,X.Y.&Wee,A.T.S.Surface transfer p-type 38.Cancado,L.G.et al.Quantifying defects in graphene via Raman spectroscopy at doping of epitaxial graphene.J.Am.Chem.Soc.129,10418-10422 (2007). different excitation energies.Nano Lett.11,3190-3196(2011). 8.Wang.Q.H.Hersam,M.C.Room-temperature molecular-resolution 39.Ferrari,A.C.Robertson,J.Interpretation of Raman spectra of disordered and characterization of self-assembled organic monolayers on epitaxial graphene. amorphous carbon.Phys.Rev.B 61,014095 (2000). Nature Chem.1,206-211(2009). 40.Lucchese.M.M.et al.Quantifving ion-induced defects and Raman relaxation 9.Jin,Z.et al.Click chemistry on solution-dispersed graphene and monolayer length in graphene.Carbon 48,1592-1597 (2010). CVD graphene.Chem.Mater.23,3362-3370(2011). 41.Das,A.et al.Monitoring dopants by Raman scattering in an electrochemically 10.Niyogi,S.et al.Spectroscopy of covalently functionalized graphene.Nano Lett. top-gated graphene transistor.Nature Nanotech.3,210-215 (2008). 10,4061-4066(2010). 42.Lazzeri,M.Mauri,F.Nonadiabatic Kohn anomaly in a doped graphene 11.Sharma,R.,Baik,J.H.,Perera,C.J.Strano,M.S.Anomalously large reactivity monolayer.Phys.Rev.Lett.97,266407 (2006). of single graphene layers and edges toward electron transfer chemistries. 43.Casiraghi,C.Probing disorder and charged impurities in graphene by Raman Nano Let.10,398-405(2010). spectroscopy.Phys.Status Solidi RRL 3,175-177 (2009). 12.Hossain,M.Z,Walsh,M.A.Hersam,M.C.Scanning tunneling microscopy, 44.Mohiuddin,T.M.G.et al.Uniaxial strain in graphene by Raman spectroscopy: spectroscopy,and nanolithography of epitaxial graphene chemically modified G peak splitting.Gruneisen parameters,and sample orientation.Phys.Rev.B79, with aryl moieties.J.Am.Chem.Soc.132,15399-15403 (2010). 205433(2009). 13.Farmer,D.B.et al.Chemical doping and electron-hole conduction asymmetry 45.Yoon,D.et al.Interference effect on Raman spectrum of graphene on SiO,/Si. in graphene devices.Nano Lett.9,388-392 (2009). Ph5.Re.B80,125422(2009). 14.Lomeda,J.R.,Doyle,C.D.,Kosynkin,D.V.,Hwang.W.F.Tour,J.M. 46.Qin,D.,Xia,Y.Whitesides,G.M.Soft lithography for micro-and nanoscale Diazonium functionalization of surfactant-wrapped chemically converted patterning.Nature Protoc.5,491-502 (2010). graphene sheets.J.Am.Chem.Soc.130,16201-16206 (2008). 47.Xia,Y.,Mrksich,M.,Kim,E.Whitesides,G.M.Microcontact printing of 15.Sinitskii,A.et al.Kinetics of diazonium functionalization of chemically octadecylsiloxane on the surface of silicon dioxide and its application in converted graphene nanoribbons.ACS Nano 4,1949-1954 (2010). microfabrication.J.Am.Chem.Soc.117,9576-9577(1995). 16.Englert,J.M.et al.Covalent bulk functionalization of graphene.Nature Chem.3, 48.Kodali,V.K.et al.Nonperturbative chemical modification of graphene for 279-286(2011). protein micropatterning.Langmuir 27,863-865(2010). 17.Farmer,D.B..Lin,Y.M..Afzali-Ardakani,A.&Avouris,P.Behavior of a 49.Nair,N.,Kim,W.-J,Usrey,M.L Strano,M.S.A structure-reactivity chemically doped graphene junction.Appl Phys.Lett.94,213106 (2009). relationship for single walled carbon nanotubes reacting with 4-hydroxybenzene 18.Fan,X.Y.,Nouchi,R.,Yin,L.C.&Tanigaki,K.Effects of electron-transfer diazonium salt.J.Am.Chem.Soc.129,3946-3954(2007) chemical modification on the electrical characteristics of graphene. 50.Mahouche-Chergui,S.,Gam-Derouich,S.,Mangeney,C.Chehimi,M.M. Nanotechnology 21,475208 (2010). Aryl diazonium salts:a new class of coupling agents for bonding polymers, 19.Zhang.H.et al.Aryl functionalization as a route to band gap engineering in biomacromolecules and nanoparticles to surfaces.Chem.Soc.Rev.40, 4143-4166(2011). single layer graphene devices.Nano Lett.11,4047-4051 (2011). 20.Chen,J.H.et al.Charged-impurity scattering in graphene.Nature Phys.4, 51.Galli,C.Radical reactions of arenediazonium ions:an easy entry into the chemistry of the aryl radical.Chem.Rev.88,765-792 (1988). 377-381(2008). 52.Bard,A.J.Faulkner,L.R.Electrochemical Methods:Fundamentals and 21.Zhang.Y..Brar,V.W,Girit,C.,Zettl,A.Crommie,M.F.Origin of spatial charge inhomogeneity in graphene.Nature Phys.5,722-726(2009). Applications 2nd edn (Wiley,2001). 22.Du,X.Skachko,I.,Barker,A.Andrei,E.Y.Approaching ballistic transport in 53.Sharma,R.,Nair,N.Strano,M.S.Structure-reactivity relationships for suspended graphene.Nature Nanotech.3,491-495 (2008). graphene nanoribbons.J.Phys.Chem.C113,14771-14777 (2009). 54.Casiraghi,C.Doping dependence of the Raman peaks intensity of graphene 23.Bolotin,K.I.et al.Ultrahigh electron mobility in suspended graphene.Solid close to the Dirac point.Phys.Rev.B 80,233407(2009). State Commun..146,351-355(2008) 55.Taniguchi,T.Watanabe,K.Synthesis of high-purity boron nitride single 24.Dean,C.R.et al.Boron nitride substrates for high-quality graphene electronics. crystals under high pressure by using Ba-BN solvent.I.Cryst.Growth 303, Nature Nanotech.5,722-726 (2010). 525-529(2007). 25.Xue,J.M.X.I.M.et al.Scanning tunnelling microscopy and spectroscopy 56.Thomsen,C.Reich,S.Double resonant Raman scattering in graphite.Phys of ultra-flat graphene on hexagonal boron nitride.Nature Mater.10, Re.Let.85,5214-5217(2000). 282-285(2011). 57.Royant,A.Noirclerc-Savoye,M.Stabilizing role of glutamic acid 222 in 26.Lafkioti,M.et al.Graphene on a hydrophobic substrate:doping reduction the structure of enhanced green fluorescent protein.J.Struct.Biol.174, and hysteresis suppression under ambient conditions.Nano Left.10, 385-390(2011). 1149-1153(2010). 27.Liu,Z.H.,Bol,A.A.Haensch,W.Large-scale graphene transistors with Acknowledgements enhanced performance and reliability based on interface engineering by This work was primarily funded by a 2009 US Office of Naval Research Multi University phenylsilane self-assembled monolayers.Nano Lett.11,523-528(2011) Research Initiative(MURI)grant on Graphene Advanced Terahertz Engineering (GATE) 28.Yokota,K.,Takai,K.Enoki,T.Carrier control of graphene driven by the at MIT,Harvard and Boston University.J.D.S.-Y.and P.J.-H.acknowledge support from an proximity effect of functionalized self-assembled monolayers.Nano Lett.11 NSF CAREER award(DMR-0845287).K.K.K.acknowledges an NSF award (DMR- 3669-3675(2011. 0845358)and support from the Materials,Structures and Device(MSD)Center of the 29.Yan,Z.et al.Controlled modulation of electronic properties of graphene by self- Focus Center Research Program(FCRP)at the Semiconductor Research Corporation assembled monolayers on SiO,substrates.ACS Nano 5,1535-1540(2011). The authors thank M.K.Mondol of the MIT Scanning Electron Beam Lithography 30.Fan,X,Nouchi,R.Tanigaki,K.Effect of charge puddles and ripples on the facility for assistance. chemical reactivity of single layer graphene supported by SiO,/Si substrate. 1.Ph5.Chm.C115,12960-12964(2011) Author contributions 31.Li,X.et al.Large-area synthesis of high-quality and uniform graphene films on Q.H.W.designed and conducted the substrate and patterning experiments,performed copper foils.Science 324,1312-1314 (2009). Raman spectroscopy.AFM and data analysis.ZJ.performed protein attachment,ATR-IR 32.Reina,A.et al.Large area,few-layer graphene films on arbitrary substrates by and fluorescence imaging.A.J.H,Q.H.W.and M.S.S.devised the model.K.K.K. chemical vapor deposition.Nano Lett.9,30-5(2008). synthesized the CVD graphene.K.W.and T.T.synthesized the hBN crystal.J.D.S.-Y 33.Huang.P.Y.et al.Grains and grain boundaries in single-layer graphene atomic exfoliated the hBN crystal.G.L.C.P..C.-I.S.and M.-H.H.conducted additional patchwork quilts.Nature 469,389-392(2011). experiments.Q.H.W.and M.S.S.wrote the manuscript.All authors contributed to the 34.Kim,K.et al.Grain boundary mapping in polycrystalline graphene.ACS Nano 5, discussion and interpretation of results. 2142-2146(2011. 35.Ferrari,A.C.et al.Raman spectrum of graphene and graphene layers.Phys.Rev. Additional information Lett.97,187401(2006). The authors declare no competing financial interests.Supplementary information 36.Ferrari,A.C.Raman spectroscopy of graphene and graphite:disorder,electron- accompanies this paper at www.nature.com/naturechemistry.Reprints and permission phonon coupling.doping and nonadiabatic effects.Solid State Commun.143, information is available online at http://www.nature.com/reprints.Correspondence and 47-57(20070. requests for materials should be addressed to M.S.S. NATURE CHEMISTRY|ADVANCE ONLINE PUBLICATION www.nature.com/naturechemistry 2012 Macmillan Publishers Limited All riahts reserved
6. Wang, Q. H. & Hersam, M. C. Nanofabrication of heteromolecular organic nanostructures on epitaxial graphene via room temperature feedback-controlled lithography. Nano Lett. 11, 589–593 (2010). 7. Chen, W., Chen, S., Qi, D. C., Gao, X. Y. & Wee, A. T. S. Surface transfer p-type doping of epitaxial graphene. J. Am. Chem. Soc. 129, 10418–10422 (2007). 8. Wang, Q. H. & Hersam, M. C. Room-temperature molecular-resolution characterization of self-assembled organic monolayers on epitaxial graphene. Nature Chem. 1, 206–211 (2009). 9. Jin, Z. et al. Click chemistry on solution-dispersed graphene and monolayer CVD graphene. Chem. Mater. 23, 3362–3370 (2011). 10. Niyogi, S. et al. Spectroscopy of covalently functionalized graphene. Nano Lett. 10, 4061–4066 (2010). 11. Sharma, R., Baik, J. H., Perera, C. J. & Strano, M. S. Anomalously large reactivity of single graphene layers and edges toward electron transfer chemistries. Nano Lett. 10, 398–405 (2010). 12. Hossain, M. Z., Walsh, M. A. & Hersam, M. C. Scanning tunneling microscopy, spectroscopy, and nanolithography of epitaxial graphene chemically modified with aryl moieties. J. Am. Chem. Soc. 132, 15399–15403 (2010). 13. Farmer, D. B. et al. Chemical doping and electron–hole conduction asymmetry in graphene devices. Nano Lett. 9, 388–392 (2009). 14. Lomeda, J. R., Doyle, C. D., Kosynkin, D. V., Hwang, W. F. & Tour, J. M. Diazonium functionalization of surfactant-wrapped chemically converted graphene sheets. J. Am. Chem. Soc. 130, 16201–16206 (2008). 15. Sinitskii, A. et al. Kinetics of diazonium functionalization of chemically converted graphene nanoribbons. ACS Nano 4, 1949–1954 (2010). 16. Englert, J. M. et al. Covalent bulk functionalization of graphene. Nature Chem. 3, 279–286 (2011). 17. Farmer, D. B., Lin, Y. M., Afzali-Ardakani, A. & Avouris, P. Behavior of a chemically doped graphene junction. Appl. Phys. Lett. 94, 213106 (2009). 18. Fan, X. Y., Nouchi, R., Yin, L. C. & Tanigaki, K. Effects of electron-transfer chemical modification on the electrical characteristics of graphene. Nanotechnology 21, 475208 (2010). 19. Zhang, H. et al. Aryl functionalization as a route to band gap engineering in single layer graphene devices. Nano Lett. 11, 4047–4051 (2011). 20. Chen, J. H. et al. Charged-impurity scattering in graphene. Nature Phys. 4, 377–381 (2008). 21. Zhang, Y., Brar, V. W., Girit, C., Zettl, A. & Crommie, M. F. Origin of spatial charge inhomogeneity in graphene. Nature Phys. 5, 722–726 (2009). 22. Du, X., Skachko, I., Barker, A. & Andrei, E. Y. Approaching ballistic transport in suspended graphene. Nature Nanotech. 3, 491–495 (2008). 23. Bolotin, K. I. et al. Ultrahigh electron mobility in suspended graphene. Solid State Commun. 146, 351–355 (2008). 24. Dean, C. R. et al. Boron nitride substrates for high-quality graphene electronics. Nature Nanotech. 5, 722–726 (2010). 25. Xue, J. M. X. J. M. et al. Scanning tunnelling microscopy and spectroscopy of ultra-flat graphene on hexagonal boron nitride. Nature Mater. 10, 282–285 (2011). 26. Lafkioti, M. et al. Graphene on a hydrophobic substrate: doping reduction and hysteresis suppression under ambient conditions. Nano Lett. 10, 1149–1153 (2010). 27. Liu, Z. H., Bol, A. A. & Haensch, W. Large-scale graphene transistors with enhanced performance and reliability based on interface engineering by phenylsilane self-assembled monolayers. Nano Lett. 11, 523–528 (2011). 28. Yokota, K., Takai, K. & Enoki, T. Carrier control of graphene driven by the proximity effect of functionalized self-assembled monolayers. Nano Lett. 11, 3669–3675 (2011). 29. Yan, Z. et al. Controlled modulation of electronic properties of graphene by selfassembled monolayers on SiO2 substrates. ACS Nano 5, 1535–1540 (2011). 30. Fan, X., Nouchi, R. & Tanigaki, K. Effect of charge puddles and ripples on the chemical reactivity of single layer graphene supported by SiO2/Si substrate. J. Phys. Chem. C 115, 12960–12964 (2011). 31. Li, X. et al. Large-area synthesis of high-quality and uniform graphene films on copper foils. Science 324, 1312–1314 (2009). 32. Reina, A. et al. Large area, few-layer graphene films on arbitrary substrates by chemical vapor deposition. Nano Lett. 9, 30–35 (2008). 33. Huang, P. Y. et al. Grains and grain boundaries in single-layer graphene atomic patchwork quilts. Nature 469, 389–392 (2011). 34. Kim, K. et al. Grain boundary mapping in polycrystalline graphene. ACS Nano 5, 2142–2146 (2011). 35. Ferrari, A. C. et al. Raman spectrum of graphene and graphene layers. Phys. Rev. Lett. 97, 187401 (2006). 36. Ferrari, A. C. Raman spectroscopy of graphene and graphite: disorder, electron– phonon coupling, doping and nonadiabatic effects. Solid State Commun. 143, 47–57 (2007). 37. Basko, D. M., Piscanec, S. & Ferrari, A. C. Electron–electron interactions and doping dependence of the two-phonon Raman intensity in graphene. Phys. Rev. B 80, 165413 (2009). 38. Canc¸ado, L. G. et al. Quantifying defects in graphene via Raman spectroscopy at different excitation energies. Nano Lett. 11, 3190–3196 (2011). 39. Ferrari, A. C. & Robertson, J. Interpretation of Raman spectra of disordered and amorphous carbon. Phys. Rev. B 61, 014095 (2000). 40. Lucchese, M. M. et al. Quantifying ion-induced defects and Raman relaxation length in graphene. Carbon 48, 1592–1597 (2010). 41. Das, A. et al. Monitoring dopants by Raman scattering in an electrochemically top-gated graphene transistor. Nature Nanotech. 3, 210–215 (2008). 42. Lazzeri, M. & Mauri, F. Nonadiabatic Kohn anomaly in a doped graphene monolayer. Phys. Rev. Lett. 97, 266407 (2006). 43. Casiraghi, C. Probing disorder and charged impurities in graphene by Raman spectroscopy. Phys. Status Solidi RRL 3, 175–177 (2009). 44. Mohiuddin, T. M. G. et al. Uniaxial strain in graphene by Raman spectroscopy: G peak splitting, Gruneisen parameters, and sample orientation. Phys. Rev. B 79, 205433 (2009). 45. Yoon, D. et al. Interference effect on Raman spectrum of graphene on SiO2/Si. Phys. Rev. B 80, 125422 (2009). 46. Qin, D., Xia, Y. & Whitesides, G. M. Soft lithography for micro- and nanoscale patterning. Nature Protoc. 5, 491–502 (2010). 47. Xia, Y., Mrksich, M., Kim, E. & Whitesides, G. M. Microcontact printing of octadecylsiloxane on the surface of silicon dioxide and its application in microfabrication. J. Am. Chem. Soc. 117, 9576–9577 (1995). 48. Kodali, V. K. et al. Nonperturbative chemical modification of graphene for protein micropatterning. Langmuir 27, 863–865 (2010). 49. Nair, N., Kim, W.-J., Usrey, M. L. & Strano, M. S. A structure–reactivity relationship for single walled carbon nanotubes reacting with 4-hydroxybenzene diazonium salt. J. Am. Chem. Soc. 129, 3946–3954 (2007). 50. Mahouche-Chergui, S., Gam-Derouich, S., Mangeney, C. & Chehimi, M. M. Aryl diazonium salts: a new class of coupling agents for bonding polymers, biomacromolecules and nanoparticles to surfaces. Chem. Soc. Rev. 40, 4143–4166 (2011). 51. Galli, C. Radical reactions of arenediazonium ions: an easy entry into the chemistry of the aryl radical. Chem. Rev. 88, 765–792 (1988). 52. Bard, A. J. & Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications 2nd edn (Wiley, 2001). 53. Sharma, R., Nair, N. & Strano, M. S. Structure–reactivity relationships for graphene nanoribbons. J. Phys. Chem. C 113, 14771–14777 (2009). 54. Casiraghi, C. Doping dependence of the Raman peaks intensity of graphene close to the Dirac point. Phys. Rev. B 80, 233407 (2009). 55. Taniguchi, T. & Watanabe, K. Synthesis of high-purity boron nitride single crystals under high pressure by using Ba–BN solvent. J. Cryst. Growth 303, 525–529 (2007). 56. Thomsen, C. & Reich, S. Double resonant Raman scattering in graphite. Phys. Rev. Lett. 85, 5214–5217 (2000). 57. Royant, A. & Noirclerc-Savoye, M. Stabilizing role of glutamic acid 222 in the structure of enhanced green fluorescent protein. J. Struct. Biol. 174, 385–390 (2011). Acknowledgements This work was primarily funded by a 2009 US Office of Naval Research Multi University Research Initiative (MURI) grant on Graphene Advanced Terahertz Engineering (GATE) at MIT, Harvard and Boston University. J.D.S.-Y. and P.J.-H. acknowledge support from an NSF CAREER award (DMR-0845287). K.K.K. acknowledges an NSF award (DMR- 0845358) and support from the Materials, Structures and Device (MSD) Center of the Focus Center Research Program (FCRP) at the Semiconductor Research Corporation. The authors thank M.K. Mondol of the MIT Scanning Electron Beam Lithography facility for assistance. Author contributions Q.H.W. designed and conducted the substrate and patterning experiments, performed Raman spectroscopy, AFM and data analysis. Z.J. performed protein attachment, ATR–IR and fluorescence imaging. A.J.H., Q.H.W. and M.S.S. devised the model. K.K.K. synthesized the CVD graphene. K.W. and T.T. synthesized the hBN crystal. J.D.S.-Y. exfoliated the hBN crystal. G.L.C.P., C.-J.S. and M.-H.H. conducted additional experiments. Q.H.W. and M.S.S. wrote the manuscript. All authors contributed to the discussion and interpretation of results. Additional information The authors declare no competing financial interests. Supplementary information accompanies this paper at www.nature.com/naturechemistry. Reprints and permission information is available online at http://www.nature.com/reprints. Correspondence and requests for materials should be addressed to M.S.S. NATURE CHEMISTRY DOI: 10.1038/NCHEM.1421 ARTICLES NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry 9 © 2012 Macmillan Publishers Limited. All rights reserved