Near room-temperature synthesis of transfer-free graphene films

Abstract

Large-area graphene films are best synthesized via chemical vapour and/or solid deposition methods at elevated temperatures (~1,000 °C) on polycrystalline metal surfaces and later transferred onto other substrates for device applications. Here we report a new method for the synthesis of graphene films directly on SiO(2)/Si substrates, even plastics and glass at close to room temperature (25-160 °C). In contrast to other approaches, where graphene is deposited on top of a metal substrate, our method invokes diffusion of carbon through a diffusion couple made up of carbon-nickel/substrate to form graphene underneath the nickel film at the nickel-substrate interface. The resulting graphene layers exhibit tunable structural and optoelectronic properties by nickel grain boundary engineering and show micrometre-sized grains on SiO(2) surfaces and nanometre-sized grains on plastic and glass surfaces. The ability to synthesize graphene directly on non-conducting substrates at low temperatures opens up new possibilities for the fabrication of multiple nanoelectronic devices.

ARTICLE

Received 31 Aug 2011 | Accepted 15 Dec 2011 | Published 24 Jan 2012 DOI: 10.1038/ncomms1650

Jinsung Kwak1, Jae Hwan Chu1, Jae-Kyung Choi1, Soon-Dong Park1, Heungseok Go2, Sung Youb Kim1,

Kibog Park2,3, Sung-Dae Kim4, Young-Woon Kim4, Euijoon Yoon4,5,6, Suneel Kodambaka7 & Soon-Yong Kwon1,2,3

Large-area graphene lms are best synthesized via chemical vapour and/or solid deposition methods at elevated temperatures (~1,000 C) on polycrystalline metal surfaces and later transferred onto other substrates for device applications. Here we report a new method for the synthesis of graphene lms directly on SiO2/Si substrates, even plastics and glass at close to room temperature (25160 C). In contrast to other approaches, where graphene is deposited on top of a metal substrate, our method invokes diffusion of carbon through a diffusion couple made up of carbon-nickel/substrate to form graphene underneath the nickel lm at the nickel substrate interface. The resulting graphene layers exhibit tunable structural and optoelectronic properties by nickel grain boundary engineering and show micrometre-sized grains on SiO2

surfaces and nanometre-sized grains on plastic and glass surfaces. The ability to synthesize graphene directly on non-conducting substrates at low temperatures opens up new possibilities for the fabrication of multiple nanoelectronic devices.

Near room-temperature synthesis of transfer-free graphene lms

1 School of Mechanical and Advanced Materials Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea.

2 School of Electrical and Computer Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea. 3 Opto-Electronics Convergence Group, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea. 4 Department of Materials Science and Engineering, Seoul National University, Seoul 151-742, Republic of Korea. 5 Energy Semiconductor Research Center, Advanced Institutes of Convergence Technology, Seoul National University, Suwon 443-270, Korea. 6 Department of Nano Science and Technology, Graduate School of Convergence Science and Technology, Seoul National University, Suwon 443-270, Korea. 7 Department of Materials Science and Engineering, University of California Los Angeles, Los Angeles, California 90095, USA. Correspondence and requests for materials should be addressed to S.-Y.K. (email: [email protected]).

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

Graphene has generated enormous scientic curiosity owing to its ultrathin geometry1,2, quantum electronic transport2,3, bandgap opening4,5, excellent thermal conductivity6 and

high mechanical strength7. Among all of the potential applications for graphene, realization of graphene-based exible electronic display technology appears to be closer to reality8. Currently, large-area graphene lms are best synthesized via pyrolytic cracking of hydro-carbon gases811 and/or a solid deposition method12 at elevated temperatures (~1,000 C) on polycrystalline metal surfaces and later transferred onto other substrates for the fabrication of devices. However, the high growth temperatures impose limitations on the choice of substrates and the subsequent transferability of as-grown layers inuence device performance. Recently, bilayer graphene has been grown directly on insulating substrates at elevated temperatures (~1,000 C) using spin-coated polymer lms as solid source of carbon (C) with nickel (Ni) thin lms on top or bottom of the polymer lm13,14.

Here we describe a very low-temperature and transfer-free approach to controllably deposit graphene lms onto desired substrates. Our synthesis methodology exploits the properties of a diffusion couple, wherein a Ni thin lm is deposited rst on the substrate, and solid C is then deposited on top of the Ni, and allowed

to diuse predominantly along grain boundaries (GBs) to create a thin graphene lm at the Nisubstrate interface. To tune the structural and optoelectronic properties of the resulting graphene layers, we have engineered grain sizes of the Ni lms on the substrate of choice. Our method allows for uniform and controllable deposition of wrinkle-free graphene lms with micrometre-sized grains on SiO2 surfaces, and with nanometre-sized grains on plastic and glass.

This study suggests that large-scale, device-ready graphene layers can be simply prepared without a transfer process on any arbitrary substrate at low temperatureshighly desirable for electronic and optoelectronic applications.

ResultsDiusion-assisted synthesis (DAS) method. Figure 1a

schematically illustrates our process, which we refer to as DAS. We demonstrate the applicability of this method using thermally oxidized Si(100), poly(methyl methacrylate) (PMMA)-covered SiO2, and commercially available glass as substrates. First, ~100-nm-thick lm of polycrystalline Ni (poly-Ni) is deposited via electron-beam evaporation at temperatures as low as ~25 C on PMMA and glass and at temperatures as high as ~400 C on SiO2/Si(100). Ni thin lms deposited on SiO2/Si(100) were annealed at ~1,000 C

a

Ni

Si

SiO2 Graphite powder

1,200

Graphene

b

c

d

1.3 a.u.

0.2 a.u.

Intensity (a.u.)

1,600 2,000

Raman shift (cm1)

2,400 2,800

e

f g 6.0

5.5

Au/Cr

I DS(106 A)

5.0

4.5

0 20 40

VG (V)

60 80 100

Figure 1 | Structural and electrical properties of the DAS-grown graphene lms on SiO2/Si substrates at low temperatures. (a) Schematic drawing of the DAS process for directly depositing graphene lms on nonconducting substrates. The diagrams represent (from left to right) the elementary steps in the DAS process, including deposition (and annealing) of Ni thin lms on desired substrates (SiO2/Si or PMMA, glass), preparation of diffusion couple of

CNi/substrate, annealing in Ar or air (25260 C) to form CNi/graphene/substrate and formation of graphene on desired substrates by etching away CNi diffusion couple, respectively. Representative (b) optical microscopy (OM) image, (c) Raman spectra from red, blue and green spots showing the presence of one, two and three layers of graphene, respectively, (from bottom to top) and (d) Raman map image of the G/2D bands of graphene grown at temperature T=160 C for 5 min on SiO2(300 nm)/Si substrate. Scale bars, 4 m (b and d). (e) Representative SEM image of graphene grown at T=160 C for 5 min on

SiO2(300 nm)/Si substrate (scale bar, 100 m). (f) Higher magnication SEM image of e, showing the presence of 1 ML (red dot), 2 ML (blue dot) graphene and multilayer graphene ridges (white dot); (scale bar, 10 m). (g) Typical room temperature IDS-VG curve at VDS = 10 mV from a DAS-grown ML graphene-based back-gated FET device. The estimated carrier (hole) mobility is ~667 cm2 V 1s 1 at room temperature and it shows weak p-type behaviour. The inset shows optical microscopy image of this device and the scale bar is 10 m. IDS, drain-source current; VDS, drain-source voltage; VG, gate voltage.

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

ARTICLE

a

b

in a H2 ambient to yielding a strong 111-texture and 520 m sized grains with smooth surfaces. Using Raman spectroscopy, we conrm that the surfaces of Ni (aer annealing) and SiO2 (aer removal of the metal) are free of crystalline C. In case of Ni thin lms deposited at room temperature, the average grain size is about 4050 nm. Details of the deposition, morphology and crystallinity of the Ni lms are presented in Methods and Supplementary Figures S1 and S2. The choice of Ni is motivated by the facts that it has a high solubility for C and Ni(111) is nearly lattice-matched with basal planes of graphite, both of which facilitate commensurate epitaxial growth of graphene15,16. Solid C is supplied from a paste composed of graphite powder (Aldrich, product 496596) dispersed in ethanol. (Highly oriented pyrolytic graphite can also be used as C source but we obtained better results with the paste.) To ensure efficient contact at the CNi interface and stimulate the C-C bond breaking on catalytic Ni(111) surface17,18, a pressure ( < 1 MPa) is uniformly applied by mechanically clamping the CNi/substrate diusion couple using a molybdenum holding stage. The samples are then sealed in a quartz tube lled with inert argon gas (or with air) and heated for short periods (t = 110 min) at temperatures T between 25 and 260 C. Substrate temperatures are measured using a k-type thermocouple directly connected to the sample holder and are accurate to within 10 C. Following graphene growth, the C-Ni diusion couple is etched away using an aqueous solution of FeCl3

leaving behind a graphene lm on the desired substrate. The details of this process are presented in Methods.

Synthesis and characterization of graphene lms on SiO2Si substrates. Figure 1bf shows typical results of surface morphology and

Raman spectra, respectively of the graphene layers on SiO2/Si(100) obtained at T = 160 C and t = 5 min. The three Raman spectra,

colour coded for clarity, in Figure 1c are obtained from the three corresponding coloured circles. They show three primary features: a D band at ~1351 cm 1, a G band at ~1592 cm 1 and a 2D band at ~2685 cm 1, all expected peak positions for graphene. Film thicknesses are determined by measuring the ratio of G-to-2D peak intensities (IG/I2D) and the full-width at half-maximum (FWHM) values of the symmetric 2D band. For the spectra of monolayer (ML)-area graphene, we obtain (1) IG/I2D < 0.5 and (2) the FWHM value of ~38 cm 1 for the 2D band, consistent with previous reports9,11. The associated G/2D band map in Figure 1d illustrates the uniformity of graphene lms over large areas (~320 m2) covered mostly with 1ML and 2ML graphene as identied by the IG/I2D < 0.5 and

IG/I2D1, respectively. There are, however, a few regions with multilayer graphene (for example, the region highlighted by the white dot in Fig. 1f) and we refer to them as graphene ridges. From optical and scanning electron microscopy (OM and SEM) images acquired from the samples, we note that the as-synthesized graphene lms are wrinkle-free and smooth over large areas. Interestingly, we nd that the morphologies of regions covered with mono- and bi-layer graphene resemble those of the grains, and the multilayer graphene ridges, the GBs in the Ni thin lms. These observations suggest that multilayer graphene growth is favoured at defects such as GBs. This is plausible as GB diusion is typically faster than bulk diusion and GBs can serve as nucleation sites19,20.

The electrical properties of graphene layers on SiO2/Si(100) obtained at T = 160 C and t = 5 min have been evaluated with back-gated graphene-based eld-eect transistor (FET) devices and typical data for the FET devices are shown in Figure 1g and Supplementary Figure S3a. The V shape of the ambipolar transfer characteristics and the shi of neutrality point to positive gate voltage, that is, weak p-type behaviour1,12,21, in ambient conditions are observed, all expected electrical properties for graphene-based FETs. The estimated carrier (hole) mobility of this particular device in ambient conditions is ~667 cm2 V 1 s 1 at room temperature, suggesting that the as-synthesized graphene lms are of reasonable quality.

Using our approach, we have synthesized graphene layers over a range of temperatures between 25 and 260 C. OM images and Raman spectroscopic measurements of G-to-2D intensity ratio (IG/I2D)

and FWHM of G bands9 (Fig. 2a,b and Supplementary Fig. S4) indicate that few-layer graphene lms with similar quality can be obtained at all T, even at room temperature (Fig. 2a). However, the surface coverage of graphene on SiO2 shows a strong dependence on

T, increasing linearly from ~60 to ~98% with increasing T from 25 to 260 C as shown in Figure 2b. (The corresponding OM images of graphene on SiO2 grown at dierent T are shown in Supplementary

Fig. S4.) We note that continuous graphene layers over large areas can only be obtained at T160 C. The sheet resistance of graphene layers grown at T = 160 C, measured using transmission line model method22,23, is found to be ~1,000 per square (Supplementary Fig. S3b), which makes the as-synthesized graphene promising as a transparent electrode material. We have also explored the possibility of using our approach to grow graphene in air instead of inert Ar atmospheres. Surprisingly, we nd that the surface morphology, areal coverage and Raman structure (Fig. 2c) of the graphene lms grown in Ar as well as in air are similar, demonstrating the potential of our approach to grow graphene in air at low temperatures.

Intensity (a.u.)

G FWHM (cm1 )

100

80

60

40

15

10

5

0 00 50

Growth temperature (C)

100

20

Coverage (%)

1,200

1,600

2,000

Raman shift (cm1)

30

25

20

2,400

2,800

150 200 250

c

d

Intensity (a.u.)

1,200

1,600

2,000

Raman shift (cm1)

2,400

2,800

Figure 2 | Structural and optical characterization of DAS-grown graphene lms versus growth temperature. (a) Raman spectra showing the presence of one (red), two (blue) and three (green) layers (from bottom to top) of graphene lm grown at room temperature for 10 min on SiO2(300 nm)/Si substrate. (b) FWHM of G bands in Raman spectra (black squares) and the surface coverage (red circles) of graphene lms grownfor 10 min on SiO2 as a function of growth temperature T. For comparison, the FWHM of G band for a highly oriented pyrolytic graphite substrateis 15 cm 1. The insets show the optical microscopy images of few-layer graphene lms grown at room temperature (left; scale bar, 20 m) and

at 205 C (right) (scale bar, 100 m) on SiO2(300 nm)/Si substrates. (c) Comparison of the Raman spectra of 1 ML-area graphene lms grown at

T=25 C (black), 60 C (red) and 160 C (green) in argon and at T=25 C (blue), 160 C (purple) in air (from bottom to top). (d) Representative lowmagnication plan-view TEM image of graphene lm grown at T=160 C on SiO2 then transferred onto a TEM support grid (scale bar, 100 nm). Inset is a selected area diffraction pattern obtained from a wrinkle-free region highlighted by the dotted circle. Scale bar in inset is 5 1/nm.

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

a

C

Ni

GB

GB

SiO2/Si template

SiO2/Si template

b

C

Ni

Figure 3 | Schematic diagrams of graphene growth mechanisms in DAS process. The diagrams represent (from left to right) the elementary steps of the process including dissociation of CC bonds at C/Ni interface, diffusion of carbon atoms, followed by nucleation and growth of graphene. Graphene growth mechanism in case (a) 460 C T 600 C; bulk diffusion of C atoms through Ni crystallites, leading to homogeneous nucleation at multiple sites resulting in the formation of nanocrystalline graphene lm by precipitation and (b) T 260 C; preferential diffusion of C atoms via GBs in Ni, followed by heterogeneous nucleation at the defect sites and growth via lateral diffusion of C atoms along Ni/substrate interface.

Growth mechanism of graphene lms in DAS process. So, how does graphene form at such low temperatures? We propose a mechanism for the growth of graphene layers in DAS as follows: to form graphene at the NiSiO2 interface, C from the solid source will have to diuse through the Ni lm and crystallize as graphene at the interface. Given that our temperatures are well below 260 C, we rule out the possibility of graphene growth via C precipitation from the bulk that requires temperatures 460 C or higher24. (In case of 460 C T 600 C, we observed the formation of nanocrystalline graphene layers and the layers contain no graphene ridges at all, as shown in Supplementary Fig. S5.) Instead, we suggest the following: C atoms from solid C source have to diuse through the Ni lm. Previous experimental17,24 and theoretical17,18 studies indicate that Ni surface can catalyze the dissociation of C-C bonds and promote diusion even at room temperature. The resulting C atoms are transported across the Ni lm primarily along the GBs to the NiSiO2 interface19,20,25. Following Fisher19 and Balluffi and Blakely20, we estimate the ratio of two transport uxes of

C atoms through the Ni lattice (L) and along the GBs (GB) in 111-textured Ni lms and it is evident that lower temperatures greatly favour GB diusion relative to lattice diusion, that is, GBL

(details of the calculations are presented in Methods). It is noted that the C atom transport along GBs in Ni lms is highly susceptible to residual impurity concentrations such as hydrogen (Supplementary Fig. S6). Upon reaching the NiSiO2 interface, C atoms precipitate out as graphene at the GBs. This is consistent with the fact

that for relatively short growth times (t13 min), independent of T (25260 C), we obtained graphene-free surfaces with traces of graphene ridges presumably along the GBs in Ni lms. We veried the ridge composition to be graphene and not amorphous C using Raman spectroscopy and cross-sectional transmission electron microscopy (TEM) (Supplementary Fig. S7). Excess C atoms reaching the graphene ridges, diuse laterally along the graphene-Ni(111) interface and lead to the growth of graphene over large areas, driven by the strong affinity of C atoms to self-assemble and expand the sp2 lattice26. To understand the origin of the interfacial processes, we have performed density functional theory calculations on the transport of C atoms along the graphene-Ni(111) interface (details of the calculations are presented in Methods). We obtain an energy barrier of ~0.51 eV for diusion of C atoms at the graph-eneNi(111) interface in agreement with earlier reports27,28, which indicates that the tracer diusion coefficient (Dt) of C atoms is

~200 nm2 s 1 at room temperature, according to the Arrhenius equation.

The suggested graphene growth mechanism in DAS process is schematically illustrated in Fig. 3. Eventually, a continuous but poly-crystalline graphene lm forms at the NiSiO2 interface. Using a combination of TEM techniques including selected area diraction pattern and dark-eld TEM, we nd that the grain sizes in graphene layers vary from a few hundred nanometres to a few micrometres (Fig. 2d and Supplementary Fig. S8), and are comparable to those of Ni grains. Our results imply that we have more room for improvement in graphene quality by Ni GB engineering, that is, micrometre-scale or larger grain sizes in graphene layers can be obtained by creating macrocrystalline Ni lms with GBs located far enough from each other and/or by wet chemical doping8,29 of the lms with high controllability.

Large-scale synthesis of graphene lms on plastic and glass substrates. We demonstrate the applicability of our approach to prepare large-area graphene on non-conducting, plastic and glass substrates. To this purpose, we use T160 C and do not anneal the Ni thin lms so as to minimize thermal degradation of the substrates. Figure 4a shows typical surface morphology of graphene layers grown at T = 60 C and t = 10 min on PMMA then transferred to SiO2(300 nm)/Si substrate (details of the transfer process are described in the Supplementary Methods). In contrast to graphene grown on SiO2, the graphene lms on plastic and glass substrates are continuous over large areas at all T (even at room temperature), possibly due to the decrease in distance between GBs. The as-grown layers are nanocrystalline graphene, whose Raman structure has the following characteristics (Fig. 4b): two peaks centred at 1,359 4 cm 1 (the D band) and 1,594 2 cm 1 (the G band) with a relatively large FWHM and a ID/IG ratio of ~0.7 0.1, all expected for nanocrystalline graphene30. We have measured the thicknesses of graphene lms transferred on to SiO2 substrates by atomic force microscopy. Figure 4c shows lm thickness plotted as a function of

T for samples grown at t = 10, 30 and 60 min and all of our samples grown for t = 10 min yield lm thicknesses of 1.3 0.3 nm. In case of graphene layers grown on glass substrates at T60 C for t = 10 min, we obtain ~97.4% transmittance (Fig. 4d). Considering 2.3% absorption of incident white light in an individual graphene layer31, it can be inferred that these lms are ML thick, in good agreement with atomic force microscopy data of as-grown graphene layers.

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

ARTICLE

a b

Intensity (a.u.)

1,200

1,600

Raman shift (cm1)

2,000

2,400

2,800

Thickness (nm)

c 6

4

2

0 0

Growth temperature (C)

Transmittance (%)

d 100

95

90

85

80

40 80 120 160

400 500 600 700 800

Wavelength (nm)

Figure 4 | Large-scale growth of graphene lms on plastic and glass substrates. Representative (a) optical microscopy image of graphene lm grown at T=60 C for 10 min on PMMA then transferred to SiO2(300 nm)/

Si substrate (scale bar, 100 m). (b) Typical Raman spectra, with each of the colours corresponding to the coloured spots on the sample. Folded regions such as those highlighted by the green open circle are occasionally observed and are formed during transfer process. (c) The thicknesses of graphene lms transferred onto SiO2(300 nm)/Si substrates after growth at

T=25160 C for 10 min (black squares), 30 min (red circles) and 60 min (blue triangle) on PMMA. Error bars are the standard deviation of the thickness measured in 30 different edge regions in transferred graphene lms. (d) Transmittances of graphene lms grown at T=25 C (black), 45 C (red), 60 C (blue) and 110 C (green) for 10 min on glass substrates. The inset shows a photograph of graphene lm grown at temperature T=60 C.

Discussion

In our DAS process, the GBs in poly-Ni(111) lms aord the C atoms the chance to preferentially diuse to other surface at low temperatures. Upon reaching the Nisubstrate interface, C atoms precipitate out as graphene at the GBs (Supplementary Fig. S7) and growth occurs via lateral diusion along the interface. This process gives rise to a continuous but polycrystalline graphene lm. This nding indicates that we can control the average grain size of resulting graphene lms by Ni GB engineering, that is, control the positions where graphene grains start to grow.

For graphene on Ni/SiO2, using thermal annealing process before DAS process, we enlarged the grain size of poly-Ni lms by up to 520 m (that is, create GBs located far enough from each other); thereby, the macrocrystalline graphene grains up to several micrometres in size are obtained, as conrmed by plan-view TEM analysis. However, for Ni on plastic and glass substrates, the as-deposited poly-Ni lms were not annealed further, thereby the grain sizes were ~4050 nm. As a result, resulting graphene lms consist of grains of the order of nanometres. Similar results are also obtained in case we grow graphene lms on SiO2 using as-deposited, unannealed poly-Ni(111) lms.

Finally, we have also used polycrystalline Ni foils and obtained large-area graphene layers via DAS process at low temperatures. The details of synthesis and structural and optoelectronic characterization results will be presented elsewhere. We found that a precise control over the foil thickness, surface roughness and crystalline quality

of the foils is critical to obtaining graphene lms over large areas. We note that similar experiments carried out using single-crystal Ni foils do not yield graphene on either side of the foils, indicating that the GBs are necessary for low-temperature synthesis of graphene.

In summary, we have demonstrated transfer-free, large-area growth of graphene lms via DAS method at close to room temperature. Our approach can, in principle, be used to grow device-ready graphene layers on any arbitrary substrate even in air and the resulting graphene layers exhibit controllable structural and opto-electronic properties by Ni GB engineering. This relatively simple method of synthesizing graphene lms is potentially scalable and opens up new possibilities for a variety of electronic and optoelectronic applications.

Methods

Predeposition of Ni lms on nonconducting substrates. All of our experiments were carried out using SiO2/Si, PMMA-coated SiO2/Si, poly-dimethylsiloxane and glass substrates (12 cm2 size). Polycrystalline Ni (poly-Ni) lms with thicknessof 100 nm are deposited using an electron-beam evaporator (operating pressure: ~510 6 Torr) with solid Ni (99.99% purity) as the source. The substrate temperature is set to 400 C for SiO2/Si and room temperature (~25 C) for plastics (PMMA and poly-dimethylsiloxane) and glass. Microstructure of the Ni thin lms is determined using atomic force microscopy (Veeco Multimode V), scanning electron microscopy (SEM; FEI Quanta 200) and X-ray diraction (Bruker D8 Advance) techniques.

As-deposited Ni thin lms on SiO2/Si substrate have an average Ni grain size of ~100 nm (Supplementary Fig. S1a). To further control the grain size, crystallinity and morphology of poly-Ni lms, the as-deposited samples were transferredto an ultrahigh vacuum chamber ( < 10 9 Torr) and annealed at temperatures as high as ~1,000 C for times between 1 and 10 min in vacuum (~10 9 Torr) or ina H2 ambient (99.9999% purity) at pressures of 10 810 5 Torr. Aer annealing, we obtain Ni lms with a predominantly 111-texture, with large crystalline grains of sizes around 520 m and atomically at terraces as shown in Supplementary Figures S1b and S1c. We nd that H2 ambient is preferred to vacuum to obtain larger grains (Supplementary Fig. S1d). Using Raman spectroscopy (WiTec alpha 300R M-Raman), we rule out the presence of any crystalline C on the Ni lms due to residual C segregation aer annealing in a H2 ambient. In addition, we do not observe any crystalline C on SiO2 surfaces aer etching away the metal lms inan aqueous solution of FeCl3. In case of the Ni thin lms deposited at room temperature on plastic and glass substrates, the average grain size is around 4050 nm (Supplementary Fig. S2). Owing to the limited thermal stability of these materials, we did not anneal these samples.

Synthesis of graphene lms by DAS process. As C sources, we used a paste composed of graphite powder (Aldrich, product 496596) dispersed in ethanol.

The average size of graphite powder is ~40 m. The paste was plastered to the Ni surface to make CNi/substrate diusion couples and dried by heating the samples on a hot plate ( < 50 C). Pressure ( < 1 MPa) is uniformly applied by mechanically clamping the CNi/substrate diusion couple using a molybdenum holding stage. Then, the samples are sealed in a quartz tube and heated for 160 min while owing inert Ar gas (1 standard l min 1) or air at temperatures between 25 and 600 C. Following graphene growth, the samples are cleaned via sonication in deionized water, and the Ni lms are removed by etching in an aqueous solution of FeCl3, leaving behind a graphene lm on desired substrates. The etching time was found to be a function of the etchant concentration and type of the substrates. Typically, a 1 cm2 by 100-nm thick Ni lm on a SiO2/Si substrate can be dissolved by 1 M FeCl3

solution within 30 min.

Fisher model. A highly idealized poly-Ni lm matrix with square-shaped grains of side l, GB slabs of width and lm thickness d is considered. Among all competing diusion mechanisms, we only consider the lattice and GB diusion quantities and ignore the dislocation quantity as insignicant dislocations threading normal to surface exist in face-centered cubic (FCC) Ni(111) lm. Under these conditions, the number of atoms () that ow per unit time is essentially equal to the product of the appropriate diusivity (Di), concentration gradient (dc/dx) and transport area involved and the ratio of two uxes can be estimated for FCC metals as GB/

L = DGB/lDL = [(310 8)/l (cm)]exp(8.1TM/T)20. Assuming l=10 m = 10 3 cm and using TM of Ni = 1726 K, we have GB/L = 2.06108 at 473 K, 2.15103 at773 K. Even if we consider the facts that real polycrystalline lms contain various types and orientations of GBs and the lattice diusion of C atoms in Ni lm would occur by interstitial diusion rather than substitutional diusion used in the model, it is evident that lower temperatures greatly favour GB diusion relative to lattice diusion.

Density functional theory calculation. All calculations were performed using the VASP package based on the spin-polarized density functional theory32,33. We used a projector augmented wave potential34, and the generalized gradient approximation

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

of Perdew and Wang35. All congurations were fully relaxed until the maximum residual forces were less than 0.01 eV/. The kinetic energy cuto was 400 eV and the MonkhorstPack36 k-point grid was 551.

We calculated the stable adsorption site of a single C atom on a free-standing graphene sheet. The graphene model consisted of 32 C atoms with periodic boundary conditions along the in-plane directions. We found that a bridge-like site, where the added C atom sits on the centre of the bond between two neighbouring C atoms of graphene, is most favourable. To obtain the prefactor for the Arrhenius equation, the vibrational frequency of the added C atom on the adsorption site was calculated37. The adsorption energy and the vibrational frequency of the added

C atom on the site were found to be 1.65 eV and 22.8 THz, respectively. In addition, the minimum energy pathway for the adatom diusion was examined by the nudged elastic band method38. The additional C atoms in both initial and nal congurations were located at the most energetically favourable sites, which were neighbours. Nine intermediate images, initially constructed by a linear interpolation between the initial and nal congurations, were optimized along the diusion pathway, which enables the determination of the minimum energy barrier. This minimum energy barrier was found to be 0.24 eV.

We then performed the same calculation of a single C atom diusion on a free Ni(111) surface. The Ni thin lm model consisted of 6 Ni(111) layers with 16 Ni atoms in each layer for a total of 96 Ni atoms. Periodic boundary conditions were applied along the in-plane directions, whereas the three bottom layers of the Ni lm were held xed during the diusion process. We found that the C atom is most stable on the hexagonal close-packed (HCP) site, for which the adsorption energy is 7.01 eV, just 0.06 eV higher than on FCC site. In addition, the calculated energy barrier for the C atom diusion on the Ni(111) surface is 0.49 eV.

To examine the diusion of a single C atom along the grapheneNi(111) interface, we obtained the stable conguration of a graphene sheet on a free Ni(111) surface. For this, eight C atoms for a graphene sheet and 24 Ni atoms for the Ni(111) surface were used. The graphene sheet was biaxially stretched using 1.22% tensile strain to remove the lattice mismatch between graphene and the Ni(111) surface. The equilibrium distance between the graphene sheet and the Ni(111) surface is 3.286 , and the interaction energy is 0.02 eV per each C atom. This is in good agreement with the earlier calculations27.

When we introduce an additional C atom at the interface between graphene and the Ni(111) surface, the energetically preferred position for the C atom is in the HCP site on the Ni(111) surface. The distance between the C atom and the Ni surface is 1.03 and the distance between the C atom and the graphene sheet is 2.62 , which indicates that the diusion of a C atom along the interface will be more strongly inuenced by the Ni surface than the graphene ML. The calculated diusion barrier from the HCP site to the FCC site of the Ni(111) surface along the interface is found to be ~0.51 eV, which is slightly higher than that ona free Ni(111) surface. During the diusion, there is no signicant change in the distance between the C atom and the Ni surface. The hopping rate of a C atom can be found by the Arrhenius equation = 0exp( Eb/kT), which results in a room-temperature hopping rate of about 20,000 s 1 with an attempt frequency of 0 = 1.01013 Hz. In addition, the tracer diusion coefficient can be expressed in terms of the hopping rate and the mean square jump length < l2 > : Dt = < l2 > /

(2d), and is found to be ~200 nm2 s 1 using the lattice constant of Ni 3.52 . If the chemical diusion or the presence of graphene edges were considered, the diusion coefficient would be higher than the calculated value of Dt.

References

1. Novoselov, K. S. etal. Electric eld eect in atomically thin carbon lms. Science 306, 666669 (2004).

2. Novoselov, K. S. etal. Two-dimensional gas of massless Dirac fermions in graphene. Nature 438, 197200 (2005).

3. Zhang, Y., Tan, Y.- W., Stormer, H. L. & Kim, P. Experimental observation of the quantum Hall eect and Berrys phase in graphene. Nature 438, 201204 (2005).

4. Zhou, S. Y. etal. Substrate-induced bandgap opening in epitaxial graphene. NatureMaterials 6, 770775 (2007).

5. Kwon, S.- Y. etal. Growth of semiconducting graphene on palladium. Nano Lett. 9, 39853990 (2009).

6. Balandin, A. A. etal. Superior thermal conductivity of single-layer graphene. NanoLett. 8, 902907 (2008).

7. Lee, C., Wei, X. D., Kysar, J. W. & Hone, J. Measurement of the elastic properties and intrinsic strength of monolayer graphene. Science 321, 385388 (2008).

8. Bae, S. etal. Roll-to-roll production of 30-inch graphene lms for transparent electrodes. NatureNanotechnol. 5, 574578 (2010).

9. Reina, A. etal. Large area, few-layer graphene lms on arbitrary substrates by chemical vapor deposition. NanoLett. 9, 3035 (2009).

10. Kim, K. S. etal. Large-scale pattern growth of graphene lms for stretchable transparent electrodes. Nature 457, 706710 (2009).

11. Li, X. etal. Large-area synthesis of high-quality and uniform graphene lms on copper foils. Science 324, 13121314 (2009).

12. Sun, Z., Yan, Z., Beitler, E., Zhu, Y. & Tour, J. M. Growth of graphene from solid carbon sources. Nature 468, 549552 (2010).

13. Yan, Z. etal. Growth of bilayer graphene on insulating substrates. ACSNano 5, 81878192 (2011).

14. Peng, Z., Yan, Z., Sun, Z. & Tour, J. M. Direct growth of bilayer graphene on SiO2 substrates by carbon diusion through nickel. ACSNano 5, 82418247 (2011).

15. Shelton, J. C., Patil, H. R. & Blakely, J. M. Equilibrium segregation of carbon to a nickel(111) surface: a surface phase transition. Surf.Sci. 43, 493520 (1974).

16. Isett, L. C. & Blakely, J. M. Segregation isosteres for carbon at the (100) surface of nickel. Surf.Sci. 58, 397414 (1976).

17. Vang, R. T. etal. Controlling the catalytic bond-breaking selectivity of Ni surfaces by step blocking. NatureMaterials 4, 160162 (2005).

18. Tan, X. & Yang, G. W. Catalytic bond-breaking selectivity in the ethylene decomposition on Ni surfaces: kinetic Monte Carlo simulations. J.Phys.Chem. C 112, 42194225 (2008).

19. Fisher, J. C. Calculation of diusion penetration curves for surface and grain boundary diusion. J.Appl.Phys. 22, 7477 (1951).

20. Balluffi, R. W. & Blakely, J. M. Special aspects of diusion in thin lms. ThinSolidFilms 25, 363392 (1975).

21. Schedin, F. etal. Detection of individual gas molecules absorbed on graphene. NatureMaterials 6, 652655 (2007).

22. Reeves, G. K. & Harrison, H. B. Obtaining the specic contact resistance from transmission line model measurements. IEEEElectronDev.Lett. 3, 111113 (1982).

23. Xia, F., Perebeinos, V., Lin, Y., Wu, Y. & Avouris, P. The origin and limits of metal-graphene junction resistance. NatureNanotechnol. 6, 179184 (2011).

24. Lamber, R., Jaeger, N. & Schulz-Eklo, G. Electron microscopy study of the interaction of Ni, Pd, and Pt with carbon: I. Nickel catalyzed graphitization of amorphous carbon. Surf.Sci. 197, 402414 (1988).

25. Berry, B. S. Diusion of carbon in nickel. J.Appl.Phys. 44, 37923793 (1973).26. Ajayan, P. M. & Yakobson, B. I. Graphene: pushing the boundaries. Nature Materials 10, 415417 (2011).

27. Helveg, S. etal. Atomic-scale imaging of carbon nanobre growth. Nature 427, 426429 (2004).

28. Abild-Pederson, F. & Nrskov, J. K. Mechanism for catalytic carbon nanober growth studied by ab initio density functional theory calculations. Phys.Rev.B 73, 115419 (2006).

29. Geng, H.- Z. etal. Eect of acid treatment on carbon nanotube-based exible transparent conducting lms. J.Am.Chem.Soc. 129, 77587759 (2007).30. Ferrari, A. C. & Robertson, J. Interpretation of Raman spectra of disordered and amorphous carbon. Phys.Rev.B 61, 1409514107 (2000).

31. Nair, R. R. etal. Fine structure constant denes visual transparency of graphene. Science 320, 1308 (2008).

32. Kresse, G. & Furthmller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput.Mater.Sci. 6, 1550 (1996).

33. Kresse, G. & Furthmller, J. Efficient iterative schemes for ab-initio total energy calculations using a plane-wave basis set. Phys.Rev.B 54, 1116911186 (1996).

34. Kresse, G. & Joubert, D. From ultraso pseudopotentials to the projector augmented-wave method. Phys.Rev.B 59, 17581775 (1999).

35. Perdew, J. P. etal. Atoms, molecules, solids, and surfaces: applications of the generalized gradient approximation for exchange and correlation. Phys.Rev.B 46, 66716687 (1992).

36. Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Phys.Rev.B 13, 51885192 (1976).

37. Ala-Nissila, T., Frerrando, R. & Ying, S. C. Collective and single particle diusion on surfaces. Adv.inPhys. 51, 9491078 (2002).

38. Mills, G., Jonsson, H. & Schenter, G. K. Reversible work transition state theory: application to dissociative adsorption of hydrogen. Surf.Sci. 324, 305337 (1995).

Acknowledgements

We gratefully acknowledge support from Research Fund of the Ulsan National Institute of Science and Technology (Grants No. 1.080014.01, 1.110019.01) and from the Basic Science Research Program (Grants No. 2009-0074599, 2010-0009824, 2011-0005467, 2011-0013440) and from National Nuclear R&D Program (Grant No. 2011-0006387)and from World Class University Program (Grant No. R31-2008-000-10075-0) through the National Research Foundation (NRF) of Korea funded by the Ministry of Education, Science and Technology. SK gratefully acknowledges support from UCLA COR-FRG. This work has beneted from the use of the facilities at UNIST Central Research Facilities.

Author contributions

S.-Y.K. planned and supervised the project; E.Y. and S.K. advised on the project; J.K., J.H.C., J.-K.C. and S.-Y.K. designed and performed experiments; S.-D.K. and Y.-W.K. made TEM samples and the structural measurements; J.K, H.G. and K.P. performed the electrical measurements; and S.-D.P. and S.Y.K. performed density functional theory calculations and the analyses; S.-Y.K., S.K., S.Y.K., K.P. and J.K. analysed data and wrote the manuscript; All authors discussed and commented on the manuscript.

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1650

ARTICLE

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/

How to cite this article: Kwak, J. etal. Near room-temperature synthesis of transfer-free graphene lms. Nat.Commun. 3:645 doi: 10.1038/ncomms1650 (2012).

NATURE COMMUNICATIONS | 3:645 | DOI: 10.1038/ncomms1650 | www.nature.com/naturecommunications

2012 Macmillan Publishers Limited. All rights reserved.