High-Electrical-Conductivity Multilayer Graphene Formed by Layer Exchange with Controlled Thickness and Interlayer

The layer exchange technique enables high-quality multilayer graphene (MLG) on arbitrary substrates, which is a key to combining advanced electronic devices with carbon materials. We synthesize uniform MLG layers of various thicknesses, t, ranging from 5 nm to 200 nm using Ni-induced layer exchange at 800 °C. Raman and transmission electron microscopy studies show the crystal quality of MLG is relatively low for t ≤ 20 nm and dramatically improves for t ≥ 50 nm when we prepare a diffusion controlling Al2O3 interlayer between the C and Ni layers. Hall effect measurements reveal the carrier mobility for t = 50 nm is 550 cm2/Vs, which is the highest Hall mobility in MLG directly formed on an insulator. The electrical conductivity (2700 S/cm) also exceeds a highly oriented pyrolytic graphite synthesized at 3000 °C or higher. Synthesis technology of MLG with a wide range of thicknesses will enable exploration of extensive device applications of carbon materials.


Results
As shown in Fig. 1a, all samples started with a stacked structure of an a-C/Ni/SiO 2 substrate and were then annealed at 800 °C for 1 h in ambient Ar. We prepared the samples with and without a diffusion controlling Al 2 O 3 IL, which improves the crystal quality of MLG 42 . The thickness of the initial Ni layer determines the resulting MLG thickness, t, after layer exchange 41,42 . We varied t from 5 nm to 200 nm to investigate the effects of t on the crystal and electrical properties of the MLG. For all samples, MLG layers were obtained on the substrate via layer exchange. As representatively shown in Fig. 1b, thin MLG layers (t ≤ 10 nm) exhibited transparency. The scanning electron microscopy (SEM) image in Fig. 1c shows that although the MLG layer has submicron-size voids, it covers nearly the entire substrate. We note the Ni concentration in the MLG layer is below the detection limit of EDX (~1%). Figure 2a,b show that Raman spectra of the back side of the samples have peaks at approximately 1350, 1580, and 2700 cm −1 , corresponding to the D (disordered mode), G (graphitic mode), and 2D (D mode overtone) peaks in the graphitic structure, respectively. This means layer exchange between the C and Ni layers occurred and MLG formed on the SiO 2 glass substrate in all samples. In the Raman spectra, the intensity ratio of the G to D peaks (I G /I D ) in the spectra corresponds to the crystal quality of MLG 43 . Figure 2c indicates the I G /I D ratio strongly depends on t, and with or without the IL. For the samples without the IL, the I G /I D ratio shows the highest value of 7 at t = 100 nm. For the samples with the IL, the I G /I D ratio is as low as for the samples without the IL for t ≤ 20 nm, while it dramatically increases for t ≥ 50 nm and exceeds 20. The crystal quality of the MLG, estimated from the I G /I D ratio, likely reflects the contamination from the substrate and the grain size enlargement by inserting the IL 42 . Although the I G /I D ratio of 20 is still lower than that of MLG formed by the transfer technique or high-temperature CVD 4 , this value is the highest among MLG synthesized on insulators by metal-induced solid-phase crystallization 18,19,23,25 .
The detailed cross-sectional structure of the sample for t = 10 nm, which is a typical thin transparent film, was investigated using an analytical transmission electron microscope (TEM). Figure 3a shows the MLG formed on the entire substrate. Figure 3b,c show Ni on the MLG layer is particle-shaped, whereas the sample with t = 50 nm has a continuous Ni layer 41,42 . Considering that the MLG layer is uniformly formed, the Ni particles are likely attributed to the agglomeration of a thin Ni layer moved on MLG after layer exchange. Figure 3d,e show {002} oriented MLG forms on the SiO 2 substrate in both the Ni-contacted and non-Ni-contacted regions. The MLG in the non-Ni-contacted region is slightly thinner than that in the Ni-contacted region, probably because of the C evaporation during annealing. Figure 3d,e indicate the crystal structure of the thin MLG is disordered compared with the sample with an IL 42 . We note that, for t = 50 nm, the grain size of the MLG layer without an IL was a few hundred nm and that with an IL was a few μm 41,42 . These results account well for the Raman study (Fig. 2c).
For high-quality MLG, such as that synthesized by thermal decomposition of a SiC substrate, TEM observation can identify the stacking structure (e.g. AA and AB stacking) 44,45 . Unfortunately, because the current MLG is  www.nature.com/scientificreports www.nature.com/scientificreports/ small-grained polycrystalline where the grains are randomly oriented in the in-plane direction 42 , it is difficult to identify the stacking. Figure 4 shows the electrical properties of the MLG strongly depend on t, and with or without the IL. Figure 4a shows the carrier concentration of the MLG decreases with the increase of t and approaches the value of a highly oriented pyrolytic graphite (HOPG) with a low mosaic degree of 0.4°, especially for the sample with the IL. Figure 4b shows the carrier mobility dramatically increases for the t ≥ 50 nm samples with the IL, whereas it remains low in the whole range of t for the samples without the IL. The carrier mobility exhibits the maximum value of 550 cm 2 /Vs for the t = 50 nm sample with the IL, which is the highest Hall mobility among MLGs directly formed on an insulator. Figure 4c shows the behavior of the electrical conductivity reflects that of the carrier concentration and carrier mobility. The electrical conductivity exhibits the maximum value of 2700 S/cm for the t = 50 nm sample with the IL, which exceeds that of the HOPG synthesized at 3000 °C or higher.

Discussion
We have controlled the thickness of MLG on an insulator formed by layer exchange with and without the IL, resulting in a high electrical conductivity. Considering the fact that the behavior of the carrier mobility (Fig. 4b) is the same as that of the I G /I D ratio in the Raman spectra (Fig. 2c), the carrier mobility clearly reflects the crystal quality of the MLG. The carrier mobility, that is, the crystal quality of the MLG is much higher than any other MLG directly formed on an insulator; however, does not reach the HOPG. The electrical conductivity of the MLG exceeds that of HOPG with the aid of autodoping. Although the reasons for the excessive carriers for the MLG are still unclear, when we assume defects in MLG produce carriers, the carrier concentration behavior is almost consistent with the Raman study.
In conclusion, the crystal quality and electrical properties of the MLG, formed on an insulator by layer exchange at 800 °C, strongly depended on t, especially when preparing the IL. The crystal quality was relatively low for t ≤ 20 nm; however, it dramatically improved for t ≥ 50 nm with the IL. As a result, the sample for t = 50 nm with the IL exhibited the maximum carrier mobility (550 cm 2 /Vs), and the highest Hall mobility among MLG directly formed on an insulator. The electrical conductivity (2700 S/cm) even exceeded the high-quality HOPG (mosaic degree: 0.4°). Thus, the layer exchange method allows us to form MLG with a wide range of thicknesses on arbitrary substrates. This opens the door for a broad range of applications that combine advanced electronic devices with carbon materials.  www.nature.com/scientificreports www.nature.com/scientificreports/ Methods sample preparation. Ni thin films with thicknesses of 5, 10, 20, 50, 100, and 200 nm were prepared on SiO 2 glass substrates. Subsequently, Al 2 O 3 (thickness: 2 nm) was employed as the diffusion control IL 42 . Then, a-C thin films were prepared, wherein the thickness ratio of C:Ni = 3:2 (e.g., C:Ni = 300:200 nm). All depositions were carried out using radio-frequency (RF) magnetron sputtering (base pressure: 3.0 × 10 −4 Pa) with an Ar plasma. During the deposition, the substrate temperature was room temperature for Al 2 O 3 and 200 °C for Ni and C. The RF power was set to 100 W for C and 50 W for Ni and Al 2 O 3 . For comparison, we prepared the samples without the Al 2 O 3 IL. Samples were annealed at 800 °C for 1 h in an Ar ambient to induce layer exchange growth. The sample was dipped in a diluted HNO 3 solution (10% HNO 3 ) for 30 min to remove Ni moved to the top layer.
Material characterization. SEM analyses were performed using a JEOL JSM-7001F with an EDX spectrometer (JEOL JEO-2300). Raman spectroscopy was performed using a JASCO NRS-5100, wherein the laser wavelength was 532 nm and the spot size was 5 µm. TEM analyses were performed using an analytical TEM, FEI Tecnai Osiris, operating at 200 kV, equipped with an EDX spectrometer (FEI Super-X system). The cross-sectional TEM sample was prepared using the conventional focused ion beam method. Hall effect measurement was performed with the Van der Pauw method using the Bio-Rad HL5500PC system. Carrier mobility, carrier concentration, and electrical conductivity were averaged over five measurements for each sample.

Data Availability
The data used in this study are available upon reasonable request from the corresponding author K.T. (toko@ bk.tsukuba.ac.jp).