Ambipolar charge transport in quasi-free-standing monolayer graphene on SiC obtained by gold intercalation

We present a study of quasi-free-standing monolayer graphene obtained by intercalation of Au atoms at the interface between the carbon buffer layer (Bu-L) and the silicon-terminated face (0001) of 4H-silicon carbide. Au intercalation is achieved by deposition of atomically thin Au on the Bu-L followed by annealing at 850 {\deg}C in an Argon atmosphere. We explore the intercalation of Au and decoupling of the Bu-L into quasi-free-standing monolayer graphene by surface science characterizations and electron transport in top-gated electronic devices. By gate-dependent magnetotransport we find that the Au-intercalated buffer layer displays all properties of monolayer graphene, namely gate tunable ambipolar transport across the Dirac point, and n- or p-type doping depending on the Au content.


Results and Discussion
To prepare Au-intercalated quasi-free-standing graphene, we start with 7 mm × 7 mm SiC/Bu-L substrates onto which we deposit Au on about half of the chip area (5 mm × 2.5 mm) (Fig. 1b).
We have studied deposition of two different Au thickness, tAu = 4 Å and 8 Å, in two SiC/Bu-L substrates. Our deposition conditions result in a cluster-free Au layer on the surface, as evidenced by atomic force microscopy (See Fig. S1), 19 with an RMS roughness of 1 Å. Despite the homogeneous coverage of Au on the Bu-L surface, such layers are electrically not conductive. 19 After Au deposition, the intercalation step at T = 850 °C results in an electrically conductive surface, showing resistance values of the order of few kOhms (~3 -20 kΩ). This onset of electrical conductivity occurs not only on the Au deposited area, but on the rest of the surface as well, where Au was originally not present, pointing to diffusion of Au across the surface. Together with the change in electrical conductance, optical inspection 20 of the chip allows to see that the thermal intercalation step leads to a modification in the transparency of the surface, which becomes more opaque millimeters away from the Au deposited area (Fig 1b). We inspected the surface with Raman spectroscopy with a laser wavelength of 638 nm, which readily shows the emergence of the graphene 2D peak at 2662 cm -1 and G peak at 1583 cm -1 everywhere on the surface, with a FWHM = 63 cm -1 for the dark and electrically conductive area, and FWHM = 68 cm -1 on the Au-deposited area ( Fig. 1(c,d)). 14 Both the Raman spectra and the electrical conductivity of the surface after the annealing step serve as a strong indication that the Bu-L has decoupled from the substrate and transformed into monolayer graphene everywhere on the surface that becomes more opaque. Scanning tunneling microscopy (STM) analysis shows that the Au-deposited area displays a granular morphology consistent with the presence of Au clusters on the surface (Fig. 1e), which gradually disappear as one moves away from the border with the diffused Au area. In the Au-diffused area, the surface appears clean and free of Au clusters, revealing a terraced surface similar to the typical SiC/Bu-L morphology (Fig.1f). As we describe below, the dark and electrically conductive area correspond to areas where Au has diffused over the surface and intercalated at the buffer-SiC interface, and we name this area the Au-diffused region as shown in Fig. 1b.
Studies at synchrotron facilities confirm the nature of Bu-L on the Au-deposited and Audiffused area. Fig. 2a is the low energy electron microcopy (LEEM) image at the boundary of the Au-deposited and Au-diffused areas after thermal drive-in of 4 Å Au. The contrast in the two regions is due to the slightly different amount of intercalated gold (the darker grey color scale corresponds to lower gold content). Decoupling of the Bu-L occurs on both the gold-deposited and gold-diffused areas, which is confirmed through the quenching of the 6√3×6√3 R30° pattern in low energy electron diffraction (LEED) ( Fig. 2(b,c), see Fig. S4). The diffractograms also show the hexagonal honeycomb structure of graphene after the thermal drive-in, visible for both gold-deposited and gold-diffused areas. The LEEM intensity IV curves from both sides of the boundary are shown in Fig. 2d: the dip at 5.5 eV is resulting from the intercalation and consequent formation of free-standing graphene. The intercalation of Au as the reason for the decoupling is further confirmed by the x-ray photoelectron microspectroscopy results (micro-XPS), showing the presence of Au in both the Au-deposited and Au-diffused areas and the gradual disappearance of Au signal far from the boundary (Fig. 2e). Another evidence of Auintercalation on both sides of the boundary can be seen in the C1s micro-XPS spectra (Fig. 2f), in which a charge transfer from gold to graphene increases the separation between carbon peaks from graphene and the SiC substrate. 13 To assess the electrical properties of the Au-intercalated quasi-free-standing graphene we have fabricated micro-sized devices on the two substrates (tAu = 4 Å, 8 Å). In total, we have studied 8 devices, placed on the Au-deposited and Au-diffused areas of the substrates. Devices were made by conventional electron beam lithography (EBL) and oxygen plasma etching. For gated devices, we have used dry-transferred hexagonal boron nitride (h-BN) (thickness ~20 nm) followed by atomic layer deposition of Al2O3 (38 nm) as a dielectric, and Ti/Au as a gate electrode. Fig. 3(ac) depicts the schematic structure of the top-gated devices in the top and edge contact configurations together with an optical micrograph of D1. To avoid further processing steps, h-BN was not patterned, and the geometry of the devices is dictated by the shape of the transferred h-BN flakes. Magneto transport properties of the devices were measured by the van der Pauw method in a gas flow cryostat down to T = 2 K. We quantified the Hall carrier density and mobility as nH = 1/eRH and µH = RH × σxx, respectively.
We found that devices made on the Au-deposited area are insensitive to the gate voltage, likely due to screening of the electric field by the Au layer present directly atop the graphene layer.
Gate response is only observed in those devices fabricated on the Au-diffused area, where Au is absent on top of the graphene layer according to STM scans.  Table I shows a summary of transport properties measured in all the devices. In general, the zero-gate doping of Au-intercalated devices is p-type, p = 5 × 10 12 -2 × 10 13 , being consistently higher in magnitude for the Au-deposited regions. The notable exception is for gated devices fabricated on the Au-diffused areas, which show n-type doping. According to Gierz et al., 14 Au contents corresponding to 3/8 monolayer (Au-ML) and 1 Au-ML results in highly n-doped (n = 5 × 10 13 cm -2 ) and slightly p-doped 15 (p = 7 × 10 11 cm -2 ) MLG, respectively. 24 Therefore, the ntype (p-type) doping in Au-diffused (deposited) areas could arise from the Au contents less (higher) than 1 ML. However, we cannot exclude the possibility of doping of the decoupled MLG from environment during the fabrication processes. While gated devices allow us to explore the carrier mobility as a function of the gate voltage, the mobilities of other devices at a fixed carrier density do not exceed 100 cm 2 /Vs. We note that the reported mobility of hydrogen intercalated Bu-L is of the order of 1000 cm 2 /Vs, [9][10][11][12] suggesting the possibility that the relatively low mobility is intrinsic in Bu-L as a consequence of defects on the as-grown Bu-L 25 or of an imperfect intercalation process. Finally, measurements in magnetic field allow us to analyze the nature of microscopic scattering processes in Au-intercalated Bu-L. Fig. 3f shows the symmetric part of the longitudinal magneto conductivity (Δσxx = σxx (B) -σxx (0)) of D1 at T = 2 K (See to the graphene lattice. 15,[28][29][30][31] Band calculations predict that the optimal configuration for SOI enhancement is that of Au atoms located at hollow sites of graphene. 15,28 Therefore, it is possible that the absence of spin-orbit scattering effects in our samples might be related to the Au atoms occupying random sites (i.e. disorder) at the Bu-L/SiC interface. Moreover, it appears that the excess of gold contributes to enhancing the disorder in the samples. The temperature dependence of devices fabricated with quasi-free-standing graphene with higher Au content (tAu = 8 Å) follows the characteristic dependence of granular metals 32 and variable range hopping 33

Conclusions
In conclusion, we show that Au-intercalated quasi-free-standing monolayer graphene can be

Growth of Bu-L on SiC:
The carbon buffer is an integral part of the epitaxial graphene-SiC material system and is the first to form when SiC substrate is exposed to high temperature (T > 1500 °C). More specifically, this is the carbon rich surface reconstruction (63 × 63) characteristic of Si face SiC at elevated temperatures. To prevent the growth of graphene and grow only the carbon zero-layer, here we used 7 × 7 mm 2 4H-SiC substrates and applied gradual (inductive) heating in Argon atmosphere until T ≈ 1700 °C was reached and that kept for 30 seconds. Then the furnace was switched off and the samples were taken out at room temperature.
Prior to growth, the chamber was pumped down to a base pressure of P0 = 1 × 10 -6 mbar in order to minimize oxygen contamination which is detrimental for a complete carbonization.          The shape of the Au intercalated Bu-L is dictated by the shape of h-BN and Ti/Au contacts are extended to samples. D2 is the edge contact and the others are top contact devices (See Fig.   3a,b). The scale bar is 10 µm.