In-plane anisotropy of graphene by strong interlayer interactions with van der Waals epitaxially grown MoO₃

Abstract

van der Waals (vdW) epitaxy can be used to grow epilayers with different symmetries on graphene, thereby imparting unprecedented properties in graphene owing to formation of anisotropic superlattices and strong interlayer interactions. Here, we report in-plane anisotropy in graphene by vdW epitaxially grown molybdenum trioxide layers with an elongated superlattice. The grown molybdenum trioxide layers led to high p-doping of the underlying graphene up to p = 1.94 × 10¹³ cm⁻² regardless of the thickness of molybdenum trioxide, maintaining a high carrier mobility of 8155 cm² V⁻¹ s⁻¹. Molybdenum trioxide–induced compressive strain in graphene increased up to −0.6% with increasing molybdenum trioxide thickness. The asymmetrical band distortion of molybdenum trioxide–deposited graphene at the Fermi level led to in-plane electrical anisotropy with a high conductance ratio of 1.43 owing to the strong interlayer interaction of molybdenum trioxide–graphene. Our study presents a symmetry engineering method to induce anisotropy in symmetric two-dimensional (2D) materials via the formation of asymmetric superlattices with epitaxially grown 2D layers.

Epitaxially grown MoO₃ on graphene induces in-plane conductance anisotropy as well as high strain, doping, and carrier mobility.

INTRODUCTION

van der Waals (vdW) epitaxy is a widely used technique for fabricating two-dimensional (2D) epitaxial layers on a variety of growth templates with crystallographic alignment (1–3). The major advantages of vdW epitaxy originate from the dangling bond–free interface and relatively weak vdW interactions between the epilayers and growth templates (4), enabling the growth of epilayers with a remarkable mismatch of lattice (5) and symmetry (6–8) without substantial interfacial strain or misfit dislocations. The interlayer interaction is of particular importance in vdW epitaxy with symmetry mismatch due to the formation of anisotropic superlattices (9–12) and directional strain during a cooling procedure (13).

Here, we report the in-plane anisotropy in graphene induced by the strong interlayer interaction with vdW epitaxially grown α-MoO₃. The strong interlayer interaction at the vdW interface helped modulate the mechanical and electrical properties of monolayer graphene. The epilayer and graphene template with different thermal expansion coefficients (TECs) exerted a high compressive strain on graphene, which increased up to −0.6% with increasing MoO₃ thickness. Meanwhile, the graphene was highly p-doped (p = 1.94 × 10¹³ cm⁻²) by the charge transfer from MoO₃ regardless of the MoO₃ thickness. Notably, MoO₃-deposited graphene exhibited in-plane anisotropy (conductance ratio of 1.43) in the electrical conductance owing to the crystal orientation–related periodic potentials by MoO₃, maintaining a high carrier mobility of 8155 cm² V⁻¹ s⁻¹. Our study shows that vdW epitaxial interface can induce the strong interlayer interaction and in-plane anisotropy, which can be used to modulate the crystal orientation–dependent properties of 2D materials.

RESULTS

Crystal structure of MoO₃/graphene heterostructure

To investigate the interlayer interaction at a symmetry-mismatched epitaxial interface, we used orthorhombic α-MoO₃ (Pnma space group) with in-plane anisotropic mechanical, electrical, and optical properties (14–18). MoO₃ layers comprising MoO₆ octahedral double layers (a = 3.96 Å and c = 3.70 Å) were stacked along the b axis with weak vdW forces (14, 19–22). MoO₃ was grown on exfoliated graphene by the thermal evaporation of the Mo film in the air (see Materials and Methods) (23). Figure 1A shows MoO₃-grown monolayer graphene (MoO₃/Gr). Figure 1B shows the Raman spectra of the as-exfoliated graphene and MoO₃/Gr samples. Two main Raman peaks of MoO₃ can be observed at ~817 and ~ 995 cm⁻¹, corresponding to the stretching modes of doubly coordinated oxygen along the a axis (24, 25). The absence of D peak, an indicator of defects in graphene, implies that there was no recognizable damage in graphene during MoO₃ growth (26, 27). However, G and 2D peaks, which are sensitive to doping and strain, shifted markedly after the deposition of MoO₃. This indicates that the deposited MoO₃ induced doping and strain in the underlying graphene. This is discussed later in the paper. The morphology and thickness of MoO₃ were measured using atomic force microscopy (AFM), as shown in Fig. 1 (C and D). The graphene was fully covered by bilayer MoO₃ (2L-MoO₃) with a few thick islands, as reported previously (fig. S1) (23, 28). The MoO₃ surface was clean and flat without contamination or damage. The height profile in Fig. 1D (from the white dashed line in Fig. 1C) shows that the thickness of the monolayer MoO₃ corresponds to half of the unit cell (~0.7 nm) (14, 19–22).

Fig. 1. vdW epitaxial growth of MoO₃ on monolayer graphene.

(A) Optical microscope image of MoO₃/Gr. (B) Raman spectra of as-exfoliated Gr (black) and MoO₃/Gr (red). The notable features after MoO₃ growth are the emergence of MoO₃ Raman peaks at ~820 and ~990 cm⁻¹, blueshift in the G and 2D peaks, and absence of D peak. a.u., arbitrary units. (C) AFM image of MoO₃/Gr of the dashed area in (A). (D) AFM height profile along the white dashed line in (C). The heights of MoO₃ islands correspond to multiple layers of MoO₃ (~0.7 nm)

Epitaxial relationship between MoO₃ and graphene

To reveal the crystal structures and epitaxial relationship between MoO₃ and Gr, we used spherical aberration-corrected transmission electron microscopy (Cs-TEM). The TEM image in Fig. 2A shows MoO₃/Gr transferred onto a TEM grid with holes, where 2L-MoO₃ is partially grown on the graphene. The high-resolution TEM image in Fig. 2B confirms the epitaxial growth of crystalline MoO₃ on the graphene. The diffraction pattern obtained by fast Fourier transform (FFT), shown in the inset of Fig. 2B, verifies that the rhombic pattern (blue dashed line) of MoO₃ is aligned with the hexagonal pattern (orange dashed line) of graphene. The (200) plane of MoO₃ is parallel to the $(1\overline{1}00)$ plane of graphene (29, 30). The FFT-filtered TEM images of graphene (left) and MoO₃ (right) in Fig. 2C, obtained from each diffraction pattern, clearly show the growth of rectangular MoO₃ unit cells with lattice parameters of 3.97 and 3.75 Å on the graphene with an epitaxial relationship: a and c axes of MoO₃ are parallel to the armchair (ac) and zigzag (zz) directions of graphene, respectively. The atomic model in Fig. 2D shows the formation of a superlattice (indicated by yellow dashed lines) in the MoO₃/Gr heterostructure with an epitaxial relationship. The long superlattice unit cell in MoO₃/Gr can induce distinct periodicities along two directions of ac and zz in graphene (9–12).

Fig. 2. Epitaxial relationship between MoO₃ epilayers and monolayer graphene growth template.

(A) Low-magnification TEM image of MoO₃/Gr. (B) High-resolution TEM image of MoO₃/Gr. The inset shows the corresponding FFT image. Only a single set of rhombus pattern indicates the growth of single-crystal MoO₃ (blue). In addition, the perfect alignment of MoO₃ and graphene (orange) patterns demonstrates the epitaxial growth of MoO₃ on monolayer graphene. (C) FFT-filtered images of graphene (left) and MoO₃ (right). (D) Schematic of MoO₃/Gr superlattices (yellow dashed boxes) based on their epitaxial relationship. The periodicity along the horizontal direction (a axis of MoO₃ and ac direction of graphene) is approximately eight times greater than that along the vertical direction (c axis of MoO₃ and zz direction of graphene).

Modulation of doping and strain in graphene by MoO₃ epilayer

To investigate the interlayer interaction between MoO₃ and graphene, we measured the Raman spectra of the graphene regions covered by MoO₃ islands of different thicknesses (1L to 5L), as shown in Fig. 3A. For comparison, the Raman spectra of the as-exfoliated graphene (gray) and uncovered graphene (black) are also shown in Fig. 3A. The uncovered graphene is the region with no deposition of MoO₃ after epitaxial growth. Although the uncovered graphene had no MoO₃, there was a blue shift in the G and 2D peaks owing to the annealing effect (31). Notably, the MoO₃/Gr region showed a marked blue shift in the G and 2D peaks compared to the as-exfoliated and uncovered graphene. Furthermore, the two peaks substantially blue-shifted with increasing MoO₃ thickness. The marked shifts in the Raman peaks indicate a strong interlayer interaction between the MoO₃ epilayers and graphene, leading to considerable doping and strain in graphene depending on the MoO₃ thickness (26, 27).

Fig. 3. Hole concentration and strain in the MoO₃-grown graphene.

(A) Raman spectra of the as-exfoliated Gr (gray), uncovered Gr (black), and MoO₃/Gr with various MoO₃ thicknesses (rainbow colored, red to purple with increasing MoO₃ thickness). For MoO₃/Gr, the apexes of the G and 2D peaks are connected by dashed lines for visual guidance. (B) Correlation plot of the ω_2D − ω_G for the samples shown in (A). (C) AFM image of MoO₃/Gr with various MoO₃ thicknesses. (D and E) Mapping images of doping and strain of MoO₃/Gr, respectively. The strain shows distinctive difference depending on the MoO₃ thickness, whereas doping exhibits a difference depending only on the presence of MoO₃ epilayers. (F and G) Plots of hole concentration and strain as a function of the number of MoO₃ layers.

To separately measure the doping concentration (p) and strain (ε) of MoO₃/Gr, we marked the positions of the G (ω_G) and 2D (ω_2D) peaks in the correlation plot of Fig. 3B. The doping and strain of graphene can be quantified by projecting ω_G and ω_2D to the doping concentration (magenta) and strain axes (gray), respectively (fig. S2) (32). As shown in Fig. 3B, the uncovered graphene is p-doped (p = 7.8 × 10¹² cm⁻²) with a small compressive strain of −0.07% after the MoO₃ growth process, which is due to the conformal contact of graphene on the SiO₂ substrate by annealing under ambient condition (31, 33, 34). In the case of MoO₃/Gr, the graphene is highly p-doped (p = 1.94 × 10¹³ cm⁻²) regardless of the MoO₃ thickness, while the compressive strain increases with the number of MoO₃ layers. Different MoO₃ thickness dependence of the doping concentration and strain can be clearly visualized in Fig. 3 (C to E). Although MoO₃ regions of different thicknesses were deposited on the graphene, as shown in the AFM image of Fig. 3C, the p-doping concentration of the graphene measured by the Raman peaks was maintained irrespective of the number of MoO₃ layers, as shown in Fig. 3D. Meanwhile, thicker MoO₃ epilayers resulted in a higher compressive strain, as shown in Fig. 3E.

To clearly show the MoO₃ thickness dependence of doping and strain in graphene, we plotted the doping concentration and compressive strain as a function of the MoO₃ thickness in Fig. 3 (F and G), respectively. The p-doping concentration of 1L-MoO₃/Gr was almost twice higher than that of the uncovered graphene and almost invariant for 1L-MoO₃ to 5L-MoO₃ (Fig. 3F). This indicates that a high p-doping concentration of graphene can be achieved using only monolayer MoO₃. This is attributed to the efficient charge transfer between MoO₃ and graphene by the large and thickness-insensitive work function of MoO₃ (28, 35–37). In contrast, the compressive strain of MoO₃/Gr clearly shows a dependency on the MoO₃ thickness (Fig. 3G). The compressive strain exerted on graphene by MoO₃ is due to the difference in the TECs between MoO₃ and graphene (14, 38, 39) as well as the high lateral stiffness and friction of MoO₃ (28, 40, 41). The strain of graphene increased, and the increment gradually decreased with the number of MoO₃ layers. This may be due to the formation of a more rigid MoO₃ structure with increasing thickness. Both the doping and strain of graphene induced by the MoO₃ epilayer indicate a strong interlayer interaction between MoO₃ and graphene with an epitaxial relationship.

In-plane anisotropy of MoO₃/Gr

The formation of an elongated superlattice and strong interlayer interaction at the MoO₃-Gr heterointerface can induce crystal orientation–dependent modulation of the properties of graphene. To verify the orientation-dependent structural modulation, we measured the angle-resolved polarized Raman spectra of MoO₃/Gr by rotating an analyzer (θ_out) at a fixed polarizer (θ_in) (see Materials and Methods for detailed information on angle-resolved polarized Raman spectroscopy). Figure 4A shows the G peaks of 3L-MoO₃/Gr measured at four specific angles. The G peaks were deconvoluted into two peaks of G⁺ (red area) at 1621.7 cm⁻¹ and G⁻ (blue area) at 1623.2 cm⁻¹, which have eigenvectors parallel and orthogonal to the strain direction, respectively (fig. S3) (42–44). The intensities of these peaks vary as a function of θ_out, resulting in a periodic shift in the convoluted G peak, as shown in the contour plot in Fig. 4B. The peak intensities of G⁺ (red circles) and G⁻ (blue circles) are plotted in polar coordinates in Fig. 4C. They show well-defined cos² θ_out patterns that repel each other in accordance with theoretical predictions (solid lines in Fig. 4C), indicating the presence of a uniaxial strain in graphene (42–44). Therefore, our observations from the polarized Raman spectra show that the vdW epitaxially grown MoO₃ generates an anisotropic strain in graphene (13, 14, 38, 39).

Fig. 4. Anisotropy in graphene induced by MoO₃ epilayers.

(A) Fitting results of four representative angle-resolved polarized Raman spectra. The measured data are represented by the hollow circles, while the red and blue shaded areas correspond to the fitted results of the G⁺ and G⁻ peaks, respectively. The peak positions of G⁺ and G⁻ are indicated by the red and blue dotted lines, respectively. (B) Contour plot of the angle-resolved polarized Raman spectra of G peak as function of analyzer angle (θ_out). (C) Intensities of G⁺ (red) and G⁻ (blue) peaks plotted in the polar coordinate. Spheres correspond to the intensities, and solid lines correspond to the results fitted to the theoretical expectation (42, 43). (D) Orientation-dependent transfer characteristics of MoO₃/Gr electric device. The inset shows an optical image of the MoO₃/Gr electric device. (E to G) Polar coordinate plots of the hole concentration at zero-gate voltage (E), conductance (F), and field-effect mobility (G) of the MoO₃/Gr device in (D), respectively. Conductance and field-effect mobility are measured under various carrier concentrations, which are represented with different colors.

We also measured the angle-resolved electrical transport in the 1L-MoO₃/Gr device. The inset of Fig. 4D shows a device in which multiple electrodes of different angles are deposited around the 1L-MoO₃ island. Figure 4D shows the transfer curves (I_DS-V_G) measured with different electrodes at specific angles, indicating a highly p-doped MoO₃/Gr, as observed in the Raman measurement. In addition, the current levels and slopes of the transfer curves strongly depend on the measured orientation. Figure 4 (E to G) shows the p-doping concentration at V_G = 0 V, conductance (G), and field-effect mobility (μ_FE) of MoO₃/Gr in polar coordinates. The p-doping concentration was calculated using the equation $p=\frac{{\mathrm{\epsilon }}_0{\mathrm{\epsilon }}_{\mathrm{r}}{V}_{\mathrm{CNP}}}{e{t}_{\mathrm{ox}}}$, where ε₀, ε_r, V_CNP, e, and t_ox are the vacuum permittivity, relative permittivity of SiO₂, charge neutrality point of the MoO₃/Gr device, elemental charge, and thickness of SiO₂, respectively. The doping level showed no orientation dependence because the doping of graphene was generated by the charge transfer between MoO₃ and graphene (Fig. 4E). In contrast, both conductance and μ_FE exhibit anisotropy with 180° periodicity, as shown in Fig. 4 (F and G), respectively. Anisotropy conductance ratio (G_max/G_min) was 1.43 at V_G = 0 V, which is comparable to that of a representative in-plane anisotropic 2D material, black phosphorus (~1.5) (45–47). Similarly, μ_FE of MoO₃/Gr showed a large anisotropy ratio (μ_FE,max/μ_FE,min) of 1.77, maintaining a high mobility of 8155 cm² V⁻¹ s⁻¹ at p = 1 × 10¹² cm⁻². This result shows that in-plane electrical anisotropy can be achieved by epitaxially grown MoO₃ even in symmetrical graphene with isotropic Dirac cones at K and K′ points of the Brillouin zone (BZ) (48, 49). The insulating properties of MoO₃ indicate no electrical contribution to the conductivity of MoO₃/Gr (fig. S4). Note that the vdW epitaxially grown MoO₃ led to a high p-doping concentration of graphene without substantial degradation in the carrier mobility. As a result, MoO₃/Gr shows low sheet resistance of ~133 ohms per square (figs. S5 and S6). Although the contact resistance of the MoO₃/Gr is reduced by doping compared to graphene device, influence of contacts on the conductivity of MoO₃/Gr was excluded because of small values (fig. S7).

Anisotropic distortion of graphene band structure by MoO₃ epilayer

To clarify the origin of the unconventional conductance anisotropy of MoO₃/Gr, we performed first-principles calculations based on density functional theory (DFT). Anisotropic periodic potentials on graphene cause anisotropic deformation of its band structure (50, 51). Accordingly, we expected that the epitaxially grown MoO₃ layers would create anisotropic periodic potentials in graphene in close contact. The formation of a rectangular superlattice and uniaxial compressive strain, which was experimentally observed in MoO₃/Gr, can induce atomic-scale corrugation in monolayer graphene. Therefore, the MoO₃ epilayers can exert anisotropic periodic potentials on corrugated graphene, leading to an anisotropic deformation of the band structure in graphene.

To verify our hypothesis, we constructed a MoO₃/Gr heterostructure by considering the strain induced by thermal expansion. The optimized lattice constants were 2.47 Å for monolayer graphene, and a = 3.96 Å and c = 3.70 Å for MoO₃ monolayer. As observed in the TEM images in Fig. 2, the a and c axes of MoO₃ are aligned with the ac and zz directions of graphene, respectively. A MoO₃/Gr heterostructure with supercells of MoO₃ (16 × a + 2 × c) and graphene (15 × ac + 3 × zz) was constructed by minimizing the mismatch between the lattice parameters of MoO₃ and graphene and by applying orientation dependent TECs for MoO₃ and graphene (14, 39). In the constructed MoO₃/Gr heterostructure, the graphene had anisotropic compressive strains of 1.2 and 0.2% along the ac and zz directions under the constraints of the MoO₃ lattice, respectively. Figure 5A shows the out-of-plane deformation of graphene in contact with epitaxially grown 1L-MoO₃. Within the superlattice (indicated by a black box), the MoO₃-deposited graphene was corrugated with a maximum out-of-plane displacement of 0.09 Å. The carbon atoms that were aligned (misaligned) with the terminal oxygen atoms of MoO₃ were relocated below (above) the mass center of the graphene layer. The upward and downward displacements of the carbon atoms in graphene are indicated in red and blue, respectively. Because of the atomic arrangement, graphene exhibited a 1D-like corrugation parallel to the zz direction, and the periodicity of the corrugation corresponded to the length of the superlattice.

Fig. 5. Anisotropic band distortion in MoO₃/Gr.

(A) Out-of-plane deformation of graphene for optimized MoO₃/Gr heterostructure. Blue (red) color represents the displacement of carbon atoms below (above) the center of mass of the graphene layer. (B) Unfolded energy bands of the MoO₃/Gr heterostructure, projected onto the graphene layer, along the x axis (left) and y axis (right) in the vicinity of the K point. The gray dashed lines represent the Fermi level of MoO₃/Gr. The green dashed lines indicate the band structure of isolated graphene with the same compressive strain as that in the heterostructure, 1.2% along the ac direction and 0.2% along the zz direction. The color scale indicates the k-dependent spectral weight for the primitive BZ of the graphene layer. (C) Calculated in-plane electrical conductivities using the unfolded spectral functions projected on graphene at 300 K along the x and y axes.

The directional corrugation of graphene results in periodic potentials on graphene depending on the distance between MoO₃ and graphene. In the supercell geometry, the electronic energy bands fold into small BZ of the cell so that the modifications of Dirac energy bands of graphene are hardly noticeable. To avoid this, we projected or unfolded the dense energy bands in the supercell BZ into the original BZ of graphene (52–54). Figure 5B shows the unfolded energy bands of the optimized MoO₃/graphene heterostructure (fire contour) along two directions of x (ac) and y (zz), shown in Fig. 5A, projected onto the graphene layer. The Fermi level (E_F; gray dashed lines) is downshifted by ~0.7 eV below the Dirac point of graphene, which supports our experimental observation of high p-doping concentration of graphene by the deposition of MoO₃. We also compared the band structures of the isolated graphene layer (green dashed lines), which had the same compressive strain as the MoO₃/graphene heterostructure. We noted that the apparent gapful (gapless) Dirac point along q_x (q_y) direction in Fig. 5B is caused by strain induced shift of Dirac point along q_y direction. The effective band structure along the y axis on the right side of Fig. 5B is strongly altered at certain energy ranges, while that along the x axis on the left side of Fig. 5B is similar to that of isolated graphene (white dashed boxes). Around the Fermi level, band splitting is noticeable, indicating strong hybridization with the MoO₃ layer. It is estimated that the directional electronic difference shown in Fig. 5B induces anisotropy in the electrical transport properties of the MoO₃/graphene heterostructure.

On the basis of these band structures, the conductivities (σ) along the x and y axes could be estimated using the Boltzmann transport equation under a simple constant relaxation time (τ) approximation (55). Considering the crystal symmetry of the current system, the resulting conductivity along α(=x,y) direction (σ_α) is given by

$${\mathrm{\sigma }}_{\mathrm{\alpha }}(\mathrm{\mu },T)=\frac{e^2\mathrm{\tau }}{8{\mathrm{\pi }}^3}\int \int {\sum }_n{[{\mathbf{v}}_{n,\mathbf{k}}^{\mathrm{\alpha }}]}^2\mathrm{\delta }(\mathrm{\epsilon }-{\mathrm{\epsilon }}_{n,\mathbf{k}})\times [-\frac{\mathrm{\partial }f(\mathrm{\epsilon },\mathrm{\mu },T)}{\mathrm{\partial }\mathrm{\epsilon }}]d\mathbf{k}d\mathrm{\epsilon }$$ (1)

where v^α_n,k, ε_n,k, τ, and f are the α-directional component of k-dependent group velocity, energy of the nth band, relaxation time (considered a constant for simplicity), and Fermi distribution function as a function of the energy (ε), chemical potential (μ), and temperature (T), respectively. As shown in Fig. 5C, the electrical conductivity along the y axis is lower than that along the x axis around the Fermi level, consistent with our experimental results. Although the simulation results partly support observed anisotropic conductance originated from structural disparities for two orthogonal directions, the obtained anisotropy is still smaller than the measurement. We may consider further detailed characteristics reflecting actual experiment situations, such as larger supercell geometries and different scattering times along the direction, to improve the simulation results later.

DISCUSSION

In conclusion, we systemically investigated the interlayer interaction between MoO₃ epilayers and monolayer graphene growth templates in the symmetry-mismatched epitaxy. Our results demonstrated that the extreme modulation of doping and strain in graphene was generated by the strong interlayer interaction with the MoO₃ epilayers. First, the hole concentration in graphene could be markedly increased by the deposition of a single layer of MoO₃. Our results were comparable with those obtained using other similar doping methods (deposition of a charge-transfer layer on graphene) (56, 57). In addition, by modulating the thickness of the MoO₃ epilayers, the strain exerted on graphene could be controlled, preserving a high hole concentration. Furthermore, the directional deformation of the graphene band structure leads to anisotropy in the electrical conductance of symmetric graphene. Overall, our work shows that the strong interlayer interaction between vdW epitaxially grown 2D oxides and 2D materials can be used as an approach for symmetry engineering of 2D materials, while preserving their outstanding electrical properties. Our findings have promising applications in optoelectronic devices that require optical and electrical anisotropy.

MATERIALS AND METHODS

Sample preparation

A MoO₃/graphene heterostructure was prepared using our previously reported methods (23). As the Mo source, a 100-nm-thick Mo film was deposited on a SiO₂ (285 nm)/Si substrate using an e-beam evaporator or a DC magnetron sputter. The quality and morphology of MoO₃ were irrelevant to the metal deposition method. As the target substrate, graphene was mechanically exfoliated on another SiO₂/Si substrate. Only monolayer graphene flakes were selected and used as target templates after their thicknesses were identified using Raman spectroscopy. To synthesize MoO₃ on monolayer graphene, the Mo film was placed on a preheated heater (525°C), and, shortly after, the target substrate was located 0.5 mm above the Mo film upside down. The Mo film was oxidized and sublimated into MoO_x and condensed on the graphene-exfoliated substrate because of the temperature difference between the source and target substrates. After 10 min, the target substrate was immediately removed from the heater. The thickness and coverage of MoO₃ could be roughly controlled by varying the deposition time.

Raman spectroscopy

Raman spectra were acquired using a Raman spectroscope (Renishaw Raman, inVia Reflex Confocal Raman Microscope, 532 nm, 600 gratings). Angle-resolved polarized Raman measurements were performed using a home-built confocal micro-Raman system with excitation sources of the 2.33 eV (532 nm) line from a diode-pumped solid-state laser. A 50× objective lens (0.8 numerical aperture) was used to focus the laser beam onto the sample and collect the scattered light (backscattering geometry). The Raman signal was dispersed using a Jobin-Yvon HORIBA iHR550 spectrometer (2400 grooves mm⁻¹) and detected by a liquid nitrogen–cooled back-illuminated charge-coupled device detector. The laser power was kept below 0.1 mW. The polarizer was fixed at constant angle (θ_in), and the analyzer angle (θ_out) was rotated to select specific polarization of the scattered light. An achromatic half-wave plate was placed in front of the spectrometer to keep the polarization direction of the signal entering the spectrometer constant with respect to the groove direction of the grating.

Transmission electron microscopy

MoO₃/Gr was transferred to a holey carbon Au TEM grid using a poly(methyl methacrylate) (PMMA)–based transfer method. PMMA was spin-coated on the SiO₂/Si substrate where MoO₃/Gr was located and immersed in a 2 weight % KOH solution after scouring the edges of the substrate. The PMMA/MoO₃/Gr film floated on the solution because SiO₂ was etched by KOH. The film was rinsed with deionized water several times and transferred onto a holey carbon Au TEM grid, and PMMA was removed by placing the TEM grid in acetone overnight. High-resolution TEM images were captured using a Cs-TEM (JEOL JEM-ARM 200F Cs-TEM).

Atomic force microscopy

AFM images were measured using NX-10 (Park Systems). Both contact and noncontact modes were performed considering the status of the samples and environment.

Device fabrication and electrical measurements

e-beam lithography was performed to define the patterns of the source and drain electrodes surrounding MoO₃/Gr. Subsequently, Cr/Pd/Au (2 nm/30 nm/40 nm) was deposited on MoO₃/Gr using an e-beam evaporator. The electrical measurements of the devices were performed using a parameter analyzer (Keithley 2400) under ambient conditions. The two-probe field-effect mobility (μ_FE) of MoO₃/Gr was calculated using the following equation

$${\mathrm{\mu }}_{\mathrm{FE}}=g_{\mathrm{m}}\frac{L}{W{V}_{\mathrm{DS}}C}$$ (2)

where c is the unit back-gate capacitance of 285-nm SiO₂, g_m = dI_DS/dV_G is the transconductance, V_DS is the drain voltage, and L and W are the channel length and width, respectively. The channel width is defined as the width of the metal electrodes. The transconductance was obtained by linearly fitting the transfer curve.

DFT calculations

To investigate the electronic and transport properties of the MoO₃/graphene heterostructure, we performed first-principles calculations based on DFT (58, 59) using the VASP code (60, 61). Projector-augmented wave potentials (62, 63) were used to describe the valence electrons. The cutoff energy for the plane wave basis was set to 450 eV, and atomic relaxation was performed until the Hellmann-Feynman force acting on every atom decreased below 0.01 eV Å⁻¹. Dipole correction was included for a more precise calculation. For the exchange-correlation function, the rev-vdW-DF2 method (64) was adopted to consider vdW interactions. The BZ was sampled using a 2 × 8 × 1 k-grid for the rectangular supercell of MoO₃/graphene. To avoid spurious interlayer interaction along the out-of-plane direction, a vacuum region of 15 Å was introduced.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S7

References