Preparation of graphene bilayers on platinum by sequential chemical vapour deposition

TL;DR

This paper presents a cost-effective, flexible method for growing bilayer graphene on platinum using sequential chemical vapour deposition, involving thermal decomposition, platinum film deposition, and annealing to achieve bilayer stacking.

Contribution

It introduces a novel sequential CVD process for epitaxial bilayer graphene growth on platinum, enabling large-area bilayer formation with detailed microscopic analysis.

Findings

Successful growth of bilayer graphene confirmed by microscopy

Moiré patterns explained by stacking arrangements

Method is scalable and cost-effective

Abstract

A cheap and flexible method is introduced that enables the epitaxial growth of bilayer graphene on Pt(111) by sequential chemical vapour deposition. Extended regions of two stacked graphene sheets are obtained by, first, the thermal decomposition of ethylene and the subsequent formation of graphene. In the second step, a sufficiently thick Pt film buries the first graphene layer and acts as a platform for the fabrication of the second graphene layer in the third step. A final annealing process then leads to the diffusion of the first graphene sheet to the surface until the bilayer stacking with the second sheet is accomplished. Scanning tunnelling microscopy unravels the successful growth of bilayer graphene and elucidates the origin of moir\'e patterns.

Full text

1 Introduction

Along with the ongoing interest in graphene and its derivatives comes the demand for reliable and scalable preparation methods for the eventual application in devices. For graphene-based layered systems such as bilayer graphene (BLG), whose properties depend strongly on stacking and interlayer twist angles, 1, 2 the epitaxial growth directly on the sample yielded high structural quality.

SiC substrates 3, 4, 5 and transition metal surfaces 6, 7 have widely been used for the preparation of monolayer graphene (MLG) as well as multilayer stackings. For the fabrication of closed monolayers on metals, the thermal decomposition of hydrocarbons assisted by the catalytically active surface is typically performed as a chemical vapour deposition (CVD). 8 The growth is self-limiting to one monolayer and, thus, offers precise thickness control. Additionally, this technique yields high-quality sheets with only a few rotational domains depending on substrate and growth temperature 7, 9, 10 and may readily be applied in industrial production. However, due to the self-limitation, its use for the preparation of multilayers is not straightforward. Instead, the growth of MLG and multilayer graphene on metals benefits from an alternative approach, namely the segregation of C at elevated temperatures. To this end, the bulk metal is artificially enriched with C, either via doping from the gas phase of hydrocarbons 6, 7, 11, 12, 13 or solid-state diffusion. 14, 15, 16, 17 Alternatively, high-quality multilayer graphene on metals may also be achieved by intercalation of C under MLG, which requires sources of atomic C. 18, 19, 20

A preparation protocol that enables the growth of homogeneous as well as heterogeneous stackings of two-dimensional materials in a controlled surface science approach would be highly desirable. Ideally, a CVD-based layer-by-layer preparation would combine the synthesis of large-scale graphene sheets with high structural quality and with precise thickness control owing to the self-limitation to one monolayer per growth cycle.

Here, we present a sequential CVD method that achieves the high-quality preparation of extended BLG regions on Pt(111). In a four-step process, MLG is first prepared on Pt(111) via thermal decomposition of C2H4, followed by the deposition of a thick Pt film and subsequent fabrication of a second graphene layer. During the postannealing in the last step, the diffusion of buried MLG to the surface is completed and leads to the formation of extended BLG domains. This last step can also be viewed as the intercalation of the deposited thick metal layer. Various materials were previously demonstrated to intercalate graphene, e. g., H, 21 Au, 22, 23 Co, 24, 25 Fe, 26 Ni, 26, 27 Ag, 28 Eu, 29 Cs, 29, 30 as well as Li. 31 Therefore, Pt is very likely to intercalate, too. A method akin to the one suggested here has recently been applied to prepare MLG flakes on Au(111). 23 Since Au does not offer the convenient CVD synthesis, MLG flakes grown by CVD on Ir(111) were subsequently intercalated by several monolayers of Au to achieve graphene flakes on top of a Au(111) surface. All prepared graphene layers are thoroughly characterised by atomically resolved scanning tunnelling microscopy (STM).

2 Results and discussion

While in principle the presented preparation protocol should be applicable to Ir, Ru, Ni, 6, 7 single-crystalline (111) surfaces of Pt have been chosen in this work since on this substrate the segregation after a CVD process is typically insufficient to already produce BLG. Moreover, graphene on Pt(111) has the appealing property of the lowest graphene–metal coupling. 32

Figure 1 illustrates the preparation process by STM images (Fig. 1a–d) and sketches (Fig. 1e–h). First, MLG results from epitaxial growth of thermally decomposed C2H4 (Fig. 1a,e). In most regions, moiré patterns due to the lattice mismatch between graphene and Pt(111) were absent in STM images, which hints at large rotation angles of graphene with respect to Pt(111). 33, 34 Occasionally, moiré superstructures were observed and indicated smaller rotation angles.35 Second, approximately $45$ atomic layers of Pt were deposited on MLG-covered Pt(111).

STM images show resulting Pt clusters (Fig. 1b,f). In the third step, the second layer of graphene was grown by thermal decomposition of C2H4 on the Pt film. The required annealing of the sample at $1000\,\text{K}$ led to the incomplete diffusion of the first graphene layer to the surface giving rise to large MLG and many small BLG regions (Fig. 1c,g). Further evidence that the buried graphene sheet forms the bottom layer of BLG in this preparation method is provided in the SI. The intercalation of Pt under MLG due to the high temperature during the CVD possibly breaks the MLG apart. Indeed, domain boundaries of MLG assist the intercalation process 36, 30 and may likewise facilitate the rupture of MLG. 37 Individual patches can then diffuse separately to the surface (Fig. 1g) as observed during the preparation of graphene nanoflakes on Au(111)/Ir(111). 23 Additionally, individual C atoms may detach from the boundary of buried graphene layers, segregate and contribute to the graphene growth at the surface. 14, 15, 16, 17 The third step clearly demonstrates that a sufficiently thick Pt film is crucial to avoid completion of the Pt intercalation before the actual growth of the second graphene layer can start. Subsequent annealing of the sample yielded flat and extended MLG and BLG regions (Fig. 1d,h). Second-layer graphene flakes hybridise to form extended BLG regions, while buried graphene patches continue their diffusion to the surface and may either be incorporated in the bottom graphene layer or create additional BLG or even trilayer graphene (TLG) regions.

Evidence for the proposed growth is given by the analysis of spatial periods ($\delta$) and the orientations of the moiré pattern ($\varphi$) and of the observed graphene lattice ($\vartheta$) with respect to Pt(111). The details of the analysis are summarised in the SI. Moiré domains that do not match the expected characteristics of MLG are attributed to BLG.

Figure 2 summarises this analysis. MLG after the first step of the preparation protocol (Fig. 2a,b) occasionally exhibits moiré patterns with, e.g., $\delta=2.37\,\text{nm}$, $\vartheta=4.8^{\circ}$ (Fig. 2a) and $\delta=0.82\,\text{nm}$, $\vartheta=16.7^{\circ}$ (Fig. 2b). After the complete preparation moiré patterns of MLG with $\delta=0.79\,\text{nm}$, $\vartheta=-15.8^{\circ}$ (Fig. 2c) are observed. The dark depressions, which most clearly appear in the STM image of Fig. 2c, may be due to C atoms in the Pt surface 17 or vacancies. 38 Figure 2d shows an STM image where BLG and TLG are present as adjacent regions. The dashed line marks the boundary between the different domains. All moiré spatial periods and orientations (symbols) together with the expected trends for MLG (solid and dashed lines) are summarised in Fig. 2e. MLG data (triangles, squares) are reasonably well captured by the expected variation of $\delta$ with $\vartheta$ for unstrained MLG (Eq. S2). The dashed lines depict $\delta$ as a function of $\vartheta$ for a stretched (upper curve) and compressed (lower curve) C lattice, where the lattice constant deviates by $2\,\%$ from the unstrained case. Deviations up to $7\,\%$ were reported for stable configurations of MLG on Pt(111). 35 Figure 2f defines the angles enclosed by crystallographic directions of the moiré superlattice and graphene with respect to Pt(111).

Some data, however, deviate significantly from the expected variation; that is, in an MLG model the observed moiré spatial period is not compatible with the observable $\vartheta$. The corresponding domains are only found after the second CVD process and are therefore assigned to BLG. The gallery of STM images in Fig. 3 further corroborates this assignment. Figure 3a shows an extended BLG region that contains both moiré patterns, BLGα ($\delta=1.6\,\text{nm}$, $\varphi=-28.8^{\circ}$) and BLGβ ($\delta=2.7\,\text{nm}$, $\varphi=0.8^{\circ}$). BLGα and BLGβ each exhibit two subdomains with identical $\delta$ and $\varphi$ but different $\vartheta$. Examplarily, this is demonstrated in Fig. 3b,c for the BLGα domain with atomic resolution. The Fourier transforms (insets to Fig. 3b,c) reveal that the graphene lattices of the subdomains are rotated by $36^{\circ}$ with respect to each other, while the moiré pattern is unaffected. These observations unambiguously demonstrate the presence of two graphene sheets in BLGα and BLGβ domains.

Remarkably, an atomically resolved close-up view of the domain boundary between BLGα and BLGβ (Fig. 3d) reveals a unique graphene lattice covering both domains. Here, subdomains of BLGα and BLGβ with $\vartheta=18^{\circ}$ are connected. The uniformly oriented top layer of graphene demonstrates that the different moiré patterns must be due to rotational domains of the bottom graphene sheet. The STM image depicted in Fig. 3e corroborates this important aspect. It shows a BLGα area with a line defect in the top graphene sheet. The moiré pattern ($\delta=1.6\,\text{nm}$, $\varphi=-28.8^{\circ}$), however, remains unaffected across the defect as evidenced by the superimposed lattice. Therefore, the observed moiré superstructure is due to the graphene/Pt(111) interface and, intriguingly, the upper graphene layer does not contribute to the moiré pattern.

Previously, moiré lattices on BLG/Ir(111) were attributed to the graphene/graphene interface. 20 The apparent contradiction with findings reported here is resolved when both the graphene/substrate and graphene/graphene twist angles are considered. For both these interfaces, low twist angles ($<10^{\circ}$) give rise to an elevated hybridization, while large twist angles ($>10^{\circ}$) decouple the layers effectively. 39, 40, 41, 20 In BLGα and BLGβ domains the bottom graphene layer is rotated by only $4.6^{\circ}$ and $0.1^{\circ}$ with respect to the Pt(111) surface, respectively (Eq. S4), while all graphene/graphene twist angles exceed $10^{\circ}$. In contrast, most of the reported domains of BLG/Ir(111) exhibit large twist angles at both interfaces. 20 Consequently, the elevated (low) graphene-Pt (graphene-Ir) hybridization favours the observation of the moiré pattern due to the graphene/Pt (graphene/graphene) interface. The twist angles at both interfaces and the resulting interlayer coupling therefore determine which moiré superstructure is visible in STM images.

3 Conclusions

In conclusion, a preparation protocol for the preparation of extended BLG on Pt(111) has been introduced. The applicability of the method to other transition metal substrates and homogeneous as well as heterogeneous multistackings of two-dimensional materials may be anticipated. The presented atomically resolved STM data unveil that the moiré superlattices observed in BLG regions result from the graphene/Pt(111) interface.

4 Experimental methods

Pt(111) surfaces were prepared by Ar+ bombardment and annealing at $1200\,\text{K}$ in O2 atmosphere ($4\cdot 10^{-5}\,\text{Pa}$). C2H4 at a partial pressure of $4\cdot 10^{-5}\,\text{Pa}$ was used as a molecular precursor in the CVD process. After MLG growth via thermal decomposition of C2H4 at $1200\,\text{K}$, Pt was evaporated from a hot filament at a rate of $\approx 1\,\text{ML}/\text{min}$ (ML: monolayer), which was calibrated from STM images of several independent depositions on Au(100) and MLG/Pt(111). The preparation of the second graphene layer was performed by annealing the sample at $1000\,\text{K}$ for $3\,\text{min}$, directly followed by exposure to C2H4 for $4\,\text{min}$. The sample was kept at $1000\,\text{K}$ for $5\,\text{min}$ before cooling to room temperature. Additional annealing cycles were performed at $1200\,\text{K}$. The experiments were performed in ultrahigh vacuum (base pressure $1\cdot 10^{-8}\,\text{Pa}$) with a low-temperature STM operated at $6\,\text{K}$. Tips were cut from Au wire and cleaned in situ by annealing and field emission. STM images were recorded in the constant-current mode with the bias voltage applied to the sample. Topographic data were processed using the WSxM software. 42

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Financial support by the Deutsche Forschungsgemeinschaft through Grants No. KR $2912/10-1$ and KR $2912/12-1$ is acknowledged.