Hadi
Arjmandi-Tash
Kavli Institute of Nanoscience, Delft University of Technology, Lorentzweg 1, 2628 CJ Delft, The Netherlands. E-mail: h.arjmanditash@tudelft.nl
First published on 28th November 2019
Transferring graphene flakes onto hexagonal boron nitride (h-BN) has become a common approach for fabricating graphene/h-BN heterostructures. Controlling the alignment between graphene and h-BN lattices is difficult to achieve and the h-BN/graphene interface is prone to contamination in this complicated process. Direct synthesis of graphene on h-BN is a rapidly growing alternative. In situ grown graphene is individually tailored to conform to the specific h-BN flake, hence the limitations of the conventional transfer-based fabrication approach are overcome. Developed processes promise improved scalablity of the device fabrication, eventually suitable for industrial applications. The developments in the field, from inception to current status is the focus of this review. How the field is progressing to overcome existing challenges is discussed together with its future prospects.
In all these reports, high quality chemically grown8,9 (via chemical vapor deposition (CVD) process) or exfoliated graphene is first isolated on an intermediate substrate and then transferred onto the h-BN flakes. The process has disadvantages: (i) contamination of the graphene/h-BN interface by polymeric materials10,11 (used to assist the transfer) and/or air or water trapping5 is highly possible. (ii) Graphene can be damaged or wrinkled and (iii) the process does not provide any control over the relative orientation of graphene and h-BN: the random orientation of the lattices may lead to non-reproducible electronic transport properties. (iv) Macroscopic alignment of microscale graphene on h-BN flakes is another issue which is time-consuming and troublesome in practice. Techniques which are transfer-free, involving direct growth of graphene on h-BN offer a solution to overcome the above limitations. Particularly, in such techniques the graphene/h-BN interface is realized in situ and thus no external contaminant can be trapped in between.
In this review, the progress of techniques to grow graphene on thick h-BN flakes is outlined. Standard CVD of graphene benefits from the presence of a catalyst (e.g., copper) as a growth promoter; the absence of such a catalyst hinders direct growth of graphene on h-BN. Different approaches to tackling this limitation are discussed and different aspects of in situ growth methods are reviewed here. Note, there are some methods developed for growing both graphene and mono- (few-) layered h-BN together to form thin (≲10 nm) graphene/h-BN stacks or patchworks;12–18 Interestingly, the growth of the first sample of this type was reported even before the first isolation of graphene by exfoliation.19 The thickness of h-BN layers achieved in such approaches, is however, not enough to smooth the roughness of the underlying substrates and diminish the effect of the random potentials resting on the wafer; such approaches are out of the scope of this review.
The first paper about directly growth CVD graphene on h-BN was submitted for publication just four months after the first realization of graphene/h-BN stacking.4 The rapid progress implies the importance of the in situ growth approaches in the first place. Published by Ding et al.,20 this paper confirmed the possibility of chemically growing a few-layer graphene on h-BN powders. The importance of the catalyst was, however, overlooked as no remarkable measure was considered to compensate its absence. The size of the graphene domains remain unclear. Fig. 1a and b shows some of the reported results. Recently and as the field started to develop, a few approaches have been introduced to overcome the absence of the catalyst.
Fig. 1 Early reports on the direct growth of graphene on thick h-BN flakes (a) and (b) results of Ding et al.: graphene was grown on h-BN powders through a CVD process. The Raman spectra on (b) are taken from bare h-BN powder and after the growth. Reprinted from ref. 20, Copyright (2011), with permission from Elsevier. (c and d) Results of Son et al.: the effect of the growth temperature on the density of the obtained flakes was reported as AFM mappings in this work. The density raised a lot as the temperature increased from 900 °C to 1000 °C. This growth lasts for 2 hours. The inset in (d) shows an example of the Raman spectrum reported on this sample. Adapted from ref. 21 with permission of The Royal Society of Chemistry. (e–g) Results of Tang et al.: AFM measurements shown in these figures were performed on the samples with 1, 3 and 6 hours of growth. Reprinted from ref. 22, © 2012, with permission from Elsevier. |
This approach was followed later by Tang et al.22 Like Son's experiment, graphene was grown on hexagonal boron nitride flakes exfoliated on silicon wafer, albeit through a low pressure CVD process. They noticed that screw dislocations on the flakes are favorable nucleation sites. The slow growth rate due to the absence of catalyst was evident in the results reported; a growth duration of 6 hours only led to the formation of graphene grains of a maximum 270 nm in diameter (Fig. 1e–g). The graphene domains were mostly single layer.
Elongating the growth – although simple – has certain drawbacks: the reported graphene domains – even after several hours of growth – barely reached a few hundred nanometers and are incompatible with typical device fabrication processes. The long operation time and high energy consumption are unfavorable for industrial application.
Fig. 2 Various approaches to improve the growth rate (a) effect of increasing the growth pressure on the grain size: SEM images showing graphene grains synthesized on h-BN flakes at 1150 °C under various chamber pressures. Reprinted from ref. 23, © 2016, with permission from Elsevier. (b and c) Gaseous phase catalyst to improve the growth rate: (b) comparison of the experimentally measured grain size of graphene flakes grown at 1280 °C with and without gaseous catalysts. (c) Schematic illustration showing the mechanism of growing monolayer graphene onto h-BN: silicon atoms (shown as red color spheres) achieved from the decomposition of SiH4 bound to the edge of the graphene and boosting the growth. Adapted from ref. 24. (d and e) Proximity driven over growth of graphene onto h-BN flakes pre-exfoliated on the copper foil: (d) AFM mapping showing an h-BN flake (covered with graphene) on the copper foil at the end of the growth course. The color code shows the height, ranging between 0 nm and 500 nm. (e) Optical image (top) and Raman G mode position mapping (bottom) of a millimeter scale h-BN flake on the copper foil, fully covered by graphene. The color code shows the Raman frequency ranging between 1584 cm−1 and 1600 cm−1. Adapted from ref. 28. |
Fig. 3 Known mechanisms governing the growth of graphene on h-BN flakes. (a) Comparison of the epitaxial growth between materials with (left) and without (right) dangling bonds, the latter has been dubbed van der Waals epitaxy; in both cases the atoms, corresponding to the substrate and to the deposited film are shown in blue and pink, respectively. (b) AFM and (c) Raman (intensity of 2D peak of graphene) mapping of a h-BN flake with deposited graphene via molecular beam epitaxy; the mappings show that carbon atoms are freely migrated and accumulated in some preferential spots on the surface of h-BN. Reprinted from ref. 31, © 2012, with permission from Elsevier. (d) Schematic representation of the mechanism (inset) and SEM image of graphene (both mono- and bi-layer) nucleated on the copper foil and extended over the h-BN flake via proximity driven growth. Adapted from ref. 28. |
Garcia et al.31 first reported van der Waals epitaxy in in situ growth of graphene on h-BN. Unlike previous experiments, graphene was grown by molecular beam epitaxy (MBE) using solid carbon sources. Combined Raman and AFM analysis revealed that the growth is independent of the flux of carbon atoms; instead, carbon atoms deposed on the surface, migrate freely and accumulate in selective spots on the h-BN surface (Fig. 3b and c). This observation indicated that the carbon atoms are of high mobility on the neutral h-BN, confirming that van der Waals epitaxy is achieved. Recently, van der Waals epitaxy was reported in CVD growth of graphene on h-BN also.23 The lack of any covalent bond results in weak interfacial interaction between graphene and h-BN.16
A separate but complementary growth mechanism involves extending graphene – already nucleated on the copper foil – onto nearby h-BN flakes. This mechanism was first observed in graphene grown on few-layer chemically grown h-BN sheets;15 our recent work,28 however, confirmed that the thickness of the h-BN is not any limitation as graphene can overgrow on hundreds-of-nanometers-thick h-BN flakes, mechanically pre-exfoliated on the copper foil. Inset to Fig. 3d explains our hypothesized model for this growth. Precursors are cracked on the copper foil. The achieved carbon radicals move randomly in different directions and are energetic enough to continuously jump over the h-BN flake and bond as-growing graphene. The presence of the copper catalyst and high mobility of carbon atoms – as a result of the van der Waals epitaxy – guarantees a high growth rate atop h-BN flakes. The main panel of Fig. 3d shows a SEM image of a h-BN flake covered with graphene.
Controlling and minimizing the misorientation angle in graphene/h-BN heterostructures is of great importance to lower structural uncertainties. Indeed the traditional method of transferring exfoliated graphene on h-BN leads to a random orientation of the lattices. Although recent progress in the field has revealed that post-treatment of the samples at elevated temperatures can drive graphene to rotate and follow h-BN lattices,35,36 such approaches are more efficient in sub-micrometer flakes. In situ grown graphene on h-BN, however, has proven capabilities for controlling Φ in much larger samples.
Yang et al. utilized the plasma-enhanced CVD technique to grow graphene on mechanically exfoliated h-BN flakes.27 Large area, epitaxial and single crystal graphene domains directly grown on the h-BN flakes were obtained. Breaking down the methane molecules with a remote plasma source eliminated the need for a catalyst and enhanced the growth rate and the domain size. The cleanness of the flakes was sufficient that they managed to observe the moiré pattern associated with the superposition of graphene and h-BN crystals by AFM analysis (Fig. 4a and b). This analysis showed that the graphene lattice follows the orientation of the underlying h-BN. The size of the graphene was limited by the size of the h-BN flake, large enough to fabricate devices for transport experiments. The signature of the superposition of the lattices as extra Dirac points in the resistivity and quantum Hall effect measurements was revealed at low temperature (Fig. 4c and d).
Fig. 4 Crystalline biased growth of graphene on h-BN (a) moiré pattern due to the superposition of the graphene and h-BN lattices, (b) filtered inverse fast Fourier transform of the pattern is visible in the dashed square in a, height profile along the dashed line is shown in the lower part. The periodicity of the oscillations can be used to calculate the rotation angle between the lattices. Reprinted by permission from Macmillan Publishers Ltd: Nature Materials,27 © 2013. (c) Gate dependence of the resistivity measured at different temperatures: satellite peaks shown at the left and right side of the Dirac point are due to the formation of the superlattice. (d) This effect can also be seen as the pattern at the left side of the fan diagram of Rxy in the quantum hall measurement. Reprinted by permission from Macmillan Publishers Ltd: Nature Materials,27 © 2013. (e) Topography of the small graphene flakes and (f) moiré pattern associated with graphene/h-BN superposition. Reprinted with adaptations from ref. 37. The mappings shown in a, b, e and f are obtained with atomic force microscopy. |
Note that a similar alignment was reported later by Tang et al.24,37 (Fig. 4e and f) and Mishra et al.23
Report | Process | Precursor | Temperature (°C) | Duration | Size | Thickness | Orientateda | Mobility |
---|---|---|---|---|---|---|---|---|
ML: monolayer, BL: bilayer.a If graphene follows the orientation of underlying h-BN.b The flow rate of the precursor was not reported in this work. | ||||||||
Ding20 | CVD | CH4 (50–90 sccm) | 1000 | 3–8 min | Not reported | >6 L | Unclear | Not reported |
Son21 | CVD | CH4 (30–50 sccm) | 900–1000 | 2 h | 100 nm | ≈0.5 nm | Unclear | Not reported |
Tang22 | CVD | CH4 (5 sccm) | 1200 | 1–6 h | <270 nm | ML | unclear | Not reported |
Garcia31 | MBE | Solid carbon | 600–930 | 40.6 min | nm-scale | ML | Unclear | Not reported |
Yang27 | PECVD | CH4b | ≈500 | ≫3 h | μm-scale | ML & BL | Yes | ∼5000 cm2 V−1 s−1 (at 1.5 K) |
Tang37 | CVD | CH4 (5 sccm) | 1200 | 1–5 h | μm-scale | ML & BL | Yes | 20000 cm2 V−1 s−1 (at 300 K) |
Tang24 | CVD | C2H2b | 1280–1350 | 5–40 min | μm-scale | ML | Yes | 20000 cm2 V−1 s−1 (at 300 K) |
Mishra23 | Cold-wall CVD | CH4b | 1000–1150 | 30 min | μm-scale | ML | Yes | Not reported |
Arjmandi-Tash28 | CVD | CH4 (5 sccm) | 1050 | 90 s | mm-scale | ML | Unclear | 20000 cm2 V−1 s−1 (at 80 K) |
Plaut38 | MBE | Solid carbon | 500–1000 | 65 min | μm-scale | ML | Unclear | Not reported |
The size of the resultant graphene samples and growth rate have been improved gradually over the last years. Similar improvements in controlling the thickness (number of layers) of graphene is also detectable. While the orientation of graphene in early reports was unclear, recent works have reported a trend in graphene/h-BN lattice alignment. The best mobility reported for in situ grown samples is still much inferior than that achieved in transfer-fabricated samples, even with CVD graphene5 which highlights the affect of crystalline defects. Indeed, the techniques employed to compensate for the lack of catalyst – even-though successful to preserve the growth rate – have still failed to yield crystalline qualities comparable to that of graphene grown on a catalyst.
Chemical vapor deposition (CVD) has been qualified as an effective in situ growth approach. The lack of a catalyst material – otherwise existing in conventional CVD of graphene on a metallic foil – is a great hindrance in the development of the process. In-fact early attempts compensated the slow chemical reaction by considerably elongating the growth up to hours and days. The new development in which the catalyst material indirectly promotes the graphene growth on h-BN28 is a breakthrough as the full coverage of graphene on millimeter scale h-BN flakes can now be achieved at rates identical to those on a catalyst. The electronic transport characteristics of graphene achieved by in situ growth approaches are still inferior than the devices achieved in transfer-fabricated methods. In fact, while the charge carrier mobility of in situ grown graphene fails to exceed 20000 cm2 V−1 s−1, ballistic transport with tens to hundreds times higher mobilities are now commonly achieved with transfer-fabricated graphene/h-BN heterostructures.6,7 Indeed, largely found crystalline defects serve as charge carrier scattering centers and strongly suppress the carrier mobility in in situ grown graphene. Certain measures, including the optimization of the growth parameters have to be taken to improve the crystalline order of graphene in the future. The electronic band structure of van der Waals heterostructures can be customized by engineering the twist angle between crystals (moiré superlattice potential). In-fact precise control of the twist angle achieves insulating, semi-metallic and superconducting states in such heterostructures.40,41 Herein, the in situ growth approaches offer great potential as the orientation of the growing graphene is governed by the crystalline orientation of the background h-BN. Sandwiching graphene in between two h-BN flakes provides the optimum device quality; as yet in situ growth methods have failed to yield such structures. Additionally, direct growth of van der Waals hetero-structures with multiple two-dimensional materials is on the horizon. The choice of CVD for the in situ growth of graphene on h-BN is largely motivated by the proven qualification of the method in growing graphene on copper. Unlike on copper, the growth is not self-limited on h-BN and does not guarantee a uniform monolayer film thickness. Although the observation of restricted monolayer growth on an h-BN ridge in MBE38 is promising, customizing the number of layers is so far out of reach. Besides CVD and MBE, certain efforts have to be made to evaluate the efficiency of other exiting approaches (e.g., bottom-up synthesis using polycyclic aromatic hydrocarbons) or developing novel modalities in in situ growing graphene on h-BN in the future.
This journal is © The Royal Society of Chemistry 2020 |