Graphene shows promising applications in electronic, photonic and mechanical devices due to its excellent physical and chemical properties [1
]. Usually, graphene overlayers need to be transferred onto SiO2
surfaces for device applications, while performances of the graphene-based devices are often limited by the SiO2
surfaces such as surface roughness and charging effect [5
]. Hence, it is urgent to find suitable supports to replace SiO2
. Hexagonal boron nitride (h-BN) has a similar structure to graphene, while it is an insulator with a bandgap of 5.8 eV, having desirable chemical inertness and mechanical strength [8
]. h-BN can assemble with graphene to form h-BN/graphene heterostructures. Integration of h-BN and graphene in a vertically stacked way may achieve high-performance graphene-based devices, in which the atomically flat and dangling-bond-free h-BN surface enables them to avoid the charge trapping at the interfaces [10
Construction of well-defined h-BN/graphene heterostructures becomes critical for future potential applications. Chemical vapor deposition (CVD) is widely used to prepare hybrid systems. For example, growing h-BN and graphene in sequence on Cu or Ni substrate has been attempted [12
]. Some researchers have reported plasma-enhanced CVD to grow graphene overlayers directly on exfoliated h-BN [17
]. In addition, the co-segregation method and temperature-triggered chemical conversion have been successfully applied to prepare the heterostructures [18
]. The synthesis of h-BN/graphene heterostructures has been extensively explored, and interface interaction in the heterostructures is worth further investigation. In situ investigations become important to understand growth processes and interface structures.
In this study, vertically stacked h-BN/graphene heterostructures on Ni substrates were prepared by combining CVD with segregation methods, which were monitored by in situ X-ray photoemission electron microscopy (XPEEM) and local X-ray photoemission electron spectrum (local XPS). It has been demonstrated that the graphene interlayer decouples the interaction between the h-BN overlayer and the metal substrate, and the decoupling effect is enhanced with the increasing graphene layer number. Additionally, there is a confinement effect of h-BN domain boundaries on the graphene growth underneath such that the monolayer graphene domains cannot extend beyond the h-BN boundaries.
Most experiments were carried out in a photoemission electron microscopy and low-energy electron microscopy (PEEM/LEEM) system from Elmitec-GmbH (Clausthal-Zellerfeld, Germany), which includes a preparation chamber and a main chamber with the base pressure of 3 × 10−10
Torr. In the LEEM mode, micro-region low electron energy diffraction (μ-LEED) can be performed to investigate surface structures from local regions. Low energy electron reflectivity (LEER) spectra, known as the intensity of elastically backscattered electron beam as a function of incident electron beam energy, can be applied to detect surface work function and the layer number of two-dimensional (2D) materials [20
]. XPEEM measurements were conducted at the XPEEM endstation (BL09U, Dreamline) of Shanghai Synchrotron Radiation Facility (SSRF) in China. Energy-filtered XPEEM image stacks were collected to construct local XPS of small regions of interest (ROI) by measuring the integrated intensity of the ROI through the stack as a function of energy. The energy resolution of the local XPS is 0.2 eV, and binding energy positions were calibrated by the Fermi edge of each sample. A lateral resolution of XPEEM images down to about 20 nm can be achieved.
Ni foils (99.3%, Goodfellow at Cambridge in England) were used as the substrates, which were cleaned by cycles of Ar+
ion (2.0 kV, 7 × 10−6
Torr Ar) sputtering and annealing in ultrahigh vacuum (UHV) to 900°C. The substrates contain a high concentration of carbon impurity at the near-surface region, acting as the carbon reservoir [21
]. Carbon species segregated to the surface forming graphene overlayers when the surface temperature was decreased to 650 °C. If the temperature was increased to 700 °C, the graphene overlayers were dissolved again. All h-BN overlayers were synthesized by exposure of the substrate surfaces to 1.0 × 10−6
Torr borazine at 850 °C. When the temperature was decreased to 650 and 600 °C successively, graphene layers were segregated at h-BN/Ni interfaces, forming the vertically stacked h-BN/graphene heterostructures.
Graphene overlayers were synthesized on the Ni substrates by surface segregation of near-surface carbon species as the temperature was decreased from 900 to 650 °C (Figure 1
a). μ-LEED patterns acquired from the forming graphene domains display a typical six-fold symmetric structure (Figure 1
b) due to the 2D hexagonal lattice [22
]. The LEER spectra (Figure 1
c) show that graphene domains have one dip at the incident energy close to 2.0 eV, suggesting the monolayer feature [23
]. All the results indicate that graphene can be prepared through the surface segregation of carbon species from near-surface regions of the Ni substrates.
The growth of h-BN overlayers was conducted by the CVD process (1.0 × 10−6
Torr borazine, 850 °C). After keeping the growth time for 30 min, a full h-BN layer was obtained (Figure 2
a). μ-LEED patterns (Figure 2
b) of h-BN overlayers show a remarkable three-fold symmetric structure owing to its intrinsic symmetry [25
]. LEER spectra indicate that h-BN overlayers have one dip at the incident energy of 2.0 eV, suggesting the monolayer feature [27
]. Similarly, previous studies have confirmed that h-BN growth on Ni substrates is a monolayer self-limited process in which borazine molecules decompose to B and N containing intermediates and successively reassemble into h-BN overlayers [29
From the surface covered by a full monolayer h-BN, graphene interlayers were segregated at the h-BN/Ni interface when decreasing the surface temperature, producing vertically stacked h-BN/graphene heterostructures. When the substrate temperature was cooled to 650 °C for several minutes, graphene domains with bright contrast appear at the h-BN/Ni interfaces, as displayed in the LEEM image (Figure 3
a). The pure h-BN zone and the graphene segregation zone are labeled as zone 1 and zone 2 in Figure 3
a, respectively. The two different zones can be easily distinguished by the PEEM image which is sensitive to the surface work function. The contrast of the graphene segregation zone (zone 2) is darker than the pure h-BN zone (zone 1), suggesting the higher work function of the graphene segregation zone (zone 2) (Figure 3
e). Besides, LEER spectra results (Figure 3
b) indicate that the difference in their work function is 0.5 eV, which agrees well with PEEM results.
Local XPS measurement was conducted to analyze the chemical state of the surface. Figure 3
c,d illustrates local C 1s and N 1s core level spectra acquired from the two zones. There are no C 1s signals from the pure h-BN zone (zone 1), while a strong C 1s peak with binding energy at 285.1 eV is seen from the graphene segregation zone (zone 2) which is from the single-layer epitaxial graphene [30
]. N 1s binding energy of zone 1 is located at 399.0 eV, which adopts a typical monolayer feature of h-BN on Ni (Figure 2
]. The N 1s binding energy of zone 2 is located at 398.6 eV, demonstrating −0.4 eV binding energy shift compared to that of zone 1. These results confirm that h-BN in zone 2 interacts with the Ni substrates less strongly than zone 1, suggesting that the h-BN/Ni interfaces are decoupled by the graphene interlayers. XPEEM images of C 1s with binding energy of 285.1 eV and N 1s with binding energy of 399.0 and 398.6 eV are shown in Figure 2
f–h, and the element distribution with the certain chemical state can be clearly observed in the field of view. Overall, we demonstrate the formation of vertical h-BN/monolayer graphene heterostructures on the Ni substrates.
For the device applications, it is necessary to grow large-area heterostructures. We find that decreasing the segregation temperature is an efficient way to increase the coverage of graphene interlayers, i.e., the vertically stacked h-BN/graphene heterostructures. Our previous studies have shown that nickel carbides form on Ni surfaces when the substrate temperature is lower than 450 °C, but the carbide overlayers decompose and transform to graphene as the temperature is higher than 500 °C [30
]. Accordingly, in order to drive more carbon species to segregate on the surface, the successive segregation temperature of 600 °C was chosen. When keeping this temperature for 30 min, graphene overlayers almost approach a full coverage (Figure 4
a). In such a case, graphene segregation zones display two different contrasts in the LEEM image, which are labeled by zone 4 and zone 5, respectively. The remaining pure h-BN zones are marked by zone 3. LEER spectra results (Figure 4
b) indicate that the work function of zone 5 is 0.4 eV higher than zone 4 and that of zone 4 is 0.5 eV higher than zone 3.
Local XPS and XPEEM characterization were applied to analyze the three representative zones. N 1s binding energy in zone 3 is located at 399.0 eV (Figure 4
d) in accordance with the pure h-BN structure. In zone 4, binding energies of N 1s and C 1s peaks are 398.6 eV (Figure 4
d) and 285.1 eV (Figure 4
c), respectively, which are corresponding to those of h-BN/monolayer-graphene heterostructures abovementioned. In zone 5, C 1s spectrum is peaked at 284.5 eV (Figure 4
c), shifting down by 0.6 eV compared with that in zone 4, which is close to that of highly oriented pyrolytic graphite (HOPG) [32
]. Therefore, graphene overlayers in zone 5 can be ascribed to multilayer structure, which further combine with the upper h-BN layer to form h-BN/multilayer graphene heterostructures [32
]. The binding energy peak of N 1s at 397.9 eV in zone 5 (Figure 4
d) shows a BE shift of −0.7 eV compared with N 1s in the h-BN/monolayer-graphene heterostructure zone, implying that the h-BN/Ni interface interaction in zone 5 is weaker than that in zone 4—that is to say, multilayer graphene overlayers enhance the decoupling effect on h-BN/Ni interfaces in comparison with monolayer graphene. In addition, XPEEM images of C 1s with binding energy of 285.1 and 284.5 eV, and N 1s with binding energy of 398.6 and 397.9 eV are shown in Figure 4
e–h, and the element distribution with the certain chemical state can be clearly displayed in the field of view.
Similar graphene segregation was performed on the Ni surface covered by sub-monolayer h-BN. First, h-BN domains were grown by exposure to 1.0 × 10−6
Torr borazine at 850 °C. The domains display a triangle shape or blade shape due to the intrinsic three-fold symmetry structure, as shown in Figure 5
]. When the temperature was decreased to 680 °C, dark contrast domains appear at the h-BN domains (Figure 5
b), indicating the formation of the segregated graphene domains. Upon decreasing the temperature to 600 °C, higher coverage graphene domains form (Figure 5
Graphene domains form at three different sites, namely, the h-BN/Ni interfaces (labeled as zone B in Figure 5
d), the h-BN domain boundaries (labeled as zone C in Figure 5
d) and the bared Ni surfaces (labeled as zone D in Figure 5
d). Hence, vertically stacked h-BN/graphene heterostructures and in-plane h-BN/graphene heterostructures are simultaneously synthesized on the surfaces, respectively. Especially for the vertical h-BN/graphene heterostructures, the graphene domains under h-BN cannot extend out of each upper h-BN domain, even after a segregation time of 153 min, as shown by Figure 5
c,d. While on the bared Ni surfaces, graphene domains grow without impedance like the zone B. The results suggest that graphene domains underneath h-BN are trapped by h-BN domain boundaries.
Carbon solubility at the near-surface region of Ni substrates plays an important role in the formation of the vertically stacked h-BN/graphene heterostructures. On the other hand, the segregation temperature is crucial to the heterostructure formation as well. Carbon atoms cannot migrate to the surface and form graphene with temperatures higher than 700 °C. When the temperature is lower than 450 °C, carbon atoms react with nickel to form the surface nickel carbide phase [30
]. Hence, temperatures between 600 and 700 °C are the most appropriate for the graphene segregation. Similar to Ni substrates, other metal substrates which can dissolve a certain amount of C such as Co and Ru can be used to achieve the vertical h-BN/graphene heterostructures by the method combining CVD and segregation under the appropriate temperature [35
]. We also reveal how the interlayer decouples the interface interaction between metal substrates and 2D materials. As shown in Scheme 1
a, the interface between h-BN and the Ni substrate is strongly decoupled with the increase in graphene layer number. Therefore, the control of the segregated graphene layers is the key step to modulate the decoupling at the h-BN/Ni interface.
Additionally, on the surface covered by sub-monolayer h-BN, segregated graphene layers are confined within each h-BN domain. The h-BN domain acts like a clap over graphene, restricting the graphene domain as shown in Scheme 1
b. Previous reports have proved that boundaries of h-BN or graphene domains are coordinatively unsaturated and thus bind more strongly with the metal substrate surface such that boundaries tightly attach to the surface and twist downward [36
]. These twisting domain boundaries combine with the substrate surfaces to form a confinement space. Accordingly, the h-BN domain boundary serves as a barrier to impede the growth of the underneath graphene, showing a boundary confinement effect. If the concentration of near-surface C species is high enough, it is possible to form the multilayer graphene, which enlarges the h-BN/Ni interface distance. The interaction between the h-BN domain boundary and Ni surfaces can be weakened such that forming graphene multilayers may extend outside the h-BN domains. Therefore, the domain boundary confinement effect is expected to be weakened as the concentration of near-surface C species increases.