Ultrathin epitaxial Bi film growth on 2D HfTe2 template

Among ultrathin monoelemental two-dimensional (2D) materials, bismuthene, the single layer of heavier group-VΑ element bismuth (Bi), has been predicted to have large non trivial gap. Here, we demonstrate the growth of Bi films by molecular beam epitaxy on 2D-HfTe2 template. At the initial stage of Bi deposition (1–2 bilayers, BL), both the pseudocubic Bi(110) and the hexagonal Bi(111) phases are formed. When reaching 3 BL Bi, a transformation to pure hexagonal Bi(111) occurs. The electronic band structure of 3 BL Bi(111) films was measured by angle-resolved photoemission spectroscopy showing very good matching with the density functional theory band structure calculations of 3 BL free standing Bi(111). The grown Bi(111) thin film was capped with a protective Al2O3 layer and its stability under ambient conditions, necessary for practical applications and device fabrication, was confirmed by x-ray photoelectron spectroscopy and Raman spectroscopy.


Introduction
Two-dimensional (2D) topological insulators (TIs) are solid state materials that have both an insulating electronic structure in the bulk and dissipationless conducting channels along the edges that are protected against back-scattering by timereversal symmetry [1]. The TIs are closely related to the quantum spin Hall state, which also has unique edge states, since in these materials the spin-orbit coupling (SOC) takes the role of the applied magnetic field. In TIs, the SOC is so strong that the spin is 'locked' perpendicular to the momentum yielding boundary metallic states robust against disorder, scattering or any other excitation [2].
Following the first experimental realization of graphene, other ultrathin monoelemental 2D materials with novel electronic properties have been explored over the last years [3]. In particular, the 2D Xenes [4], where X stands for the heavier elements of group IV, single-element sp 2 -hybridized silicene [5], germanene [6], stanene [7,8] and phosphorene [9] were successfully synthesized [10], while the application of silicene was demonstrated in field-effect transistors operating at room temperature (RT) [11]. Within the framework of topological matter, particular attention should also be given to the heavier group-VΑ element bismuth (Bi), since ultrathin Bi films exhibit a number of interesting properties such as strong SOC, high carrier mobility at RT and large Fermi wavelength [12]. These properties constitute Bi films as possible candidates for electronic and spintronic applications.
On the other hand, the PC-Bi on the so-called distorted A17 phase adopts the black phosphorous like (BP-like) structure and crystallizes in a rectangular unit cell consisting of two paired layers with an atom near the middle of the unit cell [13-16, 23, 24]. Contrary to the hex-phase, the topological properties of the Bi(110) edge states depend on in-plane or out-of plane strain [24] and therefore on the interactions between the Bi films and the substrates.
It has been shown both theoretically and experimentally that the PC phase transforms into the hex Bi phase above a critical thickness [23,24]. When grown on a Si substrate Bi(110) is stable up to 4 ML (2 BL), whereas a critical thickness up to 60 ML is observed in the case of an Au substrate [25].
Both Bi(111) and Bi(110) structures are of high importance because their edges may support topologically protected edge states [13,26].
The 'fusion' between non-trivial topological order and 2D crystals could radically change nanoelectronics bringing spin into play in very unique ways, paving the way to revolutionary exploitation of dissipationless spin channels for both electronics and spintronics.
Our goal was to grow in situ ultrathin Bi films on 2D materials, in order to achieve 2D layered heterostructures [27,28]. Our team has experience in growing tellurium based transition metal dichalcogenides and given that the Bi bilayer on Bi 2 Te 3 has been established [20,21] we chose the HfTe 2 semimetal [29] as possible substrate, aiming to the formation of Bi/HfTe 2 van der Waals heterostructures. Therefore, in this work, we grew ultrathin Bi films on InAs(111)/HfTe 2 templates by molecular beam epitaxy (MBE) and measured the atomic and electronic structures by in situ reflection high energy electron diffraction (RHEED), x-ray photoelectron spectroscopy (XPS), angle-resolved photoemission spectroscopy (ARPES), scanning tunneling microscopy (STM) and Raman techniques. Addittionally we performed density functional theory (DFT) calculations of the band structure of bismuthene, including the effect of SOC, in order to further analyze and interpret our results.
Ηigh quality epitaxial growth is expected for Bi on top of the 2D-HfTe 2 substrate, in agreement with the van der Waals heteroepitaxial nature of HfTe 2 material [29].

Experimental
Thin Bi films were prepared by MBE in an ultrahigh-vacuum chamber at RT. High-purity Bi (99.999%) was evaporated by Knudsen cells with deposition rate of ∼0.1 Å min −1 . Prior to Bi growth, Si(111)/InAs(111) substrates were cleaned according to the procedure described in detail in [26], whereas thin (2 ML) HfTe 2 semimetallic films were deposited at 400°C as a template [29,30]. In-situ lab-based ARPES measurements were performed using a SPECS PHOIBOS 100 hemispherical analyzer equipped with a 2D CCD detector and He I radiation at 21.22 eV, with total energy resolution of 100 meV (due to thermal broadening at RT). In-situ XPS was performed with excitation by Mg Kα radiation (1253.6 eV) using a SPECS XR50 source, at a take-off angle of 52°. Insitu STM measurements were carried out at RT, with an Omicron LS-STM using a Pt-Ir tip. STM images processing was made with the SPIP software. For ex situ measurements, Bi on HfTe 2 was a capped by 2.5 nm of amorphous Al 2 O 3 layer. Ex-situ Raman spectroscopy was performed with a Renishaw InVia Raman microscope equipped with 514 nm Ar+laser. The DFT calculations were performed using the Vienna Ab Initio Simulation Package [31,32] and projector augmented waves [33]. The generalized-gradient approximation with Perdew-Burke-Ernzerhof [34] parametrization was used as the exchange correlation functional. The kinetic energy cutoff was set at 500 eV, using the Monkhorst-Pack scheme [35] employing a 9×9×1 k-point mesh. Also, a vacuum of 15 Å was used to avoid interaction between the periodically repeated layers. The atomic positions were fully optimized by the conjugate gradients method, using a force threshold of 10 −3 eV Å −1 . SOC was included in the band structure calculations. In this paper, we adopt the nomenclate of Yaginuma et al [16] and Sadowski et al [36] where the 4 MLs/or 2 BLs of BP-like pseudocubic (PC) correspond to 13.1 Å height and the 3 BLs of hexagonal Bi structures correspond to 11.8 Å (height of 1 BL Bi(111) ∼ 3.93 Å), as shown in figure 2. Thus, the thickness of 6.5 Å corresponds to 1 BL Bi(110) (figure 2). Using the HfTe 2 RHEED pattern as a reference and knowing that the lattice constant of hexagonal epitaxial HfTe 2 is 4.0 Å [29], the lattice constant of the tetragonal Bi(110) is found 4.58 Å which is very close to the reported value of the PC Bi(110) paired layer structure [13][14][15].

Results and discussion
Additional faint streaks are shown for 6.5 Å of Bi ( figure 1(c)), which indicate also, the presence of the hexagonal Bi(111) phase. Upon extra deposition of Bi, at total thickness of 12 Å, the Bi(111) phase dominates with intense streaks in RHEED and the Bi(110) streaks faint away (figures 1(d), (h)). The lattice constant of hexagonal Bi(111) on HfTe 2 template was found 4.55 Å in good agreement with the lattice reported for thicker Bi(111) films on Si(111)-(7×7) [13][14][15]. The thickness of 12 Å corresponds to 3 BL of Bi(111) which is also verified by STM measurements ( figure 3).
In-situ RT STM measurements were carried out to the Bi films grown on HfTe 2 as shown in figure 3. At 6.5 Å of Bi deposition ( figure 3(a)) atomically flat Bi islands with triangular and rectangular shape are formed. The large triangular areas consist of 2 layers with a step of ∼(0.40±0.05) nm (line profiles A and B) while the respective atomic resolution STM image indicate the hexagonal β-Bi phase ( figure 3(b)). At the rectangular islands, the step height is (0.67±0.07) nm (line profile C) which corresponds to one layer of α-Bi. The atomic resolution images on the rectangular islands indicate the presence of both the BP-distorted α-Bi and the BP α-Bi phases ( figure 3(b)). According to the literature [13,18,24] the BP-distorted α-Bi is the most common tetragonal phase where the top two Bi atoms present a bucking>0.1 Å with trivial topological properties. On the other hand, in the BP α-Bi phase the top two Bi atoms have bucking<0.1 Å and present non-trivial topological properties [24]. Upon increasing the Bi thickness, at 12 Å ( figure 3(c)), the film morphology becomes very smooth and the film transforms to the hexagonal β-Bi phase (verified also by RHEED). The line profile D corresponds to 3 BL with total thickness of ∼(11.80±0.08) Å. Therefore, the hexagonal Bi(111) film is successfully formed at 3 BL Bi film on 2D HfTe 2 template.

Electronic band structure
The Bi electronic band structure of (1-2) BL (6.5 Å) and 3 BL (12 Å) Bi films on HfTe 2 template was imaged by in situ ARPES (He I) at RT. The energy dispersion diagrams are along the Γ-Κ-Μ direction of Bi(111). For (1-2) BL Bi (figure 4(b)) due to the low Bi thickness, HfTe 2 bands are shown around Γ-point at E Binding ∼ 0.5 eV and higher and in the region of k x ∼ (1.2-1.5 Å −1 ) for E Binding ∼ 1 eV and higher. The unresolved signal at k x ∼ 1 Å −1 is attributed to pseudocubic Bi(110) phase [15] while the bands near Fermi level at Γ and near Μ-point of Bi(111) are Bi(111) bands. The weak pseudocubic Bi(110) signal is more pronounced in the second derivative image (figure S1 (available online at stacks. iop.org/NANO/33/015701/mmedia)). Thus, the Fermi surface of 6.5 Å film is similar to that of pure hexagonal Bi ( figure 4(a)), although the respective RHEED data show mainly pseudocubic phase streaks with faint hexagonal. ARPES data of 6.5 Å film verify the findings of RHEED and STM, where the pseudocubic and hexagonal phases co-exist in the first stages of Bi growth. At 3 BL (12 Å), the Fermi surface (figure 4(c)) and the band structure (figure 4(d)) is clearly of Bi(111) phase in accordance to the structural RHEED and STM measurements, indicating the transition to pure hexagonal Bi. In addition, the electronic band structure of 3 BL free standing Bi(111) calculated by DFT along the K-Γ-Κ direction of the Brillouin zone shows a very good agreement with the measured ARPES data of 3 BL Bi(111) (12 Å) on HfTe 2 , as presented in (figure 4(f)).
When increasing the number of Bi(111) layers in the DFT calculations, additional bands appear in the region below Γ-point. These are not observed in our experimental data, therefore this is a further indication of the agreement of the experimental results with theory.
In the literature, the most studied case of epitaxial bismuth growth is that of Bi on Si(111)-(7×7) substrate, where the PC Bi phase is more stable than the hexagonal Bi phase for thickness <4 BL, while the critical thickness for the transformation from tetragonal to pure hexagonal phase occurs above 6 BL of Bi [15,20]. The photoemission studies on Si(111)-(7×7) [15] demonstrated the coexistence of PC and hexagonal Bi over a wide range of film thicknesses. In particular, it has been shown that the two phases have different topologies. The PC phase exhibits a surface band at Γ near the Fermi level, whereas the hexagonal phase shows a Dirac-like sub-band around M. The critical thickness for the  transformation from tetragonal to pure hexagonal phase occurs above 6 BL of Bi. Here, for Bi growth on 2D HfTe 2 template the transformation to pure hexagonal Bi(111) phase occurs at 12 Å thickness resulting in 3 BL Bi(111) film.
In order to clarify the effect of HfTe 2 substrate on the critical thickness of the structural transformation we calculated the total energy per atom [23] for both PC and hexagonal phases (figure S2), using the experimental lattice constants obtained by RHEED. At single BL the PC Bi(110) phase is more stable than the hexagonal Bi(111) phase. At 2 BLs the Bi(111) becomes more stable which is in close agreement with our observations that the two phase co-exist and at 3 BLs the Bi(111) becomes more stable than PC Bi(110). The lattice constant of Bi(111) on HfTe 2 is 1.54% larger than the respective lattice of Bi(111) on Si(111)-(7×7). This difference in the lattice constant seems to favor the Bi(111) stability.
The above results are very interesting since theoretical work on 3 BLs Bi(111) films [22] report the nontrivial Z 2 value of 1 which remains unchanged independent of the interbilayer distance of Bi films, concluding that the 3 BLs Bi(111) films are possible 2D-TI candidates.

Protective capping for ambient exposure
At the low limit of atomic layer number of Xenes, such as silicene, germanene, stanene and bismuthene, the films need to be protected from possible destructive reactivity under ambient conditions. The protective layer is mandatory for the incorporation of Xenes films in a device configuration for exploiting the devices electrical performance or other applications. The in situ growth of amorphous Al 2 O 3 was successfully used in the past to encapsulate silicene on Ag(111) [37] and germanene on Ag(111)/AlN [6]. Under that framework, the 3 BL of hexagonal Bi on HfTe 2 films, were capped by 2.5 nm Al 2 O 3 with MBE growth of Al in an oxygen overpressure of 10 -5 mbar. The chemical stability of bismuthene layers, was assessed by ex situ Raman (figure 5(a)) and XPS measurements ( figure 5(b)) after one-week exposure in ambient air. In the Raman spectrum, the characteristic E g and A 1g modes of Bi are present at 75.2 cm −1 and 100.2 cm −1 , respectively [38][39][40], while the peak at ∼105.5 cm −1 match in plane mode of HfTe 2 layers [41] and the peaks at (200-250) cm −1 are from the substrate. Raman modes attributed to Bi oxidation were not observed [39,40]. The XPS spectra of Bi 4f core levels are found at binding energies of 156.87 eV and 162.15 eV for Bi 4f 7/2 and Bi 4f 5/2 , respectively, which are characteristic for Bi-Bi bonds. The XPS spectra do not show any signal of oxidation of Bi-Bi bonds after the capping with Al 2 O 3 layer. Additionally, the quality of Bi film on HfTe 2 after the air exposure is preserved, thus making Al 2 O 3 a suitable capping for the ultrathin Bi(111) films.

Conclusions
Ultrathin Bi films were epitaxial grown on top of the 2D-HfTe 2 template by MBE. At the first stage of Bi growth, for (1-2) BL (6.5 Å) atomically flat Bi islands with triangular and rectangular shapes are formed with the presence of both the BP-distorted α-Bi and the BP α-Bi phases on the rectangular islands and the hexagonal β-Bi phase on the triangular islands, respectively. The coexistence of PC Bi (100) and hexagonal Bi(111) phases for 1-2 BL of Bi is also evident in the ARPES data of the electronic band structure. The transformation to the pure hexagonal phase was already observed for the thickness of 12 Å, resulting in 3 BL Bi(111)  on 2D-HfTe 2 , with the electronic band structure of 3 BL Bi(111) on 2D-HfTe 2 in agreement with the theoretical calculations of 3 BLs free standing Bi(111). The deposition of Al 2 O 3 capping layer on top of the ultrathin 3 BL Bi layers, seems to prevent their oxidation, which would allow potential application of ultrathin Bi layers for electronic devices such as spintronics and energy applications.