Preservation of pristine Bi₂Te₃ thin film topological insulator surface after ex situ mechanical removal of Te capping layer

V-VI compounds, like Bi₂Te₃, Bi₂Se₃, and Sb₂Te₃, belong to a new class of materials called three-dimensional topological insulators (TIs), which exhibit insulating characteristics inside the bulk and gapless metallic states on the surface.^1–4 The topological properties of bismuth chalcogenides have been theoretically predicted^3,5 and the conducting surface states, consisting of a single Dirac cone at the Γ point, have been experimentally confirmed by angle resolved photoelectron spectroscopy (ARPES) measurements in bulk crystals.^6–9 In these experiments fresh surfaces are obtained by cleaving the bulk crystals inside the ARPES ultra-high vacuum (UHV) chamber. To inhibit bulk conduction, extrinsic counter doping (Ca for Bi₂Se₃⁷ and Sn for Bi₂Te₃⁹) has been used to tune the Fermi level inside the band gap in a position where it crosses only the surface states, obtaining an insulating bulk behavior. However, even in UHV conditions the properties of the surface states degrade on a time scale of hours.⁹ 

Molecular beam epitaxy (MBE) has been applied to grow thin films of bismuth chalcogenides presenting high crystalline quality and well controlled properties on different substrates.¹⁰ Intrinsic conduction only through the topological surface states has been achieved on these films by carefully controlling the MBE growth kinetics.^11–13 All these experiments have been performed on thin films directly transferred after growth from the MBE system into the ARPES measurement chamber through UHV lines. Air exposure of the topological insulator crystals leads to the formation of oxide layers.¹⁴ This oxidation process degrades the TI surface, making ex situ surface analysis unfeasible, which hinders the characterization of TI materials using other techniques at different facilities, including synchrotron based end-stations. Therefore, different methods have been proposed to protect the surface of TI films for ex situ surface measurements.

One method consists of removing the oxidized layer through Ar⁺ ion sputtering followed by thermal annealing to reconstruct the sample surface.^15,16 Although this method enables ARPES measurements, the sputtering process usually leads to rough or damaged surfaces and stoichiometry can be modified during annealing due to Te or Se re-evaporation. In addition, this method is not appropriate to clean heavily oxidized film surfaces.¹⁷ 

A second technique employs the deposition of a protective Te or Se capping on the bismuth chalcogenide films just after the MBE growth. After transferring the capped sample to the ex situ UHV measuring system, passing through unavoidable air exposure, these protective cappings have to be removed to restore the film surface. Thermal desorption inside the UHV chamber is the most used procedure and consists of elevating the sample temperature until the film surface is completely exposed. Se capping is preferably used due to its low desorption temperature, but it normally changes the surface stoichiometry of Te-based TI compounds during thermal decapping.^18,19 Refinements of this technique have shown that it is possible to restore pristine surfaces after thermal removal of sputtered Se capping on Bi₂Se₃, which was exposed to air for a few hours²⁰ or thermal desorption of ultra-thin Te capping on Bi₂Te₃ films after exposition to air for 5 min.²¹ 

A third technique applied to layered van der Waals materials consists of cleaving thin films (thicker than 50 nm) inside the UHV chamber to remove the oxidized surface, exposing a new fresh surface to perform the aimed analysis.¹⁷ However, this procedure changes both film thickness and surface morphology, leading to uncontrollable parameters for surface characterization.

This work presents a technique that combines the deposition of a protective capping layer after the epitaxial growth with its mechanical removal inside the UHV chamber for ex situ surface investigation. The idea is to apply the adhesive tape technique, commonly used to cleave bulk crystals, for removing the cap layers to expose fresh surfaces. This method is applied here to Te capped bismuth telluride films with varied film and cap thicknesses. The Te cap is completely removed by the adhesive tape, exposing an unaltered surface without changing the Bi₂Te₃ film thickness. Analyses performed on the exposed films prepared by the method described here demonstrate that a pristine Bi₂Te₃ surface is totally preserved, exhibiting protected topological surface states of intrinsically insulating Bi₂Te₃, as reported for as-grown films measured in situ without breaking vacuum.^11–13 It is important to emphasize that the ex situ UHV surface analyses were performed after keeping the samples at room conditions during 3 months.

Bismuth telluride epitaxial films were grown in a Riber 32P MBE system on freshly cleaved (111) BaF₂ substrates, which have a lattice mismatch of only 0.04% to Bi₂Te₃, see Figs. 1(a) and 1(b). A nominal Bi₂Te₃ solid source, which determines the growth rate, and two extra Te cells, to compensate the loss of tellurium during growth, were used. The beam equivalent pressure (BEP) of Bi₂Te₃ and Te fluxes were measured by a Bayer-Alpert ion gauge and the Te to Bi₂Te₃ BEP ratio was adjusted according to the substrate temperature to obtain single phase Bi₂Te₃ films. For the samples investigated here, the substrate temperature ranged from 240 °C to 270 °C and the BEP ratio from 1 to 3. A growth rate of approximately 1 nm/min was used and the deposition time was set to obtain Bi₂Te₃ films with thickness from 8 to 170 nm. Details about MBE growth of these films can be found elsewhere.²² To protect the films for the ex situ UHV surface analyses, Te capping layers were deposited just after growth at a substrate temperature around 30 °C with a deposition rate of 6 nm/min. The thickness of the protective cap ranged from 100 to 200 nm.

To monitor in situ the film surface during growth, reflection high-energy electron diffraction (RHEED) patterns were acquired using a 35 keV RHEED system. The RHEED analysis indicated a layer-by-layer growth mode since the early stages of bismuth telluride epitaxy on (111) BaF₂ substrates.²² Figs. 1(c) and 1(d) show streaky RHEED patterns along two symmetry azimuths in a 150 nm Bi₂Te₃ film surface just after growth, evidencing a highly ordered film. Ex situ atomic force microscopy (AFM) images were taken to evaluate surface morphology using a Veeco SPM Multimode in tapping mode. AFM image of this film, displayed in Fig. 1(e), exhibits the spiral-like triangular domains, which reflects the threefold symmetry of Fig. 1(b), with terraces steps of 1 nm in height corresponding to one quintuple layer (QL) of Bi₂Te₃.

X-ray diffraction was measured at the XRD2 beamline of the Brazilian Synchrotron Light Laboratory (LNLS) with a photon energy of 8004 eV (λ = 1.54904 Å). The Q_z-scan measured along $0001$ film direction is plotted in the graph of Fig. 1(f). Besides the $111$ and $222$ BaF₂ substrate peaks, only the $000 L Bi 2 Te 3$ Bragg peaks with L = 3n (n = integer) are observed, as expected for epitaxy with the Bi$2 Te 3 0001$ hexagonal plane parallel to the BaF$2 111$ surface. Fig. 1(f) also plots two diffraction curves calculated using a set of recursive equations²³ and fitted to the experimental data considering a pure Bi₂Te₃ and a Bi-richer Bi₂Te_2.95 phase. It evidences that a pure phase Bi₂Te₃ film is obtained. Reciprocal space maps (RSMs) measured in the vicinity of the symmetric $00018$ and the asymmetric $1 1 ̄ 020 Bi 2 Te 3$ reciprocal lattice points are shown in Figs. 1(g) and 1(h), respectively. In both RSMs, scattering along the crystal truncation rod, parallel to Q_z, and along the Ewald sphere surface (oblique streaks perpendicular to the diffracted wavevectors) are visible. The low scattering in the in-plane direction Q_x on the RSM around L = 18 peak indicates a film with low mosaicity and a small density of structural defects, demonstrating that the Bi₂Te₃ thin films grown here are of very high structural quality. In Fig. 1(h), the asymmetrical $1 1 ̄ 020 Bi 2 Te 3$ epilayer and $313 BaF 2$ substrate reciprocal lattice points are well aligned along the crystal truncation rods, i.e., their maxima are in the same Q_x position, indicating that the epilayer hexagonal lattice is perfectly matched to the substrate cubic lattice.

The Te capped samples and the decapping procedure are investigated ex situ at room conditions by cross section images acquired using a TESCAN MIRA3 field-emission gun (FEG) microscope, AFM surface images, and x-ray reflectivity curves measured in a PANalytical X’Pert MRD high-resolution x-ray diffractometer. Fig. 2(a) shows a FEG microscopy cross section image of a 25 nm thick Bi₂Te₃ film capped with a 100 nm Te layer. To illustrate the method proposed here, a schematic diagram of the mechanical removal technique using adhesive tape is depicted in Fig. 2(b). The sample is glued to a holder and an adhesive tape is attached to the cap surface. The capping layer can be removed either by pulling the tape or by fixing the adhesive tape and moving the holder. Fig. 2(c) exhibits the AFM surface image of the 25 nm Bi₂Te₃ film just after removing the Te cap layer. For comparison, the surface of a non-capped Bi₂Te₃ film grown at the same conditions and with the same thickness is shown in the AFM image of Fig. 2(d). These images demonstrate that the film surface is completely preserved after the cap removal, exhibiting the same terraced triangular domains.

Fig. 2(e) shows the x-ray reflectivity curves of a 50 nm Bi₂Te₃ film covered with 100 nm of Te protective cap before (bottom curve) and after (top curve) decapping procedure. The reflectivity curve before decapping exhibits interference fringes from the cap layer at lower angles followed by fringes due to the Bi₂Te₃ film at higher angles. After cap removal, only the interference fringes from the Bi₂Te₃ film are visible. The red solid lines correspond to the calculated curves that best fitted to the experimental data. The thickness (t) and roughness (σ) values obtained by fitting indicate that the film thickness and flatness are totally preserved by the cap layer removing method described here. Since the films surface are terminated in complete QLs and the protective cap is grown at room temperature with high deposition rates, no ionic bonds between the Te cap and the film surface are expected. The weak coupling between the uppermost Te¹ monolayer of the QL and the Te atoms from the capping facilitates the removal of the protective cap exactly at the film/cap interface.

To characterize the exposed film surface after removing the Te cap layer in UHV conditions, an integrated UHV surface analysis system composed of low-energy electron diffraction (LEED) from Omicron, x-ray photoelectron spectroscopy (XPS), and ARPES facilities was used. The photoelectron experiments were performed at room temperature with a Scienta R4000 hemispherical analyzer using He I_α radiation (21.2 eV) with a resolution of 15 meV for ARPES and the Al K_α emission line (1486.7 eV) for XPS measurements.

The protective cappings were removed by fixing the adhesive tape to the wall of the load-lock chamber (pressure ∼5 × 10⁻⁸ Torr) and transferring the sample holder to the measurement chamber (pressure ∼5 × 10⁻¹¹ Torr). Figs. 3(a) and 3(b) show the LEED images of a 150 nm Bi₂Te₃ film acquired at 50 and 67 eV, respectively. Bright spots forming the hexagonal pattern characteristic of the crystal surface threefold symmetry are clearly visible in both images, demonstrating that an intact Bi₂Te₃ surface is completely exposed. The XPS spectrum of the freshly exposed Bi₂Te₃ film surface presents only the peaks corresponding to Te and Bi core levels and their plasmon satellites, indicating the absence of contaminants on the surface after Te capping removal. In particular, even at higher emission angles there is no signal from the Te capping which appears at characteristic binding energies of 585 and 575 eV for Te 3d_3/2 and 3d_5/2, respectively. The Te 3d and Bi 4f core levels XPS peaks, shown in Fig. 3(c), are very narrow and symmetric, evidencing no excess of Te or Bi.

ARPES spectrum in the vicinity of $Γ ̄$ point of a Bi₂Te₃ fresh surface recorded just after the Te cap removal is shown in Fig. 4(a). In addition to the M-shaped bulk valence band (BVB), the linear dispersion of the topological surface states (TSS) is clearly visible, evidencing the massless Dirac fermions.⁹ In this ARPES spectrum, the Fermi level lies inside the band gap intersecting only the TSS, which is an indication of a bulk insulating film. The Dirac point is located 205 meV below the Fermi level and the obtained Fermi velocity is 2.2 eV ⋅ Å (3.3 × 10⁵ m/s), which is very close to the values obtained for an intrinsic Bi₂Te₃ topological insulating thin film.¹³ These results demonstrate that the technique presented here is very effective to protect the surface of topological insulators from contamination during air exposure, preserving the TSS. Fig. 4(b) shows the ARPES spectrum measured after keeping this sample at UHV environment during 80 h. After this period, the Dirac point shifted downwards to 230 meV below the Fermi level, indicating that the bulk conduction band starts to be populated due to the residual gas atoms adsorbed on the surface. Fig. 4(c) displays the constant energy contour maps at different binding energies from the Fermi level to the BVB for this Bi₂Te₃ epitaxial film measured 24 h after decapping. The Fermi surface, visible in the uppermost layer of the map, exhibits the TSS hexagonal warping characteristic of the Bi₂Te₃ material. This warping becomes less pronounced as the binding energy increases passing along the Dirac cone (represented by red lines) in the direction of the BVB, and for binding energies higher than 100 meV, the BVB contribution starts to be visible.

In summary, the mechanical removal of Te capping layers on Bi₂Te₃ films by using adhesive tape proved to be a simple and robust technique to expose pristine surfaces for ex situ analyses. In contrast to other methods, this technique preserves film thickness, surface morphology, and stoichiometry, and does not require sputtering and/or thermal annealing facilities inside ex situ UHV analysis systems. It was possible to observe topological surface states of intrinsically bulk insulating films when applying this capping removal method to Bi₂Te₃ films grown at optimized conditions. These results suggest that the technique proposed here can be applied to other topological insulator materials with different protective layers.

The authors wish to thank Sebastian Fiedler for valuable discussions. The authors acknowledge CNPq (Grant Nos. 142191/2014-0, 302134/2014-0, 307933/2013-0, and 306982/2012-9), FAPESP (Grant No. 2014/04150-0), DFG via SFB1170 (Project No. A01) for financial support, and LNLS (Proposal No. XRD2-20150037) for measurement assistance.