Epitaxial 2D SnSe2/ 2D WSe2 van der Waals Heterostructures
SUPPORTING INFORMATION
Epitaxial 2D SnSe2/ 2D WSe2 van der Waals Heterostructures Kleopatra Emmanouil Aretouli,†,‡ Dimitra Tsoutsou,† Polychronis Tsipas,† Jose MarquezVelasco,†,§ Sigiava Aminalragia Giamini,†,‡ Nicolaos Kelaidis,† Vassilis Psycharis,† and Athanasios Dimoulas*,†
†
Institute of
Nanoscience and Nanotechnology, National Center for Scientific Research
“DEMOKRITOS”, 15310, Athens, Greece ‡
University of Athens, Department of Physics, Section of Solid State Physics, 15684 Athens,
Greece §
National Technical University of Athens, Department of Physics, 15780 Athens, Greece
AUTHOR INFORMATION Corresponding Author *[email protected]
S -1
(a)
(b)
Figure S1 Schematic representation of the structure of 1T-SnSe2 crystal: (a) side view (b) top view. The unit cell contains only one layer.
S -2
Figure S2 For thickness calibration purposes, X-Ray Reflectivity measurements were performed on a Rigaku Smart Lab diffractometer at thick (nominal 8ML) control samples SnSe2 (A1) and WSe2 (A2), grown directly on AlN/Si(111) substrates in order to avoid interference with reflection oscillations from the Bi2Se3 template. The analysis of measurements was performed using the Global Fit program1. At the early stage of analysis the Global optimization method was used and the final models were obtained using the nonlinear least squares fitting method1. In the analysis of the reflectivity curves for each sample a layer deposited on an AlN substrate it was considered. The best results were derived by considering a linear variation of the density of the top layer from the surface down to the interface with AlN layer.
Figure S3 Spectroscopic ellipsometry (SE) (FLS-300) J. A. WOOLAN) measurements for the thickness determination of thin samples A3 and A4, which are 4ML WSe2 and SnSe2 respectively, and for the thick A2 sample in order to compare XRR with SE technique.The analysis of the data was done using the CAUCHY model. In Figs.S3(a-c) the experimental amplitude ratio Ψ and phase difference Δ versus wavelength (solid lines) and the fitting curves(dashed lines) are plotted and the results are summarized in Table S1. S -3
structure
Nominal
Thickness
Thickness
thickness (nm)
XRR (nm)
Ellipsometry (nm)
A1
8ML SnSe2
4.9
3.74
-
A2
8ML WSe2
5.2
4.2
4.73
A3
4ML WSe2
2.6
-
3.0
A4
4ML SnSe2
2.45
-
2.67
Table S1 XRR and SE results for thickness determination. In the case of SnSe2, a thickness of 2.67 nm is deduced which is in good agreement with the nominal thickness of 2.45 nm (4ML). In the case of WSe2 there is a small deviation: a thickness of 3.0 nm is deduced instead of the nominal value of 2.6 nm expected based on XRR calibration from thicker samples. This discrepancy may be explained based on the fact that SE overestimates thickness with respect to XRR as has been previously reported2. To verify this in our samples, we measured the thicker sample A2 of 8ML WSe2 by both XRR and SE and we found that the latter technique yields a thickness by 0.53 nm larger than that obtained by XRR (see table S1). This indicates that the discrepancy in the measured thickness for the nominal 4ML WSe2 sample is within the uncertainty which is inherent in the measurement of small thickness by XRR and SE techniques. Therefore we estimate that both thicknesses measured by SE are close to the nominal value. It's worth noting that, the refractive indices (an) of SnSe2 and WSe2 are estimated to be 2.99 and 3.7, respectively at 632 nm, which are close to reported values for these materials3, 4.
S -4
Figure S4 Raman spectrum of SnSe2/WSe2/Bi2Se3 heterostructure. The Bi2Se3 buffer layer gives a very strong Raman spectrum overlapping with the Raman peaks of SnSe2 and WSe2, eventually masking them.
Figure S5 XRD scans recorded for the thick (8 ML) WSe2 and SnSe2 layers grown directly on AlN substrates which were grown for the purpose of thickness estimation by XRR. XRD and XRR on WSe2 and SnSe2 layers grown on Bi2Se3 template suffer from complications due to the presence of Bi2Se3 related peaks and interference oscillations. Therefore we can only present the XRD patterns recorded for the 8 ML WSe2/AlN and 8 ML SnSe2/AlN structures, as shown S -5
below. In both cases, sharp diffraction peaks originating from the AlN/Si(111) substrate are observed, as expected. Despite the inferior quality of WSe2 and SnSe2 films on AlN as compared to those grown on Bi2Se3 (see related discussion in RHEED of Fig. 2), well resolved peaks, especially for WSe2 are observed such as (001),(002) and higher order diffraction indicating that [001] WSe2(SnSe2)//[0001]AlN along the growth direction.
Figure S6 RHEED patterns along [10-10] azimuth of the 4ML WSe2(SnSe2)/ 5QL Bi2Se3/AlN Si(111) samples. The patterns are streaky at both azimuths so there is no evidence of 3D island formation. Nevertheless, the RHEED patterns of SnSe2 although they are streaky in general, their intensity is not fully uniform along the streaks which may indicate 2D island formation in compatibility with the morphology imaged by STM.
S -6
(a)
(b)
(c)
(d)
(e)
S -7
Figure S7 SEM images showing the surface of SnSe2/WSe2/Bi2Se3 heterostructures, with areas (a) 330x330μm2, as recorded from the corner of the sample showing and AlN substrate. This area is masked by the sample holder and remains uncovered. (b) 1500x1500μm2, (c) 270x270μm2, (d) 130x130μm2 and (e) 55x55μm2as were recorded from the middle of the sample. It can be seenthat the films are continuous over large areas on the substrate.
S -8
Figure S8 DFT band structure calculations of (a) 4 ML 2H-SnSe2 with D6h (P63/mmc) symmetry and (b) 1ML orthorhombic SnSe (Pnma) phase free-standing slab with spin orbit coupling. DFT predicts that 2H-SnSe2 is metallic in contrast to the case of 1T-SnSe2 bandstructure where a semiconducting band gap of 0.4eV (GGA) was obtained (Fig. 3(c)). The orthorhombic SnSe band structure indicates a semiconductor material. Both phases in S8 (a) and (b) below are distinctly different from our experimentally derived SnSe2 band structure, presented in Fig. 4(a) in the main text. The corresponding Brillouin zone for SnSe2 hexagonal(c) and SnSe orthorhombic (d) crystal lattices.
S -9
Figure S9 Enhanced contrast ARPES image of the valence band structure along the ΓΜdirection of the Brillouin zone for the 4 ML SnSe2/Bi2Se3/AlN/Si(111) sample, the complete mapping of which is presented in Fig. 4(a).The onset of the conduction band minimum is visible at the M point (kx,//=0.9 Å-1) and corresponds to the small peak near the Fermi level observed in the linescan (kx=0.9 Å-1) shown in Fig. 4(d). From S9 below, it can be seen that EFis located at–or just above– the conduction band minimum indicating that the MBE-grown SnSe2 is a strongly ntype material.
(a)
(b)
Figure S10a-b Schematic illustration of the proposed device layer structure consisting of the fully epitaxial 2D WSe2/2D SnSe2/Bi2Se3/AlN/Si(111) materials combination. As already
S -10
mentioned in the main text, the Bi2Se3 buffer layer was introduced in order to achieve a good crystalline quality of the epilayers. However, since Bi2Se3 is conductive, it can serve as a bottom electrode in TFET device implementation. The structure in reverse order where WSe2 is first grown followed by SnSe2 top layer is also a suitable combination yielding similar band alignments. The final choice should be made taking into account layer stability in air and device processing flow.
REFERENCES
(1) Rigaku Corporation, X-Ray Reflectivity analysis Program, Global Fit Version 2.05.2, 20082013. (2) Kohli, S.; Rithner, C. D.;Dorhout, P. K.; Dummer, A. M.; Menoni, C. S.;Comparison of nanometer-thick films by x-ray reflectivity and spectroscopic ellipsometry,Rev. Sci. Instrum2005, 76, 023906.
(3) Garg, A. K.; Agnihotri, O. P. ; Jain A. K.; Tyagi, R.C.; Optical absorption spectrum of tin diselenide single crystals, J. Appl. Phys. 1976, Vol. 47, 997. (4) Eichfeld, S. M.; Eichfeld, C.M.; Lin, Y-C; Hossain, L.; Robinson, J. A.; Rapid, nondestructive evaluation of ultrathin WSe2 using spectroscopic ellipsometry, APL Mater.2014, 2, 092508.
S -11
Epitaxial 2D SnSe2/ 2D WSe2 van der Waals Heterostructures Kleopatra Emmanouil Aretouli,†,‡ Dimitra Tsoutsou,† Polychronis Tsipas,† Jose MarquezVelasco,†,§ Sigiava Aminalragia Giamini,†,‡ Nicolaos Kelaidis,† Vassilis Psycharis,† and Athanasios Dimoulas*,†
†
Institute of
Nanoscience and Nanotechnology, National Center for Scientific Research
“DEMOKRITOS”, 15310, Athens, Greece ‡
University of Athens, Department of Physics, Section of Solid State Physics, 15684 Athens,
Greece §
National Technical University of Athens, Department of Physics, 15780 Athens, Greece
AUTHOR INFORMATION Corresponding Author *[email protected]
S -1
(a)
(b)
Figure S1 Schematic representation of the structure of 1T-SnSe2 crystal: (a) side view (b) top view. The unit cell contains only one layer.
S -2
Figure S2 For thickness calibration purposes, X-Ray Reflectivity measurements were performed on a Rigaku Smart Lab diffractometer at thick (nominal 8ML) control samples SnSe2 (A1) and WSe2 (A2), grown directly on AlN/Si(111) substrates in order to avoid interference with reflection oscillations from the Bi2Se3 template. The analysis of measurements was performed using the Global Fit program1. At the early stage of analysis the Global optimization method was used and the final models were obtained using the nonlinear least squares fitting method1. In the analysis of the reflectivity curves for each sample a layer deposited on an AlN substrate it was considered. The best results were derived by considering a linear variation of the density of the top layer from the surface down to the interface with AlN layer.
Figure S3 Spectroscopic ellipsometry (SE) (FLS-300) J. A. WOOLAN) measurements for the thickness determination of thin samples A3 and A4, which are 4ML WSe2 and SnSe2 respectively, and for the thick A2 sample in order to compare XRR with SE technique.The analysis of the data was done using the CAUCHY model. In Figs.S3(a-c) the experimental amplitude ratio Ψ and phase difference Δ versus wavelength (solid lines) and the fitting curves(dashed lines) are plotted and the results are summarized in Table S1. S -3
structure
Nominal
Thickness
Thickness
thickness (nm)
XRR (nm)
Ellipsometry (nm)
A1
8ML SnSe2
4.9
3.74
-
A2
8ML WSe2
5.2
4.2
4.73
A3
4ML WSe2
2.6
-
3.0
A4
4ML SnSe2
2.45
-
2.67
Table S1 XRR and SE results for thickness determination. In the case of SnSe2, a thickness of 2.67 nm is deduced which is in good agreement with the nominal thickness of 2.45 nm (4ML). In the case of WSe2 there is a small deviation: a thickness of 3.0 nm is deduced instead of the nominal value of 2.6 nm expected based on XRR calibration from thicker samples. This discrepancy may be explained based on the fact that SE overestimates thickness with respect to XRR as has been previously reported2. To verify this in our samples, we measured the thicker sample A2 of 8ML WSe2 by both XRR and SE and we found that the latter technique yields a thickness by 0.53 nm larger than that obtained by XRR (see table S1). This indicates that the discrepancy in the measured thickness for the nominal 4ML WSe2 sample is within the uncertainty which is inherent in the measurement of small thickness by XRR and SE techniques. Therefore we estimate that both thicknesses measured by SE are close to the nominal value. It's worth noting that, the refractive indices (an) of SnSe2 and WSe2 are estimated to be 2.99 and 3.7, respectively at 632 nm, which are close to reported values for these materials3, 4.
S -4
Figure S4 Raman spectrum of SnSe2/WSe2/Bi2Se3 heterostructure. The Bi2Se3 buffer layer gives a very strong Raman spectrum overlapping with the Raman peaks of SnSe2 and WSe2, eventually masking them.
Figure S5 XRD scans recorded for the thick (8 ML) WSe2 and SnSe2 layers grown directly on AlN substrates which were grown for the purpose of thickness estimation by XRR. XRD and XRR on WSe2 and SnSe2 layers grown on Bi2Se3 template suffer from complications due to the presence of Bi2Se3 related peaks and interference oscillations. Therefore we can only present the XRD patterns recorded for the 8 ML WSe2/AlN and 8 ML SnSe2/AlN structures, as shown S -5
below. In both cases, sharp diffraction peaks originating from the AlN/Si(111) substrate are observed, as expected. Despite the inferior quality of WSe2 and SnSe2 films on AlN as compared to those grown on Bi2Se3 (see related discussion in RHEED of Fig. 2), well resolved peaks, especially for WSe2 are observed such as (001),(002) and higher order diffraction indicating that [001] WSe2(SnSe2)//[0001]AlN along the growth direction.
Figure S6 RHEED patterns along [10-10] azimuth of the 4ML WSe2(SnSe2)/ 5QL Bi2Se3/AlN Si(111) samples. The patterns are streaky at both azimuths so there is no evidence of 3D island formation. Nevertheless, the RHEED patterns of SnSe2 although they are streaky in general, their intensity is not fully uniform along the streaks which may indicate 2D island formation in compatibility with the morphology imaged by STM.
S -6
(a)
(b)
(c)
(d)
(e)
S -7
Figure S7 SEM images showing the surface of SnSe2/WSe2/Bi2Se3 heterostructures, with areas (a) 330x330μm2, as recorded from the corner of the sample showing and AlN substrate. This area is masked by the sample holder and remains uncovered. (b) 1500x1500μm2, (c) 270x270μm2, (d) 130x130μm2 and (e) 55x55μm2as were recorded from the middle of the sample. It can be seenthat the films are continuous over large areas on the substrate.
S -8
Figure S8 DFT band structure calculations of (a) 4 ML 2H-SnSe2 with D6h (P63/mmc) symmetry and (b) 1ML orthorhombic SnSe (Pnma) phase free-standing slab with spin orbit coupling. DFT predicts that 2H-SnSe2 is metallic in contrast to the case of 1T-SnSe2 bandstructure where a semiconducting band gap of 0.4eV (GGA) was obtained (Fig. 3(c)). The orthorhombic SnSe band structure indicates a semiconductor material. Both phases in S8 (a) and (b) below are distinctly different from our experimentally derived SnSe2 band structure, presented in Fig. 4(a) in the main text. The corresponding Brillouin zone for SnSe2 hexagonal(c) and SnSe orthorhombic (d) crystal lattices.
S -9
Figure S9 Enhanced contrast ARPES image of the valence band structure along the ΓΜdirection of the Brillouin zone for the 4 ML SnSe2/Bi2Se3/AlN/Si(111) sample, the complete mapping of which is presented in Fig. 4(a).The onset of the conduction band minimum is visible at the M point (kx,//=0.9 Å-1) and corresponds to the small peak near the Fermi level observed in the linescan (kx=0.9 Å-1) shown in Fig. 4(d). From S9 below, it can be seen that EFis located at–or just above– the conduction band minimum indicating that the MBE-grown SnSe2 is a strongly ntype material.
(a)
(b)
Figure S10a-b Schematic illustration of the proposed device layer structure consisting of the fully epitaxial 2D WSe2/2D SnSe2/Bi2Se3/AlN/Si(111) materials combination. As already
S -10
mentioned in the main text, the Bi2Se3 buffer layer was introduced in order to achieve a good crystalline quality of the epilayers. However, since Bi2Se3 is conductive, it can serve as a bottom electrode in TFET device implementation. The structure in reverse order where WSe2 is first grown followed by SnSe2 top layer is also a suitable combination yielding similar band alignments. The final choice should be made taking into account layer stability in air and device processing flow.
REFERENCES
(1) Rigaku Corporation, X-Ray Reflectivity analysis Program, Global Fit Version 2.05.2, 20082013. (2) Kohli, S.; Rithner, C. D.;Dorhout, P. K.; Dummer, A. M.; Menoni, C. S.;Comparison of nanometer-thick films by x-ray reflectivity and spectroscopic ellipsometry,Rev. Sci. Instrum2005, 76, 023906.
(3) Garg, A. K.; Agnihotri, O. P. ; Jain A. K.; Tyagi, R.C.; Optical absorption spectrum of tin diselenide single crystals, J. Appl. Phys. 1976, Vol. 47, 997. (4) Eichfeld, S. M.; Eichfeld, C.M.; Lin, Y-C; Hossain, L.; Robinson, J. A.; Rapid, nondestructive evaluation of ultrathin WSe2 using spectroscopic ellipsometry, APL Mater.2014, 2, 092508.
S -11
