Photosensing and Characterizing of the Pristine and In-, Sn-Doped Bi2Se3 Nanoplatelets Fabricated by Thermal V–S Process

Pristine, and In-, Sn-, and (In, Sn)-doped Bi2Se3 nanoplatelets synthesized on Al2O3(100) substrate by a vapor–solid mechanism in thermal CVD process via at 600 °C under 2 × 10−2 Torr. XRD and HRTEM reveal that In or Sn dopants had no effect on the crystal structure of the synthesized rhombohedral-Bi2Se3. FPA–FTIR reveals that the optical bandgap of doped Bi2Se3 was 26.3%, 34.1%, and 43.7% lower than pristine Bi2Se3. XRD, FESEM–EDS, Raman spectroscopy, and XPS confirm defects (In3+Bi3+), (In3+V0), (Sn4+Bi3+), (V0Bi3+), and (Sn2+Bi3+). Photocurrent that was generated in (In,Sn)-doped Bi2Se3 under UV(8 W) and red (5 W) light revealed stable photocurrents of 5.20 × 10−10 and 0.35 × 10−10 A and high Iphoto/Idark ratios of 30.7 and 52.2. The rise and fall times of the photocurrent under UV light were 4.1 × 10−2 and 6.6 × 10−2 s. Under UV light, (In,Sn)-dopedBi2Se3 had 15.3% longer photocurrent decay time and 22.6% shorter rise time than pristine Bi2Se3, indicating that (In,Sn)-doped Bi2Se3 exhibited good surface conduction and greater photosensitivity. These results suggest that In, Sn, or both dopants enhance photodetection of pristine Bi2Se3 under UV and red light. The findings also suggest that type of defect is a more important factor than optical bandgap in determining photo-detection sensitivity. (In,Sn)-doped Bi2Se3 has greater potential than undoped Bi2Se3 for use in UV and red-light photodetectors.


Introduction
Bi 2 Se 3 is a well-known second-generation topological insulator (TI) with a narrow bandgap of 0.35 eV and a rhombohedral crystal structure [1]. Se vacancies (v Se ), which act as electron donors, are the main defects in the Bi 2 Se 3 structure, making it an n-type topological insulator [2]. The crystalline Bi 2 Se 3 is composed of layered structures; each layer consists of five stacked monoatomic layers, as in Se-Bi-Se'-Bi-Se, and is thus known as a quintuple layer (QL) [3]. Covalent bonds dominate the QL [4], whereas Van der Waals' forces dominate between QLs; hence, the dopants can be adequately intercalated among them [5]. A TI has an insulating bulk state and a topologically protected gapless surface state in three dimensions and an edge state in two dimensions, owing to spin-orbital coupling (SOC) and time-reversal symmetry (TRS) [6,7]. Both SOC and TRS suppress backscattering and reduce the sensitivity to surface impurities or defects when electrons are transported on the surface of a TI. These gapless states thus lead to a high electronic conductivity [8,9] and the following features: (1) photon-like electrons, (2) low power dissipation, (3) spin-polarized electrons, and (4) the quantum spin Hall effect [10][11][12][13]. Owing to TIs' unique electronic properties, they have many potential applications, including photodetectors [14], lasers [15], gas sensors [16], spintronic devices [17], magnetoelectronic devices [18], quantum computers [19], and topological superconductors [20]. Several methods are commonly used to synthesize TIs; they include chemical vapor deposition [21], for 60 min. Al 2 O 3 (100) substrate was placed upstream in the quartz tube at about 150 • C, about 21 cm away from the alumina boat. The pristine Bi 2 Se 3 nanoplatelets thus formed were then deposited on the Al 2 O 3 (100) substrate. After the 60 min deposition process, the deposition system was subsequently cooled to room temperature. The starting materials of In-doped Bi 2 Se 3 were 0.1 g Bi, 0.1 g Se, and 25 mg of high-quality In as dopant (Alfa Aesar, 99.99%, 2.17 × 10 −4 mole, USA); those of Sn-doped Bi 2 Se 3 included 25 mg Sn as dopant (Alfa Aesar, 99.8%, 2.11 × 10 −4 mole, USA); that of (In, Sn)-doped Bi 2 Se 3 included 12.5 mg co-dopants In (1.09 × 10 −4 mole) and 12.5 mg Sn (1.05 × 10 −4 mole). The In-, Sn-, and (In, Sn)-dopedBi 2 Se 3 nanoplatelets were synthesized in the same manner as the pristine Bi 2 Se 3 nanoplatelets.

Characterization of Nanoplatelets
The phase and crystal structures of the undoped, as well as the In-, Sn-, and (In, Sn)doped Bi 2 Se 3 nanoplatelets were determined using a mass absorption coefficient glancing incident X-ray diffractometer with an incidence angle of 0.5 • (λ = 0.154 nm, 30 A, 40 kV, Bruker D2 PHASER) and a high-resolution transmission electron microscope (JEOL, JEM-3000 F, Tokyo, Japan). The chemical binding energies and vibration modes of the chemical bonds were obtained using an X-ray photoelectron spectroscope (XPS, Perkin-Elmer model PHI1600 system, Waltham, MA, USA) and a Raman spectroscope (3D Nanometer Scale Raman PL Microspectrometer, Tokyo Instruments, Inc., Tokyo, Japan) with a semiconductor laser at an excitation wavelength of 633 nm. The surface morphology and EDS spectra were obtained using FESEM (ZEISS ULTRA PLUS, Carl Zeiss Microscopy GmbH, Oberkochen, Germany). The optical absorbance values were recorded using a focal plane array-FTIR spectrometer (FPA-FTIR, Bruker Vertex 70V, Hyperion 3000, 64 × 64 MCT Focal Plane Array, Bruker Optik GmbH, Ettlingen, Germany).

Photocurrent Analysis
Photocurrents were measured using a semiconductor I-V property analyzer (Keysight B2901A Precision Source/Measure Unit 100 fA, Keysight Technologies, Santa Rosa, CA, USA) under irradiations of UV or the red light at atmospheric pressure and room temperature, while the bias voltage was kept at 0 V during the photocurrent measuring. The irradiation sources were 30 cm long UV (8 W, λ = 365 nm) and red (5 W, λ = 700-900 nm) LED lamps. The distance between each lamp and the sample was 20 cm. Supplementary Figure S2 schematically depicts the photocurrent measuring system. The sample was placed in a closed box/darkroom to eliminate any effect of ambient light. The silver paste was dropped and deposited onto the surface of each sample of the nanoplatelets and connected to the photocurrent analyzer using copper wires. The photocurrent of each sample was measured in five runs; in each run, the light was on for 10 s and off for 10 s. Figure 1a presents the XRD patterns of the pristine and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets. These nanoplatelets have the typical rhombohedral Bi 2 Se 3 structure (JCPDS 89-2008). Table 1 presents the lattice constants a, b, c, and the c/a ratio. The lattice constants are calculated as the formula 1

XRD Analysis
where h, k, and l are the Miller indices, and a and c are the lattice constants. For the In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets, lattice constant a deviates by −1.312, −0.037, and −1.776%, respectively, and lattice constant c deviates by 0.315, −2.339, and 0.148%, respectively, from their values for the undoped nanoplatelets. The decreases in a are attributed to the substitution of Bi (covalent radius: 148 ± 4 pm) rather than Se (covalent radius: 120 ± 4 pm) by the In (covalent radius: 142 ± 5 pm) or Sn (covalent radius: 139 ± 4 pm) dopant. The lowering of c by doping with Sn is attributed to the substitution of Bi by Sn [42]. The bond length of Bi-Se is 2.86/3.05 Å, and the gap between each QL is 2.62 Å [43]; therefore, the increase in c upon doping with In or (In, Sn) is attributable to the intercalation of In atoms between the QLs [44]. The decrease and increase of the lattice constants a and c, respectively, imply that the In and Sn dopants affected the crystallinity of the Bi 2 Se 3 nanoplatelets. Defects are formed by In at Bi lattice sites (In Bi ), Sn at Bi lattice sites (Sn Bi ), and In in vacancies (In V ).   The In- (Figure 2b and Figure S3b) and (In,Sn)-doped Bi 2 Se 3 ( Figure 2d and Figure S3d) reveal less well-defined hexagonal structures; however, the Sn-doped Bi 2 Se 3 ( Figure 2c and Figure S3c) exhibits a very well-defined hexagonal structure. On average, the Bi 2 Se 3 nanoplatelets are unequivocally hexagonal-like in shape, typical of the rhombohedral structure. The average thickness (40 nanoplatelets) and average diameter (40 nanoplatelets) of the pristine and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets are, respectively, listed in Table 2.   Table 2 shows EDS results for Bi, Se, In, and Sn, which reveal that the ratio Bi:Se increases upon the addition of dopants. This result is attributable to the substitution of Bi by In and/or Sn dopants. Under the Se-rich condition (mole ratio, Bi/Se = 0.755) in this work, the formation energy of V Se defects increases from 1.14 to 2.16 eV, and that of V Bi decreases from 4.13 to 2.60 eV [42]; Bi is thus determined to be substituted by In and/or Sn dopants, consistent with the XRD results. Figure Table 3 provides d-spacings and diffraction planes. These results are consistent with the rhombohedral Bi 2 Se 3 structure and confirm that the dopants, such as In and Sn, have no effect on the crystal structure of Bi 2 Se 3 .   Figure 4 presents the XPS results for Bi 4f, Se 3d, In 3d, and Sn 3d. Figure 4a shows the binding energy of Bi 4f. Peaks at 157.9 and 163.2 eV are attributed to the Bi 4f 7/2 and Bi 4f 5/2 orbitals in the Bi 2 Se 3 phase [45]. Binding energies 159.1 and 164.3 eV are associated with the Bi 4f 7/2 and Bi 4f 5/2 orbitals in the Bi 2 O 3 phase [46,47]. Figure 4b presents the XPS spectra of Se 3d 5/2 and Se 3d 3/2 and the binding energies of 53.6 and 54.6 eV are associated with the Bi 2 Se 3 phase [45]. The peak at 58.9 eV is attributed to the SeO 2 phase [48,49]. The samples are stored in the ambient environment, causing the Bi 2 O 3 and SeO 2 phases to form on the surface of the nanoplatelets. These results confirm the formation of the Bi 2 Se 3 phase by the thermal V-S mechanism. Figure 4c displays the binding energies of the In 3d 5/2 and In 3d 3/2 orbitals, 444.7 and 452.3 eV, associated with the In-Se bond [50]. This finding reveals that In 3+ was doped into the Bi 2 Se 3 structure. The typical Bi 2 Se 3 structure consists of a stack of several quintuple layers (QLs). Each QL comprises Se-Bi-Se'-Bi-Se. In 3+ has two possible positions as a dopant: (1) In 3+ may substitute at the Bi 3+ lattice sites, producing the neutral defect (In 3+

XPS Analysis
Bi 3+ ) and (2) In 3+ intercalates between pairs of QLs, ind icating the possible formation of potentially forming the donor defect (In 3+ V 0 ), where V is the vacancy in the Van der Waals gap. Therefore, In-Se bonds form inside the Bi 2 Se 3 structure or between QLs. The peak at 441.5 eV is ascribed to the Bi 3+ 4d 5/2 orbital [51]. Figure 4d presents XPS spectra of Sn 3d. Peaks at 485.1 and 493.7 eV are associated with the Sn 2+ 3d 5/2 and Sn 2+ 3d 3/2 orbitals, respectively, of the SnSe phase [52]. Peaks at 486.6 and 495.1 eV are associated with the Sn 4+ 3d 5/2 and Sn 4+ 3d 3/2 orbitals of the SnSe 2 phase [53]. Accordingly, the Sn dopants substitute at some of the Bi lattice sites within the Bi 2 Se 3 crystal structure and bond with Se to form Sn-Se bonds. The integral area of the Sn 4+ in the XPS spectrum exceeds that of Sn 2+ (as shown in Figure 4d), implying that the Sn 4+ contents are higher than the Sn 2+ content. The XRD results reveal that the defect (Sn Bi ) is formed in the Bi 2 Se 3 structure during the thermal V-S process. The concentration of the donor defects (Sn 4+ Bi 3+ ) should be higher than those of the acceptor defects (Sn 2+ Bi 3+ ).  Figure 4 presents the XPS results for Bi 4f, Se 3d, In 3d, and Sn 3d. Figure 4a shows the binding energy of Bi 4f. Peaks at 157.9 and 163.2 eV are attributed to the Bi 4f 7/2 and Bi 4f 5/2 orbitals in the Bi2Se3 phase [45]. Binding energies 159.1 and 164.3 eV are associated with the Bi 4f 7/2 and Bi 4f 5/2 orbitals in the Bi2O3 phase [46,47]. Figure 4b presents the XPS spectra of Se 3d 5/2 and Se 3d 3/2 and the binding energies of 53.6 and 54.6 eV are associated with the Bi2Se3 phase [45]. The peak at 58.9 eV is attributed to the SeO2 phase [48,49]. The samples are stored in the ambient environment, causing the Bi2O3 and SeO2 phases to form on the surface of the nanoplatelets. These results confirm the formation of the Bi2Se3 phase by the thermal V-S mechanism. Figure 4c displays the binding energies of the In 3d 5/2 and In 3d 3/2 orbitals, 444.7 and 452.3 eV, associated with the In-Se bond [50]. This finding reveals that In 3+ was doped into the Bi2Se3 structure. The typical Bi2Se3 structure consists of a stack of several quintuple layers (QLs). Each QL comprises Se-Bi-Se'-Bi-Se. In 3+ has two possible positions as a dopant: (1) In 3+ may substitute at the Bi 3+ lattice sites, producing the neutral defect ( + + ) and (2) In 3+ intercalates between pairs of QLs, ind icating the possible formation of potentially forming the donor defect ( + ), where V is the vacancy in the Van der Waals gap. Therefore, In-Se bonds form inside the Bi2Se3 structure or between QLs. The peak at 441.5 eV is ascribed to the Bi 3+ 4d 5/2 orbital [51]. Figure 4d presents XPS spectra of Sn 3d. Peaks at 485.1 and 493.7 eV are associated with the Sn 2+ 3d 5/2 and Sn 2+ 3d 3/2 orbitals, respectively, of the SnSe phase [52]. Peaks at 486.6 and 495.1 eV are associated with the Sn 4+ 3d 5/2 and Sn 4+ 3d 3/2 orbitals of the SnSe2 phase [53]. Accordingly, the Sn dopants substitute at some of the Bi lattice sites within the Bi2Se3 crystal structure and bond with Se to form Sn-Se bonds. The integral area of the Sn 4+ in the XPS spectrum exceeds that of Sn 2+ (as shown in Figure 4d), implying that the Sn 4+ contents are higher than the Sn 2+ content. The XRD results reveal that the defect ( ) is formed in the Bi2Se3 structure during the thermal V-S process. The concentration of the donor defects ( + + ) should be higher than those of the acceptor defects ( + + ).    Figure 5a presents the typical Raman active mode at A 1g 1 , E g 2 , and A 1g 2 of the rhombohedral Bi 2 Se 3 structure [6,54]. No other peaks besides Bi-Se vibrational modes are observed; hence, the dopant does not change the crystal structure of the nanoplatelets or form a second-phase compound, consistent with the XRD results. The formation of Bi 2 Se 3 nanoplatelets is thus confirmed.

Raman Spectra
The typical structure of Bi 2 Se 3 is layered; each layer comprises five monoatomic planes and is, therefore, a quintuple layer (QL). The QL is denoted as A1-B1-A1'-B1-A1, as shown in Figure 5b [55]. A1 and A1' are the Se atoms; B1 is the Bi atom. Covalent bonds dominate the binding within each QL; Van der Waals' forces dominate the bonds between QLs [4,5]. The inset in Figure 5a schematically depicts the Raman peaks of A 1g 1 , E g 2 , and A 1g 2 [56]. A 1g is a symmetric out-of-plane stretching mode associated with the vibration of A1-B1 atoms in the same (A 1g 1 mode) or the opposite (A 1g 2 mode) direction. A 1g 2 has a shorter atomic displacement than A 1g 1 . Therefore, the A 1g 2 mode has higher phonon energy than the A 1g 1 mode [56]. E g 2 is a symmetric in-plane bending mode and shearing the upper two layers of A1-B1 atoms that vibrate in the opposite direction increasing the atomic displacement to a value greater than that in the A 1g 2 mode but smaller than that in the A 1g 1 mode. Thus, the E g 2 mode has a phonon energy between those of the A 1g 2 and A 1g 1 modes [56]. Figure 5c shows the variations in characteristic Raman peaks at A 1g 1 , E g 2 , and A 1g 2 with the species of dopant. A comparison with pristine Bi 2 Se 3 nanoplatelets reveals that both In and Sn dopants cause a redshift of the peaks of A 1g 1 , E g 2 , and A 1g 2 , whereas (In, Sn) co-dopants do not shift the A 1g 2 or E g 2 peak but do cause a redshift in the A 1g 1 peak. Table 4 presents the Raman peaks of pristine and doped Bi 2 Se 3 nanoplatelets, showing the redshifts with the various dopants. The redshifts of the Raman peaks are frequently suggested to involve the heavier atomic weight and/or high-electronegativity dopant to be doped in [57]. The atomic weights of Bi, Se, In, and Sn are 209.0, 78.76, 114.8, and 118.7 (g/mole), and their electronegativities are 2.02, 2.55, 1.96, and 1.78, respectively. The Raman peaks of the doped Bi 2 Se 3 nanoplatelets thus exhibit a redshift. As shown in Table 4, E g 2 , A 1g 2, and especially A 1g 1 peaks are significantly redshifted by the addition of different dopants. The redshift is attributed to the substitution of Bi with dopant In or Sn, which has less weight and a lower electronegativity.   Figure 6a shows the photocurrents in undoped, as well as In-, Sn-, and (In, Sn)doped Bi 2 Se 3 nanoplatelets under UV light. All samples pass a photocurrent (I photo ) that increases rapidly from I dark (~0.62 × 10 −11 -1.65 × 10 −11 A) to the maximum I max (~8 × 10 −10 -1 × 10 −9 A) and then suddenly falls to a stable value (I stable ) when the light is turned on. I stable is clearly independent of the exposure time and, for the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets, it is 0.40 × 10 −10 , 0.65 × 10 −10 , 1.60 × 10 −10 , and 5.20 × 10 −10 A, respectively, indicating that the dopants can increase the photocurrent of the pristine Bi 2 Se 3 nanoplatelets. In particular, the co-dopants In and Sn increase it by a factor of more than 13 to, for example, 5.20 × 10 −10 A. The I photo /I dark ratios in the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets are, respectively, estimated as 7.66, 9.29, 15.8, and 30.7, as shown in Figure S4a and listed in Table 5, implying that the codopants of In and Sn enhance the photoresponsibility of the Bi 2 Se 3 nanoplatelets 4.01 times, which is higher than the pristine one. The decay of the current ∆I decay (I max -I stable ) in the undoped, as well as In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets, is 8.33 × 10 −10 , 9.33 × 10 −10 , 7.76 × 10 −10 , and 3.21 × 10 −10 A, respectively. A smaller ∆I decay is attributed to a longer electron lifetime and a higher concentration of electrons. Both the rise time (τ r ) and fall time (τ f ) are taken by the photocurrent to rise or to fall from 10% to 90% or from 90% to 10%, respectively, of its maximum photocurrent value, as an example of the (In,Sn)doped Bi 2 Se 3 in Figure 6c. The average τ r for the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets are, respectively, evaluated as 0.053, 0.051, 0.044, and 0.041 sec; the average τ f are 0.049, 0.050, 0.054, and 0.066 sec. A shorter τ r is attributable to the higher photosensitivity. The decay time (t decay ) is taken from the I max to I stable , which depends on the recombination rate of the photo-induced electrons and holes [59]. The average t decay of the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets are estimated as 0.091, 0.097, 0.099, and 0.105 sec. A longer t decay corresponds to slower recombination. Therefore, the co-dopants of In and Sn suppress the recombination rate of the photoinduced electrons and holes in the pristine Bi 2 Se 3 nanoplatelets. The detailed variations of the log 10 (time) versus the photocurrents, which are recorded in the first run of the light-on/light-off cycle, between the pristine, and In-, Sn-, and (In,Sn)-doped Bi 2 Se 3 nanoplatelets are presented in Figure 6c. Table 5 presents relevant details. These results suggest that the dopants, and especially the co-dopants In and Sn, in Bi 2 Se 3 nanoplatelets, have various favorable effects, which are (1) extending the electron lifetime, (2) increasing the electron concentration, (3) promoting surface electronic transportation, and (4) improving the photo-sensitivity of the Bi 2 Se 3 nanoplatelets.   Figure 6b presents variations of the photocurrent of the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets under red light. Both undoped and In-doped Bi 2 Se 3 generate no photocurrent, whereas Sn-and (In, Sn)-doped Bi 2 Se 3 generate a photocurrent of 0.5 × 10 −10 and 3.5 × 10 −10 A when the red light is turned on. These results reveal that the photosensitivity of Bi 2 Se 3 nanoplatelets to a red light is greatly improved by the dopants, and especially by the co-dopants In and Sn. The I photo /I dark ratios of the undoped, and In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets are, respectively, estimated as 1, 1, 20.9, and 52.2, as shown in Figure S4b. The co-dopants of In and Sn enhance the Bi 2 Se 3 nanoplatelets 52.2 times higher than that of the undoped one. Figure 6d shows that the optical bandgaps, estimated from the Tauc plot [60], of the undoped, as well as In-, Sn-, and (In, Sn)-doped Bi 2 Se 3 nanoplatelets, are 0.973, 0.641, 0.717, and 0.548 eV, respectively. These bandgaps are estimated by the following equation of (αhν) n = A hν − E g , where α is the absorption coefficient, h is the Planck's constant, ν is the light frequency, n is the characteristic coefficient of materials, A is a constant, and E g is the bandgap. For the direct bandgap of the Bi 2 Se 3 , n is 2. Their absorbance spectra, which were recorded by FPA-FTIR, are shown in Figure S5. Each individual Tauc plot of the pristine, In-, Sn-, and (In,Sn)-doped Bi 2 Se 3 nanoplatelets is demonstrated in Figure S6. These small bandgaps show that the incident UV (~3.4 eV) and red (~1. 37 V 0 ) and (Sn 4+ Bi 3+ ) can supply more additional electrons than undoped, as well as In-and Sn-doped Bi 2 Se 3 nanoplatelets, to generate the highest photocurrent. Therefore, the (In, Sn)-doped Bi 2 Se 3 nanoplatelets have the highest photocurrents under UV and red light than do undoped, as well as In-, and Sn-doped Bi 2 Se 3 nanoplatelets. Based on the above discussion, the reduced optical bandgap of the doped Bi 2 Se 3 nanoplatelets is a minor factor that affects the photocurrent. The type of defect that is generated by doping has a greater effect on the photodetection sensitivity than the corresponding reduction of the optical bandgap. Bi 3+ ), and (Sn 2+ Bi 3+ ) were formed during the synthesis of nanoplatelets, and structural defects (In 3+ V 0 ) and (Sn 4+ Bi 3+ ) significantly improved the photocurrent of (In, Sn)-doped Bi 2 Se 3 nanoplatelets under UV and red light. This work also reveals that In or Sn dopant has no effect on the crystal structure of rhombohedral Bi 2 Se 3 . These results suggest that the photodetection sensitivity of Bi 2 Se 3 nanoplatelets is dominated by the defect structures that are generated by doping, as well as by the consequent reduction of optical bandgap energy. The co-dopants In and Sn further enhance the ability of Bi 2 Se 3 nanoplatelets to respond to UV and red light.