The Highly Uniform Photoresponsivity from Visible to Near IR Light in Sb2Te3 Flakes

Broadband photosensors have been widely studied in various kinds of materials. Experimental results have revealed strong wavelength-dependent photoresponses in all previous reports. This limits the potential application of broadband photosensors. Therefore, finding a wavelength-insensitive photosensor is imperative in this application. Photocurrent measurements were performed in Sb2Te3 flakes at various wavelengths ranging from visible to near IR light. The measured photocurrent change was insensitive to wavelengths from 300 to 1000 nm. The observed wavelength response deviation was lower than that in all previous reports. Our results show that the corresponding energies of these photocurrent peaks are consistent with the energy difference of the density of state peaks between conduction and valence bands. This suggests that the observed photocurrent originates from these band structure peak transitions under light illumination. Contrary to the most common explanation that observed broadband photocurrent carrier is mainly from the surface state in low-dimensional materials, our experimental result suggests that bulk state band structure is the main source of the observed photocurrent and dominates the broadband photocurrent.


Introduction
The interaction of light with matter has been a widely used technique for light sensors and is widely used in our daily life and scientific research. A system with higher photointeraction efficiency would lead to higher sensitivity. Therefore, there has been increasing interest in finding a material with higher photoresponsivity. The material band structure is a critical factor to finding the efficiency of the light response and light wavelength. As well as the band structure, the light penetration depth is short, so the photon interaction mainly occurs on the material surface. To enhance the photon interaction, a system with higher surface area is preferred [1][2][3][4][5][6]. In addition to the higher surface ratio, earlier studies have demonstrated that carrier mobility is one of the most critical factors for determining the light responsivity [7,8]. With these considerations in mind, nanostructures and low-dimensional systems with high mobility have been widely investigated [9][10][11][12][13][14][15][16]. Experimental results have revealed that most materials only show high photoresponsivity at particular wavelengths due to their specific band structure, and this limits their potential application. A broadband response along with a high photoresponsivity is important for the potential application across a wide range of wavelengths.
A system with a linear energy-momentum band structure, such as graphene, graphenebased heterostructures, and topological materials, could satisfy these requirements and provide a high photoresponsivity over a broadband wavelength range. Many studies have demonstrated high photoresponsivity over a wide range of wavelengths in these systems. The graphene/transition metal dichalcogenide system shows an extremely fast photoresponse [17]. InGaAs shows a photoresponse for light wavelengths ranging from 400 to 16,000 nm [18]. However, previous reports have revealed that the photoresponses show a strong wavelength-dependent deviation in these systems [10,15,. This inhomogeneous response would limit the potential applications of these materials as broadband photosensors. The search for a system with uniform light responsivity over a wide range of wavelengths is one of the on-going topics in the field of light sensor applications. The light penetration into the material and carriers from the bulk state of the topological material might also contribute to the measured photocurrent. This might be one of the reasons behind the observed strong wavelength dependence in the previous reports [50]. To solve this problem, as well as the surface state band structure, one might take the bulk state band structure into account. We checked the bulk state band structure of BixSb(2−x)TeySe(3−y) topological materials and found that Sb 2 Te 3 might be a good candidate to providing uniformity of the photocurrent over a wide range of wavelengths.
In this work, we study the photoresponse of Sb 2 Te 3 flakes. Our results show that the photocurrent is linear with the light power and the applied electric voltage, and that the photoresponsivity is proportional to the conductance of the flake. According to previous reports, the photoresponsivity is stable for light wavelengths ranging from visible to nearinfrared light, and the normalized root means that the square photoresponsivity deviation is about 6%, which is much lower than reported results in low-dimensional systems. The results show that the corresponding energy of the measured photocurrent peaks and dips at different light wavelengths are consistent with the energy difference of the carrier density of state peaks and dips between the conduction and valence band of Sb 2 Te 3 bulk state. This result supports that the observed photoresponsivity is dominated by the carriers of the bulk state and not from the topological surface state. The low-deviated band structure of bulk state leads to the behavior of the observed uniform broadband light responsivity.

Materials and Methods
High purity elements Sb (99.995%) and Te (99.995%) were mixed according to the stoichiometric ratio in a glove box with low oxygen and water to avoid oxidation. The mixed raw materials were put into a vacuum quartz glass tube (pressure~10 −3 torr) and sealed inside the quartz glass tube. After that, the materials were melted at 750 • C for 20 h, and then slowly cooled down to 625 • C in 5 h. It took another 60 h to cool the materials down to 565 • C at the cooling rate 1 • C/h. The melting point of Sb 2 Te 3 is 617 • C, so this slow cooling rate ensured that Sb 2 Te 3 could be slowly crystalized. Then, it took 10 h to cool the materials down to room temperature. The crystallized Sb 2 Te 3 was further purified and crystallized by a homemade resistance-heated floating zone furnace (RHFZ). The Sb 2 Te 3 was locally heated up to 617 • C and the Sb 2 Te 3 formed an interface of solid phase and liquid phase. At the same time, we slowly moved the quartz glass tube and changed the position of the solid/liquid interface. During this process, the impurities were moved with the liquid phase and were excluded from the recrystallized Sb 2 Te 3 . After this recrystallization process, single crystals of Sb 2 Te 3 were naturally cooled down to room temperature. The as-grown crystals were cleaved along the basal plane, producing a silvery shining mirror-like surface, and then prepared for further experiments. The Raman spectrum and electron probe microanalyzer (EPMA) spectrum support that the produced crystal was Sb 2 Te 3 [51]. Details of Raman spectrum and EPMA spectrum are given in Appendix A. Sb 2 Te 3 flakes were obtained by exfoliating bulk crystals using dicing tape and were then dispersed on the insulating SiO 2 (300 nm)/n-Si templates with prepatterned Ti/Au circuits. Two platinum (Pt) metal contacts were subsequently deposited on the selected Sb 2 Te 3 flakes using focused-ion beam (FIB) technique. The Chemical Abstracts Service (CAS) numbers of all the chemical and substrates are given in Appendix B. Details of the x-ray diffraction (XRD), Raman, EPMA, SEM, and electrical measurement system are given in Appendix C. Figure 1 shows the XRD pattern of the produced Sb 2 Te 3 . The Sb 2 Te 3 crystal showed sharp peaks. It revealed the Laue diffraction peaks at the c-axis. The crystal was grown on the [001] direction. The full width at half maximum (FWHM) was between 0.055 and 0.06 deg. The average crystallite size at the c-axis was about 124 nm and the strain was −0.018%. These results indicate that the Sb 2 Te 3 was highly single crystallized. The Sb 2 Te 3 powder was ground from the Sb 2 Te 3 crystal. The XRD pattern of the Sb 2 Te 3 powder was consistent with the Joint Committee on Powder Diffraction Standards (JCPDS) card. Figure 1. The X-ray diffraction (XRD) pattern of the produced Sb 2 Te 3 . The sharp peaks indicate that the Sb 2 Te 3 is highly single crystalline. The Sb 2 Te 3 powder was ground from the Sb 2 Te 3 crystal. The XRD pattern of the Sb 2 Te 3 powder is consistent with the JCPDS card. Figure 2 shows the SEM image of a Sb 2 Te 3 flake. It is 5 µm long and ∼5 µm wide. The thickness of a flake is determined by the atomic force microscopy; here, the flake was 260 nm thick (shown in the left-bottom inset of Figure 2).

Results and Discussion
The left-top inset of Figure 3 shows the linear current-voltage relationship of the flake; the conductivity is about 20.4 S/cm. The right-bottom inset of Figure 3 shows the on-off photocurrent response at different applied electric voltages under 808 nm light wavelength and 50 mW laser power. In this work, we define the photocurrent as the measured current difference between the condition with and without light illumination. The main image from Figure 3 shows that the extracted photocurrent, I P , is linear with the applied electric voltage. These results indicate the ohmic contacts between the Sb 2 Te 3 flake and Pt electrodes, and the metallic behavior of the Sb 2 Te 3 flake. Figure 4 shows the measured current of our Sb 2 Te 3 flake under light illumination with light power ranging from 1 to 50 mW, which corresponds to the power intensity of 40 to 2000 Wm −2 . The photocurrent increases with increasing light power and shows a similar power-dependent photocurrent under three different light wavelengths. The left-top inset of Figure 4 reveals the extracted photocurrent as a function of the light power at three different light wavelengths. It clearly reveals that the photocurrent varies linearly with the light power for three light wavelengths. The higher light intensity means higher photon interaction. The increasing amount of photon interaction will induce more excited carriers, leading to a higher photocurrent. The light intensity is not over the limit of the excited carrier number, and the measured photocurrent will be proportional to the increasing light intensity, resulting in a positive slope for the three different light wavelengths.   To further identify this characteristic in the three different wavelengths, the photocurrent is expressed as the simple power-law relation I P = αP β , where the α is a constant for the wavelength of the illuminating light, P is the illuminated light power on the Sb 2 Te 3 flake, and β is a constant related to the transport mechanism and the intrinsic optic characteristics. Table 1 lists α and β for the three different wavelengths. The result reveals that β is 1.03 ± 0.02, which indicates that the behavior of the laser power-dependent photocurrent of the system remains the same at different light wavelengths. The complex carrier scattering process might lead to the non-integer β. Our experiment shows β∼1, which indicates that the carrier transport process is insensitive to the extrinsic light power and interaction between excited carriers. This characteristic is more flexible and stable for a wide range of potential applications. The broadband light response has been observed in many lowdimensional systems [10,15,. These reports show that the observed photocurrent is strongly dependent on light wavelength, which would limit the application potential of the broadband light sensor. Advancing the findings of previous reports, our experimental results reveal that the photocurrent is weakly wavelength dependent [10,15,. The α roughly deviates by 30% at the three different wavelengths, which is much smaller than the reported values. The light power and the measured photocurrent depended upon the effective geometric area of the material. In order to quantitatively determine the performance of the Sb 2 Te 3 flake under illumination, the responsivity, R, is calculated through the following equation: where I P , P, and S are the photocurrent, the light intensity, and the effective area, respectively. The obtained R values are 38, 31, and 43 AW −1 for the wavelengths of 405, 532, and 808 nm at a constant applied voltage of 0.1 V. The deviation ratio is about 34%. To further examine the photocurrent response characteristic at wide wavelength, a photocurrent spectrum measurement was performed. As shown in the top-right inset of Figure 4, the R reveals weak wavelength dependence over a range of wavelengths. The normalized root means square is used to identify the deviation level. The normalized root means square is 0.06, which indicates a 6% deviation for the light wavelength ranging from 300 to 1000 nm. This behavior is more stable than previously reported results in low-dimensional systems. Along with the responsivity, the responsivity stability is a critical factor for the application of broadband photosensors. Table 2 lists the reported broadband light responsivity of various kinds of materials. It shows that most of these materials reveal a large light response deviation on the broadband light wavelength. The photo deviation of the Sb 2 Te 3 flake is a few orders smaller than the reported values of various materials. The left side of Figure 5 reveals the density of state of the Sb 2 Te 3 bulk state, and it reveals several peaks at valence and conduction bands. The right side of Figure 5 shows the measured photoresponsivity as a function of light wavelength, and it exhibits several photocurrent peaks at different wavelengths. Our analysis shows that the corresponding energies of these measured photoresponsivity peaks and dips at different light wavelengths are consistent with the energy difference of Density of states (DOS) peaks between conduction and valence bands. This suggests that the observed photocurrent originates from these band structure peak transitions under light illumination. Contrary to most previous explanations that the observed broadband photocurrent carrier is mainly from the surface state in topological materials, our experimental result suggests that bulk state carriers are the main source of the observed photocurrent and dominate the broadband photocurrent. The photocurrent is directly related to the effective carrier mobility, which is determined by both the carriers from the bulk state and surface state. Figure 6 shows the photocurrent responsivity as a function of the conductance of the Sb 2 Te 3 flakes. It shows that the photocurrent responsivity is proportional to the conductance of the Sb 2 Te 3 flakes. This implies that effective conductance is a critical factor to determining the photocurrent responsivity. Detectivity, which is an important figure of merit in evaluating the ability of a photodetector to detect a weak signal, is another important index used to characterize the performance of photodetectors. The specific detectivity (D * ) is calculated through the following equation: where R, S, q, and I d are the responsivity, effective area of light illumination, electronic charge, and dark current, respectively. By using the experimental data, the obtained detectivity is about 2 × 10 9 Jones. As shown in the top-right inset of Figure 6, the detectivity is almost insensitive to the flake thickness. The top-left inset of Figure 6 shows that the sensitivity is about 0.02. Our experimental results show that the detectivity and the sensitivity is insensitive to the flake thickness. However, it should be noted that the light penetration length is about 20 nm, which is much shorter than the thickness of our Sb 2 Te 3 flakes. This might be the reason behind the observed thickness-independent detectivity and sensitivity.

Conclusions
The photocurrent measurement was performed on Sb 2 Te 3 flakes at various wavelengths ranging from visible to the near IR region. The photocurrent was observed to vary linearly with the applied electric voltage, light power, and conductance of the flakes. Advancing the findings of previous reports that the photoresponse reveals strong wavelength dependence, our measured photocurrent was insensitive to wavelengths from 300 to 1000 nm. The wavelength response deviation was lower than that in all previously reported photoresponse materials and it has the potential to play a role in broadband photosensors. Our analysis showed that the corresponding energies of these photocurrent peaks were consistent with the energy difference of the density of state peaks between conduction and valence bands. This further suggests that the observed photocurrent originated from these band structure peak transitions under light illumination. Contrary to the most common explanation that the observed broadband photocurrent carrier is mainly from the surface state in low-dimensional materials, our experimental result suggests that the bulk state band structure is the main source of the observed photocurrent and dominates the broadband photocurrent. As well as the linear energy-momentum dispersion, one has to take the bulk state band structure into account for the broadband photo-detection material.