Photoconduction Properties in Tungsten Disulfide Nanostructures

We reported the photoconduction properties of tungsten disulfide (WS2) nanoflakes obtained by the mechanical exfoliation method. The photocurrent measurements were carried out using a 532 nm laser source with different illumination powers. The results reveal a linear dependence of photocurrent on the excitation power, and the photoresponsivity shows an independent behavior at higher light intensities (400–4000 Wm−2). The WS2 photodetector exhibits superior performance with responsivity in the range of 36–73 AW−1 and a normalized gain in the range of 3.5–7.3 10−6 cm2V−1 at a lower bias voltage of 1 V. The admirable photoresponse at different light intensities suggests that WS2 nanostructures are of potential as a building block for novel optoelectronic device applications.


Introduction
In the modern technology era, optoelectronic devices have been established as one of the most ambitious fields of study. Photodetectors are the sub-class of optoelectronic devices that can convert incident light into electrical signals precisely. Photodetectors are vital components to achieve devices with multi-functionality, and hence gained more attention in many applications such as imaging, optical communications, light sensing, and biomedical instruments [1][2][3]. Photodetectors can be divided into two categories based on detection mechanism, namely, photon or quantum detectors and thermal detectors. The photon detectors that include photoconductors, photodiodes, and photo-field effect transistors (photo-FETs) are widely studied due to the existence of band gaps and fast inter-band optical transition. The thermal detectors are either bolometers or thermopiles. Due to their indirect photoelectric conversion, thermal detectors have a relatively slow photoresponse speed [4]. A photoconductor is a fundamental photodetector that is simply a semiconductor channel with ohmic contacts on both ends that works on a photoconductive effect. The photoconductive effect is a process in which the conductivity of a semiconductor material increases due to photon absorption when illuminated by light energy larger than the bandgap of the semiconductor. A photoconductor possesses a gain that can be greater than unity. The high gain will reduce the response speed of the photoconductor. In order to achieve a photoconductor's desired overall performance, a trade-off between gain and response speed must be made [5].
The advent of nanomaterials leads to improving the performance and shrinking the size of novel devices due to their exceptional properties governed by high surface-to-In this work, the WS 2 nanoflakes are exfoliated from the chemical vapor transport (CVT) grown crystals using a conventional mechanical exfoliation technique. For the fabrication of a photoconductor-type photodetector, the platinum (Pt) electrodes were deposited on a WS 2 nanoflake using the focused ion beam (FIB) technique. The photoconduction properties of the device were investigated under the laser wavelength of 532 nm with different powers. The fabricated device shows good performance at a lower bias voltage of 1 V. The photodetector parameters such as responsivity, gain, and normalized gain were estimated and discussed.

WS 2 Crystal Growth
Single crystals of WS 2 were grown by the CVT method using the fine powders of sulfur (99.99%) and tungsten (99.99%) with the help of iodine (I 2 ) as a transporting agent. At first, sulfur and tungsten powders were mixed with I 2 and transferred into the quartz ampoule with a length of 30 cm. The inner and outer diameter of the quartz ampoule is 1.3 and 1.6 cm, respectively. Later, the quartz ampoule was evacuated to 10 −5 Torr and sealed at one end. Next, the sealed ampoule was kept in the two-zone horizontal furnace maintained at temperatures of 1020 and 960 • C. The precursor powder was kept at the higher temperature of 1020 • C zone; once the powder started to melt and vaporize, the I 2 transported the vaporized precursor to the other end of the tube; the temperature was maintained at 960 • C. After ten days of the process, the vaporized precursors were deposited as single crystals of 1-2 cm in length.

Fabrication of WS 2 Photodetector
The WS 2 nanoflakes were exfoliated from the bulk crystal using a conventional mechanical exfoliation technique using dicing tape. For the fabrication of a photoconductortype photodetector, the WS 2 nanoflakes were transferred onto a SiO 2 (300 nm)/n + -Si substrate with pre-patterned Ti/Au electrodes. Next, two Pt metal contacts with a thickness of 100 nm were deposited on WS 2 nanoflakes using the FIB technique. Finally, the electrical wires were connected to the Ti/Au electrodes using a silver paste to characterize the fabricated device. The Ti/Au electrodes are the interconnection between the Pt microelectrode and the millimeter-sized bonded wire.

Measurements and Characterization
The X-ray diffraction (XRD) pattern was measured using a D2 Phaser X-ray diffractometer, and the Raman spectroscopy was measured with an excitation wavelength of 532 nm using a Raman microscope (Renishaw InVia, Wotton-under-Edge, UK); these measurements were used to confirm the crystal structure of CVT-grown WS 2 crystals. The height profiles were carried out to find the thickness of nanoflakes using atomic force microscopy (AFM, Bruker-ICON2-SYS, Billerica, MA, USA). Scanning electron microscopy (SEM, Hitachi S3000H, Tokyo, Japan) was used to capture the image of the nanoflake device to obtain the dimensions of the conduction channel. Focused ion beam (FIB, FEI Quanta 3D FEG) was utilized for the deposition of Pt contacts. The dark current-voltage (i d -V) curves and photoconductive measurements of the photodetector were carried out in a four-point probe electrical measurement system using Keithly 4200-SCS. A 532 nm laser source was used for illumination and the incident laser power was measured using a calibrated power meter (Ophir Nova II) with a silicon photodiode head (Ohir PD300-UV). A holographic diffuser was utilized to minimize the error in the power density calculation by broadening the laser beam size (~20 mm 2 ).

WS 2 Crystal Characterization
The XRD pattern of CVT-grown WS 2 crystal is shown in Figure 1a (006), and (008) planes, respectively. The positions of sharp Bragg reflections confirm the 2H phase of WS 2 crystals according to JCPDS card no. 08-0237 [42,43]. The 2H WS 2 crystal lattice belongs to the P6 3 /mmc (D 4 6h ) hexagonal space group that has space inversion symmetry [44]. The observed sharp and narrow peaks are an indication of the high crystal quality of WS 2 crystals grown by the CVT technique. All diffraction peaks along the (00l) direction denote that the crystal growth is along the c-axis and the major preferential orientation is along the (002) plane. The absence of any binary or impurity phases in the XRD pattern demonstrates the exceptional quality of the CVT-grown crystals. culation by broadening the laser beam size (20 mm 2 ).

WS2 Crystal Characterization
The XRD pa ern of CVT-grown WS2 crystal is shown in Figure 1a. The observed diffraction peaks at 2 values of 14.3, 28.9, 43.9, and 59.8° are assigned to the (002), (004), (006), and (008) planes, respectively. The positions of sharp Bragg reflections confirm the 2H phase of WS2 crystals according to JCPDS card no. 08-0237 [42,43]. The 2H WS2 crystal la ice belongs to the P63/mmc ( ) hexagonal space group that has space inversion symmetry [44]. The observed sharp and narrow peaks are an indication of the high crystal quality of WS2 crystals grown by the CVT technique. All diffraction peaks along the (00l) direction denote that the crystal growth is along the c-axis and the major preferential orientation is along the (002) plane. The absence of any binary or impurity phases in the XRD pa ern demonstrates the exceptional quality of the CVT-grown crystals.   [45][46][47]. The first-ordered dominant modes () and () are most commonly observed for 2H WS2 crystals [43,44,48]. The mode is due to the in-plane vibrations of tungsten and sulfur atoms in the opposite direction, and the mode is due to the out-of-plane vibrations in sulfur atoms. The separation between these two modes is 65.1 cm −1 , which is consistent with the bulk WS2, and the separation reduces gradually with the decrease of the number of layers [48,49]. The second-order longitudinal acoustic mode 2LA (M) is very close to the () and sometimes it overlaps the () mode [45]. The full-width half maxima of 2LA, , and modes are 8.3, 3.4, and 3.7 cm −1 , respectively, and it denotes the high crystallinity of WS2 crystals grown by the CVT technique.

WS2 Nanoflake Device Characterization
The thickness of the WS2 nanoflakes was calculated using the AFM height profile measurement as shown in Figure 2a. The thickness of a typical nanoflake is 155 ± 5 nm.
The inset of Figure 2a shows the AFM picture of the WS2 nanoflake device with Pt contacts.  , and A 1g (Γ) modes of WS 2 crystal, respectively [45][46][47]. The first-ordered dominant modes E 1 2g (Γ) and A 1g (Γ) are most commonly observed for 2H WS 2 crystals [43,44,48]. The E 1 2g mode is due to the in-plane vibrations of tungsten and sulfur atoms in the opposite direction, and the A 1g mode is due to the out-of-plane vibrations in sulfur atoms. The separation between these two modes is 65.1 cm −1 , which is consistent with the bulk WS 2 , and the separation reduces gradually with the decrease of the number of layers [48,49]. The second-order longitudinal acoustic mode 2LA (M) is very close to the E 1 2g (Γ) and sometimes it overlaps the E 1 2g (Γ) mode [45]. The full-width half maxima of 2LA, E 1 2g , and A 1g modes are 8.3, 3.4, and 3.7 cm −1 , respectively, and it denotes the high crystallinity of WS 2 crystals grown by the CVT technique.

WS 2 Nanoflake Device Characterization
The thickness of the WS 2 nanoflakes was calculated using the AFM height profile measurement as shown in Figure 2a. The thickness of a typical nanoflake is 155 ± 5 nm. The inset of Figure 2a shows the AFM picture of the WS 2 nanoflake device with Pt contacts. The blue dotted line across the device denotes the position of the height profile measurement. Figure 2b depicts the i d -V characteristics of a typical WS 2 nanoflake device in the range of −0.1 to +0.1 V. The linear i d -V curve confirms the ohmic contact between the WS 2 nanoflake and FIB-deposited Pt contacts. The inset of Figure 2b represents the SEM image of the WS 2 nanoflake device that is used to calculate the dimensions of the device. The conductivity (σ) of the WS 2 nanoflake with a thickness of 230 nm was calculated using the relation [50,51] where G is the electrical conductance and l, w, and t are the length, width, and thickness of the conduction channel. G is given by I/V, which is obtained from the slope of i d -V curve, and the value is 1.69 × 10 −4 Ω −1 . The l and w are 4.22 µm and 2.65 µm, respectively, obtained from the SEM image of the nanoflake device with a thickness of 230 nm. The calculated conductivity of a typical nanoflake is 12 The blue do ed line across the device denotes the position of the height profile measure ment. Figure 2b depicts the id-V characteristics of a typical WS2 nanoflake device in the range of −0.1 to 0.1 V. The linear id-V curve confirms the ohmic contact between the WS nanoflake and FIB-deposited Pt contacts. The inset of Figure 2b represents the SEM image of the WS2 nanoflake device that is used to calculate the dimensions of the device. The conductivity (σ) of the WS2 nanoflake with a thickness of 230 nm was calculated using the relation [50,51] = = (1) where is the electrical conductance and l, w, and t are the length, width, and thickness of the conduction channel.
is given by I/V, which is obtained from the slope of id-V curve, and the value is 1.69 × 10 −4 Ω −1 . The l and w are 4.22 µm and 2.65 µm, respectively obtained from the SEM image of the nanoflake device with a thickness of 230 nm. The calculated conductivity of a typical nanoflake is 12 Ω −1 cm −1 .  Figure 3 depicts the photoresponse of a WS2 nanoflake with a thickness of 25 nm modulated by light power at an excitation wavelength of 532 nm. The photocurrent was measured for different light powers varying from 2 to 100 mW. A constant biasing voltage of 1 V was applied for the measurement of photocurrent as a function of time. First, we have recorded one cycle of photocurrent response for each light power separately. Next the photocurrent measurements of different powers were combined to clearly present the change in photocurrent with respect to different powers. The ON and OFF states denote the laser light conditions for single light power. When the laser was turned on, the photo current increased quickly, and we waited until it saturated. Once it reached saturation the laser was turned off, and the photocurrent was dropped immediately and then reached saturation. The background dark current was subtracted from photoresponse curves to represent the photocurrent curves. It is clear from the photoresponse curves tha the photocurrent increases with the increase of light power. Generally, a large number of photons of high light intensity create a higher number of electron-hole pairs, and thus the photocurrent increases. The periodic nature of the photoresponse curve under differen light powers is an indication of good stability and reproducibility of fabricated WS2 de vices. With the increase of light power up to 100 mW, we did not observe any photocurren saturation in the WS2 nanoflake, and hence the WS2 photodetectors can be suitable for operation in the linear region.  Figure 3 depicts the photoresponse of a WS 2 nanoflake with a thickness of 25 nm modulated by light power at an excitation wavelength of 532 nm. The photocurrent was measured for different light powers varying from 2 to 100 mW. A constant biasing voltage of 1 V was applied for the measurement of photocurrent as a function of time. First, we have recorded one cycle of photocurrent response for each light power separately. Next, the photocurrent measurements of different powers were combined to clearly present the change in photocurrent with respect to different powers. The ON and OFF states denote the laser light conditions for single light power. When the laser was turned on, the photocurrent increased quickly, and we waited until it saturated. Once it reached saturation, the laser was turned off, and the photocurrent was dropped immediately and then reached saturation. The background dark current was subtracted from photoresponse curves to represent the photocurrent curves. It is clear from the photoresponse curves that the photocurrent increases with the increase of light power. Generally, a large number of photons of high light intensity create a higher number of electron-hole pairs, and thus the photocurrent increases. The periodic nature of the photoresponse curve under different light powers is an indication of good stability and reproducibility of fabricated WS 2 devices. With the increase of light power up to 100 mW, we did not observe any photocurrent saturation in the WS 2 nanoflake, and hence the WS 2 photodetectors can be suitable for operation in the linear region. Nanomaterials 2023, 13, x FOR PEER REVIEW 6 of 12 To interpret the significant dependence of the photocurrent on the illumination intensity, the plot drawn between the photocurrent and light intensity in the range of 80-4000 Wm −2 is shown in Figure 4a. The photocurrent strongly depends on the light intensity, and the experimental data can be fi ed using a power law given by = , where is the photocurrent, a is the scaling constant, P is the light power, and β is an exponent [52]. The power law is well-fi ed to the experimental data with β = 0.99. Generally, β values are in the range of 0 to 1. The deviation of the β value from unity is the indication of the presence of complex processes such as generation, trapping, and recombination of electron-hole within the semiconductor [53]. In our case, the β value is near unity, which indicates that the exfoliated WS2 nanoflake was of high quality with very few defects [54].  To interpret the significant dependence of the photocurrent on the illumination intensity, the plot drawn between the photocurrent and light intensity in the range of 80-4000 Wm −2 is shown in Figure 4a. The photocurrent strongly depends on the light intensity, and the experimental data can be fitted using a power law given by i p = aP β , where i p is the photocurrent, a is the scaling constant, P is the light power, and β is an exponent [52]. The power law is well-fitted to the experimental data with β = 0.99. Generally, β values are in the range of 0 to 1. The deviation of the β value from unity is the indication of the presence of complex processes such as generation, trapping, and recombination of electron-hole within the semiconductor [53]. In our case, the β value is near unity, which indicates that the exfoliated WS 2 nanoflake was of high quality with very few defects [54]. To interpret the significant dependence of the photocurrent on the illumination intensity, the plot drawn between the photocurrent and light intensity in the range of 80-4000 Wm −2 is shown in Figure 4a. The photocurrent strongly depends on the light intensity, and the experimental data can be fi ed using a power law given by = , where is the photocurrent, a is the scaling constant, P is the light power, and β is an exponent [52]. The power law is well-fi ed to the experimental data with β = 0.99. Generally, β values are in the range of 0 to 1. The deviation of the β value from unity is the indication of the presence of complex processes such as generation, trapping, and recombination of electron-hole within the semiconductor [53]. In our case, the β value is near unity, which indicates that the exfoliated WS2 nanoflake was of high quality with very few defects [54].  The photodetectors were characterized by several crucial parameters such as responsivity (R), gain (Γ), and normalized gain (Γ n ) to evaluate their performance. The R is one of the most important figure-of-merits, which is a measure of the photodetector's electrical response to the incident light and is obtained from the formula [7]

Photoconduction Properties of WS 2 Nanoflake
where, i p is the photocurrent and P is the laser power incident on the projected area (A) of a photodetector, and it is given by P = I A = Iwl, where I is the light intensity and w and l are the width and length of the conducting channel, respectively [55]. The R values as a function of light intensity are shown in Figure 4b. We noted that R is sensitive to the lower light intensity (80-320 Wm −2 ) and insensitive to the higher light intensities (400-4000 Wm −2 ). R decreases with the increase of light intensity from 80 to 400 Wm −2 , and a further increase of light intensity up to 4000 Wm −2 results in an almost constant R value. A similar dependency of R on light intensity was observed in WS 2 monolayer [56] and SnS/rGO [57] photodetectors. The calculated R values are in the range of 36-73 AW −1 , and this high responsivity may be due to the efficient absorption and optimized WS 2 nanoflake device configuration. These values are higher than the other photodetectors based on 2D materials such as NbSe 2 nanoflakes (R~2.3-3.8 AW −1 ) [50], MoS 2 nanoflakes (R~20-30 AW −1 ) [58], and NbS 2 nanoflakes (R~0.6 AW −1 ) [59]. The largest R value (73 AW −1 ) at lower light intensity (80 Wm −2 ) is owed to the weak recombination of photo-excited carriers [60]. Gain (Γ) is another figure-of-merit of photodetectors that determine the circulating number of photo carriers moving through a photoconductor per unit time before recombination. It is given by the ratio of the carrier lifetime (τ) to the transit time (τ t ) between the electrodes [50,61].
where, l is the electrodes inter distance, µ is the mobility, and V is the applied voltage. Γ linearly depends on R and it can be calculated using the formula [62,63] where, η is the external quantum efficiency, q is the charge of an electron, h is Planck's constant, and ν is the frequency of the photon. The η value was calculated using the formula η = 1 − e −αt , where α is the absorption coefficient at the wavelength of 532 nm (2.33 eV) and t is the thickness of the nanoflake [50,58]. The reflection loss was eliminated to simply the η calculation. The α value of WS 2 bulk is~2 × 10 5 cm −1 at a photon energy of 2.33 eV [64,65]. By considering the nanoflake thickness of 25 nm, the calculated η value is 0.39 (39%). The determined Γ values as a function of light intensity are shown in Figure 5a. The Γ values follow a similar trend to R. The obtained Γ values are in the range of 215-436, with variation of light intensity from 80 to 4000 Wm −2 . The Γ decreases with the increase of light intensity up to 400 Wm −2 , and a further increase of light intensity up to 4000 Wm −2 leads to saturation in the Γ values. This may be attributed to the continuous filling of trap states upon illumination. Suppose the trap states are filled completely at a certain intensity of light, the excess electrons created by the higher light intensity cannot be trapped and thus decrease the average carrier lifetime. Hence, the photoconductive gain was reduced [4]. The Γ values (215-436) of our photodetector are superior to the reported photodetectors based on the MoS 2 nanoflake (Γ~66-103) [58], WS 2 /Au NPs (Γ~30) [66], and the phototransistors based on MoS 2 (Γ~0.2) and MoSe 2 (Γ~5 × 10 −4 ) [67].  The photodetector parameters  and  were rarely investigated for the 2D material photodetectors. Hence, the comparison of devices based on  and  becomes tough. We have compared our device performance with other reported WS2 photodetectors fabricated by different methods based on their operation region of wavelength, biasing voltage, and responsivity as summarized in Table 1. It is noticed that the WS2 nanoflake device obtained in the present work shows be er responsivity at a lower bias voltage of 1V. The high responsivity of our device also outperformed some CVD-grown WS2 monolayerbased photodetectors, and hence the WS2 nanoflakes can be a potential candidate for fabricating novel optoelectronic devices. Normalized gain (Γ n ) is independent of device geometry and can be considered a fair figure-of-merit to compare the performance of other devices. The photoconduction process is mainly involving the light absorption and movement of carriers between the electrodes. Numerous factors including carrier mobility, lifetime, applied bias, distance between electrodes, and efficiency of light absorption may affect the performance of photodetectors. Γ n is a measure of the intrinsic photoconductivity of the device and is given by the product of η, τ, and µ [61,68]. The photodetector parameters Γ and Γ n were rarely investigated for the 2D material photodetectors. Hence, the comparison of devices based on Γ and Γ n becomes tough. We have compared our device performance with other reported WS 2 photodetectors fabricated by different methods based on their operation region of wavelength, biasing voltage, and responsivity as summarized in Table 1. It is noticed that the WS 2 nanoflake device obtained in the present work shows better responsivity at a lower bias voltage of 1 V. The high responsivity of our device also outperformed some CVD-grown WS 2 monolayer-based photodetectors, and hence the WS 2 nanoflakes can be a potential candidate for fabricating novel optoelectronic devices.

Conclusions
We successfully fabricated a visible photodetector using exfoliated WS 2 nanoflakes and explored its photoconduction properties. The photocurrent increases with the increase of light intensity from 80 to 4000 Wm −2 and is well-fitted to the power law with an exponent value of 0.99. The photoresponsivity decreases with the increase of light intensity from 80 to 400 Wm −2 , and a further increase of light intensity up to 4000 Wm −2 results in an almost constant R value. The fabricated device showed a stable photoresponse with some reproducible characteristics. The device exhibited good responsivity in the range of 36-73 AW −1 , and the normalized gain was in the range of 3.5-7.3 × 10 −6 cm 2 V −1 at a lower biasing voltage of 1 V. The obtained excellent photodetector parameters suggest a promising application of WS 2 nanoflakes in future novel optoelectronic devices.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.