High-performance ultra-violet phototransistors based on CVT-grown high quality SnS2 flakes

van der Waals layered two-dimensional (2D) metal dichalcogenides, such as SnS2, have garnered great interest owing to their new physics in the ultrathin limit, and become potential candidates for the next-generation electronics and/or optoelectronics fields. Herein, we report high-performance UV photodetectors established on high quality SnS2 flakes and address the relatively lower photodetection capability of the thinner flakes via a compatible gate-controlling strategy. SnS2 flakes with different thicknesses were mechanically exfoliated from CVT-grown high-quality 2H-SnS2 single crystals. The photodetectors fabricated using SnS2 flakes reveal a desired response performance (Rλ ≈ 112 A W−1, EQE ≈ 3.7 × 104%, and D* ≈ 1.18 × 1011 Jones) under UV light with a very low power density (0.2 mW cm−2 @ 365 nm). Specifically, SnS2 flakes present a positive thickness-dependent photodetection behavior caused by the enhanced light absorption capacity of thicker samples. Fortunately, the responsivity of thin SnS2 flakes (e.g. ∼15 nm) could be indeed enhanced to ∼140 A W−1 under a gate bias of +20 V, reaching the performance level of thicker samples without gate bias (e.g. ∼144 A W−1 for a ∼60 nm flake). Our results offer an efficient way to choose 2D crystals with controllable thicknesses as optimal candidates for desirable optoelectronic devices.


Introduction
5][6][7] Furthermore, graphene shows a zero-band gap which traditionally has limited its application in optoelectronic devices. 85][16] Nowadays, various layered semiconductors can be prepared via top-down (e.g.mechanical exfoliation) or bottom-up (e.g. chemical vapour deposition, CVD) methods. 17,18Nevertheless, in view of sustainable development, earth-abundant 2D materials are more essential for widespread use in modern devices.Tin disulde (SnS 2 , belonging to the IV-VIA group) comprises earth-abundant constituents (Sn and S), which could vastly support the industrial requirements. 19Bulk SnS 2 crystals (usually representing a bandgap of 2.1-2.31eV) have long been explored for possible applications in photovoltaics and photoelectrochemistry. 20,21In addition, recent evidence has already pointed out the fascinating progress in applying 2D SnS 2 structures in lithium ion batteries, 22 eld-effect devices 23 and photodetectors, 24 benetting from the simple exfoliation from bulk crystals and the controlled bottom-up synthesis. 25,262][33] Therefore, some prospective strategies should be proposed to address this problem and make these prototypes appealing for practical applications.Unfortunately, this issue has not been well considered in previous optoelectronic applications.
In this paper, we have achieved high-performance UV photodetectors established on SnS 2 akes and employed a gatetunable strategy to address the challenge posed by the positive thickness-dependent light sensing behaviour of 2D materials.SnS 2 akes with different thicknesses (that can be thinned to $7 nm in the work) were mechanically exfoliated from CVT-grown high-quality 2H-SnS 2 single crystals.We describe several characterization methods to identify the phase and microstructure of these multilayer SnS 2 samples.The photodetectors fabricated by using individual SnS 2 akes reveal a desired response performance (R l z 112 A W À1 , EQE z 3.7 Â 10 4 %, and D* z 1.18 Â 10 11 Jones) under UV light with a very low power density (0.2 mW cm À2 @ 365 nm).Specically, the SnS 2 akes present a relatively low photodetection ability in the thinner akes.But it is found that the responsivity of thin akes (e.g.$15 nm) could be enhanced to $140 A W À1 aided by a positively gated bias (+20 V), and it is comparable to the performance of a nongated $60 nm ake.This indeed provides a potential way to address the positive thickness-dependent detection capability mainly caused by the enhanced light absorption capacity of thicker samples, and our ndings imply that such earth-abundant and environmentally friendly tin-based chalcogenides are desirable for sustainable "green" optoelectronics applications.

Preparation of multilayer SnS 2 akes
The starting materials here are SnS 2 bulk crystals, grown via a chemical vapor transport (CVT) route with pure iodine as the transport agent.The pre-mixed powders of Sn (99.99%,Aladdin) and S (99.99%, Alfa Aesar) at a stoichiometric ratio of 1 : 2 with additional iodine (99.8%,Aladdin, 5 mg cm À3 ) were vacuum sealed (>10 À4 Torr) in a quartz tube.The quartz tube was then placed in a two-zone furnace.The reactant zone was slowly heated up ($10 h) to 800 C while the other end was set to 750 C. The growth process was maintained for $10 h, followed by a naturally cooling process down to room temperature.Aerwards, strip-like products were obtained.Thin SnS 2 akes were mechanically exfoliated from the as-synthesized crystals aided by adhesive tape and then dry-transferred onto a freshly cleaned SiO 2 /Si substrate (with a dielectric layer, $300 nm thick SiO 2 ).Thin samples were roughly identied using an optical microscope (Olympus, CX41) in combination with a charge-coupled device.

Materials characterization
The phases of the crystals were characterized by X-ray diffraction (XRD) on a Rigaku Miniex 600x powder diffractometer.XPS measurements were performed using a Thermo ESCALAB 250XI X-ray photoelectron spectrometer.The thicknesses of the exfoliated SnS 2 samples were measured by atomic force microscopy (AFM, Agilent 5500).A home-built Raman spectroscope/microscope (iHR320, Horiba) was utilized to acquire Raman spectra and spatially resolved Raman maps with an incident laser of 532 nm while employing a 405 nm light for photoluminescence (PL) measurements.A transmission electron microscope (TEM, JEM-2100F) was employed to evaluate the morphologies and crystal structure of thin SnS 2 akes.The UV-vis spectrum was measured using a spectrophotometer (MPC-3100, Shimadzu).

Device fabrication and measurements
The fabrication of photodetectors based on individual SnS 2 akes relies on a standard UV lithography (URE-2000/25) process followed by thermal evaporation (JSD 300) of desired electrode metals (Cr/Au, $10/60 nm).The photoresponse measurements were executed on a probe station (ZFT-50T) equipped with two sourcemeters (Model 2450, Keithley).Light sources (THORLABS) of different wavelengths with tunable power were applied in the photodetection tests.

Results and discussion
To get SnS 2 samples with varied thicknesses, we employed tapeassisted mechanical isolation from raw crystals synthesized via a chemical vapour transport (CVT) approach.Previous research suggests that SnS 2 occurs as different polytypes associated with different interlayer stacking of S-Sn-S layers; a low-temperature (<800 C) synthesis process tends to produce the 2H-polytype. 34,35Fig. 1a illustrates the 3D structure model of 2H-SnS 2 ; the S-Sn-S atomic planes (the distance between adjacent planes is $0.6 nm) with covalent bonding are held together by weak van der Waals force. 36In this work, SnS 2 crystals were prepared via a CVT process in a two-zone tube furnace, as schematically shown in Fig. 1b (more information has been provided in the Experimental section).During the CVT growth, gas transport is realized by temperature gradient and thus impact on the nal products (Fig. S1 †).In this work, a 50 C temperature gradient is in favour of growing high quality SnS 2 single crystals, in agreement with a previous report, 37 and large sized crystals (given in the inset of Fig. 1c, lateral size > 3 mm) with a clean surface could be found at the lower temperate zone.The phase structure of the crystals was identied through the powder X-ray diffraction (XRD) technology, as depicted in Fig. 1c.The products exhibit a hexagonal structure (JCPDS PDF number 23-0677, a ¼ b ¼ 0.36 nm, c ¼ 0.59 nm).Strong diffraction reections emerging around $15.0 , $30.3 and $46.1 are indexed to the (0001), (0002) and (0003) planes, respectively.The predominance of the (000N) (N ¼ 0, 1, 2, etc.) peaks suggests that the Zdirection with the (0001) plane as the basal plane is the preferential orientation for these growing SnS 2 crystals. 38The valence states of the SnS 2 crystals were characterized by X-ray photoelectron spectroscopy (XPS).The spectra have been calibrated by employing absorbed C (1s) as the reference.Expected Sn and S elements from these crystals were detected in the XPS survey (Fig. 1d).Two peaks at 486.5 and 495.1 eV (Fig. 1e) originate from Sn3d 5/2 and Sn3d 3/2 of Sn 4+ . 39Fig. 1f illustrates the high-resolution core XPS spectrum of S 2p ($162.6 eV), an indication of the existence of S2p 3/2 and S2p 1/2 orbitals. 40These results evidence the formation of layered SnS 2 crystals with high quality.
Thin akes with varied thicknesses (Fig. 2a) were obtained through mechanical exfoliation from large crystals exceeding 10 mm (limited by the AFM scanning range) accompanied by a clean surface and pristine state.The height of multilayer SnS 2 akes oen presents lateral dimension changes shown in Fig. 2b measured at the edge between the thicker and thinner part indicating a thickness of $7 nm (corresponding to about ten layers).Raman spectroscopy was employed to quantify and map SnS 2 akes.Previous reports suggest the existence of different crystal polytypes for layered SnS 2 crystals, 4H-and 2Hphases, respectively. 41The most intense Raman peak for the 4H polytype emerges at 313.5 cm À1 , relative to the joint contribution of A 1 and E phonon modes (this peak is very close to the A 1g mode of 2H-SnS 2 at 315 cm À1 ), while the E-mode generates doublet at 200 and 214 cm À1 , respectively. 19In parallel, in 2H-SnS 2 crystals, the E g mode generates a single, intense line around 205 cm À1 , allowing a facile distinction against the 4H polytype. 28The observed Raman active modes in Fig. 2c, i.e. two main peaks at 205 (E g , in plane) and 315 cm À1 (A 1g , out of plane) occur in the thick akes (>200 nm, the inset of Fig. 2c), hence, providing an unambiguous ngerprinting of 2H polytypic crystals in this work.Here, the E g modes weaken and become unobservable with the reduction of ake thickness (down to the nanometer level), which could be presumably attributed to the reduction of in-plane scattering centers in the ultrathin SnS 2 akes. 42,43But the A 1g (out of plane) mode illustrates a signicant increase in the peak intensity with the increasing thickness (Fig. 2d), probably arising from the enhanced light absorption capacity of the thicker samples, 44,45 which may have an inuence on the device performance.Raman mapping (Fig. 2c) of a SnS 2 ake (the optical image is depicted in Fig. 2d) using the characteristic line at 315 cm À1 demonstrates its uniform polytype.2H-SnS 2 akes exhibit isotropy in the (0001) plane as conrmed by the polarized Raman characterization of the A 1g vibration mode.Therefore, the origination dependent photoresponsivity should not be a problem for their application in photodetection. 46The PL spectra of 2D SnS 2 akes in Fig. 2e consist of a single feature, attributed to the nature of their indirect band-gap structure, with a value of $2.20 eV similar to the reported data of 2H-SnS 2 . 28Moreover, the peak shis to lower energies with increasing thickness (Fig. 2f).During PL measurements, holes will combine with the photo-excited electrons via Coulomb interactions with the binding energy lying in the band-gap region; however, strong spatial connement and a reduced screening effect in ultrathin samples could result in a signicantly enhanced excitonic effect causing a blue shi in the bandgap. 47he microscopic structure and the compositions of the 2D SnS 2 crystals were determined using a transmission electron microscope (TEM) and an energy-dispersive X-ray detector (EDX).Fig. 3a shows a low magnication view (bright eld) of a single ake with no observable terraces.A direct view of the lattice structure (Fig. 3b, reected by the HRTEM image of the labeled region in Fig. 3a) suggests the hexagonal arrangement (atomic mode given on the top) of Sn and S atoms in 2H SnS 2 crystals.The plane distance of $0.32 nm well matches the d-  spacing of the {10 À 10} planes for hexagonal phase SnS 2 . 48The selected area electron diffraction (SAED) pattern (obtained by applying incident electrons parallel to the c-axis) in Fig. 3c shows well-sequenced diffraction spots with a six-fold symmetry, indicating high-quality crystallinity of this crystal caused by a vertically stacking layer plane along the [0001] direction.The composition of the crystals was veried by adopting energy-dispersive X-ray spectroscopy (EDX).EDX mapping results in Fig. 3d and e suggest the homogeneous distribution of Sn and S elements in the ake.In parallel, the EDX spectrum (Fig. 3f) clearly reveals signals of Sn and S with an approximate atomic ratio of 2 : 1.The aforementioned demonstrations signify the high purity of the single-crystalline SnS 2 akes.Photodetectors established on individual akes were employed to systematically estimate the optoelectronic properties of the SnS 2 crystals.The set-up diagram of a two-terminal light sensor is illustrated in Fig. S3a, † where the incident light is perpendicular to the SnS 2 ake.Fig. S3b † and 4a present the current-voltage (I DS -V GS ) characteristics of the representative SnS 2 light sensor in darkness and under illumination with varied wavelengths.The linear curves result from the near ohmic barrier between the Cr/Au electrodes and SnS 2 channel.The plot of responsivity (Fig. 4b, R l ¼ I ph /PS, I ph ¼ I light À I dark , S refers to the activated area of $5.17 mm 2 and P is the power intensity of incident light, V DS ¼ 1 V) versus wavelength provides a quantitative assessment of the photoresponse ability of the SnS 2 ake.The optical image of the SnS 2 channel with a thickness of $15 nm and a length of $3 mm is illustrated in the inset.The cut-off wavelength is about 550 nm, close to its deduced bandgap, E indirect z 2.18 eV from the UV-vis absorption spectrum in Fig. S4.† 29 The light sensor exhibits high photo responsivity in the ultra-violet range.The stability and reproducibility of the SnS 2 based photodetector (Fig. 4c, measured at a bias of 1 V) are uncovered via tracking current under periodic illumination of 0.2 mW cm À2 @ 365 nm.Enlarged views of the rising and decay sides are illustrated in Fig. S3c, † from which the response time (s rising ) and recovery time (s decay ) are calculated to be 40 and 160 ms, respectively.Fig. 4d displays the dependence of photocurrent on light intensity, which can be well described by a power law, I ph $ P b , where b is an exponent determined by trap states on the surface of the photo-sensitive media. 49In general, a low power intensity would benet the occupation of surface states by holes separated from photoinduced electron-hole pairs, followed by a rapid recombination with the negatively charged oxygen.But abundant electron-hole pairs will be generated under a higher light intensity.The reduction of the hole-trap states at the surface until the complete occupation of surface traps will contribute to faster recombination (in several picoseconds) between extra charges; subsequently, these fresh pairs will not contribute to the charge transfer process, but form non-radiative carrier-recombination centers, thus leading to a decline (Fig. 4e) of responsivity and quantum efficiency (EQE ¼ hcR l /el, h is Planck's constant). 50,51he sub-linear behaviour in Fig. 4d with a tting value b $ 0.98 (very close to 1) may be associated with a low concentration of traps or defects in these SnS 2 akes. 52The parameters of our sensors are comparable or superior to those of documented 2D SnS 2 and other 2D material-based UV sensors as summarized in Table 1, potentially offered by the high-quality, large specic surface area. 53In addition, we nd a positive thickness-  dependent photodetection capability of these 2D SnS 2 akes with a similar active zone (R l up to 144 A W À1 for a $60 nm thick SnS 2 ake, in Fig. 4f), which has also been documented in other research. 54For a given wavelength range, the photocurrent of photosensitive akes is proportional to the absorption if we consider a constant namely the internal quantum efficiency (h), which can be expressed as , where a is the absorption and d is the thickness of the ake. 55Therefore, the dependence of photoresponsivity on ake thickness may originate from enhanced absorption in thicker samples. 56he positive thickness-dependent photodetection capability may be a challenge for the system-on-a-chip design where a thinner channel is required.Previous work employed the surface sensitization of SnS 2 nanosheets using a 2 nm thick HfO 2 nanolayer grown by atomic layer deposition (ALD). 54owever, this inevitably leads to a complex procedure or a higher production cost.A phototransistor, in which the increase of gate voltage gives rise to an increase in the channel current, may pave an alternative way to address this issue. 62,63ere, the emphasis has been placed on the device (as schematically shown in the inset of Fig. 5b) performance operating under illumination (0.2 mW cm À2 @ 365 nm).The light output characteristics (thickness $ 15 nm) are displayed in Fig. 5a.The output and transfer characteristics measured in darkness are illustrated in Fig. S5.† Both the SnS 2 channels reveal a typical ntype semiconducting behaviour.The correlation between the channel current and back-gate bias (I DS -V GS curves) under illumination and dark conditions is plotted in Fig. 5b.An increase in the channel current can be found as V GS increases, possibly due to the leading role of photo-generation in comparison with tunneling or thermionic currents. 64The falsecolor plot in Fig. 5c provides direct evidence that higher photocurrent could be achieved under a high (positive) gate bias.Thereby, the indicators of light sensors relying on SnS 2 akes, such as responsivity and detectivity (Fig. 5d), could be further improved (up to 140 A W À1 ) through changing the gate voltage.Under a positive gate bias, the Fermi level in the (n-type) semiconductor will approach the conduction band and lead to a reduction of barrier height; as a result, more photo-excited charges could overcome the gate barrier and contribute to an increased carrier density. 65,66Hence, this compatible manufacturing approach could address the positive thicknessdependent sensing capability caused by the enhanced absorption capacity in thicker samples.

Conclusions
In summary, we report high-performance UV photodetectors established on SnS 2 akes and address the relatively low photodetection capability in the thinner akes via a compatible gate-tunable route.Multilayer SnS 2 akes (that can be thinned to $7 nm in this work) were mechanically isolated from CVTgrown high-quality 2H-SnS 2 single crystals whose components are both inexpensive and earth-abundant.The phase and microstructure were unambiguously identied through several characterization techniques, including XRD, XPS, AFM, Raman and TEM.When exposed to UV illumination with a very low power density (0.2 mW cm À2 @ 365 nm), the light sensors using SnS 2 akes exhibit high responsivity (112 A W À1 ), EQE (3.7 Â 10 4 %) and detectivity (1.18 Â 10 11 Jones), comparable or superior to those of reported SnS 2 and other 2D material-based UV photodetectors.Most importantly, SnS 2 akes present a positive thickness-dependent photodetection behaviour, possibly attributed to the enhanced light absorption capacity of thicker samples.However, the responsivity of thinner akes (e.g.$15  Paper nm) can be further improved (up to 140 A W À1 ) under a gate bias of +20 V, comparable with the performance of a non-gated thick ake ($144 A W À1 for a $60 nm ake).Hence, our results offer an efficient way to choose 2D materials with an optimal thickness, and such earth-abundant and environmentally friendly tin-based chalcogenides are highly desirable for sustainable "green" optoelectronics applications.

Fig. 2
Fig. 2 Characterization of SnS 2 flakes.(a and b) Typical AFM images of ultrathin SnS 2 flakes acquired via mechanical exfoliation.(c) Thickness dependent Raman spectra of SnS 2 flakes, inset: enlarged view of the characteristic peak at $205 cm À1 .(d) Change in the intensity of the A 1g peak with layer thickness, inset: an optical image of a SnS 2 flake and the corresponding Raman map of the A 1g mode.(e) PL spectrum of SnS 2 with different thicknesses.(f) The layer-dependent bandgap of few-layer SnS 2 .

Fig. 1
Fig. 1 Synthesis of SnS 2 crystals.(a) Crystal structure (side view) of layered SnS 2 .(b) Schematic diagram showing the CVT process.(c) XRD patterns of the as-synthesized products, inset: the photograph of SnS 2 single crystals.(d) XPS survey of SnS 2 crystals, (e) Sn3d XPS and (f) high resolution S 2p XPS of SnS 2 single crystals.

Fig. 3
Fig. 3 TEM identification of SnS 2 flakes.(a) Low-magnification TEM image of an ultrathin SnS 2 flake.(b) High resolution TEM characterization with atomic mode on the top and (c) the corresponding SAED pattern of the SnS 2 flake.EDX elemental mapping of (d) Sn and (e) S revealing uniform distributions.(f) EDX spectrum of the SnS 2 flake; the inset illustrates the atomic ratio of Sn and S.

Fig. 4
Fig. 4 Photodetectors established on ultrathin SnS 2 flakes.(a) I-V characteristics measured in darkness and under incident light of varied wavelengths (under comparable light intensity).(b) Spectral responsivity of the SnS 2 flake based photodetector, inset: the representative two-terminal devices designed on a thin SnS 2 flake.(c) Time resolved current of the light sensor measured at V DS ¼ 1 V under 365 nm (0.2 mW cm À2 ).(d) Power law fitting photocurrents versus light intensities.(e) Responsivity (R l ) and external quantum efficiency (EQE) plotted as a function of light intensity.(f) Thickness dependent responsivity of SnS 2 flake based devices.

Fig. 5
Fig. 5 Phototransistors based on SnS 2 flakes.(a) Output characteristics of the SnS 2 ($15 nm) transistors operated under incident light of 0.2 mW cm À2 @ 365 nm.(b) I DS -V GS curves under 365 nm light and in darkness at V DS ¼ 1 V, inset: the layout of the phototransistor.(c) Falsecolor plot of the SnS 2 phototransistor exposed to incident light, where the color reflects the intensity of the measured photocurrent.(d) Photoresponsivity and detectivity measured at V DS ¼ 1 V as a function of V GS .

Table 1
Comparison of the 2D UV photodetector performance with those reported by others.T: thickness, ME: mechanical exfoliation, D*: detectivity