Paper-based broadband flexible photodetectors with van der Waals materials

Layered metal chalcogenide materials are exceptionally appealing in optoelectronic devices thanks to their extraordinary optical properties. Recently, their application as flexible and wearable photodetectors have received a lot of attention. Herein, broadband and high-performance paper-based PDs were established in a very facile and inexpensive method by rubbing molybdenum disulfide and titanium trisulfide crystals on papers. Transferred layers were characterized by SEM, EDX mapping, and Raman analyses, and their optoelectronic properties were evaluated in a wavelength range of 405–810 nm. Although the highest and lowest photoresponsivities were respectively measured for TiS3 (1.50 mA/W) and MoS2 (1.13 μA/W) PDs, the TiS3–MoS2 heterostructure not only had a significant photoresponsivity but also showed the highest on/off ratio (1.82) and fast response time (0.96 s) compared with two other PDs. This advantage is due to the band offset formation at the heterojunction, which efficiently separates the photogenerated electron–hole pairs within the heterostructure. Numerical simulation of the introduced PDs also confirmed the superiority of TiS3–MoS2 heterostructure over the other two PDs and exhibited a good agreement with the experimental results. Finally, MoS2 PD demonstrated very high flexibility under applied strain, but TiS3 based PDs suffered from its fragility and experience a remarkable drain current reduction at strain larger than ± 0.33%. However, at lower strains, all PDs displayed acceptable performances.

With the development of smart and Internet of Things (IoT) based technologies, great efforts have been made on wearable and flexible electronic devices which are mostly implemented on Polyethylene terephthalate (PET) and Polydimethylsiloxane (PDMS) substrates 1 . However, in addition to high manufacturing costs and lack of biodegradability, their fabrication processes require advanced laboratory equipment, which obstruct their possible practical applications 2 . Flexible paper-based electronic devices are a new class of high-performance devices that offer admirable properties with facile and low-cost fabrication processes, ending in lightweight and environmentally friendly devices that are very promising for the future of smart electronics [3][4][5] . Flexible photodetectors (PDs), as a member of this family, have received a great deal of attention, suitable for optoelectronic systems such as optical communications, environmental monitoring, and imaging 6 .
Two-dimensional (2D) materials, especially the chalcogenides family, have been increasingly used in optoelectronic devices in recent years due to their adjustable band gap, high electrical conductivity, and effective lightmatter interactions 7 . Metal chalcogenide based PDs have shown high photoresponsivity, fast response time, and high quantum efficiency 8 . Interestingly, their paper-based PDs also demonstrates such a high performance 9 www.nature.com/scientificreports/ paper followed by decorating with various metal nanoparticles (Au, Pt, and Pd) 11 . The photoresponsivities and response times of the PDs were measured to be around 45-100 mA/W and 0.9-1.2 s in the visible range, respectively. Cordeiro et al. introduced near-infrared PDs by growing MoS 2 on cellulose paper using a hydrothermal technique. Their PD showed a photoresponsivity of about 200 mA/W and a response time of 3.7 s 13 . However, majority of methods used for the fabrication of paper-based PDs are time-consuming, complex, and consisting wet procedures. In contrast, rubbing of bulk crystal on ordinary paper can be a potential alternative to facile realization of paper-based PDs through a simple, fast and completely dry process. As an example, Mazaheri et al. transferred MoS 2 flakes by rubbing the corresponding crystal on paper and reported a paper based PD with a photoresponsivity of 1.5 μA/W in a wavelength range of 365 to 940 nm 14 .
In this work, MoS 2 and TiS 3 bulk crystals were employed to fabricate paper-based PDs by transferring their corresponding flakes through finger-rubbing process. TiS 3 is another class of 2D materials with MX 3 structure, which its unique properties make it very suitable for optoelectronic applications 15 . TiS 3 is an n-type semiconductor with a direct band gap of 1.1 eV independent of thickness with high reported electron mobility and photoresponsivity 16 . Unlike MX 2 , it has a chain-like structure that provides large aspect ratios with better electrical connectivity even in lower loading fractions 16 . In addition, the electronic band structure of TiS 3 could be matched with MoS 2 to provide a suitable heterostructure for optoelectronic applications. Accordingly, three types of MoS 2 , TiS 3 , and TiS 3 -MoS 2 PDs were fabricated and their optoelectronic properties were carefully evaluated in the wavelength range of 405 to 810 nm. Based on the results, the highest photoresponsivity were measured for TiS 3 PDs and the lowest for MoS 2 PDs. In the case of TiS 3 -MoS 2 (MoS 2 film was placed on top of the TiS 3 film and exposed to light), it exhibited a faster response time, and higher on-off ratio thanks to formation of band offset at semiconductor heterojunctions. Numerical simulations also confirmed the superiority of TiS 3 -MoS 2 PDs over MoS 2 PDs. Moreover, to evaluate the flexible performance of the PDs, the introduced devices were bent on both positive and negative curvatures and their I ds -V ds and photocurrent characteristics were measured under strain range of ± 0.27 to ± 0.54%. The results showed that MoS 2 PDs had high flexibility, but TiS 3 based PDs were associated with the decline of drain current and photodetection responsivity at applied strains larger than ± 0.33%.

Results and discussions
SEM image and XRD analysis of the grown TiS 3 microcrystals are presented in Fig. S1. Before transferring the flakes, the surface of paper was investigated by SEM and EDX measurements. As can be seen in Fig. S2a, the paper is formed by stacked cellulose fibers and it presents deep gaps/voids between fibers. EDX analysis also confirms the presence of carbon (C), oxygen (O), calcium (Ca), silicon (Si), aluminum (Al), iron (Fe), sodium (Na), and manganese (Mg) in the raw paper where their atomic percentages are listed in the inset of Fig. S2b. Figure 1 shows the characterization of the deposited flakes on paper substrates. The schematic illustration of the TiS 3 -MoS 2 heterostructure and its corresponding photograph are also presented in panels (a) and (b) of Fig. 1, respectively. Figure 1c exhibits the SEM image of the MoS 2 flakes on the paper. According to the figure, MoS 2 flakes cover the fibers and gaps between fibers of the paper. The deposited TiS 3 flakes are also presented in Fig. 1d in which the thickness of the film is measured to be ~ 10 μm. Moreover, the SEM image of the TiS 3 -MoS 2 heterostructure is observed in Fig. 1e. Due to softness and flexibility of MoS 2 , it forms a smoother film than TiS 3 on paper, which is visible in Fig. 1e. The EDX analysis of MoS 2 , TiS 3 , and TiS 3 -MoS 2 samples on papers is presented in Fig. 1f,g. By rubbing MoS 2 on the paper, the Mo and S elements becomes significant (Fig. 1f). The EDX analysis of TiS 3 sample also proves the presence of S and Ti elements according to Fig. 1g. SEM-EDX analysis of transferred TiS 3 flakes is also presented in Fig. S3. In the case of TiS 3 -MoS 2 film, dominant elements are S, Mo, and Ti (Fig. 1h). The C, O and Ca elements come from raw paper. Table S1 summarizes the element contents of all samples with their weight and atomic percentages. Figure 2a exhibits the magnified SEM image of the TiS 3 -MoS 2 film where the MoS 2 film is deposited on the TiS 3 film. However, some TiS 3 microcrystals may also appear on top of the film due to the rubbing process. The Raman spectra of MoS 2 , TiS 3 , and TiS 3 -MoS 2 samples are also presented in Fig. 2b. For MoS 2 , E 1 2g and A 1g peaks are found at 379 and 403 cm −1 , respectively 17 . In the case of TiS 3 , the 193, 385 and 627 cm −1 peaks refer to A g and E g modes, respectively 18 . There is also one additional peak located at 507 cm −1 , which refers to B 1g peak of TiO 2 19 . This oxide peak may have appeared during the growth of TiS 3 microcrystals in the ampoule process. The Raman spectrum of TiS 3 -MoS 2 sample contains all peaks of both structures. The corresponding Ag, E 1 2g , and A 1g peaks of TiS 3 and MoS 2 are also fitted in the Raman spectrum of the hybrid sample. Figure 2c shows the mapping analysis of the TiS 3 -MoS 2 sample where the dominant elements are separately shown in Fig. 2d. Accordingly, O, C, and Ca elements are due to the raw paper, and elements of Ti, Mo, and S originate from the TiS 3 and MoS 2 films. It can be seen that all elements are uniformly distributed on the paper.
To investigate the optoelectronic properties of the films, MoS 2 , TiS 3 , and TiS 3 -MoS 2 PDs were fabricated. Details of their fabrication are provided in the experimental section. The fabrication steps of MoS 2 and TiS 3 PDs are shown in Figs. S4 and S5, respectively. In the case of TiS 3 -MoS 2 PD, the fabrication steps are presented in Fig. 3.
Panels (a) to (c) of Fig. 4 display I-V characteristics of the MoS 2 , TiS 3 , and TiS 3 -MoS 2 PDs at a bias range of − 10 to + 10 V in dark and under 532 nm laser illumination at different power intensities. In all three PDs, significant photocurrents were generated compared to the dark state. The incident power is normalized in terms of laser spot and PD's channel areas. Accordingly, the effective incident power is calculated as: where P laser is the power of the laser, A device is the area of the PD's channel, and A laser is the area of the laser spot. The diameter of the 532 nm laser spot was 2.81 mm and its area (A laser ) was measured to be 6.22 mm 2 . As the laser intensity increases, the photocurrent also increases for all three PDs. In detail, the drain currents were measured where I ph is the photocurrent of the PDs. In the MoS 2 PD, increasing the effective power associates with decrease of the photoresponsivity. The same trends are also observed for TiS 3 , and TiS 3 -MoS 2 PDs. As the incident power increases, more photocarriers are generated, which increases the recombination rate of photogenerated carriers or the possibility of being captured by the traps, leading to a decrease in R 21,22 . In general, the photoresponsivities are measured in the range of 0.4-1.2 μA/W, 0.4-1.6 mA/W, and 90-170 μA/W for MoS 2 , TiS 3 , and TiS 3 -MoS 2 PDs, respectively. Accordingly, the photoresponsivity of the last two PDs is three and two orders of magnitude greater than that of MoS 2 PD. Figure 5a-c show the measured photocurrent in terms of the different effective powers of the 532 nm laser. Accordingly, an increase in the optical powers leads to an increase in photocurrent in all PDs. Panels (d) to (f)    In the hybrid PD, MoS 2 is at the top and is more exposed to light, so the final behavior of the hybrid is more tended toward MoS 2 than TiS 3 . For this reason, its photoresponsivity is associated with improvement compared with the MoS 2 PD due to the formation of band offset and efficient charge separation. Moreover, the channel length of TiS 3 is about 1 mm in the hybrid PD, which is half that of individual TiS 3 PD. Hence, the needle-like TiS 3 flake has lower series resistance and therefore shows a faster response time 25,26 . Furthermore, due to the growth process, TiS 3 probably has more defects than MoS 2 , so it has a slower response time 27 . However, some part of the TiS 3 is buried under (passivated by) MoS 2 in the hybrid PD and is resulted in less ambient gas absorption, which speeds up the response time 27 . Figure 6a compares the photoresponsivity of several fabricated PDs at an applied voltage of 10 V under a laser wavelength of 532 nm. It is observed that all three TiS 3 PDs have the highest photoresponsivity, followed by TiS 3 -MoS 2 and MoS 2 samples, respectively. In general, the R are measured in the range of 0.67-1.56 mA/W for TiS 3 PDs and 0.20-1.93 μA/W in the case of MoS 2 PDs. For TiS 3 -MoS 2 PDs, these values are measured in the range of 0.08 to 0.19 mA/W. Accordingly, MoS 2 has the lowest photoresponsivity and TiS 3 possess the highest photoresponsivity. However, both photoresponsivity and photocurrent of TiS 3 -MoS 2 PDs are considerable compared to the MoS 2 PD. Moreover, a similar trend is observed in the order of magnitude of all calculated R in these PDs. Figure 6b shows the photoresponsivity of all three devices in terms of different laser wavelengths at the  Since the light detection mechanism is based on the change of drain current in these PDs, less dark current provides better performance in light detection as will be discussed below 29 . Another important parameters are the rise and fall times which are presented in Fig. 6d. Accordingly, the TiS 3 -MoS 2 PD has the fastest response time to laser radiation than the other two PDs. Figure 6e shows the energy band diagram of the TiS 3 -MoS 2 heterostructure and its corresponding photodetection mechanism. TiS 3 is an n-type semiconductor with an energy band gap of 1.1 eV and MoS 2 is also an n-type semiconductor with an energy band gap in the range of 1.2-1.8 eV 30,31 . Here, the energy band gap of bulk MoS 2 is considered because most layers are thick, although the deposited film can contain single and few layers of MoS 2 . The electron affinities of MoS 2 and TiS 3 are around 4 and 4.7 eV, respectively, and hence the conduction band of TiS 3 is located of MoS 2 , forming an n-n + heterostructure 32,33 . In the MoS 2 film under laser irradiation, electron-hole pairs are generated and the photogenerated electrons enter into the TiS 3 due to its lower conduction band which prevents electrons from recombination with the holes. Moreover, the valence band of MoS 2 is located higher than that of TiS 3 resulting in transferring of minority holes from TiS 3 into MoS 2 . Hence, the recombination rate of the carriers is decreased in the heterostructure. Such charge separation in the heterostructure is responsible for faster response time, and improved photoresponsivity of the heterostructure compared with individual MoS 2 PDs. In order to investigate the effect of thicknesses, the performance of the three hybrid PDs was evaluated with three different TiS 3 thicknesses. In this regard, thicknesses of ~ 2, ~ 10 and ~ 20 µm were prepared where dark current and photoresponsivity of the PDs were measured as reported in the Fig. S7. Based on the result, as the thickness of TiS 3 increases, the dark current increases but the photoresponsivity decreases. Less thickness of TiS 3 leads to more changes in electrical current, which can be due to non-uniformity of the deposited film. Moreover, as the thickness increases, the upper layers are more involved in light absorption and the lower layers play a lesser role in the photogenerated carriers. Therefore, the ratio of photogenerated to non-photogenerated carriers decreases, which leads to a decrease in photoresponsivity. As can be seen, the thickness of ~ 10 µm generally shows better performance compared to less and more thicknesses.
To evaluate the air stability of the PDs, the performance of the fresh devices was also investigated after two weeks based on the measured photocurrent. Based on the Fig. S8, it can be seen that after two weeks of device life, the performance of the PDs is very close to the fresh states, which show their high environmental stability.  www.nature.com/scientificreports/ Since the substrate is made of paper, the performance of the introduced PDs was thoroughly investigated under applied strain. For this purpose, a homemade setup was fabricated to apply strain to the substrates as shown in Fig. S9. This setup includes a cage covered with aluminum foil to act as a Faraday cage. The sample is fixed to a motor shaft through a hook to provide upward and downward strains by moving in a clockwise or counterclockwise direction. The strain applied to the PDs was calculated according to ε = t/2R equation, where t is the thickness of the substrate and r is the radius of bending curvature 34 . Figure S10 presents the applied strains in the PDs, which are performed in the range of − 0.54% to + 0.54%˚. As can be seen, at strains larger than ± 0.33%, a significant bending is occurred in the substrates.   www.nature.com/scientificreports/ found that the I ds decreases after applying downward strains, but a more significant decrease is observed in the upward strains. Moreover, after upward strain, by returning the TiS 3 to its flat position, current does not return to its original value of before any strain. The structure of the needle-like TiS 3 flake is more fragile than that of MoS 2 , and bending can break the contact between TiS 3 and Ag paste, affecting the conductivity 37 . As a result, even after the sample has returned to the flat state, the current dramatically suffers from decline. In detail, by applying + 0.57% strain, a relative change of drain current |ΔI/I flat | is measured to be ~ 88% compared to the flat state, which indicates the high fragility of the TiS 3 flakes within the channel. A similar trend is observed in the TiS 3 -MoS 2 PD, but with a more noticeable decrease after positive strains, which could be due to the separation of the two films at the interface and the reduction of the electric field in the channel. Figure 7f-h show the photocurrent characteristics of all PDs under 532 nm laser irradiation at a drain voltage of 10 V and an incident power of 15 mW. For MoS 2 PD, the photocurrent is well switched under negative and positive strains of 0.33% similar to the flat state. The lack of change in the photocurrent as well as the slight change in the drain current under the applied strain indicate that the tensile and compressive strains have no effect on the band gap, density of state of carriers, and barrier height at the contacts of the MoS 2 PDs. This is probably due to the presence of MoS 2 flakes of different thicknesses and their polycrystalline nature which minimize the effect of strain 36 . In the case of TiS 3 , the photocurrent is about 1.00 μA in the flat state, which is reduced to 0.50 and 0.25 μA in the applied strains of − 0.33 and + 0.33%. This reduction in photocurrent is not due to a change in the energy gap or piezoresistive effect, but rather to the breaking of the needle-like TiS 3 flakes in the channel, which leads to their less contribution in the drain current and generation of photocarriers. For TiS 3 -MoS 2 , the current is about 250 nA in the flat state, which is reduced to ~ 70 and ~ 10 nA by applying positive and negative strains. As discussed, the brittle structure of TiS 3 as well as the separation of the two films at the junction, especially at a strain of + 0.33% are the main factors in the TiS 3 -MoS 2 PD performance drop under applied strains.  www.nature.com/scientificreports/ Numerical simulation is employed to present further insight about the physics of light-induced generationrecombination processes for charged carriers in these devices. We used the model to calculate the photocurrent of the devices in response to the incident light under various biasing conditions. Details of the simulation can be found in the supporting information section. Panels (a) to (c) of Fig. 8 depict the schematic of the devices consisting MoS 2 , TiS 3 and heterostructure with stacked layers of both materials. Overlapped part in the heterostructure device was exposed to the incident light with the wavelength of 532 nm. Stray electromagnetic field due to unintentional exposure is also took into the account. Corresponding light-induced electric field can be seen in Fig. 8d for all devices due to laser exposure. Vertical electric field was established by applying a voltage difference between the two terminals at both ends, where contacts are located. Figure 8e-f show the electron and hole concentrations under this biasing condition in a logarithmic scale, respectively. As can be seen, the distribution of electrons and holes in the two structures of MoS 2 and TiS 3 is almost uniform, but in the TiS 3 -MoS 2 device, the band offset accelerates the separation of electrons and holes and causes a non-uniform distribution of carriers at the two-layer boundary. These results indicate that the formed n-n + heterostructure can more efficiently separate the charge and confirms its superior optoelectronic performance compared with the individual MoS 2 case.
At various light intensities, I-V characteristic curves has been simulated for MoS 2 , TiS 3 , and TiS 3 -MoS 2 photodetectors. The obtained curves for drain currents are consistent with the measurements as illustrated in Fig. S11a-c. Relative differences between the amplitude of photocurrent is in all devices is in total agreement with the experiments where TiS 3 and MoS 2 PDs have higher and lower photocurrents, respect to the hybrid structure under identical biasing and light exposure conditions. Table 1 compares the performance of the introduced PDs with some other reported paper-based PDs. According to it, most of them have a photoresponsivity in the range of a few μA/W. In the case of MoS 2 , it is observed that our introduced PD has a slightly higher photoresponsivity than the others. The TiS 3 PD shows a much higher photoresponsivity (in the range of mA/W), which has a significant improvement in performance compared to other reported PDs. However, it shows a larger dark current and a smaller on/off ratio than MoS 2 PDs which can  www.nature.com/scientificreports/ limit its optical detection performance. Finally, the photoresponsivity of the TiS 3 -MoS 2 heterostructure shows a significant improvement, which indicates its superiority over other works.
Conclusions. The two-dimensional (2D) layered family of transition metals chalcogenides shows high potential as photodetectors because they not only interact effectively with light but also provide high carrier transport properties. In this report, a very simple, fast and promising clean method is introduced to fabricate the paper-based photodetectors using MoS 2 , TiS 3 and their integration. The all introduced photodetectors show remarkable photoresponsivities in the range of 405 to 810 nm. In the case of TiS 3 -MoS 2 heterostructure, it associates with fast response time and large on-off ratio compared to individual MoS 2 and TiS 3 photodetectors. The numerical simulation results are consistent with the experimental results and confirm the superiority of the hybrid structure over the two other PDs. Moreover, the bending results of the photodetectors indicate that in applied strains smaller than ± 0.33%, these devices still show acceptable performance.

Methods
Material. Naturally occurring molybdenite crystal, from Wolfram Camp Mine, Queensland, Australia, were used. Silver paste was purchased from Dycotec Materials, UK DM-SIP-3060S. Ordinary papers (CopiMax, Thailand) were used as substrates.
Growth of TiS 3 microcrystals. TiS 3 microcrystals were synthesized by a solid-gas reaction of titanium powder through direct sulfurization. In detail, titanium powder (Goodfellow, > 99% purity) and sulfur powder (Merck, > 99.9% purity) were vacuum sealed in an ampule and heated up to 500 °C for 20 h. Then, the ampule was cooled down to room temperature and grown microcrystals were collected.

Fabrication of MoS 2 (TiS 3 ) paper-based photodetector.
First, a piece of paper with 2 × 6 cm 2 dimensions was cut and attached on a slide glass using adhesive tape. A shadow mask (a 1 × 6 mm 2 rectangular window) was then opened in the center of the paper with adhesive tape. A piece of bulk MoS 2 (or TiS 3 ) was rubbed several times along the length and width of the open-window reign. After deposition of a homogeneous layer of MoS 2 (TiS 3 ), the mask was removed from the paper. Silver paste was then placed on the both sides of the deposited film, and two pieces of soft copper wire were mounted on the paste and kept at room temperature for 24 h for improving the adhesion. Then, the dried silver paste was covered with epoxy glue for strengthening and achieving higher mechanical stability. The photograph of the fabrication steps of MoS 2 and TiS 3 PDs are presented in the supporting file.

Fabrication of TiS 3 -MoS 2 paper-based photodetector.
First, a piece of paper with a dimension of 2 × 6 cm 2 was cut and attached to a slide glass. A rectangular shadow mask (mask 1) was opened by the adhesive tape in the center of the paper. With the use of another mask (mask 2), this part was divided into two sections with dimensions of 1 × 4 mm 2 (for TiS 3 deposition) and 1 × 2 mm 2 (for MoS 2 deposition). A piece of bulk TiS 3 was rubbed several times in the open-window region. Then, mask 2 was removed from the paper and MoS 2 crystal was finger-rubbed several times on the 1 × 2 mm 2 part of the bare paper and 1 × 1 mm 2 part of the deposited TiS 3 . Thus, the center of the sample (device) with dimensions of 1 × 1 mm 2 was a film composed of TiS 3 and MoS 2 , in which TiS 3 was placed beneath the MoS 2 film. After removing mask 1 from the paper, a homogeneous film composed of TiS 3 , MoS 2 and, TiS 3 -MoS 2 with dimensions of 1 × 3, 1 × 1, and 1 × 1 mm 2 can be achieved, respectively. Similar to previous PDs, copper wires were added to both sides of the film. The TiS 3 -MoS 2 PD is ready for testing.
Characterizations. SEM images and EDX analyses were taken by TESCAN MIRA 3 and EDX (Energydispersive X-ray) systems. Crystalline structure was identified by using a Panalytical X`Pert Pro X-ray diffractometer at glancing angle configuration (incident angle of 1.7°) with CuKα radiation. Raman spectra of the transferred flakes were measured by Avantes (AVASPEC-ULS3648-RS) system under 532 nm laser illumination. Electrical measurements were carried out by a Keithley 2450 source meter. Optical measurements were done under different laser excitations including 405, 532, 655, and 810 nm at room temperature in air ambient. Bending tests were performed with a homemade setup capable of applying both upward and strains.