Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Photo-tunable transfer characteristics in MoTe2–MoS2 vertical heterostructure

Photo-tunable transfer characteristics in MoTe2–MoS2 vertical heterostructure www.nature.com/npj2dmaterials ARTICLE OPEN Photo-tunable transfer characteristics in MoTe –MoS vertical 2 2 heterostructure 1 1 2 1 2 1 1 Arup Kumar Paul , Manabendra Kuiri , Dipankar Saha , Biswanath Chakraborty , Santanu Mahapatra , A. K Sood and Anindya Das Fabrication of the out-of-plane atomically sharp p–n junction by stacking two dissimilar two-dimensional materials could lead to new and exciting physical phenomena. The control and tunability of the interlayer carrier transport in these p–n junctions have a potential to exhibit new kind of electronic and optoelectronic devices. In this article, we present the fabrication, electrical, and opto- electrical characterization of vertically stacked few-layers MoTe (p)–single-layer MoS (n) heterojunction. Over and above the 2 2 antiambipolar transfer characteristics observed similar to other hetero p–n junction, our experiments reveal a unique feature as a dip in transconductance near the maximum. We further observe that the modulation of the dip in the transconductance depends on the doping concentration of the two-dimensional flakes and also on the power density of the incident light. We also demonstrate high photo-responsivity of ~10 A/W at room temperature for a forward bias of 1.5 V. We explain these new findings based on interlayer recombination rate-dependent semi-classical transport model. We further develop first principles-based atomistic model to explore the charge carrier transport through MoTe –MoS heterojunction. The similar dip is also observed in the 2 2 transmission spectrum when calculated using density functional theory–non-equilibrium Green’s function formalism. Our findings may pave the way for better understanding of atomically thin interface physics and device applications. npj 2D Materials and Applications (2017) 1:17 ; doi:10.1038/s41699-017-0017-3 INTRODUCTION the position of the dip in I –V curve can be modulated by DS BG changing the incident power density of light. This unique In recent years, van der Waals (vdW) heterostructures based on observation can be explained by including interlayer recombination transition metal dichalcogenides (TMDs) are being studied rate of charge carriers along with the individual layer response. We extensively, due to their excellent electronic and opto-electronic 1–4 5 also report very high photo-responsivity of ~10 A/W under forward properties with potential applications such as transistor, photo 6, 7 8–10 11, 12 bias 1.5 V at room temperature. It is seen that the photo-responsivity detector, light-emitting diode (LED), and solar cells. for blue (450 nm) and red (650 nm) light is few orders of magnitude Atomically sharp interfaces with intralayer high carrier mobility higher than near infrared (IR) (850 nm) light. We have further carried and lack of dangling bonds result in unique spatial charge out first principles-based density functional theory (DFT)- none- separation between the layers, as well as produce long-lived quilibrium Green’s function (NEGF) calculations, which do capture interlayer excitons under light exposure. These TMD-based the evolution of the above mentioned dip. vertical heterostructures have shown potentials as p–n junction and most of these p–n junctions show antiambipolar transcon- 15–21 ductance behavior. However, p–n junction made of single 22–25 RESULTS layer of TMDs, known as atomically thin p–n junctions, show Experimental results without light quite different type of charge transport mechanism compared to the conventional p–n junction. Here, transport occurs via The schematic of the device and the experimental setup are shown 6, 25 tunneling of carriers from one layer to the other layer. This in Fig. 1a. Oxidized silicon wafer with 285-nm thick SiO layer was tunneling at the interfaces is predicted to be governed by used as a substrate for the heterojunction devices. Few-layer MoTe Shockley-Read-Hall (SRH) or by Langevin recombination mechan- was exfoliated on the wafer, followed by transfer of a single-layer 25 26 isms. The band engineering of vertical heterostructures with MoS flake using well-known polydimethylsiloxane (PDMS) dry available TMDs having different band gaps and work functions transfer technique. The metal contacts on individual flakes were have paved the way to investigate the charge transport made using standard e-beam lithography, followed by thermal mechanisms in atomically thin p–n junction. deposition of Ti/Au contacts (5/70 nm). For the electrical measure- In this work, we have carried out electrical and photo-conductivity ments, DC bias (V ) was applied and the output current (I )was DS DS measurements of vertical heterostructures made of a few-layers measured with a current amplifier as shown in Fig. 1a. In order to MoTe (~6 layers) and single-layer MoS . The junction current (I ) change the carrier concentration, the gate voltage (V ) was applied 2 2 DS BG ++ vs. bias (V ) data show the characteristics of a p–njunction. to the P Si substrate. The fabrication details as well as DS Interestingly, the I as a function of back gate voltage (V )show an measurement details have been mentioned in the “Methods” DS BG unusual dip at the highest conductance value on top of the usual section. Figure 1b shows the optical as well as atomic force 15–21 antiambipolar nature. We further show that the magnitude and microscope (AFM) image of one of the devices, termed as D1. As 1 2 Department of Physics, Indian Institute of Science, Bangalore 560012, India and Department of Electronic Systems Engineering, Indian Institute of Science, Bangalore 560012, India Correspondence: Anindya Das ([email protected]) Received: 9 December 2016 Revised: 25 May 2017 Accepted: 30 May 2017 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Fig. 1 a Schematics of the device, with the experimental setup. Bias (V ) was applied to MoTe flake and current output was measured with a DS 2 ++ current amplifier (CA) connected to MoS flake. Gate voltage (V ) was applied to P Si of the substrate. LED was used to illuminate the 2 BG sample for opto-electronics measurements. b Optical image (left) and AFM image (right) of the device. The scale bar (black) for optical and AFM images are, respectively, 2 μm and 1 μm. The color scale bar for AFM image is 85 nm. Inset is MoTe height profile along the dashed line. c Raman and PL (inset) of MoS . d Raman for MoTe 2 2 seen in Fig. 1b, the measured height of the MoTe flake was 4 nm, the layer goes to the insulating state. The current is maximum 28, 29 corresponding to six layers. Figure 1d depicts the Raman near V ~−10 V, as expected from the conductance of individual BG spectra of the MoTe flake, which shows the characteristic in plane layers seen in Fig. 2a. Though this series resistance model is able −1 24–29 E peak at 234 cm . The Raman spectra for MoS flake is to explain the antiambipolar nature of the transconductance, it is 2g showninFig. 1c. Here, the MoS characteristic E and out of plane inadequate to explain the unusual dip near the conductance 2g −1 A peaks can be seen, respectively, at 386 and 405 cm . The inset maximum near −10 V. Inset of Fig. 2c shows the I –V response 2g DS BG of Fig. 1cshows theMoS photoluminescence (PL) spectra, which 2 from yet another device (D2), which also shows similar dip near shows a single peak near 625 nm. Both Raman and PL verifies that the conductance maximum (V ~−13 V). However, for D3 and D4 BG 30, 31 the MoS flakeisasinglelayer. 2 devices (Fig. 2d) no dip was observed. In the next paragraph we Electrical and opto-electronic measurements were performed at will show that the dip can be tailored with the incident light. The −5 room temperature in a vacuum of 10 mbar on seven devices. detailed characteristics of these devices are given in SI. Though the responses were similar in all the devices, the magnitude of the response varied from sample to sample, Experimental results with light depending on the quality of the contacts as well as the quality To further elucidate the I –V curves,wehavecarried outthe DS BG of the heterojunction interface. Here, we discuss four devices, for transport measurements with the exposure of light for blue which best results were obtained. Figure 2a shows conductance (wavelength λ ~450 nm), red (λ ~650 nm), and near IR (λ ~850 nm) (G) as a function of V for individual MoS (red curve) and MoTe BG 2 2 LEDs on the D3 device. It was observed that the device response (blue curve) for the D1 device. As can be seen in Fig. 2a, MoS flake changed drastically with exposure of light, and even after LED was is n-doped with a threshold voltage (V )~−20 V and MoTe flake th 2 switched off the changes were retained for a long time ~24 h. First, is p-doped with V ~+5 V. The I –V response of individual th DS DS I –V measurements are made as a function of incident light DS BG flakes, shown in the Supplementary Information (SI), indicates intensity (in increasing order), and then with the light off gate absence of any significant Schottky barriers at the contacts. response is measured at regular intervals of time for several hours Figure 2b displays the I –V response of the junction for several DS DS until the response curve is same as the initial state. Similar results V , showing rectification behavior, expected for the p–n junction BG were obtained in repeated runs of light-on/light-off cycles. Figure 3a at the interface. Even though the observed I –V curve is DS DS shows the evolution of the gate response (at V =1.5 V) under blue DS qualitatively similar to a conventional p–n junction, the current LED for the second cycle of measurements. It can be seen that with does not increase exponentially with forward bias; rather, it increasing light intensity the peak value of I increases and the DS increases almost linearly, pointing to the underlying rectification position of the peak systematically shifts toward higher negative gate mechanism being different. The I –V plots for different V DS BG DS voltages. This shift of position (ΔV )ofpeak I as a function of BG DS are shown in Fig. 2c. The overall antiambipolar shape of the curves power density is shown in Fig. 3d(green curve). As can be seen in can be understood qualitatively by considering the resistances of Fig. 3a, the V on the left side of the peak moves toward more th MoS and MoTe parts in series. At large positive V (V >20V) negative V without affecting significantly the V on the right side. 2 2 BG BG BG th and negative V (V < −20 V) the current is very small, as one of This suggests that with increasing light intensity the MoS flake gets BG BG 2 npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. (b) (a) MoS MoTe 2 2 (c) (d) Fig. 2 a Gate response of individual MoS (red curve) and MoTe (blue curve) for device D1. b I –V response of the junction. Responses for 2 2 DS DS V −20 V to −5 V shows clear rectification behavior. c I –V response of the junction from V −3 V to +3 V in 0.5 V steps. An anomalous dip BG DS BG DS can be observed near the maximum of current in forward bias. Inset shows I –V response for device D2. Here also similar dip can be DS BG observed. d I –V response for device D3 and inset shows the same for device D4. No dip was observed for D3 and D4 DS BG more n-doped, while the change in MoTe doping is very small. Most compared for several V , which shows the dependence of 2 BG interestingly, the dip in gate response curve near the maximum I photo-responsivity on V . The highest photo response is DS DS starts appearing with increasing optical power and then vanishes at achieved at V = 1.5 V, which has been plotted in Fig. 4b. The DS higher light intensities. In the inset of Fig. 3a, we show gate response nature of plots is very similar to MoS -graphene heterostruc- for three different light intensities during the first cycle of the tures. As can be seen, for red and blue light, photo-responsivity measurement. The magnitude of the dip (ΔI ) as a function of ~10 A/W has been achieved. For IR light (1.45 eV), the photo- DIP power density is plotted in Fig. 3d(blue curve). A comparison of I – responsivity is several orders of smaller (Fig. 4b) as the photon DS V characteristics under dark and illuminated condition at V = −30 energy is below the band gap of single-layer MoS . DS BG Vhas been showninFig. 3b. The overall I –V curve under DS DS illumination is similar to that of the device in on-state (SI). It may be DISCUSSION noted that the device goes from completely off-state to on-state with light. This behavior could be understood as following: at V = −30 V Understanding the dip in photo response within a classical model BG the junction does not conduct in dark state because MoS is We use the interlayer recombination model to understand the completely depleted and starts conducting on light exposure, as results presented in Fig. 3. Under forward bias the charge carriers MoS gets doped. Inset of Fig. 3b shows the response of the device conduct via tunneling-mediated interlayer recombination as I ∝ DS (at V =1V, V = −30 V) to a very short light pulse. Though DS BG R V , where R is the recombination rate. R is proportional to C DS C C switching time from off to on state is less than ~600 mS (limited by n /τ, where n is effective carrier concentration and τ is eff eff our measuring instrument), the device shows a strong memory effect interlayer tunneling time. Here, n = n p /(n + p ), where n and eff S T S T S as I reduces slowly with time. In Fig. 3c, the evolution of gate DS p are the electron and hole concentrations of individual flakes, response with time is shown after the LED was switched off (LED off which are tuned by gate voltage as well as light intensity. In Fig. 5, part of the second cycle) and it takes almost a day for the response the effect of photo-doping on R has been discussed. Here, we to come back to its initial condition. Interestingly, here also we see 25 have used the SRH mechanism for R . However, similar kind of the dip appearing and again disappearing with time. The transport characteristics can be obtained from Langevin pro- quantitative measure of ΔI with time has been plotted in Fig. 3e. 25 DIP cesses. The photo-induced doping mechanism has been illustrated schematically in Fig. 5a, where the gate voltage drops Photo-responsivity mostly across the SiO layer and few-layers MoTe . The incident 2 2 photons create electron-hole pairs in both MoS and MoTe , but Similar to the results for blue LED shown in Fig. 3, the experiments 2 2 due to the presence of electric field (negative V ) the excited were performed with red and IR LEDs. The overall transconduc- BG electrons (holes) will move toward the MoS layer (MoTe –SiO tance behavior of the device was similar, except for the magnitude 2 2 2 interface), moving the Fermi level in MoS closer to the of I , for all the LEDs, as shown in Fig. 4a. I is taken as the DS ph 2 conduction band (Fig. 5a). However, the photo-excited holes in difference of I of the gate response under illumination to the peak corresponding I in gate response without illumination. As shown MoTe get trapped mostly at the MoTe –SiO interface as well as DS 2 2 2 in Fig. 4a, I is maximum for the red light. In SI Fig. S5 I –V at the defect states inside the MoTe closer to its valence band, ph DS DS 2 responses under dark and illuminated condition has been which are mostly immobile (reflected in poor mobility <1 cm /Vs) Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17 Photo-tunable transfer characteristics AK Paul et al. (a) (b) led pulse led (d) (c) (e) Fig. 3 a Evolution of I –V response of device D3, with increasing light intensity (blue) during second cycle. The dip can be seen appearing DS BG and again disappearing with increasing optical power. Inset shows evolution of I –V during first cycle. b Comparison of I –V response DS BG DS DS under dark and illuminated condition. Inset shows led pulse experiment, which shows memory effect. c Evolution of I –V with time after led DS BG was switched off. Here also the dip can be seen appearing after few hours, and then again vanishing. d, e Evolution of dip magnitude (ΔI ) DIP with optical power density and time (a) (b) V = 1.5 V DS V = 1.5 V DS Fig. 4 For device D3. a Photo current (I ) as function of optical power density for red (red curve), blue (blue curve), and IR (green curve) lEDs. b ph Photo-responsivity for red, blue, and IR light at V = 1.5 V. Though for red and blue the photo-responsivity was similar and of the order of DS 10 A/W, for IR it is several orders of magnitude small 32–35 and hence do not participate in conduction. This also explains MoS toward left side by ΔV comparable to the experimental 2 BG why it takes almost a day to come back to its initial value after the values (Fig. 3d). To explain the transconductance data of our light is off (Fig. 3c). The photo-induced effect has been device in Fig. 3, we need to consider resistances coming from both quantitatively presented in Fig. 5b. The induced carrier concentra- the individual layers (non-overlapped) as well as the junction part tion (n) as a function of V in MoS and MoTe has been (overlapped). However, the gate response of the individual flakes BG 2 2 generated using capacitive model, n = n + C (V –V )/e, where (non-overlapped) changes less significantly with light (see SI), and 0 G BG th C is the geometrical capacitance and n is the majority carrier hence we consider only the effect of the junction. In the bottom G 0 concentration due to trap and defect states for V < V . In this panel of Fig. 5b, we have plotted R as a function of V for BG th C BG 13 −2 model, we have taken n =3×10 m , which is similar to ref. 25. different amount of photo-induced doping as shown in Fig. 5b With increasing optical power density we have shifted the V of (upper panel). This classical model qualitatively shows the th npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Fig. 5 a Schematic diagram of photo-induced doping process of MoS and MoTe flakes. b Capacitive model to simulate effect of photo 2 2 doping on the carrier concentration of individual flakes. Upper panel shows carrier concentration (n) vs. V plot for both flakes. For MoS , BG 2 n–V curves with different V has been plotted considering its pronounced photo-doping effect. On the other hand, one fixed n–V has BG th BG been plotted for MoTe , as it shows much less photo-doping effect. In the lower panel, R –V graphs has been plotted for the corresponding 2 C BG MoS n–V curves. The curves marked by “a” and “b” are used for G calculation. c G as function of V for two different V of MoS 2 BG eff L BG th 2 corresponding to the “a” and “b” n–V curves. Inset shows the individual resistance R and R as functions of V . d G as function of V .As BG S T BG eff BG can be seen G captures the dip to peak transition eff transformation from a dip to no-dip in the modulation of (perpendicular to the plane) we incorporate a vacuum region of 15 recombination rate by the photo-induced doping. Å(sufficient to avoid any spurious interaction between periodic To understand how R influences the overall transconductance images). Here, the left (t_left ~16.57 Å) and the right (t_right ~16.57 Å) electrodes are acting as the semi-infinite reservoirs. Here, after curve, we need to consider the lateral series resistance (R ) coming we will call this device structure as vdW interface1. Besides, the from the individual flakes at junction, as well as the tunneling zoomed region shown within dotted arrows portrays the top view resistance (R ), which is inversely proportional to R . R is equal to j C L of the channel of MoS –MoTe vdW interface (having an width ~1 R + R , where R and R are lateral resistances coming from 2 2 S T S T nm). The details of the calculation are given in the SI. individual MoS and MoTe , respectively. In Fig. 5c, the inset shows 2 2 Next, we consider two more variants of the aforementioned R and R as functions of V . These have been obtained by fitting S T BG device (namely, vdW interface2 and vdW interface3), where the the experimental G–V curve of MoS and MoTe , respectively BG 2 2 MoS and the MoTe flakes are of doped with different doping 2 2 (for device D3). Figure 5c shows the plot of total lead conductance −1 concentrations. In ATK (Atomistix Tool Kit), the effective doping (n- G =(R + R ) vs. V for two different V of MoS (differing by L S T BG th 2 type or p-type) can be achieved by means of incorporating ΔV ~2.5 V) corresponding to the curves marked by “a” and “b” in BG 42, 43 appropriate compensation charge to the system. This Fig. 5b(lower panel). Please note that Fig. 5c does not capture the effective doping scheme (atomic compensation charge) is really dip observed in our experiments. Next, we show the effect on the advantageous, as it does not depend either on the exact transconductance by including the series resistance R along with −1 dimensions of the system or on the specific details of the dopant the tunneling resistance R . In Fig. 5d, we plot G =(R + R ) as a j eff L j atoms. Considering moderate and high doping densities function V corresponding to the same “a” and “b” curves in BG 17 −3 19 −3 3.85 × 10 cm and 1.28 × 10 cm , we set the “atomic Fig. 5b. It can be seen that the dip can be qualitatively explained compensation charge” values, which are equivalent to two- by inclusion of the interlayer recombination rate. 14 −2 15 −2 dimensional densities 2.56 × 10 m and 8.5 × 10 m , respec- tively. For vdW interface2 and vdW interface3, the MoS flakes are DFT–NEGF calculation n-doped, whereas the MoTe flakes are p-doped, with the In order to get insights on the charge carrier transport through corresponding “atomic compensation charge” values. It is MoTe –MoS vdW interface, we have also carried out an atomistic 2 2 important to realize that, for any DFT–NEGF simulation, the 36–41 study on the two port device structures as illustrated in Fig. 6. electrodes act as the carrier reservoirs to maintain equilibrium for As shown in Fig. 6a, the length of the channel region (L) is taken as the entire system. Moreover, taking vdW interface3 into ~6.6 nm (along the transport direction, i.e., the Z-axis), whereas the consideration, we find that the energy-position resolved local MoS –MoTe overlapping distance is maintained as ~2.2 nm (which density of states (LDOS) diagram (as shown in Fig. 6) clearly 2 2 is 1/3rd of the total channel length). Moreover, along the X-axis represents the band alignment of a type-II heterostructure. Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17 Photo-tunable transfer characteristics AK Paul et al. (a) (b) Fig. 6 a (top) MoTe –MoS vdW interface, where length of the channel is ~6.6 nm and width is ~1 nm. (bottom) Energy-position resolved LDOS 2 2 diagram for the vdW interface3. b Transmission spectra of vdW interface1 (upper left), vdW interface2 (upper right), and vdW interface3 (bottom left). (bottom right) The schematic illustrates the formation of electric dipoles along the channel Figure 6b illustrates the zero bias transmission spectra of the lower-energy range values (which is ~1.8 eV, in this case). Even MoS –MoTe vdW interfaces. It is quite interesting to observe that though both classical and DFT–NEFG calculations qualitatively 2 2 the transmission spectra obtained for the vdW interface2 and vdW support our observation, a complete understanding of the interface3 structures are quite different from that of the vdW transformation of the dip need further theoretical work. interface1. For the intrinsically charge neutralized MoS and In summary, we demonstrate electrical and opto-electronic MoTe sheets of vdW interface1, we see no significant dip in measurements on MoTe –MoS vdW heterojunction devices. In 2 2 states, within the positive energy range of 0 to 2 eV. However, for normal transconductance measurements, some of the devices the samples those are effectively doped into n-type and p-type, showed an anomalous dip in current near the maximum con- we notice the distinct effect of charge separation. The excess ductance state. We showed that the dip can be modulated with light negative and positive charge carriers across the individual layers intensity. We have correlated this anomalous feature within a classical (Fig. 6b, bottom right) give rise to strong interlayer coupling and model based on interlayer recombination processes, followed by modulates the transmission states within the range of 0–2 eV. electrical transport calculations using DFT–NEGF. The first principles- Here, we emphasize on the positive energy transmission states of based quantum transport calculation qualitatively capture the the individual vdW interfaces. It is worth mentioning that for near- anomalous feature depending on the carrier concentration of the equilibrium electrical current calculation, the states within the bias individual flakes. Our devices also show large photo response of window ultimately matters. So for any finite bias, the states that ~10 A/W at room temperature, which makes it a potential candidate are closer to the Fermi energy (E ) will play the dominant role. for charge integrating type opto-electronic applications. Nonetheless, it can be seen from Fig. 6b that the states corresponding to V.B. have further been shifted away from max the energy zero (i.e., E−E = 0) for the vdW interface2 and the vdW METHODS interface3. Considering the energy range of 0–2 eV, it is evident The heterojunction devices were fabricated on Si/SiO substrate with 285-nm from Fig. 6b(upper left) that there will be no anomalous dip in the thick SiO layer. The MoS and MoTe flakes were exfoliated on PDMS and 2 2 2 maximum current (at forward bias) for the vdW interface1. substrate, respectively, from bulk crystal using scotch tape method. To make However, this is not the same for other device(s) with excess n- heterojunction devices, the thinnest flakes were identified with optical microscope. Then Raman and PL spectroscopy was used to sort out single- type or p-type carriers. Investigating the transmission spectra, as layer MoS and non-contact AFM was used to identify 6–8layersofMoTe illustrated in Fig. 6b(upper right and bottom left), we observe clear 2 2 flakes. All the AFM characterizations were performed with Park NX10 AFM. dips (pointed by arrows) around 1.8 eV. A quantitative calculation After transferring the MoS flakes on the MoTe flakes, the eletrical contacts 2 2 of transmission spectra as a function of back gate voltage is, were patterned using standard e-beam lithography. Then Ti/Au (5/70 nm) however, beyond the scope of our work at this stage. were deposited on the lithography-patterned sample, by thermal evapora- To understand the physics behind the atypical nature of the tion to realize the electrical contacts. The electrical and opto-electronic transmission spectra obtained for vdW interface2 and vdW measurements were performed inside a home-built cryostat, under vacuum interface3, we propose a plausible explanation. It is known from −5 of 10 mbar at room temperature. For all DC characterization, Keythly the literature that the excess negative and positive charge 2400 source-meter, Agilent 34401A digital multimeter, and a home-built low- carriers across the individual layers (Fig. 6b) make interlayer noise current amplifier was used. For opto-electrical measurements, the leds 13, 14, 45 electric dipoles, which could be the permanent one. were fixed at ~2 mm distance from the devices. The leds were illuminated Nevertheless, this will give rise to a strong interlayer coupling, and with Kythly 2400 in current source mode. The led power was calibrated with perhaps shift-in the trenches of the transmission spectra to the PM203 Thorlabs optical power meter. npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Data availability 17. Yi, S.-G. et al. Optoelectric properties of gate-tunable MoS /WSe heterojunction. 2 2 IEEE Trans. Nanotechnol. 15, 499–505 (2016). The authors declare that the main data supporting the findings of this 18. Wang, Z., He, X., Zhang, X.-X. & Alshareef, H. N. Hybrid van der Waals p–n het- study are available within the paper and its SI file. Other relevant data are erojunctions based on SnO and 2D MoS . Adv. Mater. 28, 9133–9141 (2016). available from the corresponding author upon request. 19. Li, Y. et al. Anti-ambipolar field-effect transistors based on few-layer 2D transition metal dichalcogenides. ACS Appl. Mater. Interfaces 8, 15574–15581 (2016). ACKNOWLEDGEMENTS 20. Jariwala, D. et al. Gate-tunable carbon nanotube–MoS heterojunction pn diode. Proc. Natl. Acad. Sci. 110, 18076–18080 (2013). The device fabrication was performed using facilities at CeNSE, funded by 21. Kim, J.-K. et al. Trap-mediated electronic transport properties of gate-tunable Department of Information Technology, Govt. of India, and located at Indian pentacene/MoS pn heterojunction diodes. Sci. Rep. 6, 36775 (2016). Institute of Science, Bangalore. A.K.S. thanks Department of Science and Technology 22. Cheng, R. et al. Electroluminescence and photocurrent generation from atomically (DST), India for financial support. A.D. thanks DST and Indian Space Research sharp WSe /MoS heterojunction p–ndiodes. Nano Lett. 14,5590–5597 (2014). 2 2 Organization (ISRO) for financial support. This work is supported by Department of 23. Deng, Y. et al. Black phosphorus–monolayer MoS van der Waals heterojunction Science and Technology (DST), Government of India, under Grant No: DSTO/PPH/ 2 p–n diode. ACS Nano 8, 8292–8299 (2014). AYD/1470, and Indian Space Research Organization (ISRO), Government of India, 24. Pezeshki, A., Shokouh, S. H. H., Nazari, T., Oh, K. & Im, S. Electric and photovoltaic under Grant No: ISTC/PPH/AYD/0343. S.M. acknowledges the support by Science and behavior of a few-layer α-MoTe /MoS dichalcogenide heterojunction. Adv. Engineering Research Board (SERB), Department of Science and Technology (DST), 2 2 Mater. 28, 3216–3222 (2016). Government of India, under Grant No: SB/S3/EECE/0209/2015. 25. Lee, C.-H. et al. Atomically thin p–n junctions with van der Waals heterointerfaces. Nat. Nanotechnol. 9, 676–681 (2014). AUTHOR CONTRIBUTIONS 26. Kang, J., Tongay, S., Zhou, J., Li, J. & Wu, J. Band offsets and heterostructures of two-dimensional semiconductors. Appl. Phys. Lett. 102, 012111 (2013). A.P., A.K.S., and A.D. conceived the idea of this research. A.D. and A.P. together 27. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensional materials designed the experimental setup. A.P. performed the heterojunction device by all-dry viscoelastic stamping. 2D Materials 1, 011002 (2014). fabrication, electrical and opto-electronic characterization of the heterojunctions, 28. Pradhan, N. R. et al. Field-effect transistors based on few-layered α-MoTe . ACS and data analysis. M.K. optimized the transfer technique. B.C. did the Raman and Pl Nano 8, 5911–5920 (2014). characterizations. D.S. and S.M. developed the atomistic device model and performed 29. Lin, Y.-F. et al. Ambipolar MoTe transistors and their applications in logic circuits. the DFT–NEFG calculations. All author contributed in writing the manuscript. Adv. Mater. 26, 3263–3269 (2014). 30. Yin, Z. et al. Single-layer MoS phototransistors. ACS Nano 6,74–80 (2011). ADDITIONAL INFORMATION 31. Late, D. J., Liu, B., Matte, H. R., Dravid, V. P. & Rao, C. Hysteresis in single-layer MoS field effect transistors. ACS Nano 6, 5635–5641 (2012). Supplementary Information accompanies the paper on the npj 2D Materials and 32. Roy, K. et al. Graphene-MoS hybrid structures for multifunctional photo- Applications website (doi:10.1038/s41699-017-0017-3). responsive memory devices. Nat. Nanotechnol. 8, 826–830 (2013). 33. Li, X. et al. Persistent photoconductivity in two-dimensional Mo1- x WxSe –MoSe 2 2 Competing interests: The authors declare no competing financial interests. van der Waals heterojunctions. J. Mater. Res. 31, 923–930 (2016). 34. Wu, Y.-C. et al. Extrinsic origin of persistent photoconductivity in monolayer MoS Publisher’s note: Springer Nature remains neutral with regard to jurisdictional field effect transistors. Sci. Rep. 5, 11472 (2015). claims in published maps and institutional affiliations. 35. Li, T., Du, G., Zhang, B. & Zeng, Z. Scaling behavior of hysteresis in multilayer MoS field effect transistors. Appl. Phys. Lett. 105, 093107 (2014). 36. Atomistix ToolKit (ATK) with Virtual NanoLab, Version 2015.1, Last accessed_Dec. REFERENCES 2016, QuantumWise A/S, http://quantumwise.com/. 1. Novoselov, K., Mishchenko, A., Carvalho, A. & Neto, A. C. 2D materials and van der 37. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made Waals heterostructures. Science 353, aac9439 (2016). simple. Phys. Rev. Lett. 77, 3865 (1996). 2. Das, S., Robinson, J. A., Dubey, M., Terrones, H. & Terrones, M. Beyond graphene: 38. Ozaki, T. Variationally optimized atomic orbitals for large-scale electronic struc- progress in novel two-dimensional materials and van der Waals solids. Ann. Rev. tures. Phys. Rev. B 67, 155108 (2003). Mater. Res. 45,1–27 (2015). 39. Ozaki, T. & Kino, H. Numerical atomic basis orbitals from h to kr. Phys. Rev. B 69, 3. Jariwala, D., Marks, T. J. & Hersam, M. C. Mixed-dimensional van der Waals het- 195113 (2004). erostructures. Nat. Mater. 16, 170–181 (2016). 40. Grimme, S. Semiempirical GGA-type density functional constructed with a long- 4. Bablich, A., Kataria, S. & Lemme, M. C. Graphene and two-dimensional materials range dispersion correction. J. Comput. Chem. 27, 1787–1799 (2006). for optoelectronic applications. Electronics 5, 13 (2016). 41. Brandbyge, M., Mozos, J.-L., Ordejón, P., Taylor, J. & Stokbro, K. Density-functional 5. Roy, T. et al. Dual-gated MoS /WSe van der Waals tunnel diodes and transistors. 2 2 method for nonequilibrium electron transport. Phys. Rev. B 65, 165401 (2002). ACS Nano 9, 2071–2079 (2015). 42. Stradi, D., Martinez, U., Blom, A., Brandbyge, M. & Stokbro, K. General atomistic 6. Zhang, K. et al. Interlayer transition and infrared photodetection in atomically approach for modeling metal-semiconductor interfaces using density functional thin type-ii MoTe /MoS van der Waals heterostructures. ACS Nano 10, 2 2 theory and nonequilibrium Green’s function. Phys. Rev. B 93, 155302 (2016). 3852–3858 (2016). 43. Soler, J. M. et al. The SIESTA method for ab initio order-n materials simulation. J. 7. Li, Y., Xu, C.-Y., Wang, J.-Y. & Zhen, L. Photodiode-like behavior and excellent pho- Phys. Condens. Matter 14, 2745 (2002). toresponse of vertical si/monolayer MoS heterostructures. Sci. Rep. 4, 7186 (2014). 2 44. Zhang, Y. et al. Thickness considerations of two-dimensional layered semi- 8. Withers, F. et al. Light-emitting diodes by band-structure engineering in van der conductors for transistor applications. Sci. Rep. 6, 29615 (2016). Waals heterostructures. Nat. Mater. 14, 301–306 (2015). 45. Yu, Y. et al. Equally efficient interlayer exciton relaxation and improved absorption 9. Ross, J. S. et al. Electrically tunable excitonic light-emitting diodes based on in epitaxial and nonepitaxial MoS /WS heterostructures. Nano Lett. 15,486–491 2 2 monolayer WSe pn junctions. Nat. Nanotechnol. 9, 268–272 (2014). 2 (2014). 10. Withers, F. et al. WSe light-emitting tunneling transistors with enhanced brightness at room temperature. Nano Lett. 15, 8223–8228 (2015). Open Access This article is licensed under a Creative Commons 11. Flöry, N. et al. A WSe /MoSe heterostructure photovoltaic device. Appl. Phys. Lett. 2 2 Attribution 4.0 International License, which permits use, sharing, 107, 123106 (2015). adaptation, distribution and reproduction in any medium or format, as long as you give 12. Furchi, M. M., Pospischil, A., Libisch, F., BurgdoÌ’Lrfer, J. & Mueller, T. Photovoltaic appropriate credit to the original author(s) and the source, provide a link to the Creative effect in an electrically tunable van der Waals heterojunction. Nano Lett. 14, Commons license, and indicate if changes were made. The images or other third party 4785–4791 (2014). material in this article are included in the article’s Creative Commons license, unless 13. Fogler, M., Butov, L. & Novoselov, K. High-temperature superfluidity with indirect indicated otherwise in a credit line to the material. If material is not included in the excitons in van der Waals heterostructures. Nat. Commun. 5, 4555 (2014). article’s Creative Commons license and your intended use is not permitted by statutory 14. Rivera, P. et al. Observation of long-lived interlayer excitons in monolayer regulation or exceeds the permitted use, you will need to obtain permission directly MoSe –WSe heterostructures. Nat. Commun. 6, 6242 (2015). 2 2 from the copyright holder. To view a copy of this license, visit http://creativecommons. 15. Jariwala, D. et al. Hybrid, gate-tunable, van der Waals p–n heterojunctions from org/licenses/by/4.0/. pentacene and MoS . Nano Lett. 16, 497–503 (2015). 16. Jariwala, D. et al. Large-area, low-voltage, antiambipolar heterojunctions from © The Author(s) 2017 solution-processed semiconductors. Nano Lett. 15, 416–421 (2014). Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png npj 2D Materials and Applications Springer Journals

Photo-tunable transfer characteristics in MoTe2–MoS2 vertical heterostructure

Download PDF
7 pages

Loading next page...
 
/lp/springer-journals/photo-tunable-transfer-characteristics-in-mote2-mos2-vertical-OzrOIpXwLX

References (46)

Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Materials Science; Materials Science, general; Nanotechnology; Surfaces and Interfaces, Thin Films
eISSN
2397-7132
DOI
10.1038/s41699-017-0017-3
Publisher site
See Article on Publisher Site

Abstract

www.nature.com/npj2dmaterials ARTICLE OPEN Photo-tunable transfer characteristics in MoTe –MoS vertical 2 2 heterostructure 1 1 2 1 2 1 1 Arup Kumar Paul , Manabendra Kuiri , Dipankar Saha , Biswanath Chakraborty , Santanu Mahapatra , A. K Sood and Anindya Das Fabrication of the out-of-plane atomically sharp p–n junction by stacking two dissimilar two-dimensional materials could lead to new and exciting physical phenomena. The control and tunability of the interlayer carrier transport in these p–n junctions have a potential to exhibit new kind of electronic and optoelectronic devices. In this article, we present the fabrication, electrical, and opto- electrical characterization of vertically stacked few-layers MoTe (p)–single-layer MoS (n) heterojunction. Over and above the 2 2 antiambipolar transfer characteristics observed similar to other hetero p–n junction, our experiments reveal a unique feature as a dip in transconductance near the maximum. We further observe that the modulation of the dip in the transconductance depends on the doping concentration of the two-dimensional flakes and also on the power density of the incident light. We also demonstrate high photo-responsivity of ~10 A/W at room temperature for a forward bias of 1.5 V. We explain these new findings based on interlayer recombination rate-dependent semi-classical transport model. We further develop first principles-based atomistic model to explore the charge carrier transport through MoTe –MoS heterojunction. The similar dip is also observed in the 2 2 transmission spectrum when calculated using density functional theory–non-equilibrium Green’s function formalism. Our findings may pave the way for better understanding of atomically thin interface physics and device applications. npj 2D Materials and Applications (2017) 1:17 ; doi:10.1038/s41699-017-0017-3 INTRODUCTION the position of the dip in I –V curve can be modulated by DS BG changing the incident power density of light. This unique In recent years, van der Waals (vdW) heterostructures based on observation can be explained by including interlayer recombination transition metal dichalcogenides (TMDs) are being studied rate of charge carriers along with the individual layer response. We extensively, due to their excellent electronic and opto-electronic 1–4 5 also report very high photo-responsivity of ~10 A/W under forward properties with potential applications such as transistor, photo 6, 7 8–10 11, 12 bias 1.5 V at room temperature. It is seen that the photo-responsivity detector, light-emitting diode (LED), and solar cells. for blue (450 nm) and red (650 nm) light is few orders of magnitude Atomically sharp interfaces with intralayer high carrier mobility higher than near infrared (IR) (850 nm) light. We have further carried and lack of dangling bonds result in unique spatial charge out first principles-based density functional theory (DFT)- none- separation between the layers, as well as produce long-lived quilibrium Green’s function (NEGF) calculations, which do capture interlayer excitons under light exposure. These TMD-based the evolution of the above mentioned dip. vertical heterostructures have shown potentials as p–n junction and most of these p–n junctions show antiambipolar transcon- 15–21 ductance behavior. However, p–n junction made of single 22–25 RESULTS layer of TMDs, known as atomically thin p–n junctions, show Experimental results without light quite different type of charge transport mechanism compared to the conventional p–n junction. Here, transport occurs via The schematic of the device and the experimental setup are shown 6, 25 tunneling of carriers from one layer to the other layer. This in Fig. 1a. Oxidized silicon wafer with 285-nm thick SiO layer was tunneling at the interfaces is predicted to be governed by used as a substrate for the heterojunction devices. Few-layer MoTe Shockley-Read-Hall (SRH) or by Langevin recombination mechan- was exfoliated on the wafer, followed by transfer of a single-layer 25 26 isms. The band engineering of vertical heterostructures with MoS flake using well-known polydimethylsiloxane (PDMS) dry available TMDs having different band gaps and work functions transfer technique. The metal contacts on individual flakes were have paved the way to investigate the charge transport made using standard e-beam lithography, followed by thermal mechanisms in atomically thin p–n junction. deposition of Ti/Au contacts (5/70 nm). For the electrical measure- In this work, we have carried out electrical and photo-conductivity ments, DC bias (V ) was applied and the output current (I )was DS DS measurements of vertical heterostructures made of a few-layers measured with a current amplifier as shown in Fig. 1a. In order to MoTe (~6 layers) and single-layer MoS . The junction current (I ) change the carrier concentration, the gate voltage (V ) was applied 2 2 DS BG ++ vs. bias (V ) data show the characteristics of a p–njunction. to the P Si substrate. The fabrication details as well as DS Interestingly, the I as a function of back gate voltage (V )show an measurement details have been mentioned in the “Methods” DS BG unusual dip at the highest conductance value on top of the usual section. Figure 1b shows the optical as well as atomic force 15–21 antiambipolar nature. We further show that the magnitude and microscope (AFM) image of one of the devices, termed as D1. As 1 2 Department of Physics, Indian Institute of Science, Bangalore 560012, India and Department of Electronic Systems Engineering, Indian Institute of Science, Bangalore 560012, India Correspondence: Anindya Das ([email protected]) Received: 9 December 2016 Revised: 25 May 2017 Accepted: 30 May 2017 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Fig. 1 a Schematics of the device, with the experimental setup. Bias (V ) was applied to MoTe flake and current output was measured with a DS 2 ++ current amplifier (CA) connected to MoS flake. Gate voltage (V ) was applied to P Si of the substrate. LED was used to illuminate the 2 BG sample for opto-electronics measurements. b Optical image (left) and AFM image (right) of the device. The scale bar (black) for optical and AFM images are, respectively, 2 μm and 1 μm. The color scale bar for AFM image is 85 nm. Inset is MoTe height profile along the dashed line. c Raman and PL (inset) of MoS . d Raman for MoTe 2 2 seen in Fig. 1b, the measured height of the MoTe flake was 4 nm, the layer goes to the insulating state. The current is maximum 28, 29 corresponding to six layers. Figure 1d depicts the Raman near V ~−10 V, as expected from the conductance of individual BG spectra of the MoTe flake, which shows the characteristic in plane layers seen in Fig. 2a. Though this series resistance model is able −1 24–29 E peak at 234 cm . The Raman spectra for MoS flake is to explain the antiambipolar nature of the transconductance, it is 2g showninFig. 1c. Here, the MoS characteristic E and out of plane inadequate to explain the unusual dip near the conductance 2g −1 A peaks can be seen, respectively, at 386 and 405 cm . The inset maximum near −10 V. Inset of Fig. 2c shows the I –V response 2g DS BG of Fig. 1cshows theMoS photoluminescence (PL) spectra, which 2 from yet another device (D2), which also shows similar dip near shows a single peak near 625 nm. Both Raman and PL verifies that the conductance maximum (V ~−13 V). However, for D3 and D4 BG 30, 31 the MoS flakeisasinglelayer. 2 devices (Fig. 2d) no dip was observed. In the next paragraph we Electrical and opto-electronic measurements were performed at will show that the dip can be tailored with the incident light. The −5 room temperature in a vacuum of 10 mbar on seven devices. detailed characteristics of these devices are given in SI. Though the responses were similar in all the devices, the magnitude of the response varied from sample to sample, Experimental results with light depending on the quality of the contacts as well as the quality To further elucidate the I –V curves,wehavecarried outthe DS BG of the heterojunction interface. Here, we discuss four devices, for transport measurements with the exposure of light for blue which best results were obtained. Figure 2a shows conductance (wavelength λ ~450 nm), red (λ ~650 nm), and near IR (λ ~850 nm) (G) as a function of V for individual MoS (red curve) and MoTe BG 2 2 LEDs on the D3 device. It was observed that the device response (blue curve) for the D1 device. As can be seen in Fig. 2a, MoS flake changed drastically with exposure of light, and even after LED was is n-doped with a threshold voltage (V )~−20 V and MoTe flake th 2 switched off the changes were retained for a long time ~24 h. First, is p-doped with V ~+5 V. The I –V response of individual th DS DS I –V measurements are made as a function of incident light DS BG flakes, shown in the Supplementary Information (SI), indicates intensity (in increasing order), and then with the light off gate absence of any significant Schottky barriers at the contacts. response is measured at regular intervals of time for several hours Figure 2b displays the I –V response of the junction for several DS DS until the response curve is same as the initial state. Similar results V , showing rectification behavior, expected for the p–n junction BG were obtained in repeated runs of light-on/light-off cycles. Figure 3a at the interface. Even though the observed I –V curve is DS DS shows the evolution of the gate response (at V =1.5 V) under blue DS qualitatively similar to a conventional p–n junction, the current LED for the second cycle of measurements. It can be seen that with does not increase exponentially with forward bias; rather, it increasing light intensity the peak value of I increases and the DS increases almost linearly, pointing to the underlying rectification position of the peak systematically shifts toward higher negative gate mechanism being different. The I –V plots for different V DS BG DS voltages. This shift of position (ΔV )ofpeak I as a function of BG DS are shown in Fig. 2c. The overall antiambipolar shape of the curves power density is shown in Fig. 3d(green curve). As can be seen in can be understood qualitatively by considering the resistances of Fig. 3a, the V on the left side of the peak moves toward more th MoS and MoTe parts in series. At large positive V (V >20V) negative V without affecting significantly the V on the right side. 2 2 BG BG BG th and negative V (V < −20 V) the current is very small, as one of This suggests that with increasing light intensity the MoS flake gets BG BG 2 npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. (b) (a) MoS MoTe 2 2 (c) (d) Fig. 2 a Gate response of individual MoS (red curve) and MoTe (blue curve) for device D1. b I –V response of the junction. Responses for 2 2 DS DS V −20 V to −5 V shows clear rectification behavior. c I –V response of the junction from V −3 V to +3 V in 0.5 V steps. An anomalous dip BG DS BG DS can be observed near the maximum of current in forward bias. Inset shows I –V response for device D2. Here also similar dip can be DS BG observed. d I –V response for device D3 and inset shows the same for device D4. No dip was observed for D3 and D4 DS BG more n-doped, while the change in MoTe doping is very small. Most compared for several V , which shows the dependence of 2 BG interestingly, the dip in gate response curve near the maximum I photo-responsivity on V . The highest photo response is DS DS starts appearing with increasing optical power and then vanishes at achieved at V = 1.5 V, which has been plotted in Fig. 4b. The DS higher light intensities. In the inset of Fig. 3a, we show gate response nature of plots is very similar to MoS -graphene heterostruc- for three different light intensities during the first cycle of the tures. As can be seen, for red and blue light, photo-responsivity measurement. The magnitude of the dip (ΔI ) as a function of ~10 A/W has been achieved. For IR light (1.45 eV), the photo- DIP power density is plotted in Fig. 3d(blue curve). A comparison of I – responsivity is several orders of smaller (Fig. 4b) as the photon DS V characteristics under dark and illuminated condition at V = −30 energy is below the band gap of single-layer MoS . DS BG Vhas been showninFig. 3b. The overall I –V curve under DS DS illumination is similar to that of the device in on-state (SI). It may be DISCUSSION noted that the device goes from completely off-state to on-state with light. This behavior could be understood as following: at V = −30 V Understanding the dip in photo response within a classical model BG the junction does not conduct in dark state because MoS is We use the interlayer recombination model to understand the completely depleted and starts conducting on light exposure, as results presented in Fig. 3. Under forward bias the charge carriers MoS gets doped. Inset of Fig. 3b shows the response of the device conduct via tunneling-mediated interlayer recombination as I ∝ DS (at V =1V, V = −30 V) to a very short light pulse. Though DS BG R V , where R is the recombination rate. R is proportional to C DS C C switching time from off to on state is less than ~600 mS (limited by n /τ, where n is effective carrier concentration and τ is eff eff our measuring instrument), the device shows a strong memory effect interlayer tunneling time. Here, n = n p /(n + p ), where n and eff S T S T S as I reduces slowly with time. In Fig. 3c, the evolution of gate DS p are the electron and hole concentrations of individual flakes, response with time is shown after the LED was switched off (LED off which are tuned by gate voltage as well as light intensity. In Fig. 5, part of the second cycle) and it takes almost a day for the response the effect of photo-doping on R has been discussed. Here, we to come back to its initial condition. Interestingly, here also we see 25 have used the SRH mechanism for R . However, similar kind of the dip appearing and again disappearing with time. The transport characteristics can be obtained from Langevin pro- quantitative measure of ΔI with time has been plotted in Fig. 3e. 25 DIP cesses. The photo-induced doping mechanism has been illustrated schematically in Fig. 5a, where the gate voltage drops Photo-responsivity mostly across the SiO layer and few-layers MoTe . The incident 2 2 photons create electron-hole pairs in both MoS and MoTe , but Similar to the results for blue LED shown in Fig. 3, the experiments 2 2 due to the presence of electric field (negative V ) the excited were performed with red and IR LEDs. The overall transconduc- BG electrons (holes) will move toward the MoS layer (MoTe –SiO tance behavior of the device was similar, except for the magnitude 2 2 2 interface), moving the Fermi level in MoS closer to the of I , for all the LEDs, as shown in Fig. 4a. I is taken as the DS ph 2 conduction band (Fig. 5a). However, the photo-excited holes in difference of I of the gate response under illumination to the peak corresponding I in gate response without illumination. As shown MoTe get trapped mostly at the MoTe –SiO interface as well as DS 2 2 2 in Fig. 4a, I is maximum for the red light. In SI Fig. S5 I –V at the defect states inside the MoTe closer to its valence band, ph DS DS 2 responses under dark and illuminated condition has been which are mostly immobile (reflected in poor mobility <1 cm /Vs) Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17 Photo-tunable transfer characteristics AK Paul et al. (a) (b) led pulse led (d) (c) (e) Fig. 3 a Evolution of I –V response of device D3, with increasing light intensity (blue) during second cycle. The dip can be seen appearing DS BG and again disappearing with increasing optical power. Inset shows evolution of I –V during first cycle. b Comparison of I –V response DS BG DS DS under dark and illuminated condition. Inset shows led pulse experiment, which shows memory effect. c Evolution of I –V with time after led DS BG was switched off. Here also the dip can be seen appearing after few hours, and then again vanishing. d, e Evolution of dip magnitude (ΔI ) DIP with optical power density and time (a) (b) V = 1.5 V DS V = 1.5 V DS Fig. 4 For device D3. a Photo current (I ) as function of optical power density for red (red curve), blue (blue curve), and IR (green curve) lEDs. b ph Photo-responsivity for red, blue, and IR light at V = 1.5 V. Though for red and blue the photo-responsivity was similar and of the order of DS 10 A/W, for IR it is several orders of magnitude small 32–35 and hence do not participate in conduction. This also explains MoS toward left side by ΔV comparable to the experimental 2 BG why it takes almost a day to come back to its initial value after the values (Fig. 3d). To explain the transconductance data of our light is off (Fig. 3c). The photo-induced effect has been device in Fig. 3, we need to consider resistances coming from both quantitatively presented in Fig. 5b. The induced carrier concentra- the individual layers (non-overlapped) as well as the junction part tion (n) as a function of V in MoS and MoTe has been (overlapped). However, the gate response of the individual flakes BG 2 2 generated using capacitive model, n = n + C (V –V )/e, where (non-overlapped) changes less significantly with light (see SI), and 0 G BG th C is the geometrical capacitance and n is the majority carrier hence we consider only the effect of the junction. In the bottom G 0 concentration due to trap and defect states for V < V . In this panel of Fig. 5b, we have plotted R as a function of V for BG th C BG 13 −2 model, we have taken n =3×10 m , which is similar to ref. 25. different amount of photo-induced doping as shown in Fig. 5b With increasing optical power density we have shifted the V of (upper panel). This classical model qualitatively shows the th npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Fig. 5 a Schematic diagram of photo-induced doping process of MoS and MoTe flakes. b Capacitive model to simulate effect of photo 2 2 doping on the carrier concentration of individual flakes. Upper panel shows carrier concentration (n) vs. V plot for both flakes. For MoS , BG 2 n–V curves with different V has been plotted considering its pronounced photo-doping effect. On the other hand, one fixed n–V has BG th BG been plotted for MoTe , as it shows much less photo-doping effect. In the lower panel, R –V graphs has been plotted for the corresponding 2 C BG MoS n–V curves. The curves marked by “a” and “b” are used for G calculation. c G as function of V for two different V of MoS 2 BG eff L BG th 2 corresponding to the “a” and “b” n–V curves. Inset shows the individual resistance R and R as functions of V . d G as function of V .As BG S T BG eff BG can be seen G captures the dip to peak transition eff transformation from a dip to no-dip in the modulation of (perpendicular to the plane) we incorporate a vacuum region of 15 recombination rate by the photo-induced doping. Å(sufficient to avoid any spurious interaction between periodic To understand how R influences the overall transconductance images). Here, the left (t_left ~16.57 Å) and the right (t_right ~16.57 Å) electrodes are acting as the semi-infinite reservoirs. Here, after curve, we need to consider the lateral series resistance (R ) coming we will call this device structure as vdW interface1. Besides, the from the individual flakes at junction, as well as the tunneling zoomed region shown within dotted arrows portrays the top view resistance (R ), which is inversely proportional to R . R is equal to j C L of the channel of MoS –MoTe vdW interface (having an width ~1 R + R , where R and R are lateral resistances coming from 2 2 S T S T nm). The details of the calculation are given in the SI. individual MoS and MoTe , respectively. In Fig. 5c, the inset shows 2 2 Next, we consider two more variants of the aforementioned R and R as functions of V . These have been obtained by fitting S T BG device (namely, vdW interface2 and vdW interface3), where the the experimental G–V curve of MoS and MoTe , respectively BG 2 2 MoS and the MoTe flakes are of doped with different doping 2 2 (for device D3). Figure 5c shows the plot of total lead conductance −1 concentrations. In ATK (Atomistix Tool Kit), the effective doping (n- G =(R + R ) vs. V for two different V of MoS (differing by L S T BG th 2 type or p-type) can be achieved by means of incorporating ΔV ~2.5 V) corresponding to the curves marked by “a” and “b” in BG 42, 43 appropriate compensation charge to the system. This Fig. 5b(lower panel). Please note that Fig. 5c does not capture the effective doping scheme (atomic compensation charge) is really dip observed in our experiments. Next, we show the effect on the advantageous, as it does not depend either on the exact transconductance by including the series resistance R along with −1 dimensions of the system or on the specific details of the dopant the tunneling resistance R . In Fig. 5d, we plot G =(R + R ) as a j eff L j atoms. Considering moderate and high doping densities function V corresponding to the same “a” and “b” curves in BG 17 −3 19 −3 3.85 × 10 cm and 1.28 × 10 cm , we set the “atomic Fig. 5b. It can be seen that the dip can be qualitatively explained compensation charge” values, which are equivalent to two- by inclusion of the interlayer recombination rate. 14 −2 15 −2 dimensional densities 2.56 × 10 m and 8.5 × 10 m , respec- tively. For vdW interface2 and vdW interface3, the MoS flakes are DFT–NEGF calculation n-doped, whereas the MoTe flakes are p-doped, with the In order to get insights on the charge carrier transport through corresponding “atomic compensation charge” values. It is MoTe –MoS vdW interface, we have also carried out an atomistic 2 2 important to realize that, for any DFT–NEGF simulation, the 36–41 study on the two port device structures as illustrated in Fig. 6. electrodes act as the carrier reservoirs to maintain equilibrium for As shown in Fig. 6a, the length of the channel region (L) is taken as the entire system. Moreover, taking vdW interface3 into ~6.6 nm (along the transport direction, i.e., the Z-axis), whereas the consideration, we find that the energy-position resolved local MoS –MoTe overlapping distance is maintained as ~2.2 nm (which density of states (LDOS) diagram (as shown in Fig. 6) clearly 2 2 is 1/3rd of the total channel length). Moreover, along the X-axis represents the band alignment of a type-II heterostructure. Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17 Photo-tunable transfer characteristics AK Paul et al. (a) (b) Fig. 6 a (top) MoTe –MoS vdW interface, where length of the channel is ~6.6 nm and width is ~1 nm. (bottom) Energy-position resolved LDOS 2 2 diagram for the vdW interface3. b Transmission spectra of vdW interface1 (upper left), vdW interface2 (upper right), and vdW interface3 (bottom left). (bottom right) The schematic illustrates the formation of electric dipoles along the channel Figure 6b illustrates the zero bias transmission spectra of the lower-energy range values (which is ~1.8 eV, in this case). Even MoS –MoTe vdW interfaces. It is quite interesting to observe that though both classical and DFT–NEFG calculations qualitatively 2 2 the transmission spectra obtained for the vdW interface2 and vdW support our observation, a complete understanding of the interface3 structures are quite different from that of the vdW transformation of the dip need further theoretical work. interface1. For the intrinsically charge neutralized MoS and In summary, we demonstrate electrical and opto-electronic MoTe sheets of vdW interface1, we see no significant dip in measurements on MoTe –MoS vdW heterojunction devices. In 2 2 states, within the positive energy range of 0 to 2 eV. However, for normal transconductance measurements, some of the devices the samples those are effectively doped into n-type and p-type, showed an anomalous dip in current near the maximum con- we notice the distinct effect of charge separation. The excess ductance state. We showed that the dip can be modulated with light negative and positive charge carriers across the individual layers intensity. We have correlated this anomalous feature within a classical (Fig. 6b, bottom right) give rise to strong interlayer coupling and model based on interlayer recombination processes, followed by modulates the transmission states within the range of 0–2 eV. electrical transport calculations using DFT–NEGF. The first principles- Here, we emphasize on the positive energy transmission states of based quantum transport calculation qualitatively capture the the individual vdW interfaces. It is worth mentioning that for near- anomalous feature depending on the carrier concentration of the equilibrium electrical current calculation, the states within the bias individual flakes. Our devices also show large photo response of window ultimately matters. So for any finite bias, the states that ~10 A/W at room temperature, which makes it a potential candidate are closer to the Fermi energy (E ) will play the dominant role. for charge integrating type opto-electronic applications. Nonetheless, it can be seen from Fig. 6b that the states corresponding to V.B. have further been shifted away from max the energy zero (i.e., E−E = 0) for the vdW interface2 and the vdW METHODS interface3. Considering the energy range of 0–2 eV, it is evident The heterojunction devices were fabricated on Si/SiO substrate with 285-nm from Fig. 6b(upper left) that there will be no anomalous dip in the thick SiO layer. The MoS and MoTe flakes were exfoliated on PDMS and 2 2 2 maximum current (at forward bias) for the vdW interface1. substrate, respectively, from bulk crystal using scotch tape method. To make However, this is not the same for other device(s) with excess n- heterojunction devices, the thinnest flakes were identified with optical microscope. Then Raman and PL spectroscopy was used to sort out single- type or p-type carriers. Investigating the transmission spectra, as layer MoS and non-contact AFM was used to identify 6–8layersofMoTe illustrated in Fig. 6b(upper right and bottom left), we observe clear 2 2 flakes. All the AFM characterizations were performed with Park NX10 AFM. dips (pointed by arrows) around 1.8 eV. A quantitative calculation After transferring the MoS flakes on the MoTe flakes, the eletrical contacts 2 2 of transmission spectra as a function of back gate voltage is, were patterned using standard e-beam lithography. Then Ti/Au (5/70 nm) however, beyond the scope of our work at this stage. were deposited on the lithography-patterned sample, by thermal evapora- To understand the physics behind the atypical nature of the tion to realize the electrical contacts. The electrical and opto-electronic transmission spectra obtained for vdW interface2 and vdW measurements were performed inside a home-built cryostat, under vacuum interface3, we propose a plausible explanation. It is known from −5 of 10 mbar at room temperature. For all DC characterization, Keythly the literature that the excess negative and positive charge 2400 source-meter, Agilent 34401A digital multimeter, and a home-built low- carriers across the individual layers (Fig. 6b) make interlayer noise current amplifier was used. For opto-electrical measurements, the leds 13, 14, 45 electric dipoles, which could be the permanent one. were fixed at ~2 mm distance from the devices. The leds were illuminated Nevertheless, this will give rise to a strong interlayer coupling, and with Kythly 2400 in current source mode. The led power was calibrated with perhaps shift-in the trenches of the transmission spectra to the PM203 Thorlabs optical power meter. npj 2D Materials and Applications (2017) 17 Published in partnership with FCT NOVA with the support of E-MRS Photo-tunable transfer characteristics AK Paul et al. Data availability 17. Yi, S.-G. et al. Optoelectric properties of gate-tunable MoS /WSe heterojunction. 2 2 IEEE Trans. Nanotechnol. 15, 499–505 (2016). The authors declare that the main data supporting the findings of this 18. Wang, Z., He, X., Zhang, X.-X. & Alshareef, H. N. Hybrid van der Waals p–n het- study are available within the paper and its SI file. Other relevant data are erojunctions based on SnO and 2D MoS . Adv. Mater. 28, 9133–9141 (2016). available from the corresponding author upon request. 19. Li, Y. et al. Anti-ambipolar field-effect transistors based on few-layer 2D transition metal dichalcogenides. ACS Appl. Mater. Interfaces 8, 15574–15581 (2016). ACKNOWLEDGEMENTS 20. Jariwala, D. et al. Gate-tunable carbon nanotube–MoS heterojunction pn diode. Proc. Natl. Acad. Sci. 110, 18076–18080 (2013). The device fabrication was performed using facilities at CeNSE, funded by 21. Kim, J.-K. et al. Trap-mediated electronic transport properties of gate-tunable Department of Information Technology, Govt. of India, and located at Indian pentacene/MoS pn heterojunction diodes. Sci. Rep. 6, 36775 (2016). Institute of Science, Bangalore. A.K.S. thanks Department of Science and Technology 22. Cheng, R. et al. Electroluminescence and photocurrent generation from atomically (DST), India for financial support. A.D. thanks DST and Indian Space Research sharp WSe /MoS heterojunction p–ndiodes. Nano Lett. 14,5590–5597 (2014). 2 2 Organization (ISRO) for financial support. This work is supported by Department of 23. Deng, Y. et al. Black phosphorus–monolayer MoS van der Waals heterojunction Science and Technology (DST), Government of India, under Grant No: DSTO/PPH/ 2 p–n diode. ACS Nano 8, 8292–8299 (2014). AYD/1470, and Indian Space Research Organization (ISRO), Government of India, 24. Pezeshki, A., Shokouh, S. H. H., Nazari, T., Oh, K. & Im, S. Electric and photovoltaic under Grant No: ISTC/PPH/AYD/0343. S.M. acknowledges the support by Science and behavior of a few-layer α-MoTe /MoS dichalcogenide heterojunction. Adv. Engineering Research Board (SERB), Department of Science and Technology (DST), 2 2 Mater. 28, 3216–3222 (2016). Government of India, under Grant No: SB/S3/EECE/0209/2015. 25. Lee, C.-H. et al. Atomically thin p–n junctions with van der Waals heterointerfaces. Nat. Nanotechnol. 9, 676–681 (2014). AUTHOR CONTRIBUTIONS 26. Kang, J., Tongay, S., Zhou, J., Li, J. & Wu, J. Band offsets and heterostructures of two-dimensional semiconductors. Appl. Phys. Lett. 102, 012111 (2013). A.P., A.K.S., and A.D. conceived the idea of this research. A.D. and A.P. together 27. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensional materials designed the experimental setup. A.P. performed the heterojunction device by all-dry viscoelastic stamping. 2D Materials 1, 011002 (2014). fabrication, electrical and opto-electronic characterization of the heterojunctions, 28. Pradhan, N. R. et al. Field-effect transistors based on few-layered α-MoTe . ACS and data analysis. M.K. optimized the transfer technique. B.C. did the Raman and Pl Nano 8, 5911–5920 (2014). characterizations. D.S. and S.M. developed the atomistic device model and performed 29. Lin, Y.-F. et al. Ambipolar MoTe transistors and their applications in logic circuits. the DFT–NEFG calculations. All author contributed in writing the manuscript. Adv. Mater. 26, 3263–3269 (2014). 30. Yin, Z. et al. Single-layer MoS phototransistors. ACS Nano 6,74–80 (2011). ADDITIONAL INFORMATION 31. Late, D. J., Liu, B., Matte, H. R., Dravid, V. P. & Rao, C. Hysteresis in single-layer MoS field effect transistors. ACS Nano 6, 5635–5641 (2012). Supplementary Information accompanies the paper on the npj 2D Materials and 32. Roy, K. et al. Graphene-MoS hybrid structures for multifunctional photo- Applications website (doi:10.1038/s41699-017-0017-3). responsive memory devices. Nat. Nanotechnol. 8, 826–830 (2013). 33. Li, X. et al. Persistent photoconductivity in two-dimensional Mo1- x WxSe –MoSe 2 2 Competing interests: The authors declare no competing financial interests. van der Waals heterojunctions. J. Mater. Res. 31, 923–930 (2016). 34. Wu, Y.-C. et al. Extrinsic origin of persistent photoconductivity in monolayer MoS Publisher’s note: Springer Nature remains neutral with regard to jurisdictional field effect transistors. Sci. Rep. 5, 11472 (2015). claims in published maps and institutional affiliations. 35. Li, T., Du, G., Zhang, B. & Zeng, Z. Scaling behavior of hysteresis in multilayer MoS field effect transistors. Appl. Phys. Lett. 105, 093107 (2014). 36. Atomistix ToolKit (ATK) with Virtual NanoLab, Version 2015.1, Last accessed_Dec. REFERENCES 2016, QuantumWise A/S, http://quantumwise.com/. 1. Novoselov, K., Mishchenko, A., Carvalho, A. & Neto, A. C. 2D materials and van der 37. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made Waals heterostructures. Science 353, aac9439 (2016). simple. Phys. Rev. Lett. 77, 3865 (1996). 2. Das, S., Robinson, J. A., Dubey, M., Terrones, H. & Terrones, M. Beyond graphene: 38. Ozaki, T. Variationally optimized atomic orbitals for large-scale electronic struc- progress in novel two-dimensional materials and van der Waals solids. Ann. Rev. tures. Phys. Rev. B 67, 155108 (2003). Mater. Res. 45,1–27 (2015). 39. Ozaki, T. & Kino, H. Numerical atomic basis orbitals from h to kr. Phys. Rev. B 69, 3. Jariwala, D., Marks, T. J. & Hersam, M. C. Mixed-dimensional van der Waals het- 195113 (2004). erostructures. Nat. Mater. 16, 170–181 (2016). 40. Grimme, S. Semiempirical GGA-type density functional constructed with a long- 4. Bablich, A., Kataria, S. & Lemme, M. C. Graphene and two-dimensional materials range dispersion correction. J. Comput. Chem. 27, 1787–1799 (2006). for optoelectronic applications. Electronics 5, 13 (2016). 41. Brandbyge, M., Mozos, J.-L., Ordejón, P., Taylor, J. & Stokbro, K. Density-functional 5. Roy, T. et al. Dual-gated MoS /WSe van der Waals tunnel diodes and transistors. 2 2 method for nonequilibrium electron transport. Phys. Rev. B 65, 165401 (2002). ACS Nano 9, 2071–2079 (2015). 42. Stradi, D., Martinez, U., Blom, A., Brandbyge, M. & Stokbro, K. General atomistic 6. Zhang, K. et al. Interlayer transition and infrared photodetection in atomically approach for modeling metal-semiconductor interfaces using density functional thin type-ii MoTe /MoS van der Waals heterostructures. ACS Nano 10, 2 2 theory and nonequilibrium Green’s function. Phys. Rev. B 93, 155302 (2016). 3852–3858 (2016). 43. Soler, J. M. et al. The SIESTA method for ab initio order-n materials simulation. J. 7. Li, Y., Xu, C.-Y., Wang, J.-Y. & Zhen, L. Photodiode-like behavior and excellent pho- Phys. Condens. Matter 14, 2745 (2002). toresponse of vertical si/monolayer MoS heterostructures. Sci. Rep. 4, 7186 (2014). 2 44. Zhang, Y. et al. Thickness considerations of two-dimensional layered semi- 8. Withers, F. et al. Light-emitting diodes by band-structure engineering in van der conductors for transistor applications. Sci. Rep. 6, 29615 (2016). Waals heterostructures. Nat. Mater. 14, 301–306 (2015). 45. Yu, Y. et al. Equally efficient interlayer exciton relaxation and improved absorption 9. Ross, J. S. et al. Electrically tunable excitonic light-emitting diodes based on in epitaxial and nonepitaxial MoS /WS heterostructures. Nano Lett. 15,486–491 2 2 monolayer WSe pn junctions. Nat. Nanotechnol. 9, 268–272 (2014). 2 (2014). 10. Withers, F. et al. WSe light-emitting tunneling transistors with enhanced brightness at room temperature. Nano Lett. 15, 8223–8228 (2015). Open Access This article is licensed under a Creative Commons 11. Flöry, N. et al. A WSe /MoSe heterostructure photovoltaic device. Appl. Phys. Lett. 2 2 Attribution 4.0 International License, which permits use, sharing, 107, 123106 (2015). adaptation, distribution and reproduction in any medium or format, as long as you give 12. Furchi, M. M., Pospischil, A., Libisch, F., BurgdoÌ’Lrfer, J. & Mueller, T. Photovoltaic appropriate credit to the original author(s) and the source, provide a link to the Creative effect in an electrically tunable van der Waals heterojunction. Nano Lett. 14, Commons license, and indicate if changes were made. The images or other third party 4785–4791 (2014). material in this article are included in the article’s Creative Commons license, unless 13. Fogler, M., Butov, L. & Novoselov, K. High-temperature superfluidity with indirect indicated otherwise in a credit line to the material. If material is not included in the excitons in van der Waals heterostructures. Nat. Commun. 5, 4555 (2014). article’s Creative Commons license and your intended use is not permitted by statutory 14. Rivera, P. et al. Observation of long-lived interlayer excitons in monolayer regulation or exceeds the permitted use, you will need to obtain permission directly MoSe –WSe heterostructures. Nat. Commun. 6, 6242 (2015). 2 2 from the copyright holder. To view a copy of this license, visit http://creativecommons. 15. Jariwala, D. et al. Hybrid, gate-tunable, van der Waals p–n heterojunctions from org/licenses/by/4.0/. pentacene and MoS . Nano Lett. 16, 497–503 (2015). 16. Jariwala, D. et al. Large-area, low-voltage, antiambipolar heterojunctions from © The Author(s) 2017 solution-processed semiconductors. Nano Lett. 15, 416–421 (2014). Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2017) 17

Journal

npj 2D Materials and Applications – Springer Journals

Published: Jun 27, 2017

There are no references for this article.