I ntegrated two-dimensional (2D) electronic circuits based on 2D layered semiconductors, including the transitionmetal dichalcogenides (TMDCs) as well as other 2D semiconductors, such as atomically thin black phosphorus and InSe, etc., promise advanced electronics and flexible devices with increased functionality, performance, and scaling in integrated circuits. [1][2][3][4][5][6][7] The electronic and optoelectronic devices performance is significantly affected by the characteristics of the electrical contacts that connect the semiconductor materials with external circuitry. 8,9 As the pristine surface of a 2D material has no dangling bonds, it is difficult to form strong interface bonds with metal, thereby increasing the contact resistance. 8 Moreover, direct metal electrode deposition on the 2D semiconductor surface causes considerable defects, strain, disorder, and metal diffusion, resulting in a glassy layer dominated by interdiffusion and strain. 10 Previous efforts, e.g., aligning the metal work function with the conduction/valence band edge of 2D layered semiconductors, showed unsatisfactory high-resistance contacts, due to Fermi level pinning. 8 The traditional method to reduce the contact resistance for silicon is to decrease the depletion region width by locally doping near the silicon-metal junction. However, the 3D doping method used in silicon technology cannot be employed in 2D devices. Recently, contacting 2D semiconductors using coplanar 2D conductors has shown promise in reducing the problematic high contact resistance. Reduced contact resistances were observed in metallic 1T-MoS 2 /semiconducting 2H-MoS 2 , metallic 1T′-MoTe 2 /semiconducting 2H-MoTe 2 , and metallic VS 2 /semiconducting 2H-MoS 2 with seamless coplanar interfaces, which were fabricated using intercalation, laser heating induced phase transition, and heteroepitaxy growth, respectively. [11][12][13][14] However, many of these methods are not ideal for scaled production. The fabrication of these coplanar contacts required the solution processing, suffering from the substitution of chalcogen atoms, or resulted in random place and orientation. It is imperative for practical 2D applications that a scalable (i.e., compatible with wafer-scale technology) method of spatially controlling the semiconductor and metal coplanar interface is realized. Previous studies revealed that MoTe 2 is particularly interesting for phase-engineering applications, as the free energy difference between semiconducting 2H-MoTe 2 and metallic 1T′-MoTe 2 is much smaller (∼35 meV per MoTe 2 formula unit) compared with those of other TMDC materials. [18][19][20][21] This small energy difference leads to the possibility of phase-controlled synthesis in a large scale.

Here, we demonstrate the large-scale, spatially controlled chemical assembly of 2H-MoTe 2 field-effect transistors (FETs) integrated with coplanar metallic 1T′-MoTe 2 contacts via phase engineering. The heterophase FETs exhibit ohmic contact behavior with a low contact resistance of ∼1.1 kΩ μm at a high doping level and high average carrier mobility of ∼23 cm 2 V -1 s -1 . We also developed a patterned growth method of 1T′-MoTe 2 and realized 1T′-MoTe 2 gated heterophase FET arrays whose components of channel, gate, and contacts are all 2D materials. Each FET has an independently patterned 1T′-MoTe 2 gate electrode, which is required for logic circuitry. Few-layer 2H-MoTe 2 , with a bandgap of ∼1 eV, is a candidate material for near-infrared (NIR) photodetector. [15][16][17] We transferred the heterophase device onto a flexible substrate and demonstrated the NIR photoresponse with enhanced photoresponsivity (∼1.02 A/W) due to the reduced contact resistance. Our study shows the promise of large-scale applications using a 2D coplanar heterophase structure for advanced electronic and optoelectronic devices.

Results and Discussion. Patterned Chemical Assembly of the Heterophase MoTe 2 on a Large Scale. To chemically assemble large-scale coplanar 2H-1T′ heterophase MoTe 2 , a high-quality few-layer 2H-MoTe 2 film was first synthesized on a p + -Si/SiO 2 substrate via chemical vapor deposition (CVD) (Figure 1a). Our previous work has found the synthesis of 2H-MoTe 2 film was driven by the solid-to-solid 1T′ to 2H-MoTe 2 phase transformation, 22 which can be well described by the time-temperature-transformation diagram. First, a 1-1.5 nm thick Mo film was prepared on the Si/SiO 2 substrate via magnetron sputtering. Then, the MoTe 2 films were grown by tellurizing the Mo film at 650 °C for 2 h. During the tellurization, 1T′ phase polycrystalline MoTe 2 film with Te vacancies was first synthesized. As Te atoms occupy the vacancies in the 1T′ phase film, the phase transformation from 1T′ to 2H-MoTe 2 occurs and diffuses into circles, where the recrystallization occurs at the boundary of 1T′-MoTe 2 and 2H-MoTe 2 . As growth time prolongs, all the circles merge and become a uniform 2H phase MoTe 2 film. By controlling the kinetic rates of nucleation and crystal growth, we were able to synthesize large-scale continuous 1T′-MoTe 2 and 2H-MoTe 2 thin films, as well as arbitrary shaped 1T′-MoTe 2 (see Supplementary Section Ia). This method allowed us to also synthesize a centimeter-scale 2H-MoTe 2 thin film with a domain size up to several hundred micrometers under 650 °C for 2 h (Figure 1e) (see details in Methods). The large-scale electron backscatter diffraction (EBSD) map of the assynthesized 2H-MoTe 2 thin film shows a uniform contrast in the out-of-plane direction (see Supplementary Section Ib), indicating that the highly textured MoTe 2 thin film was stacked in the c-axis, with an average domain size up to 486 ± 187 μm in the in-plane direction (Figure 1i). The domain size (d) is calculated based on a circularly equivalent area ( π = d s 4 / ), where s is the statistically averaged domain area. The regular Hall bar structure was fabricated to measure the intrinsic carrier mobility of the synthesized 2H-MoTe 2 (see Supplementary Section Ic). The Hall mobility μ H is extracted to be 95 cm 2 V -1 s -1 at room temperature, indicating the high quality of the 2H-MoTe 2 . The large-scale 2H-MoTe 2 film Hall mobility was also measured via a van Der Pauw method (see Supplementary Section Ic). The hall mobility μ H is extracted to be 31 cm 2 V -1 s -1 , which is lower than that measured by regular Hall bar structure. This is because of the existence of the grain boundaries in the large-scale film, which scatter the carriers and reduce the mobility.

Following the growth, large-scale 2H-MoTe 2 /Mo is fabricated into stripes using photolithography, reactive ion etching (RIE), and magnetron sputtering (Mo deposition), followed by a lift-off process (Figure 1b,f). Subsequently, the sample was sent back into the furnace at a lower temperature (530 °C) for 30 min. After the second CVD growth, the Mo film was tellurized to 1T′-MoTe 2 , while the 2H-MoTe 2 remained unaffected (see Supplementary Section Id). Ultimately, this formed striped patterns of 2H/1T′-MoTe 2 (Figure 1c,g). This was confirmed by the appearance of the well-resolved A g Raman modes (107, 127, 161, and 256 cm -1 ) of 1T′-MoTe 2 and the out-of-plane A 1g (171 cm -1 ), B 2g 1 (291 cm -1 ), and the strong in-plane E 2g (234 cm -1 ) Raman modes of 2H-MoTe 2 in the respective regions (Figure 1j). This heterophase pattern was further fabricated into isolated 1T′/ 2H/1T′-MoTe 2 arrays by photolithography and reactive ion etching (RIE, Figure 1d,h). Using electron beam lithography (EBL) in place of photolithography, we also synthesized arbitrarily controlled coplanar 2H/1T′-MoTe 2 heterophase structures with smaller feature sizes, e.g., Peking University's (PKU) logo (Figure 1k). Raman spectroscopy mapping of the representative 2H-MoTe 2 E 2g mode and 1T′-MoTe 2 A g mode (labeled by the corresponding dashed lines in Figure 1j) clearly shows that both the 2H-MoTe 2 PKU logo and 1T′-MoTe 2 background are uniform, with clear and distinct interfaces (Figure 1l).

Characterizations of the Interface of the Coplanar 2H/1T′ Heterophase MoTe 2 . The interface between 2H-MoTe 2 and 1T′-MoTe 2 was characterized using atomic force microscopy (AFM), the Kelvin probe force microscopy (KPFM) and transmission electron microscopy (TEM). The typical optical image of an isolated 1T′/2H/1T′-MoTe 2 coplanar heterostructure showed a large contrast difference between respective regions (Figure 2a), resulting from the significant dielectric constant difference between the semiconducting 2H-MoTe 2 and metallic 1T′-MoTe 2 . 23 The AFM height image at the 2H/ 1T′-MoTe 2 interface (Figure 2b) exhibited a nearly homogeneous color contrast, indicating the uniformity of the sample thickness. The line profiles of the 2H-MoTe 2 and 1T′-MoTe 2 (Figure 2c) regions showed a similar thickness of about 5 nm (corresponding to 7-layer MoTe 2 ), 24 resulting from the precise control of the Mo film thickness. Additionally, the line profile across the interface (see Supplementary Section IIa) did not show any stitches or flake overlaps, confirming the seamless coplanar contact between the heterophase.

The surface work function image measured by KPFM and the potential line profile (see Supplementary Section IIb) show a sharp demarcation between the 1T′-MoTe 2 and 2H-MoTe 2 , with a potential difference between them of ∼27.5 mV. Notably, the measured higher work function of the 1T′-MoTe 2 is preferable for the ohmic contact of the p-type 2H-MoTe 2 .

To evaluate the crystallinity of the heterophase, we transferred 1T′/2H/1T′-MoTe 2 arrays onto a copper grid (see Supplementary Section IIIa). At the interface of 2H/1T′-MoTe 2 (Figure 2d), the selected area electron diffraction (SAED) pattern (Figure 2e) comprised a single set of diffraction spots with 6-fold symmetry related to the single-crystalline 2H-MoTe 2 together with a series of diffraction rings related to the polycrystalline 1T′-MoTe 2 . 25 The high-resolution TEM (HR-TEM) image (Figure 2f) showed the interface contains polycrystalline monoclinic (1T′) structure domains of about tens of nanometers and a single hexagonal (2H) structure domain. We further used high-angle annular dark-field scanning TEM (HAADF-STEM) to resolve in-plane atomic arrangements of a typical 1T′/2H-MoTe 2 interface (Figure 2g), where an atomically smooth interface transition was observed. The corresponding FFT pattern contained a set of The room-temperature field-effect mobility was estimated to be ∼32 cm 2 V -1 s -1 . (e) Field-effect mobilities for different channel length measured on 100 devices. All the mobilities hover in a range of 20-24 cm 2 V -1 s -1 and are independent of the channel length, due to the large single-crystalline domain of the 2H-MoTe 2 . (f) Contact resistance of the heterophase and 1T′-MoTe 2 only devices extracted using the transfer length method. The total contact resistance shows strong gate dependence, while the contact resistance of 1T′-MoTe 2 is about 0.6 kΩ μm, independent of the gate voltage. The measured minimum value of the total contact resistance is about ∼1.7 kΩ μm, which was obtained at high doping, yielding a coplanar 1T′/2H-MoTe 2 contact resistance of ∼1.1 kΩ μm. monoclinic 1T′ spots and a set of hexagonal 2H spots (Figure 2h). The simulated HAADF-STEM image, as well as the schematic lattices (Figure 2g) at the interface, is provided, which agree well with the observed results. We found the 1T′ domains stitched seamlessly to single-crystalline 2H-MoTe 2 with random orientations at different locations of the interface (see Supplementary Section IIIb). Electrical Properties of the Coplanar Heterophase MoTe 2 FETs. The seamless contact allowed us to inject current into semiconducting 2H-MoTe 2 from metallic 1T′-MoTe 2 electrodes with reduced contact resistance (R c ). By using coplanar seamless contacts, the reduced contact resistance is of great significance to the development of practical 2D semiconductor applications. The large-scale coplanar heterophase MoTe 2 FETs array with different channel lengths was fabricated (Figure 3a). Pd/Au electrodes (10/50 nm) were deposited on the 1T′ parts of each 1T′/2H/1T′-MoTe 2 structure for electrical measurements. The p + -Si was used as the back gate electrode. The devices array with Pd/Au electrodes directly deposited on the 2H phase were fabricated for comparison (see Supplementary Section IVa). With the same channel length, the current injected from coplanar 1T′-MoTe 2 contact was 1 order of magnitude larger than that from the deposited Pd/Au metal contact (Figure 3b), indicative of lower contact resistance at the 1T′/2H-MoTe 2 interface. The typical sourcedrain current (I ds versus source-drain voltage (V ds ) curves measured at various gate voltages (Figure 3c) showed linear behavior, confirming the ohmic contact between the 1T′ and 2H-MoTe 2 . The gate voltage (V g ) dependence of the heterophase FET under a bias voltage of 0.5 V showed ptype channel characteristics with an on-off ratio of ∼1 × 10 4 (Figure 3d). The room-temperature field-effect mobility (μ) was ∼32 cm 2 V -1 s -1 , using μ = (dI ds /dV g ) (L/W) (1/V ds C g ), where L, W, and C g are the channel length, channel width, and the gate capacitance per unit area, respectively. By contrast, the field-effect mobilities extracted from Pd/Au contacted 2H-MoTe 2 FETs are only 2 cm 2 V -1 s -1 (see Supplementary Section IVa). We measured 100 FETs with different change lengths and found that the average mobility (in a range of 20-24 cm 2 V -1 s -1 , comparable to the reported values for exfoliated 2H-MoTe 2 single crystals) 12,13 was independent of the channel length (Figure 3e). This was not surprising, as our average 2H-MoTe 2 single-crystalline domain size (∼486 μm) was much larger than the largest channel length (60 μm) being measured. Hence, there were unlikely to be many grain boundaries that could act as scattering centers.

To determine the contact resistance between the 1T′-MoTe 2 and 2H-MoTe 2 , we used the transfer length method (TLM), 9 in which the channel length-dependent resistances of both the 1T′/2H/1T′ coplanar heterophase structure (see Supplementary Section IVb) and 1T′-MoTe 2 (see Supplementary Section IVc) were measured. The total R c of the metal/1T′/2H-MoTe 2 structure contains the R c of metal/1T′-MoTe 2 and the R c of 1T′/2H-MoTe 2 . The total R c (metal/ 1T′/2H-MoTe 2 ) shows a strong dependence on the back gate voltage due to the semiconducting nature of 2H-MoTe 2 . In contrast, due to the metallic nature of 1T′-MoTe 2 , the R c of metal/1T′-MoTe 2 was observed at a constant of ∼0.6 kΩ μm, independent of the back gate voltage (Figure 3f). The minimum total R c was measured to be ∼1.7 kΩ μm at a higher turn-on gate voltage (Figure 3f). Thus, the corresponding R c at the 1T′/2H-MoTe 2 interface is ∼1.1 kΩ μm, which is about 2 orders of magnitude smaller than those of metalcontacted 2H-MoTe 2 (reported value of 409 kΩ μm). 1T′-MoTe 2 Gated Heterophase Transistor and NIR Flexible Photodetector. To realize 2D circuitry, it is required to control each transistor independently. Technical challenges and future trends are presented for the application of transparent electrodes in flexible electronics. Here, we chemically assembled the coplanar 1T′/2H/1T′-MoTe 2 array on top of the prepatterned 1T′-MoTe 2 gate electrodes (Figure 4a). The 1T′-MoTe 2 patterns were synthesized via the developed patterned growth method (see Supplementary Section Ia). A layer of 30 nm Al 2 O 3 (via atomic layer deposition (ALD)) was used as the gate dielectric. The detailed fabrication process is provided in the Methods and Supplementary Section IVd. For a typical 1T′-MoTe 2 gated heterophase transistor (inset of Figure 4b), the I-V curves measured at various gate voltages showed linear behavior (Figure 4b), indicating the ohmic contact between the coplanar 1T′/2H-MoTe 2 . The transfer curve exhibited a clear p-type channel characteristic as well (Figure 4c). Compared with the device depicted in Figure 3a, the oncurrent of the transistor is about 1 order of magnitude lower, possibly due to n-type doping effect induced by the ALD Al 2 O 3 (see Supplementary Section IVe). 26 It is reasonable to envision that the device performance can be further improved by replacing Al 2 O 3 with hBN. 27,28 With a bandgap of ∼1.0 eV, few-layer 2H-MoTe 2 can be used for near-infrared (NIR) photodetection. 17 Due to the reduced contact resistance of our device scheme, it is reasonable to assume improved photodetection performance over a large scale. To fabricate these photodetectors, we transferred the large-scale 1T′/2H/1T′-MoTe 2 array to a flexible polyimide (PI) substrate (Figure 4d, see Methods for details). The heterostructure maintains its integrity on a large scale after the transfer (Figure 4e). We investigated the photoresponse behavior of the heterophase under 1064 nm laser illumination with various incident light powers, where we found the photocurrent increased with increasing incident light power (Figure 4f). It is worth noting that when the incident 1064 nm laser power is 50 nW, the photoresponsivity (R = ΔI/ P, where ΔI is the difference between the photocurrent and the dark current and P is the incident light power illuminated on the sample) was as high as 1.02 A/W at V ds = 0.5 V. The photoresponsivity deceased with increased incident light power (see Supplementary Section V), possibly due to the reduction of the number of carriers that could be collected under high photon flux. 29 Thus, the coplanar 1T′/2H/1T′ heterophase can be a promising candidate for future NIR flexible and transparent optoelectronics.

Conclusion. We demonstrated spatially controlled chemical assembly of integrated 1T′/2H/1T′ coplanar heterophase MoTe 2 FETs across a centimeter-scale via phase engineered growth. Such coplanar interfaces are atomically sharp and seamless across the two phases. This approach established a new type of atomic-scale electrical contact. The heterophase contact FETs exhibited ohmic contact with reduced contact resistance. We observed that the field-effect mobility of the 2H-MoTe 2 can rival that of the exfoliated single crystalline samples, due to the large domain size (up to 486 μm) of the as-synthesized 2H-MoTe 2 . With the developed spatialcontrolled patterned growth method, we fabricated 1T′-MoTe 2 gated heterophase FET arrays, whose components of the channel, gate, and contact electrodes are all 2D materials. We also transferred the heterophase device array onto a flexible substrate and demonstrated NIR photoresponse with high photoresponsivity. Our spatial-controlled large-scale chemical assembly of coplanar conductor-semiconductor heterophase structure provides a route toward industry compatible waferscale high-performance 2D electronics and optoelectronics.

Methods. Synthesis of the 2H and 1T′-MoTe 2 Films. The MoTe 2 films were synthesized by tellurizing the Mo films at atmospheric pressure using a horizontal tube furnace equipped with mass flow controllers and a vacuum pump. Mo films were deposited on Si/SiO 2 substrates using magnetron sputtering. The substrates and Te powders were placed in an alumina boat, which was later inserted into a one-inch diameter quartz tube inside the furnace. After evacuating the quartz tube to less than 10 mTorr, high purity Ar gas started to flow at a rate of 500 standard cubic centimeter per minute (sccm) until atmospheric pressure was reached. After that, Ar and H 2 flowed at rates of 4 and 5 sccm, respectively. The furnace was ramped to 650 °C for 2 h and 530 °C for 30 min to synthesize 2H and 1T′-MoTe 2 films, respectively. After growth, the furnace was cooled down to room temperature naturally. For the patterned growth of MoTe 2 , Mo film was prepatterned by either photolithography or EBL followed by magnetron sputtering and lift-off process.

Transfer of the 1T′/2H/1T′ Heterophase MoTe 2 Array. The heterophase MoTe 2 array on the Si/SiO 2 substrate was spincoated with a poly methyl methacrylate (PMMA) layer under 3000 rpm for 60 s. After being baked at 100 °C for 2 min, the sample was immersed in a dilute HF solution (1.5%) at room temperature for 10 min. Subsequently, the heterophase MoTe 2 array with PMMA was gently peeled off from Si/SiO 2 substrate in deionized water and transferred onto a mesh copper grid with carbon film on top for TEM characterization or a PI substrate for flexible device investigation. Finally, the PMMA was removed using acetone, and the sample was thoroughly rinsed with isopropyl alcohol (IPA).

Fabrication of the 1T′-MoTe 2 Gated Heterophase Transistors. First, the bottom 1T′-MoTe 2 gate electrodes were grown by the patterned growth method on the p + -Si/SiO 2 (285 nm) substrate. Second, a layer of 30 nm Al 2 O 3 was deposited by atomic layer deposition (ALD). Next, a pregrown 2H-MoTe 2 film was transferred onto the Al 2 O 3 layer by the above-described transfer method (see Methods for details). Finally, the 1T′/2H/1T′-MoTe 2 heterophase structures were obtained by the phase engineered chemical assembly method described in the main text. Spatial-controlled photolithography is needed to define the 2H-MoTe 2 channels on top of the 1T′ MoTe 2 gate electrodes. For the convenience of measurement, 10/50 nm Pd/Au metal electrodes were fabricated to contact the 1T′-MoTe 2 .

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.9b02006.

Patterned growth of 1T′-MoTe 2 ; Hall measurements of the 2H-MoTe 2 ; AFM and KPFM characterizations across the 1T′/2H-MoTe 2 interface; TEM characterizations of the coplanar 1T′/2H-MoTe 2 ; Pd/Au contacted 2H-MoTe 2 for comparison; the total contact resistance of metal/1T′/2H-MoTe 2 ; contact resistance between metal and 1T′-MoTe