Conformal Growth of Nanometer-Thick Transition Metal Dichalcogenide TiS x -NbS x Heterostructures over 3D Substrates by Atomic Layer Deposition: Implications for Device Fabrication.

The scalable and conformal synthesis of two-dimensional (2D) transition metal dichalcogenide (TMDC) heterostructures is a persisting challenge for their implementation in next-generation devices. In this work, we report the synthesis of nanometer-thick 2D TMDC heterostructures consisting of TiS x -NbS x on both planar and 3D structures using atomic layer deposition (ALD) at low temperatures (200-300 °C). To this end, a process was developed for the growth of 2D NbS x by thermal ALD using (tert-butylimido)-tris-(diethylamino)-niobium (TBTDEN) and H2S gas. This process complemented the TiS x thermal ALD process for the growth of 2D TiS x -NbS x heterostructures. Precise thickness control of the individual TMDC material layers was demonstrated by fabricating multilayer (5-layer) TiS x -NbS x heterostructures with independently varied layer thicknesses. The heterostructures were successfully deposited on large-area planar substrates as well as over a 3D nanowire array for demonstrating the scalability and conformality of the heterostructure growth process. The current study demonstrates the advantages of ALD for the scalable synthesis of 2D heterostructures conformally over a 3D substrate with precise thickness control of the individual material layers at low temperatures. This makes the application of 2D TMDC heterostructures for nanoelectronics promising in both BEOL and FEOL containing high-aspect-ratio 3D structures.

Introduction

Layered two-dimensional (2D) transition metal dichalcogenides (TMDCs) such as MoS2, NbS2, etc., have attracted much interest for their unique electrical, optical, and mechanical properties.[1,2] Lately, fabrication of nanometer-thick vertical heterostructures based on 2D TMDCs (e.g., WS2-NbS2) has gained significance over the formation of heterostructures based on 3D materials due to advantages such as the absence of dangling bonds and lattice mismatch at the layers’ interface.[3−7] Additionally, stacking different 2D materials on top of each other continues to open up unique functionalities, which in turn leads to prospects for new application in various fields.[6,8,9] For example, in 2D-material-based FETs (field effect transistors), 2D-2D heterostructures provide superior device performance due to weak Fermi level pinning at the metal–semiconductor interface along with a reduced Schottky barrier height compared to 2D-3D heterostructures.[10,11] While 2D-2D TMDC vertical heterostructures exhibit large potential, the ability to fabricate them in nanometer-thickness on the wafer-scale (over a large area) is a persisting challenge.

The most common fabrication technique for 2D heterostructures is a top-down approach known as the mechanical transfer method. In this method, the scotch tape technique is used to exfoliate 2D TMDC layers before aligning and placing them on top of another 2D TMDC layer to form the heterostructure. This approach provides significant flexibility to create various 2D heterostructure combinations.[12,3] However, it is not an industrially scalable method, and it has some important issues such as reproducibility, yield, and residue/contamination.[9] Recently, chemical vapor deposition (CVD) has been explored as an alternative approach to grow heterostructure and this is a direct, bottom-up synthesis method. With this method, a nanometer-thick CVD grown 2D TMDC is used as a substrate for the CVD growth of a second nanometer-thick 2D TMDC material. While this method overcomes most of the abovementioned limitations for 2D heterostructure synthesis,[13,14] limitations such as large-area uniformity (due to island growth leading to discontinuous films), nanometer-thickness control, and high synthesis temperatures persist.[15]

On top of the aforementioned constraints, the capability to grow nanometer-thick 2D TMDC heterostructures conformally on a 3D structure (including a high aspect ratio) have yet to be explored using the abovementioned methods. However, incompetency to grow conformally could be an additional bottleneck for implementation as the down-scaling of device dimensions continues with increasing level of structural complexity. Conformal growth means having complete coverage with uniform thickness of nanometer scale over a 3D structure. Control in growth of 2D TMDCs over complex 3D structures could also open new unique application potential, for example, conformal growth of nanometer-thick 2D TMDC as a copper metal-dielectric barrier layer in the back-end of the line (BEOL).[16] Furthermore, in the front-end of the line (FEOL), the ability to grow gate oxide and gate metal conformally around the channel in gate-all-around transistors (GAAFETs) makes continuous down-scaling of device dimensions below 5 nm feasible.[17] For further down-scaling the technology node below 5 nm or so, 2D TMDCs have emerged as strong contenders to overcome device performance limitations accompanied with down-scaling.[18] Similar to GAAFETs, conformal growth of TMDCs and their heterostructures could thus be an added benefit in the fabrication of 2D-based transistors (vdWFETs).[19] Therefore, conformality could be a valuable addition to the toolbox for 2D vertical heterostructure formation to fast track the implementation of 2D materials in nanoelectronic device process flows.

Lately, atomic layer deposition (ALD) has become of significance for the synthesis of various 2D TMDCs.[20−23] It is a method that is based on self-limiting, saturated surface reactions where precursors are dosed in a cyclic manner, separated by purge steps to avoid parasitic CVD reactions.[24,25] Due to its self-limiting nature, ALD provides precise angstrom-level thickness control over a large area along with uniformity and conformality on complex 3D substrates at low processing temperatures. These attributes of ALD would be favorable for the fabrication of 2D TMDC heterostructures as they can overcome the limitations observed in the aforementioned synthesis methods. Thus far, direct ALD of vertical heterostructures (i.e., layers of different materials stacked on top of each other) based on non-TMDC-based layered materials such as Sb2Se3, Sb2Te3, etc. has been successfully demonstrated over a large planar substrate.[26−28] With reference to layered TMDC-based heterostructure formation using ALD, there have been reports only on growing TMDCs on an exfoliated or CVD-grown TMDC substrate.[23,29] On that account, to the best of our knowledge, direct synthesis of nanometer-thick 2D heterostructures based on layered TMDCs by ALD has not been demonstrated.

In this work, we report a newly developed ALD process for niobium sulfide (NbS) growth, in addition to the previously reported process for titanium sulfide (TiS) film growth.[30] Using these ALD processes, we synthesize nanometer-thick TMDC-TMDC vertical heterostructures in the form of TiS-NbS heterostructures at low deposition temperatures (≤300 °C). Through deposition of multilayer (5 layers) TiS-NbS heterostructures composed of TiS and NbS as the alternating layers with varying numbers of ALD cycles per individual layer, we demonstrate nanometer-thickness control of the individual material layers in the heterostructure. In addition to TMDC heterostructure formation on a planar substrate, we also demonstrate the conformal growth of multilayer nanometer-thick TiS-NbS heterostructures on a 3D substrate consisting of a nanowire array. Our work shows that ALD is an excellent method to grow TMDC-TMDC vertical heterostructures over a large area with precise thickness control and conformality at low deposition temperatures.

Experimental Details

Atomic Layer Deposition

The TiS and NbS thin films were deposited by atomic layer deposition (ALD) using an Oxford Instruments Plasma Technology FlexAL ALD reactor. The base pressure of the system was 10–6 Torr. The reactor acted as a hot wall reactor for depositions up to and including 150 °C; however, the reactor acted as a warm wall reactor for depositions above 150 °C as the wall temperature was maintained at 150 °C while the table temperature was varied between 150 and 450 °C. The metal–organic precursors tetrakis-(dimethylamido)-titanium (TDMAT) (Sigma-Aldrich Chemie BV, 99.999%) and (tert-butylimido)-tris-(diethylamino)-niobium (TBTDEN) (STREM Chemical, Inc., 98%) were used for TiS and NbS growth, respectively. The TDMAT and TBTDEN precursors were kept in stainless steel bubblers at 50 and 65 °C, respectively, and were bubbled using Ar as the carrier gas. In both ALD processes, the pressure was maintained at 80 mTorr during both the precursor and coreactant exposure steps. In both the processes, the coreactant gas mixture consisted of 10 sccm of H2S gas (>99.98%) and 40 sccm of Ar gas (>99.999%).

The TiS process consists of the TDMAT precursor and the coreactant gas exposure for 4 and 30 s, respectively. In the case of the NbS process, the TBTDEN precursor exposure was 10 s, while the coreactant was dosed for 20 s. More details on ALD process optimization and saturation data for both processes are provided in Figure S1. In all cases, the precursor and coreactant exposure steps were separated by Ar purge steps. The details about the TiS-NbS heterostructure formation such as deposition temperature and growth will be addressed in the Results and Discussion section. The thin films were deposited on Si substrates covered with approximately 450 nm-thick thermally grown SiO2 (planar substrate). The conformal growth studies were performed by growing the TiS-NbS heterostructures on an ALD SiO2-coated regular array (pitch: 2 μm) of 2 μm-long GaAs nanowires with a diameter of 60 nm (3D substrate).

Characterization Techniques

Scanning electron microscopy (SEM) was employed to investigate the surface morphology of the films. A Zeiss Sigma SEM with an in-lens detector and operating at an accelerating voltage of 3 kV was used for SEM studies. The microstructures of the thin film and the heterostructures were studied using (scanning) transmission electron microscopy [(S)TEM] with a JEOL ARM 200F microscope operated at 200 kV. For cross-sectional (S)TEM studies on planar Si substrates, a lamella (∼100 nm) was prepared in an FEI Nova600i NanoLab SEM/FIB using the lift-out preparation procedure after a protective SiO2 layer was deposited on the top of the film by electron-beam-induced deposition (EBID). High-angle annular dark-field STEM (HAADF-STEM) combined with energy-dispersive X-ray spectroscopy (EDX) was used to study the chemical composition of the TiS-NbS heterostructures on both the planar Si and 3D nanowire substrates. The growth of the TiS and NbS thin films was investigated using in situ spectroscopic ellipsometry (SE) where the film thickness was measured as a function of the number of ALD cycles. Data were collected every 10 ALD cycles by a J.A. Woollam Co., Inc., M2000U spectroscopic ellipsometer with a photon energy range of 0.7–5 eV. The dielectric function of the films was modeled using a B-spline model. The crystallinity of the thin films was studied by Gonio and grazing-incident X-ray diffraction (XRD) with a PANalytical X′Pert Pro MRD analyzer using a Cu Kα (λ = 1.54 Å) X-ray source operated at 40 mA and 45 kV. The scan range was 5 to 80° 2θ with a scan rate and step size of 0.2 s/step and 0.01, respectively. The composition and purity of the thin films were determined using Rutherford backscattering spectrometry (RBS) and elastic recoil detection (ERD) by Detect 99 B.V. Eindhoven, The Netherlands, using a 2 MeV He+ beam source and with the detectors at scattering angles of 105 and 170° for RBS and 25° for ERD. X-ray photoelectron spectroscopy (XPS) data were collected on a Thermo Scientific KA1066 spectrometer with monochromatic Al Kα (hν = 1486.6 eV) X-ray radiation to determine the binding environment and valence band spectra for the deposited thin films. Resistivity measurements were performed ex situ in ambient conditions using a Signatone Four-Point Probe method in combination with a Keithley 2400 sourcemeter acting both as the current source and voltmeter. The resistivity of the thin films was determined from the slope of the obtained I–V curves.

Results and Discussion

The film growth of both TiS and NbS by ALD was investigated independently before the synthesis of the TMDC-TMDC heterostructures based on TiS-NbS was attempted. Thermal ALD processes for TiS and NbS were preferred over plasma-enhanced ALD processes for 2D TMDCs as plasma could cause interface mixing or damage during heterostructure formation.[31] In addition, thermal ALD processes could provide better conformal growth than plasma-enhanced ALD processes.[31] We previously reported a thermal ALD process for TiS films using TDMAT as the precursor and H2S gas as the coreactant.[30] A thermal ALD process for NbS films was developed in this work using TBTDEN as the precursor and H2S gas as the coreactant for deposition temperatures between 150 and 300 °C. More details on the NbS ALD process optimization and film properties as a function of the deposition temperature can be found in the Supporting Information (Figures S1 and S2).

Next, to achieve high-quality TiS-NbS heterostructures, we investigated the properties of the individual TMDCs as a function of the deposition temperatures to determine the optimal deposition temperature. In the case of TiS, the film deposited at 200 °C had the best crystallinity (see Figure a and Figure S3). Cross-sectional HAADF-STEM analysis of TiS deposited at 200 °C shows (Figure b) the layered 2D nature of the TiS film. All van der Waals layers are oriented parallel to the substrate without any out-of-plane oriented (OoPO) growth. The average distance between the two layers was measured to be ∼5.7 Å, which is in good agreement with the distance between two layers of 1T-TiS2.[32] Likewise, the top-view SEM and cross-sectional STEM in Figure a,c, respectively, of the TiS film deposited at 200 °C show a smooth surface morphology with the absence of OoPO structures. The resistivity of the TiS prepared at 200 ° C was measured to be below 0.8 mΩ cm (see Table S1). This resistivity value is similar to the reported bulk resistivity of TiS2.[33]

Figure 1

Top-view SEM images of ∼30 nm-thick TiS (a) and NbS (e) films deposited at 200 and 300 °C, respectively, on the SiO2/Si substrate. Cross-sectional HAADF-STEM image of the TiS (b,c) and NbS (d,f) films deposited at 200 and 300 °C, respectively. Note that panels (b) and (f) show cross-sectional views of the central part of the films while panels (c) and (d) show cross-sectional views of the surface of the films. (g) GI-XRD patterns of TiS and NbS films deposited at 200 and 300 °C, respectively. The dotted lines indicate the peak position corresponding to the (001) and (003) planes of TiS (blue) and NbS (black), respectively.

NbS films deposited at the highest temperature (at 300 °C) were observed to have the best crystallinity (see Figure S3). Yet, the crystallinity of the TiS film grown at 200 °C is more pronounced than that of the NbS film synthesized at 300 °C (see Figure g). The cross-sectional HAADF-STEM image of NbS (Figure f) confirms these XRD results. The layer is polycrystalline, exhibiting a grain size of only a few nanometers and random grain orientation. As a result, the layered 2D nature was only visible in selected regions. The average distance between two van der Waals layers was measured to be ∼5.9 Å, which is in accordance with the interlayer spacing value reported in the literature for NbS2.[34] The top-view SEM of the NbS film (see Figure e and Figure S4) shows that irrespective of deposition temperatures, OoPO structures were present on the film surface. The density of OoPO structures decreased with increasing deposition temperature, which led to relatively smooth surfaces for films deposited at 300 °C. Few of these OoPO structures were marked by white dotted circles in Figure e. Likewise, the cross-sectional HAADF-STEM image of the top part of the NbS film in Figure d confirms the relatively smooth surface with few vertically tapered OoPO structures as highlighted by white dotted lines (also check Figure S5). The resistivity value of the NbS deposited at 300 °C was 4.8 mΩ cm and is close to the reported bulk resistivity of NbS2.[35]

Heterostructure Growth on a Planar Substrate

As ALD is a self-limiting cyclic process, the thickness of the film/layer can be controlled precisely by the number of ALD cycles. Linear growth in terms of film thickness as a function of the number of ALD cycles was observed for the TiS and NbS thermal ALD processes at 200 and 300 °C, respectively (see Figure S1c). The corresponding growth per cycle (GPC) values were measured (by taking the slope over the linear region of the curve) to be ∼0.6 and ∼1.17 Å using in situ SE measurements (Figure S1). It is important to note that TiS could not be deposited above 200 °C due to decomposition of the TDMAT precursor above 200 °C. On the other hand, to obtain a good interface between the TiS and NbS layers, growth of OoPO structures in NbS needs to be avoided by depositing the film at 300 °C. Therefore, as schematically illustrated in Figure , two different deposition temperatures were used to grow the TiS (200 °C) and NbS (300 °C) layers, and the table temperature was increased and decreased accordingly for the TiS-NbS heterostructure formation to acquire the best individual layer quality. Also note that TiS growth on NbS layers (and vice versa) using the above optimized thermal ALD processes was confirmed to be self-limiting in nature.

Figure 2

Schematic illustration of the synthesis of the TiS-NbS heterostructure by thermal ALD. TDMAT and TBTDEN correspond to Ti and Nb precursors, respectively.

In addition, to demonstrate thickness controllability by ALD, a 5-layer TiS-NbS heterostructure with varying individual layer thicknesses was fabricated (employing a process indicated by the gray dotted line in Figure ). At 200 °C, ∼15.0 nm of TiS was deposited as the first layer in the heterostructure using 260 cycles of the TiS thermal ALD process. On top of this TiS layer, a NbS layer was then deposited for 60 cycles at 300 °C. The 60 ALD cycles of NbS would lead to a nominal thickness of ∼5.8 nm (measured using in situ SE) on SiO2 as the starting surface (see Figure S1). However, the thickness of 60 cycles of NbS ALD on TiS as the starting surface might deviate from 5.8 nm due to possible differences in nucleation behavior. Therefore, from now on, we use the term nominal thickness, which refers to the thickness of the material (for both TiS and NbS) when deposited for an N number of ALD cycles on SiO2 as the starting surface. The buildup of the heterostructure continued with the deposition of layers composed of TiS (90 cycles), NbS (60 cycles), and TiS (180 cycles) with nominal thicknesses of ∼5.1, ∼5.8, and ∼10.7 nm, respectively. Both NbS layers were limited to 60 ALD cycles to minimize the appearance of OoPO structures (marked by white circles in Figure c) at the interface. Additionally, during heating and cooling of the substrate table between 200 and 300 °C for the growth of TiS and NbS layers, respectively, the sample was transferred to the load lock chamber from the ALD chamber to reduce annealing effects at the elevated temperatures. While the number of layers in the heterostructure was restricted to five in this work, this process can be easily repeated by any number of times to grow multilayer heterostructures with varied individual layer thicknesses.

The microstructure of the synthesized 5-layer TiS-NbS heterostructure was then studied by cross-sectional (S)TEM imaging as shown in Figure . A clear difference in contrast can be observed between the TiS and NbS layers in bright-field TEM (Figure a) due to the difference in mass density between the materials: the TiS layers appear bright, while the NbS layers are dark. Furthermore, considering the width of the lamella being ∼100 nm, the sharp transition in layer contrast between the TiS (light) and NbS (dark) layers in the TEM image indicates a sharp interface between the layers, with a low degree of mixing. The average vertical thickness of each individual layer was measured by taking an average of thickness measurements from several cross-sectional HAADF-STEM images. The measured layer thicknesses of the three TiS layers (layers 1, 3, and 5) were ∼14.4 ± 0.9, 4.9 ± 0.5, and 10.2 ± 0.6 nm, respectively, and those of the two NbS layers (layers 2 and 4) were ∼4.2 ± 0.6 and 4.5 ± 0.4 nm, respectively.

Figure 3

(a) Cross-sectional bright-field TEM image of the TiS-NbS heterostructure with 5 layers on the Si wafer with native SiO2 on top. (b) Cross-sectional HAADF-STEM image of the TiS-NbS heterostructure layers. (c–e) Corresponding EDX elemental mapping of Ti, Nb, and S, respectively.

The measured thicknesses of both the TiS (layers 3 and 5) layers grown on NbS as the starting surface were comparable to the expected nominal thicknesses of 5.1 and 10.7 nm, respectively. On the contrary, the measured thicknesses of both NbS (layers 2 and 4) layers grown on TiS as the starting surface were lower than the nominal thickness of 5.8 nm. The lower-than-expected thickness in the case of NbS could be due to nucleation delay on the TiS starting surface in comparison to a SiO2 starting surface. This change in growth indicates the higher sensitivity of the NbS thermal ALD process to the starting surface chemistry over the TiS thermal ALD process. Similar nucleation delays have also been observed for several other ALD processes as ALD critically depends on the starting surface chemistry.[36,37]

Differences in crystallinity between the TiS and NbS layers can also be observed in Figure a. Here, the TiS layers display the expected layered structure with clearly visible van der Waals gaps, while the NbS layers lack significant ordering and appear much more amorphous. This is in agreement with the observations from GI-XRD and STEM in Figure for the individual TiS and NbS films of 30 nm each. The intensity of the 15° 2θ peak (corresponding to the (001) and (003) planes for both TiS2 and NbS2, respectively) in the XRD spectrum was observed to be much stronger for the TiS film than for the NbS film indicating a higher crystallinity. This shows that the observed crystallinity difference between the TiS and NbS layers is intrinsic to the respective ALD processes and not due to heterostructure growth. Although NbS is lower in crystallinity, there are regions where a continuous layered van der Waals formation is clearly visible through all five layers of the heterostructure. Therefore, provided that NbS layers were also synthesized with equally high crystallinity as TiS layers, this shows that ALD has the capability to synthesize high-quality 2D van der Waals heterostructures with relatively sharp interfaces in situ without exposure to ambient conditions.

An HAADF-STEM image of the heterostructure is shown in Figure b, and the corresponding EDX elemental maps of Ti, Nb, and S are shown in Figure c–d, respectively. The EDX maps of both Ti and Nb confirm the presence of two NbS layers of similar thickness, sandwiched between three TiS layers of different individual layer thicknesses. The EDX mapping also reveals the lack of intermixing between the two elements per layer as no Ti was observed in the NbS layer within the detection limits of EDX. Likewise, no Nb counts were detected in the TiS layers. Therefore, the EDX mapping strongly indicates that ALD can fabricate TMDC heterostructures with little interlayer mixing.

On the other hand, the S mapping shows a difference in S content between TiS and NbS layers as S counts were observed to be relatively low in the NbS layers. This could indicate differences in the chemical composition/stoichiometry between the TiS and NbS layers. Consequently, the chemical compositions of both TiS and NbS films deposited at 200 and 300 °C, respectively, were individually investigated using RBS (see Table S1). From the RBS measurements, the stoichiometry (sulfur to transition metal ratio) of both the films was calculated to be 1.41 and 1.25 for TiS and NbS, respectively. This confirms the difference in stoichiometry between the two films while also revealing relatively low S content, especially in the NbS film. This indicates the presence of S vacancies and/or excess metal in both layers. In addition, RBS revealed the presence of 6–9 at. % H, 7 at. % C, and 1 at. % O impurities in both TiS and NbS ALD-grown films (see Table S1). The detected H and C in the films could be from precursor ligands and/or coreactant. The observed O impurities in the film could be due to the presence of residual O2 and H2O in the background of the ALD chamber during deposition.

Heterostructure Growth on a 3D Substrate

Finally, to demonstrate conformality, an analogous 5-layer TiS-NbS heterostructure was deposited on an array of wurtzite GaAs nanowires (3D substrates), which were coated with SiO2 by ALD. The coverage of the 5-layer heterostructure on the nanowires was analyzed with side-view HAADF-STEM (Figure a) upon aligning the wurtzite GaAs nanowire to its <11-20> zone axis, allowing for imaging parallel to the {10-10} side facets. The HAADF-STEM image shows the complete coverage of the nanowire with the heterostructure over the length as well as the curvature (tip) of the nanowire (formed by the Au catalyst particle used for the nanowire growth), confirming conformal growth (Figure S6 shows the SEM image containing multiple nanowires). Furthermore, the thickness of the heterostructure was measured on various spots along the length of the nanowire and was observed to hardly vary, with a thickness variation of <5%. Thus, this validates the uniform growth of the 5-layer heterostructure on a 3D structure.

Figure 4

(a) Side-view HAADF-STEM image of the ALD SiO2-coated GaAs nanowire, grown from a Au seed particle, covered with the TiS-NbS heterostructure. (b–e) Higher-magnification (red box in panel (a)) HAADF-STEM image and the corresponding EDX elemental mapping of Ti, Nb, and S, respectively.

Figure b shows the magnified STEM image of the nanowire (red box), and the corresponding EDX elemental maps of Ti, Nb, and S, respectively, are shown in Figure c–e The STEM image along with EDX maps confirms the synthesis of the 5-layer TiS-NbS heterostructure. The Ti mapping shows three TiS layers with different individual layer thicknesses as expected. The Nb mapping shows two NbS layers with comparable thickness sandwiched between TiS layers. The average individual layer thicknesses were measured on various spots on the STEM image (Figure b), and it was observed that the TiS layers grown on NbS layers were measured to be 4.9 and 10.4 nm, respectively (layers 3 and 5). The thicknesses of the NbS layers on top of the TiS layers were measured to be 4.4 and 4.7 nm (layers 2 and 4), respectively. These values are comparable to the individual layer thickness observed on a planar Si wafer. Thus, this confirms the precise thickness control of all five individual layers in the TiS-NbS heterostructure over the nanowire and the absence of intermixing between the layers. Hence, this strongly establishes the capability of ALD for the scalable synthesis of 2D TMDC vertical heterostructures with sharp interfaces, excellent uniform coverage, and conformally over a 3D structure at low temperature.

Conclusions

In this work, ALD has been successfully applied as a suitable technique to synthesize nanometer-thick 2D TMDC vertical heterostructures. A TiS-NbS heterostructure consisting of five layers with different individual layer thicknesses was deposited on a planar Si wafer as well as on a 3D substrate. The thicknesses of the individual layers were controlled precisely by varying the number of ALD cycles of the corresponding thermal ALD process. We successfully demonstrated large-area uniformity and conformality over a 3D substrate at low deposition temperatures. The ability to synthesize TMDCs and their heterostructures by ALD with conformal growth over 3D structures at low temperatures could be beneficial to open fresh avenues such as fabricating complex 3D device structures (partly) consisting of 2D TMDCs both in BEOL and FEOL for nanoelectronics. Furthermore, the current work can be extended to synthesis of other 2D TMDC-based heterostructures such as metal–semiconductor heterostructures.