Low-Temperature Phase-Controlled Synthesis of Titanium Di- and Tri-sulfide by Atomic Layer Deposition.

Phase-controlled synthesis of two-dimensional (2D) transition-metal chalcogenides (TMCs) at low temperatures with a precise thickness control has to date been rarely reported. Here, we report on a process for the phase-controlled synthesis of TiS2 (metallic) and TiS3 (semiconducting) nanolayers by atomic layer deposition (ALD) with precise thickness control. The phase control has been obtained by carefully tuning the deposition temperature and coreactant composition during ALD. In all cases, characteristic self-limiting ALD growth behavior with a growth per cycle (GPC) of ∼0.16 nm per cycle was observed. TiS2 was prepared at 100 °C using H2S gas as coreactant and was also observed using H2S plasma as a coreactant at growth temperatures between 150 and 200 °C. TiS3 was synthesized only at 100 °C using H2S plasma as the coreactant. The S2 species in the H2S plasma, as observed by optical emission spectroscopy, has been speculated to lead to the formation of the TiS3 phase at low temperatures. The control between the synthesis of TiS2 and TiS3 was elucidated by Raman spectroscopy, X-ray photoelectron spectroscopy, high-resolution electron microscopy, and Rutherford backscattering study. Electrical transport measurements showed the low resistive nature of ALD grown 2D-TiS2 (1T-phase). Postdeposition annealing of the TiS3 layers at 400 °C in a sulfur-rich atmosphere improved the crystallinity of the film and yielded photoluminescence at ∼0.9 eV, indicating the semiconducting (direct band gap) nature of TiS3. The current study opens up a new ALD-based synthesis route for controlled, scalable growth of transition-metal di- and tri-chalcogenides at low temperatures.

Introduction

Since the isolation of monolayer graphene, two-dimensional (2D) materials such as the transition-metal di-chalcogenides (TMDCs) have shown interesting electrical and optical properties.[1] TMDCs consist of a transition metal (such as Mo, W, Ti, Nb, etc.) paired with a chalcogenide (S, Se, and Te) forming a layered structure in which the layers are held together by weak van der Waals forces. Besides TMDCs, there is an alternative class of low-dimensional materials that consist of similar elements to TMDCs. This class of materials is known as the transition-metal tri-chalcogenides (TMTCs) and consists of quasi-one-dimensional materials possessing an added degree of freedom that results in a strong electrical and optical in-plane anisotropy.[2,3] The TMTCs are known for the groups IV, V, and VI transition metals (such as Ti, Hf, Nb, etc.) and are predominantly semiconducting in nature.[2−6] The electrical properties can vary between di- and tri-chalcogenide systems containing analogous elements. For example, TiS2 and NbS2 are considered a semimetal and metal, respectively, whereas TiS3 and NbS3 are both semiconductors. Therefore, controlling phase transitions between TMDCs and TMTCs during synthesis allows for direct tailoring of the electrical characteristics of these low-dimensional materials. This phase control could offer new possibilities for device fabrication.

Titanium sulfide (TiS) is one of the systems that could provide an excellent framework for phase control during synthesis. TiS2 has been studied extensively for batteries,[7] optics,[8] and thermoelectric material[9] applications and has been synthesized by various techniques such as chemical vapor transport (CVT),[10] chemical vapor deposition (CVD),[11−15] atomic layer deposition (ALD),[16−18] and wet chemical synthesis.[19] Interestingly, the electrical properties of TiS2 have been under debate for the last few decades, with hypotheses from both theory and experimental results splitting between a semiconducting, metallic, or semimetallic nature.[20,21] Recently, TiS2 has been considered as a degenerate, small-gap semiconductor or a semimetal owing to its high conductivity and optical absorption properties.[22] Conversely, TiS3 has a direct band gap in the infrared region (0.9–1 eV) irrespective of its thickness and is the only known TMTC possessing a direct band gap.[3] This makes TiS3 an ideal candidate for (opto)electronics applications.[23,24] It is known from the literature that TiS3 is less common to synthesize, with reports predominantly using CVT methods[3] with a few instances of CVD.[13,25,26] CVT is a technique where elemental Ti and S are sealed inside a quartz ampoule and placed inside a furnace for synthesis, making this a nonscalable (i.e., not wafer-scale), relatively high temperature, and time-consuming technique.[3−5,27]

In TiS CVT synthesis,[10] phase control can be attained by carefully modulating both the S to Ti ratio and the process temperature. TiS2 is obtained with an initial S to Ti ratio of 2, whereas TiS3 is obtained by maintaining a S to Ti ratio above 3. Temperature is also important due to pyrolysis of S from TiS3 above 550 °C. Thus, TiS3 must be synthesized at or below 550 °C. A significant drawback of the CVT technique is that it requires days to weeks at high temperatures (>450 °C) for growing large crystals (>1 cm), which can be subsequently used to mechanically exfoliate 2D flakes of several microns in size. Thus, a technique to rapidly synthesize both TiS2 and TiS3 over a large area (wafer-scale) with high uniformity and with controlled thickness would be very beneficial for nanoelectronic applications.

In this work, using atomic layer deposition (ALD) as a synthesis technique, we address the challenges associated with other synthesis methods. The ALD method has advantages such as precise thickness control and conformal growth over a large area.[28,29] In addition, most ALD processes are performed at low temperatures (<450 °C), which are favorable for device manufacturing schemes.[29−31] Recently, interest in controlling the phase as well as the composition of metal-oxide-based materials has been made viable by ALD.[32−34] Therefore, ALD could serve as an ideal technique for synthesizing both TMDCs and TMTCs over a large area at low temperature with precise thickness control. Previously, TiS2 has been reported to be synthesized by ALD using TiCl4 and H2S with a growth per cycle (GPC) between 0.15 and 0.5 Å depending on the deposition temperature.[16,17] Recently, ALD of amorphous TiS2 has been reported by Nam et al.[18] using tetrakis(dimethylamido)titanium and H2S at a deposition temperature between 120 and 180 °C with a GPC of ∼0.5 Å. They also reported the surface oxidation of TiS2 after exposure to the ambient conditions. To the best of our knowledge, TiS3 has not been synthesized by ALD.

This work describes a scalable, low-temperature ALD synthesis route with control over the phase of titanium sulfide (TiS). The effect of varying the sulfur coreactant and the deposition temperature on the phase of the deposited material is studied using an extensive array of characterization techniques (Raman spectroscopy, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and Rutherford backscattering spectrometry (RBS)). Our work shows that ALD is well suited for obtaining both metallic TiS2 and semiconducting TiS3 in a temperature range of 100–200 °C. By tuning the temperature and the sulfur coreactant in this temperature range, we demonstrate control over the phase.

Experimental Details

Film Growth

The TiS thin films were deposited by ALD using an Oxford Instruments Plasma Technology FlexAL ALD reactor. The base pressure of the system was 10–6 Torr, and the reactor was equipped with a remote inductively coupled plasma source (ICP, 13.56 MHz). The deposition temperature was varied between 100 and 200 °C. The reactor was a warm wall reactor where the wall temperature can be varied from room temperature to a maximum of 150 °C, independently of table temperature. For depositions below 150 °C, both the table and wall temperatures were maintained at equal temperatures, whereas for depositions above 150 °C, only the table temperature was varied, while the wall temperature remained at 150 °C (maximum wall temperature). The metal–organic precursor tetrakis(dimethylamido)titanium (TDMAT) (Sigma-Aldrich Chemie BV, 99.999% pure) was kept in a stainless steel bubbler at 50 °C and was bubbled using Ar as the carrier gas. In the first half-cycle, the precursor was dosed into the reactor using Ar as the carrier gas and the pressure was maintained at 80 mTorr. During the second half-cycle, a gas mixture of Ar and H2S (both >99.98% pure) with 10 and 40 sccm, respectively, was used as a coreactant in the case of thermal ALD. In the case of plasma-enhanced ALD (PE-ALD), H2S:Ar plasma with the aforementioned gas ratio was used as a coreactant and an ICP power of 200 W was applied to ignite the plasma. Ar (300 sccm) was used as the purge gas after the precursor as well as the coreactant dose steps in both the ALD processes. The thin films were deposited on Si substrates covered with approximately 450 nm thermally grown SiO2.

In some cases, postdeposition annealing of the ALD grown film was conducted in a sulfur atmosphere in a tube furnace. The sample was placed at the center of a quartz tube. After loading sulfur powder on a quartz boat (500 mg, 99.98%, Aldrich), which was situated upstream, the tube was sealed and Ar (50 sccm) was flown downstream as the carrier gas. The furnace was heated up to the target temperature (400 °C) with a heating rate of 10 °C/min. Meanwhile, the sulfur powder situated upstream of the tube was heated to 170 °C by heating tape. The sample was annealed at 400 °C for 3 h and then cooled down naturally in Ar to room temperature.

Analysis Techniques

Spectroscopic ellipsometry (SE) was used to investigate the film growth in situ where the film thickness was measured as a function of the number of ALD cycles. A J.A. Woollam Co., Inc., M2000U with a rotating compensator having a photon energy range of 0.7–5 eV was used. The film thickness was modeled using a B-spline oscillator (the film thickness was verified using cross-section scanning electron microscopy (SEM)). Raman spectroscopy was performed on a Renishaw inVia system using a 514 nm excitation laser in the ambient environment. XPS was used to determine both the elemental composition and valence band spectra of synthesized thin films. A Thermo Scientific KA1066 spectrometer with monochromatic Al Kα (hν = 1486.6 eV) X-ray radiation was employed. The crystallinity of the thin films was studied by X-ray diffraction with a PANalytical X′Pert Pro MRD analyzer using a Cu Kα (λ = 1.54 Å) X-ray source. The surface morphology of the films was investigated by SEM using a Zeiss Sigma with an in-lens detector and operating at an acceleration voltage of 3 kV. The thin film microstructure was studied using TEM and scanning-TEM (STEM) with a JEOL ARM 200F operated at 200 kV. For TEM/STEM studies, the films were grown on Si3N4 TEM windows coated with ∼5 nm ALD SiO2. The selected-area electron diffraction (SAED) patterns were acquired from a 1.3 μm diameter area on each sample. For cross-section TEM/STEM, a thin lamella (∼100 nm) was prepared by a focused ion beam (FIB) using the lift-out preparation procedure. To protect the postdeposition annealed TiS3 film during lamella preparation, a SiO2 layer was deposited on the top by electron-beam-induced deposition. 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 postdeposition annealed TiS3 films. RBS and elastic recoil detection (ERD) measurements were used to determine the composition and purity of the thin films. The measurements were conducted by Detect 99 B.V. Eindhoven, The Netherlands with a 2 MeV He+ beam source containing two detectors at scattering angles of 105 and 170°. Optical emission spectroscopy (OES) of the H2S:Ar plasma was performed using a USB4000 spectrometer from Ocean Optics with a wavelength range of 180–1100 nm mounted vertically on top of the plasma source. Resistivity measurements were performed ex situ and at room temperature using a Signatone four-point probe in combination with a Keithley 2400 sourcemeter acting both as a current source and as a voltage meter. The resistivity of the thin films was determined from the slope of the I–V curve. The near-infrared (NIR) photoluminescence (PL) measurements were conducted using an inverted microscope in an epi-illumination geometry with an objective lens (50X Mitutoyo M Plan NIR Infinity Corrected, NA = 0.42) and a 2.040 eV excitation continuous wave laser. The PL spectra were recorded using an Andor Shamrock 163 spectrometer and an Andor iDus 1.7 μm InGaAs camera.

Results and Discussion

Phase control between TiS2 and TiS3 during CVT has previously been observed to obey the following empirical relationship[35]where P corresponds to the partial pressure of sulfur and T corresponds to the synthesis/system temperature. This relationship indicates the sulfur partial pressure at which the pyrolysis of TiS3 to TiS2 is favored at a given temperature and therefore illustrates the minimum partial pressure of S required at a given temperature to synthesize TiS3. Thus, these two variables can be modulated accordingly to achieve phase control between TiS2 and TiS3. This forms the basis for studying both the effects of H2S as a coreactant and the deposition temperature during the ALD process.

The effect of the coreactant sulfur species was studied by comparing H2S plasma (PE-ALD) and H2S gas (thermal ALD) as the coreactant at a deposition temperature of 100 °C for ∼30 nm thick TiS thin films. The schematic representation of one complete ALD cycle as a function of time is shown in Figure a. In the case of PE-ALD, power was applied to the ICP source to ignite the plasma during the H2S:Ar coreactant step, as shown in Figure a, whereas in the case of thermal ALD the ICP source was turned off. The film thickness as a function of the number of ALD cycles as measured by in situ SE is shown in Figure b. The growth per cycle (GPC) was calculated by taking the slope of the curve in the linear part. Both the PE-ALD and the thermal ALD processes have a GPC of ∼1.6 Å and exhibit typical ALD linear growth behavior with precise thickness control from monolayer to bulk. The self-limiting saturated growth behavior was observed for both ALD processes at a deposition temperature of 100 °C (Supporting Information Figure S1). In the first half-cycle, the TDMAT precursor dose reached saturation around 4 s for both ALD processes. In the second half-cycle, in the case of PE-ALD, the H2S:Ar plasma exposure reached saturation around 25 s. Although plasma exposure reached saturation around 25 s, a plasma exposure of 30 s was used during the film deposition to ensure uniform growth over a large area. In the case of thermal ALD, a 30 s dose time of H2S:Ar gas was chosen to have a fair comparison with the PE-ALD process and the chosen dose time was observed to be in the saturation regime as well.

Figure 1

(a) Schematic of one complete ALD process cycle implemented at all investigated temperatures. Plasma power was on for PE-ALD, while it was off for thermal ALD. (b) Film thickness as a function of the number of ALD cycles measured using in situ SE for PE-ALD grown films at deposition temperatures between 100 and 200 °C and also for a thermal ALD film grown at 100 °C. Raman spectra of (c) PE-ALD and (e) thermal ALD grown films at 100 °C deposition temperature and also the corresponding TiS3 and TiS2 vibration peaks. The peak at 520 cm–1 is of the Si substrate. S 2p XPS spectra of (d) PE-ALD and (f) thermal ALD grown films at 100 °C deposition temperature. In (d), the (dark gray) doublet peaks correspond to S2– species and the (light gray) doublet peaks at higher energy correspond to (S2)2– species of TiS3 (Ti4+(S2)2–S2–). (g, h) Crystal structures of two TiS2 and TiS3 layers, where the respective thickness of 1 c-axis plane is indicated.

Raman spectroscopy studies were conducted on the TiS films deposited by both ALD processes at 100 °C. For the film deposited by PE-ALD, four vibrational peaks at 170, 295, 360, and 550 cm–1 were observed. These four vibrational peaks correspond to the four A1g-type (Figure c) Raman modes of TiS3.[36] Interestingly, the Raman spectrum for the film deposited by thermal ALD had only two vibrational peaks at ∼230 and 330 cm–1 with a shoulder peak at ∼380 cm–1. The two vibrational peaks at 230 and 330 cm–1 observed for thermal ALD (Figure e) film correspond to the Eg and A1g Raman modes of 1T-TiS2, respectively.[37] These findings indicate that phase-controlled synthesis of TiS2 and TiS3 can be attained with ALD by switching between H2S gas and H2S plasma in the coreactant step at 100 °C.

The chemical composition and binding environments in the deposited films were then investigated using XPS analysis. The as-deposited films had some C and O contamination at the surface due to exposure of the films to ambient conditions after deposition. However, the bulk of the films was free from C and O impurities. The XPS spectra obtained were calibrated by setting the C–C peak to 248.8 eV in the C 1s spectrum. The S 2p spectrum of the PE-ALD grown film at 100 °C contains two doublets (Figure d): one at 161.1 and 162.3 eV (dark gray) and the other at 162.4 and 163.6 eV (light gray). In comparison to the literature, the lower energy doublet at 161.1 and 162.3 eV corresponds to the S 2p3/2 and S 2p1/2 spin–orbit doublet of an isolated S2– species and the second doublet at 162.4 and 163.6 eV corresponds to the spin–orbit doublet of (S2)2– (also known as a S–S pair). This is consistent with the TiS3 formula written as Ti4+(S2)2–S2–.[38] The presence of both doublets could also be due to the presence of a mixed phase of both TiS2 and TiS3, but the Raman spectrum from Figure c showed vibration modes of only TiS3, indicating a lack of TiS2 in the film. Additionally, the fitted peak area ratio of (S2)2–/S2– was calculated to be 2.5:1, which indicated the presence of excess (S2)2– species in the film. From the XPS S 2p peaks, it was determined that the excess S present in the film does not match the binding energy of any elementary S species such as S8, S4, etc. The S to Ti atomic ratio calculated from RBS measurements for a PE-ALD grown film at 100 °C further confirms the synthesis of TiS3 (see Table ) with the incorporation of excess S (S/Ti ratio above 3). In contrast to the PE-ALD film, the films deposited by thermal ALD (Figure f) at 100 °C exhibited only one S 2p doublet at 161.1 and 162.3 eV (spin–orbit doublet of S2–), indicating the deposition of TiS2. The S to Ti atomic ratio calculated from RBS measurements also confirms that the TiS2 film obtained by thermal ALD is slightly S-deficient (S/Ti ratio less than 2). The RBS and XPS results confirm that we can obtain phase control during ALD by tuning the gas composition in the coreactant step.

Table 1

In Situ SE, RBS, ERD, and Four-Point Probe Measurements of PE-ALD Deposited Titanium Sulfide Thin Film between 100 and 200 °C Deposition Temperaturesa

	GPC, Å	S/Ti ratio	O, atom %	C, atom %	H, atom %	resistivity, 103 μΩ cm
100 °C	1.6	3.48 ± 0.02	1.4 ± 0.6	3.7 ± 0.9	1.3 ± 0.1	-
150 °C	2.1	1.94 ± 0.01	1.7 ± 0.1	1.0 ± 0.1	10.1 ± 0.8	9
200 °C	2.3	1.93 ± 0.01	0.6 ± 0.1	0.6 ± 0.1	9.8 ± 0.8	3.5

All films were ∼30 nm thick, and the GPC was calculated by taking the slope of the linear region above 140 cycles; see Figure b. The atomic concentrations of O, C, and the S/Ti ratio were determined by RBS. The H content was determined by ERD. The resistivity was measured using the four-point probe method.

To better understand the conditions contributing to the phase-controlled synthesis of TiS2 and TiS3, OES was performed (Figure ) to obtain insight into the composition of the H2S:Ar plasma. In the H2S:Ar plasma, S-containing species, including SH+ (at around 337 nm) and S2 (peaks between 280 and 610 nm) species, were detected along with the Ar and H species.[39,40] The contribution of S2 species in the OES spectrum (between 280 and 610 nm) is significant (see Supporting Information Figure S2, where Figure is compared with the OES spectrum of H2:Ar plasma without S species). In the case of thermal ALD, the H2S:Ar gas (no optical emission in the gas phase) would most likely decompose into Ar and reactive H, SH, and S species at the reaction surface.[41,42] By comparing both the cases, we hypothesize that the generation of S2 species in the H2S:Ar plasma could lead to the synthesis of TiS3 [Ti4+(S2)2–S2–] where majority of the sulfur exists as S–S pairs. A similar observation was made by Carmalt et al. using CVD,[13] where TiS3 was synthesized only when a coreactant containing S2 species was used. These S2 species, however, are not present in the case of H2S:Ar gas utilized during thermal ALD, which yielded only TiS2.

Figure 2

Optical emission spectrum of the H2S:Ar plasma from PE-ALD with a gas flow of 10:40 sccm. It shows the corresponding peak for SH+ (blue) and the range of peaks for S2 (green) species along with H and Ar species (black).

Following the synthesis of TiS3 by PE-ALD at 100 °C, the effect of the synthesis temperature (second variable in the earlier-mentioned empirical relationship) on the material’s phase/composition was studied by varying the deposition temperature in PE-ALD between 100 and 200 °C. In Figure b, the film thickness as a function of the number of ALD cycles is shown for the investigated deposition temperatures. Linear growth was observed for deposition at 100 °C without any nucleation delay, whereas for depositions at 150 and 200 °C, nonlinear growth was observed. This nonlinear growth can be ascribed to both the change in morphology of the film and the change in material phase (as will be discussed later). The GPC of the PE-ALD films increased from 1.6 Å at 100 °C to >2.0 Å at the higher temperatures; see Table . In comparison to the 100 °C film, the increase in GPC at higher temperatures is due to fin-like, out-of-plane oriented (OoPO) growth, as will be discussed later (in Figure ). Similar growth behavior and increased GPC with OoPO growth have been previously observed for PE-ALD grown MoS2.[29]

Figure 4

Electron microscopy studies of the structure of PE-ALD films. Images (a, d, g), (b, e, h), and (c, f, i) correspond to films deposited by PE-ALD at 100, 150, and 200 °C, respectively. (a–c) Top-view SEM images showing the change in film morphology for various deposition temperatures. (d–f) Top-view TEM images showing the microstructure of the films; in (e) and (f), layered structure of the OoPO structure in the TiS2 film can be seen. The average measured distance between two layers is highlighted. (g) SAED pattern showing the amorphous nature of the film deposited at 100 °C. (h, i) SAED patterns showing the polycrystalline nature of the films deposited at 150 and 200 °C.

Raman spectra of the films grown at higher temperatures were compared to the previously studied TiS3 film grown at 100 °C. Figure a shows the Raman spectra of the films deposited at 150 and 200 °C containing two vibrational peaks at 230 and 330 cm–1 along with a shoulder at 380 cm–1 analogous to the Raman spectrum obtained for the TiS2 film grown at 100 °C by thermal ALD (Figure e). These results demonstrate that phase control within the PE-ALD process was achieved between TiS2 and TiS3 by increasing the growth temperature, similar to CVT. In the PE-ALD process, the transition temperature of ∼100 °C for TiS3 to TiS2 synthesis is very low in comparison to that in CVT (550 °C). Additionally, the XPS S 2p spectra for the 150 and 200 °C films corroborate this result with the Figure f. The S 2p spectra of both films display only one doublet at 161.1 and 162.3 eV (similar to thermal ALD grown films at 100 °C) corresponding to the spin–orbit doublet of S2– species, indicating the deposition of TiS2.

Figure 3

(a) Raman spectra showing the vibration mode peaks and (b) XPS valence band spectra of PE-ALD grown films between 100 and 200 °C deposition temperature. (c) Grazing-incidence X-ray diffraction (GI-XRD) patterns of the films deposited at various temperatures where the peaks corresponding to (001), (011), and (110) planes are mentioned.

The valence band spectra of the films deposited by PE-ALD were also investigated using XPS analysis. The Fermi level (BE = 0 eV) was calibrated by measuring the Fermi edge of gold. At 100 °C, four broad peaks (see Figure b, black) at around 1.6, 4.5, 10.7, and 15.3 eV binding energies (BE) were observed. Due to the hybridization by S–S pairs in TiS3, the two valence band peaks of TiS2 split to form four broad peaks.[38] The XPS valence band spectra for films deposited at 150 and 200 °C differed from the spectrum obtained at 100 °C showing only two broad peaks at around 3.5 and 13 eV (Figure b). In comparison to the literature, the spectra for the films deposited at 150 and 200 °C correspond to the valence band spectrum of TiS2.[38] These results further confirm that the phase-controlled synthesis of TiS2/TiS3 can be obtained by tuning the deposition temperature.

The composition and purity of the material were further determined using RBS measurements; see Table . The S/Ti ratio was measured to be 3.48 for the film deposited at 100 °C, whereas the S/Ti ratio for the films deposited at 150 and 200 °C was around 2 with a small S deficiency (or excess Ti). Additionally, contaminants such as C and O were detected within the error limit (Table ), confirming the high purity of the films similar to the XPS measurements. Interestingly, ERD revealed a significant amount of H impurities (∼10 atom %) for the 150 and 200 °C films. The H may have most likely been incorporated from the plasma coreactant and/or precursor ligands. It could be possible to reduce the H impurities in the film by further optimizing the H2S:Ar plasma coreactant exposure time during the ALD cycle.

The crystallinity of the PE-ALD grown TiS3 and TiS2 films was investigated using grazing-incidence (GI)-XRD. The GI-XRD pattern of the film deposited at 100 °C (TiS3) does not show well-defined peaks, indicating the formation of an amorphous film (Figure c). The films deposited at 150 and 200 °C, however, confirm the synthesis of crystalline 1T-TiS2.[19][19] The strong peak around 2θ of 15° corresponds to the (001) plane of 1T-TiS2 and reveals a significant orientation effect. This orientation effect can be explained by a crystalline layer in which a strong majority of crystals have their basal planes of 1T-TiS2 oriented parallel to the substrate. This implies that below the OoPO structures visible in the SEM and TEM images (Figure ) a continuous film of the aforementioned basal plane texture is present. The contribution of OoPO structures to the XRD pattern ((011) and (110) peaks) is small because of their limited volume fraction and random orientation with respect to the XRD detection geometry.

Additionally, the SAED pattern also shows the amorphous nature of the film synthesized at 100 °C (Figure g). Complementary to GI-XRD, the SAED patterns of the 150 and 200 °C films display reflections of lattice planes oriented perpendicular to the substrates. As a result, for both 150 and 200 °C films (Figure h,i) high-intensity (011), (110), and (010) rings are visible. The continuous nature of the rings confirms the polycrystalline nature of PE-ALD grown TiS2 films. The discontinuous (001) ring reflects the presence of only a few OoPO structures per probed area having their c-axis parallel to the film surface.

Top-view SEM images (Figure a–c) display the morphology of the PE-ALD grown films deposited at all three temperatures. For the film deposited at 100 °C, no surface texture was observed. Similar to SEM, top-view TEM (Figure d) shows no lateral variation in the morphology (i.e., crystallinity and density) of the film. On the other hand, for the films deposited at 150 and 200 °C, top-view SEM images show OoPO structures, which can explain the high growth rate observed in the in situ SE measurements. Analogous OoPO structures have been observed previously for other TMDCs like MoS2[29] and ReS2.[43] A similar relation between the OoPO film morphology and increased GPC has been previously observed for MoS2.[29] The top-view TEM images of films grown at 150 and 200 °C (Figure e,f) confirm the OoPO morphology of the TiS2 film. The average distance between the 2D layers was measured to be ∼5.7 Å (Figure e,f), equivalent to the distance between two 1T-TiS2 layers (Figure g).[44]

The electrical resistivity of the PE-ALD deposited TiS films was measured using a four-point probe method. The resistivity of the TiS3 film deposited at 100 °C could not be measured due to its primarily amorphous nature, which could have yielded a high resistivity value that was beyond the measurement detection limit. Table shows the measured resistivity of the other films. The deposited TiS2 films exhibit a decrease in resistivity from 9000 to 3500 μΩ cm with the increase in temperature from 150 to 200 °C. From an application perspective, these low-resistivity 1T-TiS2 films could be used as 2D contact materials with 2D semiconductors in device fabrication. (Note that the crystalline (by GI-XRD) TiS2 film synthesized by thermal ALD at 100 °C has a resistivity of 2000 μΩ cm.) In summary, the effect of temperature on material composition, phase, morphology, and resistivity was studied in detail. All of the discussed characterization techniques confirm the synthesis of TiS3 at 100 °C and a transition to TiS2 at temperatures above 100 °C by PE-ALD. We also investigated the effect of deposition temperature by thermal ALD and observed no change in phase from TiS2 at all investigated temperatures. These studies, however, will be discussed in detail in a future work.

The crystallinity of the PE-ALD grown amorphous TiS3 film at 100 °C deposition temperature was improved by a postdeposition annealing step. To avoid oxidation to TiO2 and pyrolysis to TiS2 during annealing, a sulfur-rich atmosphere was chosen for the annealing treatment of 3 h at 400 °C. The Raman spectra of TiS3 before and after annealing are compared in Figure a. The four vibration peaks observed for the annealed film confirm that the TiS3 phase was preserved. Additionally, the red shift of the vibration peaks along with the reduced width of the peaks, indicated improved crystallinity of the film. Some surface oxidation to anatase TiO2 (∼143 cm–1) was observed, probably due to the transfer of the sample through the air from the ALD chamber after deposition to the annealing furnace. The GI-XRD pattern (Figure b) of the annealed film also confirmed improved crystallinity and preservation of the TiS3 phase of the film, along with TiO2 (*) due to surface oxidation. Similar to the PE-ALD grown 1T-TiS2 crystalline films, a strong peak around 2θ of 10° corresponds to the (001) plane of TiS3 and indicates that basal planes of TiS3 are orientated parallel to the substrate. An average (vertical) crystal size of ∼16 nm was calculated using the Scherrer equation.[45]

Figure 5

(a) Comparison of the Raman spectra of PE-ALD grown TiS3 before (red, amorphous) and after annealing at 400 °C (blue). (b) GI-XRD pattern of the postdeposition annealed TiS3 film showing its crystalline nature with peaks corresponding to TiS3 (00l) planes. (c–f) TEM image showing a cross section of the postdeposition annealed TiS3 film with corresponding EDX elemental mapping of Ti, S, and O. The red (*) in (a), (b), and (c) indicates the presence of anatase TiO2 due to surface oxidation. The white dashed line in (c) highlights the interface between TiO2 and TiS3. (g) High-resolution HAADF-STEM image shows the cross section of postdeposition annealed TiS3 with atomic resolution. The average measured distance between two layers is highlighted. (h) Zoomed-in area (red box) from (g) with an atomic model of the TiS3 crystal structure overlaid on the STEM image. (i) PL spectrum demonstrating the semiconductor nature of the annealed film with an optical band gap ∼0.9 eV, compared with PL (red) of the fused silica substrate.

A cross-sectional TEM image of the postdeposition annealed TiS3 film displays the layered crystalline nature of the postdeposition annealed TiS3 film (Figure c). Similar to the results from GI-XRD, the basal planes of TiS3 are oriented parallel to the substrate. An average (lateral) crystal size of ∼50 nm was measured from the cross-sectional TEM images. The corresponding EDX elemental mapping of the S, Ti, and O are shown in Figure d–f, respectively. EDX mapping also further confirms the surface oxidation (to TiO2) of the film and also shows the presence of a pure titanium sulfide phase below the oxidized surface. Figure g shows a high-resolution HAADF-STEM image of a cross-sectional region in the postdeposition annealed crystalline TiS3 film. The average distance between the monolayers was measured to be 8.9 Å, which is equivalent to the distance between two TiS3 layers reported in the literature.[46] In the zoomed-in inset (Figure h), an overlay of the TiS3 crystal structure viewed along the [010] zone axis, is shown in the HAADF-STEM image.

Finally, the semiconducting nature of the TiS3 after annealing was confirmed using NIR PL spectroscopy. PL was observed to be centered around ∼1400 nm (∼0.9 eV) after annealing (Figure i), whereas it was not detectable before annealing. The position of the PL peak is close to the band gap of TiS3.[47][47] Thus, the postdeposition annealing treatment of the PE-ALD grown amorphous TiS3 at 400 °C improves the quality and crystallinity of the TiS3 film.

Conclusions

We report on a low-temperature ALD process with which the phase of titanium sulfide films can be precisely tuned between TiS2 and TiS3. The key to successful phase control was to carefully tune the coreactant composition and deposition temperature. The generation of S2 species in the H2S plasma has been speculated to lead to the formation of amorphous TiS3 by PE-ALD at a deposition temperature of 100 °C. Above 100 °C, a transition to the formation of TiS2 occurs by PE-ALD. Electrical measurements show the low resistive nature of the TiS2 films. The crystallinity of PE-ALD grown TiS3 films was improved by adding a postdeposition annealing step of 400 °C for 3 h. After annealing, NIR PL was observed at ∼1400 nm, confirming the direct optical band gap of TiS3 at ∼0.9 eV. This approach can potentially be extended to other transition-metal di- and tri-chalcogenides, opening up new avenues for implementation of both transition-metal di- and tri-chalcogenides in nano- and optoelectronic device fabrication schemes.