Atomic layer deposition of transition metal chalcogenide TaS_x using Ta[N(CH₃)₂]₃[NC(CH₃)₃] precursor and H₂S plasma

Transition metal chalcogenides (TMCs) have a wide variety of applications, ranging from electronics and optics to catalysts and sensors.^1–5 TMCs with a transition metal from group IV, V, or VI are especially interesting since these can occur in di-chalcogenide and tri-chalcogenide forms. Controlled synthesis of TMCs is required to tune the phase of the TMC nanolayers and with that their material properties.^6,7

One interesting TMC is tantalum sulfide (⁠ $TaS x$⁠), which has various applications, such as hydrogen evolution catalysts, Cu diffusion barriers, photodetectors, and quantum dots.^8–13 Currently available synthesis techniques for $TaS x$ are exfoliation, powder vapor deposition, chemical vapor deposition (CVD), and thermal/plasma sulfurization.^2,11,14,15 Each of these techniques has their own characteristics with respect to the control over the material properties and film thickness. Exfoliation generally results in highly crystalline nanolayers and control of the thickness at the monolayer level but offers no possibility to alter the material properties during the synthesis. Powder vapor deposition offers some control over the material properties depending on the used Ta precursor.¹⁴ For both powder vapor deposition and CVD, the thickness can be regulated to some extent by adjusting the deposition time. Sulfurization of Ta-metal films results in a predetermined thickness and limited control over the film stoichiometry. Atomic layer deposition (ALD) is a nanoscale synthesis technique that is based on self-limiting surface reactions and, thus, offers cycle-wise growth, resulting in accurate control over the film thickness. Moreover, ALD offers the opportunity to tune the composition and is very suitable for deposition on high aspect ratio structures, as is often required for catalysts and electronics.

Many different TMCs have already been deposited using ALD, such as $TiS x$⁠, $NbS x$⁠, $MoS 2$⁠, and $WS 2$⁠.^16–20 However, for tantalum sulfide (⁠ $TaS x$⁠), no ALD process has been reported yet. The Ta precursor (tert)-butylimidotris(dimethylamido)tantalum (Ta[N(CH $3$⁠) $2$] $3$[NC(CH $3$⁠) $3$], TBTDMT) was previously used for TaN^21,22 and is used here for $TaS x$⁠. Previous work on ALD of TMCs demonstrated that in addition to the growth, the material phase can be controlled; for example, both $TiS 2$⁠, $TiS 3$⁠, and both $NbS 2$⁠, $NbS 3$ were synthesized.^23,24 We expected that phase control is possible for ALD of $TaS x$ since Ta is a group V transition metal, similar to Nb.

In this work, we report the synthesis of $TaS x$ by ALD using TBTDMT as the precursor and an $H 2$S-based plasma as the coreactant. The process development and an overview of the material properties are presented. The film composition can be tuned by altering the temperature and plasma composition. It is demonstrated that polycrystalline 2H- $TaS 2$ can be deposited by using a plasma mixture containing $H 2$S, $H 2$⁠, and Ar.

Tantalum sulfide thin films were deposited using a commercial Oxford Instruments FlexAL reactor equipped with a remote inductively coupled plasma source (more detailed description by Heil et al.²⁵). The TBTDMT precursor (>98% purity, Strem Chemicals) was contained in a stainless-steel canister, which was heated to 60  $°$C. A 50 sccm Ar (>99.999% purity) carrier gas flow was applied to facilitate precursor delivery into the reaction chamber. Before the precursor dose, a 200 sccm Ar flow predose, as is shown in Fig. 1, was flown through the precursor manifold into the reactor for 4 s, while the precursor canister was closed. This prevents back-flow during the subsequent precursor dose. Due to the Ar flow, the reactor pressure builds up to the set pressure of 80 mTorr. Intermediate purge steps with Ar flow were implemented after each precursor dose and plasma coreactant exposure. An $Ar\, : \, H 2$S plasma, containing 40 sccm Ar and 10 sccm $H 2$S, was used as a coreactant. The plasma was operated at a power of 200 W at radio frequency (13.56 MHz). During the plasma step, the butterfly valve to the pump is fully open, resulting in a plasma pressure of $≈6$ mTorr. Before the 200 W plasma power is turned on, there is a 5 s gas stabilization step.

$TaS x$ films were deposited on single-side polished (100) Si substrates with 450 nm wet thermal $SiO 2$⁠. Different table temperatures were used, ranging from 100 to 400  $°$C, in order to investigate the ALD window. Based on previous TaN experiments where the same Ta precursor was used, precursor decomposition is expected above 350  $°$C. The walls of the reactor were heated to 145  $°$C for table temperatures of 145  $°$C and higher. At lower temperatures, the walls were heated to the same temperature as the table. The substrate temperature is lower than the table temperature due to limited thermal contact, especially at elevated temperatures, where the table and wall temperature lie further apart.²⁶ 

Film thickness and optical properties were measured both in situ and ex situ using a J.A. Woollam spectroscopic ellipsometer (SE). Optical properties were determined from the SE data per temperature at the end of the deposition process using a B-spline model. These optical properties were then used to extract the film thickness during film growth. Top view transmission electron microscopy (TEM) and electron diffraction was conducted using a JEOL ARM 200F transmission electron microscope. Film crystallinity was characterized using grazing incidence x-ray diffraction (GIXRD) on a Bruker Discover D8. Atomic force microscopy (AFM) measurements were performed using the Bruker Dimension Icon AFM. Elemental film composition, specifically the S:Ta ratio, was analyzed using x-ray photoelectron spectroscopy (XPS) on a Thermo Fisher Scientific K-Alpha XPS system and confirmed by Rutherford backscattering spectrometry (RBS). RBS and elastic recoil detection (ERD) were carried out by Detect99.²⁷ The electrical resistivity of the films was measured at room temperature using a Signatone four-point probe in combination with a Keithley 2400 Sourcemeter.

The recipe of the process is illustrated in Fig. 1. We investigated the self-limiting behavior of the half-reactions by independently varying the dose and purge times $t 1, t 2, t 3$⁠, and $t 4$⁠. Saturation curves are shown in Fig. 2, established from in situ ellipsometry, where each data point was determined from a separate deposition of 100 cycles at 300  $°$C.

Soft-saturation behavior was observed for the TBTDMT precursor dose [Fig. 2(a)], consistent with the earlier work with this precursor.²² For further processing, a dose time of $t 1=20$ s was selected. In a separate experiment, a thickness increase of 1.3 Å was observed when only the precursor was dosed for 480 s on a $TaS x$ film, indicating that there is only a fairly small contribution of precursor decomposition to the growth at 300  $°$C.

The H $2$S:Ar plasma dose time saturates at $t 3=10$ s. Only dosing the plasma gas mixture without igniting the plasma did not result in film growth. Both purge times after the precursor and plasma dose saturate quickly, possibly since the predose and plasma gas stabilization steps provide additional purging. Purge times were set to $t 2=5$ s and $t 4=2.5$ s.

Linear growth as a function of cycles was observed for the whole temperature range, as can be seen in Fig. 3. GPC values were calculated from the slope of a linear fit through the data. The resulting values are shown as a function of table temperature in Fig. 4(a1). In the range 150–400  $°$C, the GPC varies a little, with values between $1.17±0.03$ and $0.87±0.08$ Å, while the GPC at 100  $°$C is much higher with a value of $1.79±0.03$ Å. At high temperatures, the values for the optical constants $n$ and $k$ are higher as shown in Fig. S1 of the supplementary material, indicative of a denser film. Therefore, the small reduction in GPC as a function of temperature appears to be related to an increase in the film density.

In addition to the variation of temperature, the plasma composition was varied. $H 2$ was added to the plasma mixture, while keeping the total flow rate of $H 2$S and $H 2$ constant to 10 sccm. The GPC values as a function of the $H 2$S and $H 2$ plasma flow rates show only a minor variation as shown in Fig. 4(a2).

The uniformity of the deposition was checked by performing SE mapping on a 6 $″$ wafer after 100 ALD cycles at 300  $°$C with determined saturation times. In Fig. S2 of the supplementary material, the corresponding thickness map can be found. The deposited $TaS x$ film is uniform, with a nonuniformity value of 3.2%.

We characterized the chemical composition of the deposited $TaS x$ films in terms of S and Ta content using XPS and RBS. Part of the XPS analysis can be found in Fig. S3 of the supplementary material. No data on the O content are provided as the XPS analysis was done ex situ such that it is not possible to differentiate between O that is incorporated during the ALD process and O that is introduced by surface oxidation when exposed to ambient conditions. The number of deposited Ta atoms per cycle, measured by RBS, appears to be independent of the table temperature as shown in Table I. The determined S:Ta ratios are shown in Fig. 4(b1) as a function of the table temperature. At 150  $°$C, the sulfur to tantalum ratio is 4.1 and decreases to 2.5 at 400  $°$C, revealing that increasing the table temperature results in less sulfur in the film. RBS measurements at 150 and 300  $°$C confirm the observed trend of stoichiometry. The small difference between the XPS and RBS results can be attributed to the fact that XPS is a nonquantitative surface sensitive technique, whereas RBS is quantitative and probes the entire film and the substrate. The change in stoichiometry as a function of temperature is in line with the films being less dense at low temperatures (see Sec. III A) since more excess sulfur corresponds to less dense films. Sulfur-rich films at a low temperature were observed for $MoS 2$ ALD as well.^28,29 For $MoS 2$⁠, an explanation for the change in stoichiometry as a function of temperature was reported, being that by increasing the temperature a part of the $S 2 2 −$ species in the film can be oxidized to elemental S, while another part is reduced to $S 2 −$⁠. The elemental S evaporates from the film resulting in less sulfur in the film.²⁹ In this work, the same plasma chemistry was employed as for $MoS 2$ ALD; therefore, it is probable that this mechanism also holds for $TaS x$ thin films.

The stoichiometry as a function of $H 2$S and $H 2$ flows can be found in Fig. 4(b2). Addition of $H 2$ to the plasma results in a decrease of the S:Ta ratio as expected.^29,30 The trend of the lower S:Ta ratio with a change in plasma composition can be explained by (i) addition of $H 2$ leading to more H radicals, which are known to etch S and (ii) less $H 2$S resulting in less S plasma species. RBS measurements for 10:0 and 5:5 sccm $H 2$S:H $2$ flow rates [Fig. 4(b2)] confirm the decrease in S:Ta ratio.

Furthermore, from RBS (see Table I), it was observed that the number of Ta $atoms / cm 2$ is higher for the film prepared with additional $H 2$ in the plasma, while the number of S atoms is very similar for both. Thus, addition of $H 2$ appears to lead to more efficient TBTDMT precursor adsorption. Possibly, this is due to a change in the reactivity of the surface when $H 2$ is added to the plasma. For example, a higher density of reactive surface groups could result in elimination of more ligands during the precursor half-reaction such that sterically, more precursor molecules fit on the surface.^31,32

To determine the hydrogen content in the films, ERD measurements were performed for films deposited at 150 and 300  $°$C; see Table I. At 150  $°$C, the number of H atoms per area is higher than at 300  $°$C; the atomic percentage is also slightly higher with $≈16.1$ vs $≈14.6$ at. %. The actual atomic percentages for H in the film are expected to be slightly lower as there is possible additional H on or near the surface and on the interface with $SiO 2$⁠. A likely explanation for the higher amount of H is that at lower temperatures, the reaction with the $H 2$S coreactant is not complete and H is incorporated into the film in the form of S $x$H $y$ species. Comparing the two films with and without addition of $H 2$ to the plasma, the absolute amount of H in both films is similar, and the atomic percentage is also very similar (13.7 vs 14.6 at. %, respectively). Thus, addition of $H 2$ to the plasma does not change H incorporation in the film.

Addition of $H 2$ to the plasma leads to a higher degree of crystallinity in the resulting $TaS x$ film, as can be deduced from XRD, TEM, and selected area electron diffraction (SAED). XRD spectra, SAED patterns, and extracted intensity curves from SAED are shown in Fig. 5. The expected phase of the films is $TaS 3$⁠, 1T-TaS $2$⁠, or 2H-TaS $2$ since these phases are the most common.^10,33,34 In the XRD spectra, a broad peak is present between 10 and 20 $°$⁠. This peak can most likely be assigned to the (002) peak of 2H- $TaS 2$ at 14.6 $°$⁠.^33,34 The peak at 56 $°$ origins from the Si substrate. The sole presence of the (002) peak in the XRD spectra implies a strongly textured film. In the asymmetrical GIXRD detection configuration, no (002) peak would be expected if all 2D layers would be oriented exactly parallel to the substrate. Apparently, some degree of inclination of these 2D planes is present in the films. The larger peak intensity of the (002) 2H- $TaS 2$ peak for the layer grown with additional $H 2$ could either indicate a higher degree of disorder, implying more tilted grains that can contribute to the GIXRD pattern, or a higher degree of crystallinity. TEM images in Fig. S4 of the supplementary material confirm the latter hypothesis. Without $H 2$ in the plasma, the film displays a few crystals in an amorphous matrix, while with $H 2$ in the plasma, the film can be described as polycrystalline.

In the GIXRD analysis configuration, the full width half maximum (FWHM) of the (002) peaks probes the crystallite size in the direction of the surface normal. For the film deposited without $H 2$⁠, the FWHM is $4.1± 0.3 °$⁠, while for the film deposited with the addition of $H 2$⁠, the FWHM is $2.60± 0.04 °$⁠. Using the Scherrer equation, the found crystallite sizes are 2.4 and 3.5 nm, respectively. For both films, the crystallite size is smaller than the film thickness of 10 nm. This finding, together with the above-mentioned observation that the (002) peaks are visible in an asymmetric detection geometry, suggests that the 2D planes are oriented almost parallel to the substrate surface, but there is some disorder in the vertical stacking of the 2D layers leading to mosaicity.

SAED probes the in-plane periodicities and the in-plane grain size. The rings in the SAED patterns in Figs. 5(c) and 5(d) are significantly sharper for the film with 5:5 sccm H $2$S:H $2$ plasma composition, indicating a larger lateral grain size. The rings are continuous, implying that many grains contribute to the pattern, which is expected from the 1.3  $μ$m diameter probing area of the pattern. Radially averaged intensity curves extracted from the SAED patterns are displayed in Fig. 5(e). Four distinct lattice spacings of the 2H- $TaS 2$ phase can be recognized. The (010), (110), and (020) peaks represent the in-plane periodicities of the 2D layers. No (002) peak is present. Together, these observations confirm the textured growth of 2H-TaS $2$⁠. A small additional (011) peak implies that some crystals in the film are differently oriented. The broad, diffuse rings of the 10:0 sample indicates that there is some onset to ordering in this film, without the film being entirely crystalline.

From AFM (images in Fig. S5), no significant difference in the surface profile was observed, with both films appearing relatively smooth. The RMS roughness of both films is 0.3 nm.

Electrical properties of the $TaS x$ films were assessed using resistivity measurements. The measured resistivity values are shown in Fig. 4(c) for all films. Films deposited at temperatures below 300  $°$C have a high resistivity, above the upper measurement limit of approximately 1000  $Ω$ cm for 10 nm thick films. The addition of $H 2$ to the plasma results in less resistive films. In general, films with a low S:Ta ratio have a lower resistivity compared to films with a high S:Ta ratio. The trend in resistivity shows that the films conduct better with relatively more Ta in the film.

We developed an ALD process for $TaS x$ using as a precursor TBTDMT and as a coreactant a plasma mixture containing $H 2$S. The growth of $TaS x$ is linear over a wide temperature range, where the GPC varies from $1.17±0.03$ Å at 150  $°$C to $0.87±0.08$ Å at 400  $°$C. We observe that with addition of $H 2$ to the plasma mixture in the coreactant step, the film becomes less sulfur-rich (S:Ta $≈2$⁠). Furthermore, addition of $H 2$ to the plasma mixture leads to a change in the morphology of the film, from an amorphous matrix with some crystals to a polycrystalline film. More specifically, the 2H- $TaS 2$ phase was grown at 300  $°$C with 5:5 sccm $H 2$S:H $2$ flow. The 2H- $TaS 2$ film is strongly textured with different crystal grain orientations, mainly parallel to the surface. We showed that a lower amount of sulfur in the film corresponds to a low resistivity. The developed ALD process for $TaS x$ enables deposition of $TaS x$ for catalysis and nanoelectronics applications.

See the supplementary material for the optical constants determined by ellipsometry, the uniformity thickness map, the XPS data including the fits, and structural characterization data by AFM and TEM.

This work was performed with the financial support of ASM. We thank Detect99 for RBS and ERD measurements and analysis; S. A. Peeters for support on the XRD analysis; M. J. M. Merkx for input based on the previous TaN experiment; and C. O. van Bommel, B. Krishnamoorthy, C. van Helvoirt, J. J. L. M. Meulendijks, and J. J. A. Zeebregts for their technical assistance. We wish to acknowledge Solliance and the Dutch province of Noord-Brabant for funding the TEM facility.

The authors have no conflicts to disclose.

J. H. Deijkers: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead). H. Thepass: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal). M. A. Verheijen: Formal analysis (equal); Investigation (equal); Visualization (equal); Writing – review & editing (equal). H. Sprey: Conceptualization (supporting); Formal analysis (supporting); Writing – review & editing (supporting). J. W. Maes: Conceptualization (supporting); Formal analysis (supporting); Writing – review & editing (supporting). W. M. M. Kessels: Formal analysis (supporting); Resources (equal); Supervision (equal); Writing – review & editing (equal). A. J. M. Mackus: Conceptualization (supporting); Formal analysis (supporting); Project administration (lead); Resources (equal); Supervision (lead); Writing – original draft (supporting); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.