AVS 72 Session AP+2D+EL+EM+PS+TF-WeA: Atomic Scale Processing to Enable 2D Materials

Two-dimensional (2D) materials are of interest for a range of applications from microelectronics to sensing to energy storage and production. One of the most prominent group of 2D materials are metal dichalcogenides, some of which are semiconductors, such as the prototypical MoS₂, while others are (semi)metals such as TiS₂ and TaS₂.¹ Deposition of 2D materials with desired properties on large, often temperature-sensitive, and complexly shaped substrates is a key challenge for practical applications. Atomic layer deposition (ALD) can fulfill these requirements, but unlocking the full potential of ALD requires understanding of precursor chemistry, nucleation, substrates, and film characteristics, among other factors.^1,2

I will discuss chemistries for deposition of a range of 2D metal dichalcogenides including MoS_2,³ SnS₂,⁴ ReS₂,⁵ WS₂,⁶ HfS₂ and ZrS₂⁷ using thermal ALD. I will cover precursor challenges related to reactivity, thermal stability, and etching reactions. Controlling crystallinity and morphology and continuity at low thicknesses are also key challenges encountered in ALD of 2D materials. To this end, I highlight the role of substrate choice and the use of a two-step SnS₂ process where the deposited amorphous film is crystallized with mild-post deposition annealing.^4,8,9

Plasma-enhanced ALD (PEALD) can help overcome some of the challenges encountered in thermal ALD, such as limited reactivity of many precursors. For example, few thermal chemistries are available for direct deposition on low-cost plastic substrates for flexible applications (T ≤ 150 °C). To this end, I will discuss PEALD of TMDCs including MoS₂, TiS₂, and WS₂ at record-low temperatures down to 100 °C by controlling plasma chemistry¹⁰ I will also describe advanced ABC type processes with two plasma steps, where both the plasma chemistry (reactive radicals) and physics (low-energy ions) are used to control film growth and properties.¹¹

References:

1. Mattinen et al., Adv. Mater. Interfaces 8, 2001677 (2021).
2. Popov, Mattinen et al., JVSTA, 43, 030801 (2025).
3. Mattinen et al., Adv. Mater. Interfaces 4, 1700213 (2017).
4. Mattinen et al., Small 14, 1800547 (2018).
5. Hämäläinen, Mattinen et al., Adv. Mater., 30, 1703622 (2018).
6. Mattinen et al., JVSTA, 37, 020921 (2019).
7. Mattinen et al., Chem. Mater. 15, 5713 (2019).
8. Mattinen et al., 2D Mater., 7, 011003 (2020).
9. Mattinen et al., Adv. Mater. Interfaces, 7, 2001046 (2020).
10. Mattinen et al., Chem. Mater. 34, 7280 (2022).
11. Mattinen et al., ACS. Appl. Mater. Interfaces, 15, 35565 (2023).

Field effect transistors (FETs) are foundational components of modern electronics. For these applications, scalability is as important as device performance. Two-dimensional MoS₂ exhibits a direct band gap and enables high fidelity electrostatic control due to its reduced dimensionality and the absence of dangling-bonds at the surface. MoS₂ is therefore an ideal channel material for future FETs. Atomic layer deposition (ALD) is a promising technique for scalable, conformal, and back-end-of-line compatible MoS₂ deposition. While promising, the electrical performance of ALD grown 2D MoS₂ does not meet expectations when compared to the electrical properties of mechanically exfoliated MoS₂, due its polycrystalline nature. Polycrystalline MoS₂ contains grain boundaries that act as scattering centers, trap-rich regions, and structural discontinuities, that degrade its desirable electronic properties. In this work, ALD processes were developed using a novel molybdenum precursor, MoO₂(THD)₂. MoO₂(THD)₂ contains bulky organic ligands, which could impose steric hindrance during surface reactions, thus limiting surface chemisorption density and favoring formation of larger crystal grains.

First, MoO₂(THD)₂ volatility and decomposition behavior were studied by thermogravimetric analysis (TGA). TGA showed sublimation beginning at 120°C with 100% mass loss at 300°C. Next, plasma and thermal ALD processes were developed at 400°C. A mixed H₂S/Ar co-reactant was used for both plasma and thermal ALD. The effects of precursor and co-reactant dosing times, co-reactant ratios, as well as purge conditions were systematically investigated using in situ ellipsometry. The growth per ALD cycle (GPC) was measured to be 0.07 Å/cycle for the thermal recipe and 0.15 Å/cycle for the plasma process. The resulting films were characterized by X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, and atomic force microscopy (AFM). XPS indicates near-stoichiometric MoS₂. Raman spectra confirm MoS₂ vibrational modes and film crystallinity for both processes, while AFM reveals smooth, continuous films without out-of-plane features, with an RMS roughness of ~0.5 nm.

Overall, this work establishes MoO₂(THD)₂ -enabled thermal and plasma-enhanced ALD at 400 °C as a scalable route to smooth, continuous, and crystalline MoS₂ thin films. These results lay the groundwork for integrating conformal, large-grain MoS₂ channels into back-end-of-line compatible process flows, with ongoing grain-size quantification and electrical testing expected to link microstructural improvements to device performance.

Low-temperature magnetized plasmas generated by electron beams or non-thermal electrons offer new opportunities for atomic-scale materials processing relevant to microelectronics and quantum technologies.^1-3 Operating at low pressures (0.1–10 mTorr), the applied magnetic field enables spatial separation between a high-density plasma generation region (~10¹¹–10¹² cm⁻³, electron energies >10 eV) and a peripheral processing region characterized by lower plasma density (<10⁹ cm⁻³) and electron temperatures (<1 eV).⁴ Substrates placed in this region are primarily exposed to low-energy ion fluxes (≲ a few eV), enabling damageless processing of two-dimensional materials or processing with a controllable defect formation of such materials. However, plasma instabilities can drive anomalous cross-field transport and increase ion fluxes and energies at the surface.⁵ We present experimental results on instability mitigation in ExB plasma systems⁴ and demonstrate improved low-damage processing of graphene and transition metal dichalcogenides. Quantitative defect characterization in plasma-treated monolayered WS₂ films is presented and benchmarked against conventional RF inductively coupled plasma processing. A focus will also be made on the effect of vacuum conditions in the plasma reactor chamber on the defect formation (e.g. sulfur vacancies). Applications to hydrogen and nitrogen surface passivation of diamond for quantum devices are also discussed.²

References

¹ D. R. Boris et al., ECS J. Solid State Sci. Technol. 4, N5033 (2015)

² C. Pederson et al., Phys. Rev. Mater. 8, 036201 (2024)

³ F. Zhao et al., Carbon 117, 244 (2021)

⁴ N. S. Chopra, Y. Raitses, Appl. Phys. Lett.126, 064101 (2025)

⁵M. Tyushev et al., Phys. Plasmas 33, 013511 (2025)

Acknowledgement

Transition-metal dichalcogenides (TMDs) are a class of 2D materials consisting of covalently bonded sheets. Multi-layers are held together by cross-plane van der Waals forces. While these materials hold promise in future semiconductor applications, processes for manufacturing using TMDs are in the early stages of development. One challenge is obtaining uniform, monolayer growth of TMD films over large areas. Chemical vapor deposition results in the formation of multilayer islands. Etching of TMD films will also be required, if for no other reason than to thin multilayers to single layers. Atomic layer etching (ALE) is likely to be called upon for etching steps. TMDs are easily damaged by high temperature and energetic species bombardment, making thermal or plasma-assisted etching challenging. We have explored ALE of WS₂, a promising TMD 2D material. A remote plasma supplied reactive radicals. A switchable magnetic field allow low energetic ion bombardment to be turned on or off while having little effect on radical density. ALE of WS₂ was achieved in a two-step process. Vacuum-transfer x-ray photoelectron spectroscopy (XPS) was used to characterize the chemical composition and thickness of the remaining film at various stages in the process. Sulfur-terminated multilayer films of WS₂ were exposed to O atoms downstream of an Ar/O₂ inductively coupled plasma (ICP). Without ion bombardment thermal O atoms removed S from the surface and formed a tungsten oxide layer 1-2 monolayers thick, with a composition of roughly WO₄. These results agree favorably with parallel molecular dynamics simulations. In the presence of low energy ions (~15 eV) at a flux of ~3 x 10¹⁵cm^-2, the etching rate increased several-fold and the W-oxide stoichiometry was not changed. The Ar/O₂ plasma was followed by Ar/BCl₃ plasma to selectively etch the tungsten oxide layer relative to the underlying WS₂. This step required low energy ion bombardment to remove the WO₄ film and expose the next WS₂ layer. A detailed analysis of the film at various stages in the ALE process was carried out by using the high-resolution W(4f), S(2p), B(1s) and Cl(2p) in the film as well as the Si(2p) from the underlying substrate and O(1s) in the WS₂ film and underlying interfacial native oxide layer on Si. A complete picture of the stoichiometry of the film during ALE was obtained and will be presented.

This material is based upon work supported by the U.S. Department of Energy, Office of Science, Fusion Energy Sciences and Basic Energy Sciences, as part of the Extreme Lithography & Materials Innovation Center (ELMIC), a Microelectronics Science Research Center (MSRC), under contract No. DEAC02-09CH11466.

As semiconductor devices continuously scale down, the demand for channel materials of field effect transistors (FETs) with high mobility at low dimensions has increased. 2D transition metal dichalcogenides (TMDs) have gained interest due to relatively consistent mobility at 5 nm of body thickness compared to conventional silicon channel. However, non-uniform large area synthesis has been a challenge in integration. To improve device performance, non-uniform adlayers should be removed layer-by-layer to realize a uniform and large-area TMD films.

To investigate layer-by-layer etching of MoS₂, thermochemical analysis suggests thermodynamic feasibility of using oxygen and fluorine chemistries for MoS₂ etching. Individual and synergistic effects of Ar⁺, O radical (produced by a coaxial waveguide microwave source) and XeF₂ (3Torr) were studied in the molecular beam where each species can be independently controlled in flux and energy. The as-received MoS₂ has S/Mo and Si/Mo ratios of 1.6 and 1.0. Neither O nor XeF₂ spontaneously reacted with the as-received MoS₂. Ar⁺ ion beam (200V, 0.5mA, 10s) was used to remove surface contamination prior to etching, resulting S/Mo and Si/Mo ratios of 0.94 and 0.18 respectively. Following in-situ Ar⁺ ion beam cleaning, O radical was shown to oxidize MoS₂ decreasing S/Mo from 0.94 to 0.46, while XeF₂ removes MoS₂ increasing Si/Mo from 0.18 to 7.1. It showed surface fluorination (~3%), corresponding to formation of MoOF_x and CF_x.

Finally, the sequential process consisting of Ar⁺ ion bombardment, O radical introduction, and XeF₂ vapor exposure (0.5T, 10 pulses) was applied which could enable surface activation, self-limiting oxidation and removal of modified layer by forming oxyfluorides. After XeF₂, O atomic composition decreased and S atomic composition increased indicating removal of MoO₃ with surface fluorination (~3%) of MoF_x and MoOF_x. Increasing number of pulses to 20 increases S/Mo from 1.1 to 1.4 while reducing the MoO₃ content. This research reveals the individual and combination effects of Ar⁺ ion beam, O radicals and XeF₂ on MoS₂ etching and potentially applied to investigating mechanism of layer-by-layer etching of MoS₂.

This research is supported by the U.S. DOE, Office of Science, as part of the ELMIC MSRC through the Plasma-enabled 2D Materials project at PPPL, under contract No. DEAC02-09CH11466.

Atomic layer etching (ALE) of 2D MoS₂ nanoribbons is important for the fabrication of MoS₂ channel transistors. E-beam lithography and conventional dry etching processes face problems when etching MoS₂ films. These disadvantages include delamination for nanoribbon widths < 25 nm and inability to control line edge roughness. ALE methods may bypass these limitations and produce narrower width MoS₂ channels with smoother edges that may have better electrostatic control and on/off switching.

Narrowing the width of 2D MoS₂ nanoribbons requires lateral etching in the 2D plane. Our earlier work on the thermal ALE of 2D MoS₂ showed that sequential exposures of O₃ and SOCl₂ produced lateral etching by either “outside-in” or “inside-out” mechanisms. In this study, the “outside-in” lateral etching of 2D MoS₂ crystalline nanoribbons was utilized to reduce the widths of 2D MoS₂ nanoribbons. O₃ (ozone) was used for MoS₂ oxidation to MoO₃ and SOCl₂ (thionyl chloride) was used for the volatilization of MoO₃ as MoO₂Cl₂. The studies were performed using high quality 2D MoS₂ nanoribbons on silicon coupons. The MoS₂ nanoribbons were prepared from monolayer 2D MoS₂ films using e-beam lithography and dry etching.

The etching of the 2D MoS₂ nanoribbons was examined by atomic force microscopy (AFM) at 150°C. The lateral etching could be quantified by returning to the same nanoribbon after various numbers of etching cycles and measuring the nanoribbon width using AFM. Figure 1 shows that the width of the nanoribbons decreases progressively versus number of etching cycles. The AFM images indicate that the etching cycles also smooth the nanoribbon line edge roughness.

The AFM images were consistent with lateral etching at the edges of the 2D MoS₂ nanoribbons. The AFM measurements determined that the 2D lateral etching rate was ~3.9 Å per etching cycle at 150°C as displayed in Figure 2. This lateral etch rate represents ~1 MoS₂ units removed per etching cycle at the step edge of the 2D MoS₂ crystalline domains. Additional AFM measurements revealed that nanoribbon widths < 25 nm were possible using this thermal ALE process.