ALD/ALE 2024 Session ALE1-WeM: Plasma and Energy-Enhanced ALE

The critical dimensions of semiconductor devices are continuously shrinking with 3D device structure and are approaching to nanometer scale. The demand for dimension control in angstrom level is drastically increasing also in etching processes. In this talk, various plasma-enhanced ALE (PEALE) processes are discussed for anisotropic and isotropic patterning of metals and dielectric materials including molybdenum, ruthenium, cobalt, titanium nitride, tantalum nitride, aluminum oxide, hafnium oxide, zirconium oxides, silicon oxide, and silicon nitride. [1-10] Typical ALE processes consist of surface modification step and removal step. Various surface modification schemes were applied including fluorocarbon deposition on surface, surface fluorination, surface chlorination, surface oxidation with radicals generated with plasmas. For the removal or etching step, the modified layers were removed by ion-bombardment, heating, ligand exchange, ligand volatilization, or halogenation. The characteristics of reaction kinetics, surface roughness, surface residue in the plasma-enhanced ALE processes will be also discussed.

1) K. Koh, Y. Kim, C.-K. Kim, H. Chae, J. Vac. Sci. Technol. A, 36(1), 10B106 (2017)
2) Y. Cho, Y. Kim, S. Kim, H. Chae, J. Vac. Sci. Technol. A, 38(2), 022604 (2020)
3) Y. Kim, S. Lee, Y. Cho, S. Kim, H. Chae, J. Vac. Sci. Technol. A, 38(2), 022606 (2020)
4) D. Shim, J. Kim, Y. Kim, H. Chae, J. Vac. Sci. Technol. B., 40(2) 022208 (2022)
5) Y. Lee, Y. Kim, J. Son, H. Chae, J. Vac. Sci. Technol. A., 40(2) 022602 (2022)
6) J. Kim, D. Shim, Y. Kim, H. Chae, J. Vac. Sci. Technol. A., 40(3) 032603 (2022)
7) Y. Kim, S. Chae, H. Ha, H. Lee, S. Lee, H. Chae, Appl. Surf. Sci. 619, 156751 (2023)
8) Y. Kim, H. Kang, H. Ha, C. Kim, S. Cho, H. Chae, Appl. Surf. Sci. 627, 157309 (2023)
9) Y. Kim, H. Kang, C. Kim, H. Chae, ACS Sustain. Chem. Eng. 11(16), 6136 (2023)
10) Y. Kim, H. Kang, H. Ha, M. Choi, M. Jeon, S. Cho, H. Chae, Plasma. Process. Polym., DOI: 10.1002/ppap.202300161, (2023)

Plasma atomic layer etching (ALE) processes were developed for titanium nitride (TiN) with surface fluorination or chlorination in the modification step and ion bombardment in the removal step. In the modification step, the TiN surface was fluorinated using NF₃ plasma to form Ti-F bonds or chlorinated using BCl₃ plasma to form Ti-Cl bonds. The fluorinated or chlorinated layers were removed by ion bombardment using Ar plasma. The etch per cycle (EPC) of TiN was investigated depending on ion energy and etching time. The ALE window of constant EPC of TiN was confirmed in the energy range of 25 – 80 V for fluorination and 70 – 80 V for chlorination. The EPC of TiN was determined to be 1.5 nm/cycle in fluorination and 3.8 nm/cycle in chlorination in the ALE window region. The EPC of TiN increased with increasing Ar plasma time and exhibits self-limiting properties at 180 seconds for both fluorination and chlorination. The strong Ti-F peak was observed about 13% on the surface after ALE process with fluorination, but small Ti-Cl peaks were observed below 1% after ALE process with chlorination. The Root-mean-square (RMS) roughness of the TiN surface after ALE was measured at 0.24 nm for fluorination and 0.29 nm for chlorination, which is lower than before etching.

Nickel aluminides have attracted tremendous research attention as a potential alternative material for barrier-less interconnects and EUV mask absorber applications due to its lower resistivity than other elemental metals and higher printing resolution, respectively.^1,2 However, down-scaling these binary intermetallics brings emerging challenges in developing etch methods due to the high chemical inertness and low gas-phase volatility of Ni compounds. In addition, preserving the concentration of each element in the alloy during the etching process is essential to be useable at the nanoscale. In this work, we report a new and promising approach towards the atomic layer etching (ALE) of stoichiometric NiAl and Ni₃Al. A super-cycle ALE process has been developed, in which each sub-cycle enables the removal of the Al-rich and Ni-rich phases, respectively. Fluorine-containing plasma and trimethyl-aluminum (TMA) precursors have been utilized to etch Al in the first phase. Whereas, in the second phase, the etching of Ni has been achieved using a nitrogen-containing plasma, probably forming Ni nitride and then chelation by hexafluoro-acetylacetone (Hhfac) precursor. A fundamental ALE study on each individual ALE process has been performed on pure Ni and pure Al films to evaluate the etching efficiency and selectivity of each element versus the other. In-situ spectroscopic ellipsometry measurements were employed to elucidate the self-limiting nature of the ALE sequence on the pure metal film and thickness change during alloy etching. The saturation behavior for each precursor on the compatible pure metal films has been observed. X-ray reflectivity measurements were carried out to confirm the etch rates on Ni-aluminide films. The etch rates were varied from 0.5 ± 0.10 Å/super-cycle at 250 ˚C to 3.3 ± 0.23 Å/super-cycle at 350 ˚C. Furthermore, atomic force microscopy analysis shows that thin Ni aluminide films remained smooth during initial etching and might introduce a little roughness when the etch cycles progressed. The results obtained from x-ray photoelectron spectroscopy confirmed that, at the optimized super-cycle ALE process, the etched Ni aluminide films still preserve the same concentration of each element in the alloy. The developed ALE process enabled the etching of nickel aluminide alloys and created a paradigm for future studies on alternative etching of binary intermetallics.

Atomic Layer Etching (ALE) is a self-limiting process used in nanofabrication that allows for precise material removal on the atomic scale by a cyclic nature introducing reactive gas adsorption and removal of surface atoms by a low-energy ion bombardment. In practical realisation of ALE, e.g. using commercially available RIE tools, the ion energy in plasma may exceed the sputtering threshold, typically 20-40 eV. It will result in a quasi-ALE (Q-ALE) regime where the process steps are not self-limiting, but still highly accurate and of lower damage compared to RIE. Direct measurements of the ion energy in Q-ALE are thus instumental for studies of both the etching process and surface damage effects.

For Si etching experiments, we have used a commercial Inductively Coupled Plasma RIE Takachi™ tool from Plasma-Therm LLC, USA. The tool operates in a Cl₂-molecular activation regime using Ar RF-plasma to desorb the etch reaction products in the etch step. Unpatterned silicon-on-insulator (SOI) 10x10 mm² samples with a 50 nm thick top Si layer were used for the etching experiments and characterised by spectroscopic ellipsometry and AFM. Typically, the Q-ALE operated in a 25 cycle mode, sufficient to measure the Si thickness difference by ellipsometry using a 4-layer model. A Retarding Field Energy Analyzer (RFEA) from Impedance LLC, Ireland, was installed to measure the Ar⁺ ion energy and ion current densities at different Q-ALE conditions, such as RF-power and pressure. The surface effects of the etched SOI samples are characterised by AFM, SEM and Kelvin Probe Force Microscopy (KPFM).

In this work we present the actual data of Ar⁺ ion energy distribution and the ion current densities at the sample level in the commercial Takachi™ RIE tool operating in Q-ALE mode. In order to reach a low ion energy regime, we used RF-power in the range of 3-16 W, that corresponds to the bias voltages of 10-120 V. In this voltage range the ion energy peak shifts from ≈50 to 147 eV, respectively, with an increase in the RF-power. The calculated Ar⁺ ion flux from the measured ion currents is about 3x10¹³ – 2x10¹⁴ ion cm^-2 s^-1 that fits well with the published data. Etch per cycle (EPC) was measured in both sputtering (no Cl₂) and ALE-regime at different bias set-points (RF-power). A clear EPC plateau for the ALE regime was observed at biases of 20-40 V that corresponds to the peak Ar⁺ ion energy of 50-70 eV. The AFM and high-resolution SEM data demonstrated a decreased surface roughness of etched Si in the plateau region indicating a low surface damage. More details to be presented.

Renowned for its exceptional hardness and chemical impermeability, diamond is a wide bandgap semiconductor material with promising applications in optoelectronics, quantum optics, and beyond. Despite its desirable properties, the inherent manufacturing and fabrication challenges associated with monocrystalline diamond make it difficult to smooth, leading to lossy devices in its various applications. Moreover, existing methods for surface smoothing are often costly and time intensive.

In this context, atomic layer etching (ALE) emerges as a potentially viable solution for smoothing diamond surfaces. ALE is a cyclical process that precisely defines nanostructures by removing individual atomic layers. Recently, bias-pulsed ALE (BP-ALE), which is an augmented ALE process wherein reagent gases are not purged between steps, was shown to drastically smooth the surface of 4H-SiC substrates.

Here, we present a novel BP-ALE recipe tailored for diamond substrates. This innovative approach not only achieves precise etching of diamond but also facilitates rapid surface smoothing compared to existing conventional methods. To evaluate the efficacy of BP-ALE, diamond films were grown on etched surfaces, conventionally smoothed substrates, and commercially purchased substrates. Our findings indicate that BP-ALE not only expedites surface smoothing but also yields growth-ready surfaces suitable for subsequent diamond growth.

In summary, our study underscores the potential of BP-ALE as a cost-effective and efficient method for smoothing diamond surfaces, thereby enhancing their suitability for various optoelectronic and quantum optical applications.