Selectivity in Thermal Atomic Layer Etching Using Sequential, Self-Limiting Fluorination and Ligand-Exchange Reactions

Thermal atomic layer etching (ALE) can be achieved with sequential, self-limiting surface reactions. One mechanism for thermal ALE is based on fluorination and ligand-exchange reactions. For metal oxide ALE, fluorination converts the metal oxide to a metal fluoride. The ligand-exchange reaction then removes the metal fluoride by forming volatile products. Previous studies have demonstrated the thermal ALE of amorphous Al2O3 films. However, no previous investigations have explored the differences between the thermal ALE of amorphous and crystalline Al2O3 films. This study explored the thermal ALE of amorphous and crystalline Al2O3 films. HF, SF4, or XeF2 were used as the fluorination reactants. Trimethylaluminum (TMA) or dimethylaluminum chloride (DMAC) were used as the metal precursors for ligand-exchange. Spectroscopic ellipsometry measurements revealed that the amorphous Al2O3 films had much higher etch rates than the crystalline Al2O3 films. When using HF and TMA at 300 °C, the amorphous Al2O3 film was removed at an etch rate of 0.78 Å/cycle. For the crystalline Al2O3 film, an etch rate of 0.06 Å/cycle was initially observed prior to the stoppage of etching after removing about 10 Å of the film. Thermal ALE with HF and DMAC resulted in similar results. Etch rates of 0.60 and 0.03 Å/cycle were measured for amorphous and crystalline Al2O3 films at 300 °C, respectively. Other fluorination agents, such as SF4 or XeF2, were also used together with TMA or DMAC for Al2O3 ALE. These reactants for fluorination and ligand-exchange were able to etch amorphous Al2O3 films at 300 °C. However, they were unable to etch crystalline Al2O3 film at 300 °C beyond the initial 10–20 Å surface layer. The investigations also examined the effect of annealing temperature on the etch rate per cycle using HF and TMA as the reactants at 300 °C. Amorphous Al2O3 films were etched at approximately the same etch rate of 0.78 Å/cycle until the crystallization of amorphous Al2O3 films at ≥ 880 °C. The differences between amorphous and crystalline Al2O3 thermal ALE could be used to obtain selective thermal ALE of amorphous Al2O3 in the presence of crystalline Al2O3.

Thermal atomic layer etching (ALE) is an important technique for the precise isotropic etching of nanostructures. Thermal ALE of many materials can be achieved using a two-step fluorination and ligand-exchange reaction mechanism. Most previous thermal ALE processes have used HF as the fluorination reactant. Alternative fluorination reactants may be needed because HF is a weak nucleophilic fluorination reactant. Stronger fluorination agents may be required for the fluorination of some materials. To demonstrate the usefulness of SF4 as an alternative to HF, thermal Al2O3 ALE was compared using SF4 or HF together with Sn(acac)2 as the metal precursor for ligand exchange. SF4 and HF were observed to behave similarly as fluorination reactants during Al2O3 ALE. The mass gains during the initial SF4 and HF exposures on Al2O3 atomic layer deposition (ALD) films at 200 °C were comparable at 35 and 38 ng/cm2, respectively, using quartz crystal microbalance measurements. In addition, the etch rates were similar at 0.20 and 0.28 Å/cycle for Al2O3 ALE using SF4 and HF, respectively, at 200 °C. Thermal VO2 ALE was also performed for the first time using SF4 or HF and Sn(acac)2 as the reactants. There was evidence that SF4 is a stronger fluorination reactant than HF for VO2 fluorination. The mass gains during the initial SF4 and HF exposures on VO2 ALD films were 38 and 20 ng/cm2, respectively, at 200 °C. Thermal VO2 ALE also had a higher etch rate when fluorinating with SF4 compared with HF. Etch rates of 0.30 and 0.11 Å/cycle were measured for VO2 ALE using SF4 and HF, respectively, together with Sn(acac)2 at 200 °C. Fourier transform infrared experiments were also used to monitor fluorination of the Al2O3 and VO2 ALD films by SF4 or HF. FTIR difference spectroscopy was used to observe the increase of Al–F and V–F stretching vibrations and the loss of the Al–O and V—O/V═O stretching vibrations for Al2O3 and VO2, respectively, versus SF4 or HF exposure at 200 °C. Additional absorbance features after fluorination of the Al2O3 ALD films by SF4 were consistent with SFx surface species. SF4 is a useful fluorination agent for thermal ALE processes and can be used as an alternative to HF. In addition, SF4 may be necessary when fluorination requires a stronger fluorination reactant than HF.

Thermal atomic layer etching (ALE) can be achieved using sequential, self-limiting fluorination and ligand-exchange reactions. Previous studies have demonstrated thermal ALE of amorphous HfO2 and ZrO2 ALD films. This study explored the differences between thermal ALE of amorphous and polycrystalline films of hafnium oxide, zirconium oxide, and hafnium zirconium oxide on silicon wafers. HF, XeF2, or SF4 were used as the fluorination reactants. Titanium tetrachloride or dimethylaluminum chloride (DMAC) was employed as the metal precursor for ligand exchange. The spectroscopic ellipsometric measurements revealed that the amorphous films had much higher etch rates per cycle than the crystalline films regardless of the fluorination reactants or metal precursors for ligand exchange. The differences were most pronounced for HfO2. Using HF and TiCl4 as the reactants at 250 °C, the etch rates were 0.36 Å/cycle for amorphous HfO2 and 0.02 Å/cycle for crystalline HfO2. In comparison, the etch rates were 0.61 Å/cycle for amorphous ZrO2 and 0.26 Å/cycle for crystalline ZrO2. The etch rates were 0.35 Å/cycle for amorphous HfZrO4 and 0.04 Å/cycle for crystalline HfZrO4. When HF and DMAC were used as the reactants, the etch rates were higher than with HF and TiCl4 for every material. Using HF and DMAC as the reactants at 250 °C, the etch rates were 0.68 Å/cycle for amorphous HfO2 and 0.08 Å/cycle for crystalline HfO2. In comparison, the etch rates were 1.11 Å/cycle for amorphous ZrO2 and 0.82 Å/cycle for crystalline ZrO2. The etch rates were 0.69 Å/cycle for amorphous HfZrO4 and 0.16 Å/cycle for crystalline HfZrO4. SF4 as the fluorination reactant resulted in higher etch rates than for HF when using TiCl4 as the metal precursor for ligand exchange. XeF2 as the fluorination reactant resulted in even higher etch rates than for SF4. The differences in the etch rate with the fluorination reactant can be partially attributed to differences in thermochemistry for fluorination by HF, SF4, and XeF2. The differences in etch rates between amorphous and crystalline films may be caused by the greater degree of fluorination and subsequent ligand-exchange reaction for the amorphous films. The amorphous films have a lower density and may be able to better accommodate the large volume expansion upon fluorination.

Thermal atomic layer etching (ALE) of Al2O3 and HfO2 using sequential, self-limiting fluorination and ligand-exchange reactions was recently demonstrated using HF and tin acetylacetonate (Sn(acac)2) as the reactants. This new thermal pathway for ALE represents the reverse of atomic layer deposition (ALD) and should lead to isotropic etching. Atomic layer deposition and ALE can together define the atomic layer growth and removal steps required for advanced semiconductor fabrication. The thermal ALE of many materials should be possible using fluorination and ligand-exchange reactions. The chemical details of ligand-exchange can lead to selective ALE between various materials. Thermal ALE could produce conformal etching in high-aspect-ratio structures. Thermal ALE could also yield ultrasmooth thin films based on deposit/etch-back methods. Enhancement of ALE rates and possible anisotropic ALE could be achieved using radicals or ions together with thermal ALE.

Thermal atomic layer etching is rapidly becoming an important complementary processing technology in the manufacturing of 5 and 3 nm devices in the semiconductor industry. Critically, architectures such as 3D NAND and 3D DRAM require conformal isotropic etching to remove material such as HfO2 in hard-to-reach locations with aspect ratios that can be greater than 50:1. To achieve repeatable device performance throughout a 3D stack, the removal rate (etch per cycle) of the etched material during an etch process needs to be controlled such that the overall etch amount is the same from top to bottom of the device stack. In this work, the reaction kinetics of reactants and byproducts during a cyclical ligand exchange-based atomic layer etching (ALE) process have been modelled. This ALE process consists of two steps: a fluorination step followed by a fluorine-to-chlorine ligand exchange-based removal step. Modeling was performed for each of those steps separately. Experimental data revealed that the fluorine dosing during the fluorination step was predominantly responsible for controlling the etch rate of the ALE process but had only a minimal impact on the etch profile inside high aspect ratio holes. The ligand exchange dosing, on the other hand, predominantly controlled the etch profile (depth loading) with equal etch rates from top-to-bottom, obtained when the step was operated close to saturation. The model predicts that the chemical reaction rate of dimethylaluminum chloride (DMAC) on a fluorinated surface during the ligand exchange step is 9.1 s−1, about 46 times greater than the reaction rate of hydrogen fluoride (HF) on the hafnium oxide surface during the fluorination step (only 0.2 s−1). Furthermore, modeling results revealed that the sticking coefficient for DMAC on a hafnium fluoride surface far exceeded that of HF on a hafnium oxide surface in the conditions modelled (0.94 s−1 for DMAC vs 0.0058 s−1 for HF). With these modeling results, the different roles fluorination and ligand exchange steps have regarding the control of etch rate per cycle and profile inside high aspect ratio holes can be explained.

Wafer scale atomic layer etching (ALE), one of the key fabrication processes of both cutting-edge gate-all-around (GAA) and complementary field-effect transistors, is essential for the implementation of a 3D gate stack where a few nanometer-thick gate oxide should be etched out in an isotropic manner. In the present study, we successfully resolved the limit of conventional thermal ALE, which has been using excessively reactive and corrosive HF, by adopting plasma-assisted thermal ALE with SF6 for the source of fluorine radicals. As a result, we successfully established self-limiting etching conditions and, in turn, an optimized ALE in a 6 in. wafer scale with a uniformity of more than 99% at 5 Torr. We suggested a step-by-step ALE mechanism of surface oxide layers consisting of conversion by trimethylaluminum, fluorination by F*, and removal, which was supported by the combination of an alternative fluorination reaction and X-ray photoelectron spectroscopy analysis. After the fabrication of silicon GAA multichannel nanostructure, deposition of SiO2 by ALD, and applying the optimized ALE, we confirmed that the oxide layers are conformally etched out from cross-sectional transmission electron microscopy.

Trimethylaluminum (TMA, Al(CH3)3) was used as the metal precursor, together with HF, for the atomic layer etching (ALE) of Al2O3 using sequential, self-limiting thermal reactions. Al2O3 ALE using TMA demonstrates that other metal precursors, in addition to Sn(acac)2, can be employed for Al2O3 ALE. The use of TMA for Al2O3 ALE is especially interesting because TMA can also be used for Al2O3 atomic layer deposition (ALD). Quartz crystal microbalance (QCM) experiments monitored Al2O3 ALE at temperatures from 250 to 325 °C. The Al2O3 ALE was linear versus the number of HF and TMA reaction cycles. The QCM studies showed that the sequential HF and TMA reactions were self-limiting versus reactant exposure. The Al2O3 etching rates increased at higher temperatures. The QCM analysis measured mass change per cycle (MCPC) values that varied from −4.2 ng/(cm2 cycle) at 250 °C to −23.3 ng/(cm2 cycle) at 325 °C. These MCPCs correspond to Al2O3 etch rates from 0.14 Å/cycle at 250 °C to 0.75 Å/cycle at 325 °C. X-ray reflectivity and spectroscopic ellipsometry analyses confirmed the linear removal of Al2O3 and etching rates. Fourier transform infrared spectroscopy measurements monitored Al2O3 ALE by observing the loss of infrared absorbance from Al–O stretching vibrations. Surface intermediates were also identified after the HF and TMA exposures. Al2O3 ALE with TMA is believed to occur by the reaction Al2O3 + 4Al(CH3)3 + 6HF → 6AlF(CH3)2 + 3H2O. The proposed mechanism involves fluorination and ligand-exchange reactions. The HF exposure fluorinates the Al2O3 and forms an AlF3 surface layer and H2O as a volatile reaction product. During the ligand-exchange transmetalation reaction, TMA accepts F from the AlF3 surface layer and donates CH3 to produce volatile AlF(CH3)2 reaction products. The QCM measurements were consistent with an AlF3 surface layer thickness of 3.0 Å on Al2O3 after the HF exposures. The larger etch rates at higher temperatures were attributed to the removal of a larger fraction of the AlF3 surface layer by TMA exposures at higher temperatures.

Atomic layer etching (ALE) is set to be a vital part of the advanced semiconductor manufacturing toolbox, known for its precise control of the film thickness and minimal damage. These benefits are possible due to the sequential self-limiting half-cycles that are employed within an ALE process. Initially, ALE was underestimated due to low etch rates, but it is now experiencing a renaissance due to the requirements imposed by further downscaling.1 The ALE community is mostly divided into two groups: plasma anisotropic and thermal isotropic etching. 2 In this work, the focus is on exploring isotropic plasma ALE, which is a relatively unexplored direction. This approach involves using plasma radicals to modify and/or etch material isotropically, while minimizing the directional ion component of the plasma. Isotropic plasma ALE can offer several benefits over strictly thermal ALE, such as higher etch rates, lower operating temperatures, and the ability to etch more resistant materials.In this talk we will first address some of the different techniques employed for plasma isotropic ALE in the literature. Subsequently, we will discuss our own work, which has focused on two isotropic plasma ALE chemistries. The first involves alternating between plasma fluorination and ligand-exchange half-cycles, this ALE chemistry has been found to be effective for etching Al2O3, HfO2, AlN, and GaN films.3,4 Surface contaminants and roughness were found to be reduced post ALE, an aspect that is particularly important for AlN and GaN films that are used in quantum, power semiconductor and LED applications. The second approach involves using diketone (e.g. acetylacetone, hexafluoroacetylacetone) doses followed by either H2 or O2 plasma exposures for ALE of Al2O3 and ZnO.5 The diketone in this process acts to etch the surface, however due to competitive adsorption of different diketone orientations an etch inhibition layer forms. The plasma step is then utilized to clean the surface, removing the self-limiting inhibitor layer, enabling etching to continue in the next cycle. This ALE chemistry diverges from the typical modification/removal ALE cycle, where a modified layer if often observed on the material post ALE. In comparison the diketone ALE chemistry does not rely on modification of the surface making it an ideal process for surface cleaning. This approach provides improved selectivity over the plasma fluorination process and also enables smoothing.Plasma isotropic ALE is a growing part of the ALE community that can help accelerate the adoption of ALE in academia and industry. The potential benefits of ALE are only just beginning to be explored, and more processes are needed to etch the ever-increasing range of materials utilized in IC manufacturing. Kanarik, K. J., Tan, S. & Gottscho, R. A. Atomic Layer Etching: Rethinking the Art of Etch. J Phys Chem Lett 9, 4814–4821 (2018).George, S. M. Mechanisms of Thermal Atomic Layer Etching. Acc Chem Res 53, 1151–1160 (2020).Chittock, N. J. et al. Isotropic plasma atomic layer etching of Al2O3 using a fluorine containing plasma and Al(CH3)3. Appl Phys Lett 117, 162107 (2020).Wang, H., Hossain, A., Catherall, D. & Minnich, A. J. Isotropic plasma-thermal atomic layer etching of aluminum nitride using SF6 plasma and Al(CH3)3. 1–14 (2022) doi:arXiv.2209.00150.Mameli, A., Verheijen, M. A., Mackus, A. J. M., Kessels, W. M. M. & Roozeboom, F. Isotropic Atomic Layer Etching of ZnO Using Acetylacetone and O2 Plasma. ACS Appl Mater Interfaces 10, 38588–38595 (2018).

In thermal atomic layer etching (ALE) of Ni, a thin oxidized Ni layer is removed by a hexafluoroacetylacetone (hfacH) etchant gas at an elevated surface temperature, and etching ceases when a metallic Ni surface appears (self-limiting step). However, atomistic details of the self-limiting step was not well understood. With periodic density-functional-theory calculations, it is found that hfacH molecules barrierlessly adsorb and tend to decompose on a metallic Ni surface, in contrast to the case of a NiO surface, where they can form volatile Ni(hfac)2. Our results clarify the origin of the self-limiting process in the thermal ALE.

Thermal atomic layer etching (ALE) of metals is fundamentally constrained by the low volatility of metal halides, which has limited the scalability, selectivity, and temperature flexibility of dry etching processes....

Multivariable run-to-run control of thermal atomic layer etching of aluminum oxide thin films

The continuous driving force from the semiconductor industry for realizing smaller features and device structures with higher density and higher-aspect ratio poses increasing challenges in traditional etching techniques. Atomic layer etching (ALE) provides the opportunity to overcome the shortcomings of more conventional etching processes. Extensive research in this field has led to the development of two main classes: plasma ALE to achieve anisotropic etching and thermal ALE for the isotropic counterpart. Besides anisotropic etching, plasma can also be applied for achieving isotropic etch profiles. Yet, this process option has not been explored extensively. In this work, we demonstrate isotropic ALE of ZnO on a regular array of vertical nanowires, using the alternated doses of O2 plasma and acetylacetone (Hacac). A linear ZnO thickness decrease with increasing number of cycles was measured by in-situ spectroscopic ellipsometry, and saturation behavior was demonstrated for both Hacac and O2 plasma, at 250 oC. The etch rates per cycle were found to increase from 0.5 to 1.3 A/cycle with process temperature increasing from 100 to 250 oC. Transmission electron microscopy (TEM) studies conducted on ZnO-coated nanowires before and after ALE proved the isotropic character of the process. Moreover, the ALE process produced no surface damage as concluded from high resolution TEM inspection. In-situ infrared spectroscopy measurements were conducted to elucidate the mechanism underlying self-limited etching. Based on these measurements, a preliminary reaction mechanism is proposed, in which Zn(acac)2 is assumed to be the etching product and persisting acac-species on the ZnO surface provide the self-limiting character. Finally, a high etch selectivity over SiO2 and HfO2, and the possibility to extend the process to other oxides such as Al2O3 are also demonstrated. We believe that this plasma-based ALE process represents a valuable addition to the ALE toolbox and opens-up new possibilities for using plasma-processing to achieve isotropic etch profiles. Furthermore, similar approaches can be used to extend the process to other materials by tuning the plasma chemistry accordingly.

A dramatic increase in the Al2O3 atomic layer etching (ALE) rate versus time was demonstrated using sequential, self-limiting exposures of hydrogen fluoride (HF) and trimethylaluminum (TMA) as the reactants with no purging. The normal purging expected to be required to prevent chemical vapor etching or chemical vapor deposition (CVD) is not necessary during the Al2O3 ALE. This purgeless, rapid atomic layer etching (R-ALE) was studied from 250 to 325 °C using various techniques. In situ quartz crystal microbalance (QCM) measurements monitored Al2O3 R-ALE at 300 °C. The Al2O3 R-ALE process produced linear etching versus number of R-ALE cycles. Each HF exposure fluorinates the Al2O3 substrate to produce an AlF3 surface layer. Each subsequent dose of TMA then undergoes a ligand-exchange transmetalation reaction with the AlF3 surface layer to yield volatile products. Using reactant partial pressures of HF = 320 mTorr and TMA = 160 mTorr, the fluorination and ligand-exchange reactions produced a mass change per cycle (MCPC) of −32.1 ng/(cm2 cycle) using sequential, 1 s exposures for both HF and TMA with no purging. This MCPC equates to a thickness loss of 0.99 Å/cycle or 0.49 Å/s. Comparison experiments using the same reactant exposures and purge times of 30 s yielded nearly identical MCPC values. These results indicate that the etch rates for Al2O3 R-ALE are much faster than for normal Al2O3 ALE because of shorter cycle times with no purging. Smaller MCPC values were also observed at lower reactant pressures for both Al2O3 R-ALE and Al2O3 ALE. The QCM studies showed that the Al2O3 R-ALE process was self-limiting versus reactant exposure. Ex situ spectroscopic ellipsometry and x-ray reflectivity (XRR) measurements revealed temperature-dependent etch rates from 0.02 Å/cycle at 270 °C to 1.12 Å/cycle at 325 °C. At lower temperatures, AlF3 growth was the dominant mechanism and led to an AlF3 atomic layer deposition (ALD) growth rate of 0.33 Å/cycle at 250 °C. The transition temperature between AlF3 growth and Al2O3 etching occurred at ∼270 °C. XRR scans showed that the Al2O3 ALD films were smoothed by Al2O3 R-ALE at temperatures ≥270 °C. Additionally, patterned wafers were used to compare Al2O3 R-ALE and normal Al2O3 ALE in high aspect ratio structures. Scanning electron microscope images revealed that the etching was uniform for both processes and yielded comparable etch rates per cycle in the high aspect ratio structures and on flat wafers. The HF and TMA precursors were also intentionally overlapped to explore the behavior when both precursors were present at the same time. Similar to ALD, where precursor overlap produces CVD, precursor overlap during Al2O3 ALE leads to AlF3 CVD. However, any AlF3 CVD growth that occurs during precursor overlap is removed by spontaneous AlF3 etching during the subsequent TMA exposure. This spontaneous AlF3 etching explains why no purging is necessary during R-ALE. R-ALE represents an important advancement in the field of thermal ALE by producing rapid etching speeds that will facilitate many ALE applications.

Ligand exchange reactions at the surface of oleate- and trioctylphosphine oxide (TOPO)-capped CdS quantum dots have been studied with attenuated total reflection infrared (ATR-IR) spectroscopy, using thin films deposited from organic solvent suspensions. The oleate and trioctylphosphine capping ligands were found to form highly ordered and densely packed monolayers on the CdS surface. Adsorbed oleate is coordinated to CdS in a chelating bidentate manner through the carboxylate functional group, while adsorbed trioctylphosphine oxide is coordinated though the P=O functional group and appears to have numerous adsorption environments on the CdS surface. Exposure of such films to aqueous solution was found to cause partial delamination of the films from the ATR prism interface which was reversible upon redrying. Ligand exchange reactions on the oleate- and trioctylphosphine-capped CdS films were studied in situ at room temperature by allowing the films to be exposed to dilute aqueous solutions of thiol-containing ligands. Oleate and trioctylphosphine oxide are both strongly adsorbed to the CdS surface, and ligand exchange with monothiol-containing ligands has been found to be highly dependent upon experimental conditions, in particular pH, where exchange is only observed at solution pH where the exchanging ligand is uncharged. This is attributed to the inability of a charged ligand to penetrate the hydrophobic polymethylene layer on the CdS surface.

Unique Ligand Exchange Dynamics of Metal–Organic Polyhedra for Vitrimer-like Gas Separation Membranes