Here we present a new strategy for implementing atomic layer etching (ALE) of metals using wet chemistry at room temperature. Unlike plasma- and thermal-ALE, material removal in wet ALE is achieved through dissolution of the self-limiting passivation layer rather than volatilization. This relaxes the temperature requirements of thermal ALE, and the sputter threshold requirements for plasma ALE. Solubility is strongly dependent on solvent species, so solvent exchange can be used as the trigger for release of the self-limiting passivation layer as an alternative to ligand exchange. In the first half of the etch cycle, a self-limiting metal-ligand surface complex is formed on exposure to the first etch solution. The surface layer must be insoluble in this solution, but readily soluble in the second etch solution in order to complete the ALE cycle. Additionally, the second solution must be unreactive with the freshly exposed metal surface. Sequential exposure to both etchants is required for material removal. A diagram of this wet ALE process is shown in Figure 1.

Most transition metal complexes form around oxidized metal centers, so formation of the surface complex is governed by solution redox chemistry in the presence of appropriate ligands. Solubilization of this complex can be achieved in one of several ways. Solvent exchange can be used to dissolve the layer in an appropriate solvent. Ligand exchange in a non-redox active solution can be used to alter the solubility of the surface complex without changing the solvent or reacting with the underlying substrate. And lastly, ion exchange can be used if the surface complex is a coordination compound containing both a ligand-bound complex ion and a counterion.

Previous pathways for thermal ALE of metals, such as Cu and Co, have employed either oxidation or chlorination reactions for surface modification followed by either ligand-substitution or ligand-addition reactions for volatile release. This study introduces another mechanism for the thermal ALE of metals using sequential oxidation and chlorination reactions. This mechanism relies on the metal having volatile oxychloride complexes. The thermal ALE of Mo was studied using quartz crystal microbalance (QCM), x-ray reflectivity (XRR), and quadrupole mass spectrometry (QMS) measurements. Mo films were first deposited on quartz crystals or silicon coupons using sputtering. The mass or thickness changes were then measured versus sequential oxidation and chlorination reactions using O₃/O₂ and SOCl₂ as the reactants.

The QCM measurements observed very digital mass changes during the sequential O₃/O₂ and SOCl₂ exposures. QCM results showed pronounced mass gains during O₃/O₂ exposures and distinct mass losses during SOCl₂ exposures. QCM measurements for three cycles of Mo ALE at 175°C are displayed in Figure 1. The mass changes were also self-limiting versus O₃/O₂ and SOCl₂ reactant exposure. The mass loss per Mo ALE cycle was -1,600 ng/cm² or 15.7 Å/cycle under the saturation reaction conditions at 175°C. The Mo ALE etch rate was studied from 100°C to 250°C and larger etch rates were observed at higher temperatures.

Mo ALE is dependent on the release of volatile oxychloride etch products. For under-saturation reaction conditions, the mass changes during the O₃/O₂ exposures could either be negative or positive. Lower O₃/O₂ exposures led to negative mass changes and higher O₃/O₂ exposures led to positive mass changes. This behavior can be explained based on the volatile oxychloride etch product during Mo thermal ALE. There is chlorine on the surface after the SOCl₂ exposure. Initial O₃/O₂ exposures remove the chlorine as a volatile Mo oxychloride product and produce a mass loss. As the chlorine is removed, larger O₃/O₂ exposures can then proceed to oxidize the Mo surface and produce a mass gain.

QMS experiments were performed to identify the etch products. To improve the detection sensitivity, these experiments utilized various powder samples to model the Mo surface after the O₃/O₂ or SOCl₂ exposures. The main Mo oxychloride species detected was MoO₂Cl₂. Figure 2 shows the observation of MoO₂Cl₂ during SOCl₂ exposure on MoO₃ after a previous O₃/O₂ exposure. The parent MoO₂Cl₂⁺ ion intensities are observed in a cluster of peaks around m/z 200. All of the main ion intensities are explained by the natural isotopic abundances of Mo and Cl in MoO₂Cl₂.

Selective removal of metal gate materials such as TiC, TiAlC, TiN, or TiAl is required in the fabrication of next generation field effect transistor (FET) of logic semiconductor devices. In our development of advanced etching methods, we have proposed wet-dry etching or wet-like plasma etching that combines the advantages of wet etching (high isotropy and selectivity) and dry etching (high controllability). For atomic layer etching (ALE) processes, surface modification reduces surface energy of sample surface in the first step, and the modified layer is removed in the next step.

Surface modification of metal carbide TiAlC using floating wire-assisted non-halogen vapor plasma, that can generate a high-density plasma at medium pressure, has been demonstrated in our recent study, indicating a potential of isotropic and selective etching of TiAlC by using chemistries containing NH, H, and OH. In this study, the N-H-O containing plasmas can selectively remove TiAlC over TiN. Film thickness change of titanium compounds was measured by an in situ ellipsometer, and surface chemical bonding during the processing was analyzed by an in situ attenuated total reflectance-Fourier transform infrared spectrometer. Optical emission spectra of NH and OH molecular lines from plasmas were detected by a high-resolution optical emission spectrometer. The sample surface was analyzed before and after the plasma treatment by X-ray photoelectron spectroscopy.

By controlling the surface reactions, selective removal of TiAlC over TiN can be achieved by its reaction with N-H-O plasma to form volatile products, as firstly proposed here. We emphasized that this non-halogen etching method can be applied for highly selective etching of metal carbide, nitride, or oxide.

Metals are of increasing importance in modern electronics. For example, Ni and Ni-based alloys have been used for extreme ultraviolet lithography (EUVL) absorber layers and as components in composite multiferroics for magnetoelectric devices. In this work, we present an atomic layer etching (ALE) technique to etch Ni that relies on plasma nitridation to modify the Ni surface and organic vapor exposure to remove the modified layer. Theoretical study by density functional theory has shown that ALE of metals including Ni is possible through nitridation and formic acid (FA) exposure, through the formation of dimeric Ni₂(HCOO)₄ and NH₃ as reaction products.¹ Since surface nitridation is not a spontaneous reaction like oxidation upon exposure to ambient air (e.g., moisture), this process can be analyzed more definitively without the convolution of native oxide formation.² In this work, we experimentally examined this new method using a remote nitrogen plasma and a custom vapor chamber to modify and etch nickel and nickel based alloys.

A pure nitrogen plasma in a custom planar geometry ICP reactor was used to form the nickel nitride.The thickness of the nitridized layer is controlled by modifying the power, pressure, and exposure time of the plasma and quantified by a combination of spectroscopic ellipsometry (SE) and scanning electron microscopy (SEM).To alleviate the complication in SE analysis by the fact that the optical constants of thin metal films can vary depending on deposition method and thickness, SE measurements of samples that have undergone similar numbers of ALE cycles are fit simultaneously assuming that the optical constants over a narrow thickness range are the same. This improves the thickness measurement accuracy, as corroborated by SEM analysis. FA in the solution phase and vapor phase are both examined as a means of removing the nitridized Ni layer, with X-ray photoelectron spectroscopy (XPS) showing both phases completely removing the N 1s peak that forms after plasma nitridation. The etch rate of Ni by nitridation and FA solution is 6.1 nm/cycle.For the case of a Ni ALE cycle involving plasma nitridation and FA vapor exposure at 80°C, the etch rate was about 1 nm/cycle, indicating a relatively controllable process. For the Ni_xAl_y alloy, both solution phase and vapor phase FA at 80ׄ°C removed the nitridized surface layer. However, this led to Al surface enrichment, suggesting the ALE cycle preferentially etched Ni over Al. The selectivity to common hardmasks was also tested.

References

1. Y. Xia and P. Sautet, Chem. Mater. 33 (17), 6774-6786 (2021).

2. X. Sang, E. Chen and J. P. Chang, J. Vac. Sci. Technol. A 38 (4), 042603 (2020).

Cu is widely used for interconnect metal in semiconductor devices, but the resistance of Cu increases significantly as the interconnect width decreases below 10 nm due to electron scattering at grain boundary and surface. Ruthenium is one of the candidate materials to replace Cu as the interconnect metal because resistivity of Ru is lower than that of Cu when the interconnect width is less than 10 nm. [1] Reactive ion etching (RIE) is used for direct patterning of metals, but controlling atomic-level film thickness in nanopatterning processes is challenging. Atomic layer etching (ALE) is suggested as an alternative to RIE because ALE can remove layers with atomic scale precision, excellent uniformity, and low surface roughness. [2]

In this work, plasma ALE process was performed for Ru in an inductively coupled plasma (ICP) reactor. The ALE process consists of two steps: surface fluorination with CHF₃, C₄F₈ or CF₄ plasmas and ion bombardment with Ar plasma. The CHF₃ and C₄F₈ plasmas generate fluorocarbon layer on the Ru surface and the CF₄ plasma converts the Ru surface to RuF_x. The C₄F₈ plasma produces fluorine-rich fluorocarbon layer on the Ru surface compared to the CHF₃ plasma. Etch per cycle (EPC) of Ru was investigated according to ion energy and etching time. ALE window of Ru was observed in the ion energy range of 150 ~ 200 V for CHF₃ and C₄F₈ plasmas and 100 ~ 200 V for CF₄ plasma. The EPC of Ru was determined to be 0.6 nm/cycle for CHF₃ plasma, 1.5 nm/cycle for C₄F₈ plasma, and 0.3 nm/cycle for CF₄ plasma in the ALE window region. Chemical sputtering threshold energy of Ru was determined to be 100 V for CHF₃ plasma, 50 V for C₄F₈ and CF₄ plasmas. The EPC of Ru was increased with increasing Ar plasma time and exhibits self-limiting properties at 180 seconds for CHF₃ plasma and at 300 seconds for C₄F₈ and CF₄ plasmas. Surface roughness and etch residues of Ru were compared with RIE and ALE.

References

[1] S. Decoster, E. Camerotto, G. Murdoch, S. Kundu, Q. T. Le, Z. Tőkei, G. Jurczak and F. Lazzarino, J. Vac. Sci. Technol. B 40, 032802 (2022)

[2] Y. Kim, S. Lee, Y. Cho, S. Kim, and H. Chae, J. Vac. Sci. Technol. A 38, 022606 (2020)