As announced in our previous blog post on the 2nd area selective deposition (ASD2017) workshop, this post will highlight the new developments in the field of ASD. I will summarize my take-home messages from the workshop by discussing three topics that were addressed in several presentations:
- New categories of area selective deposition (ASD)
- The improvement of the selectivity during the deposition
- The development of plasma-assisted ASD processes.
New categories of area selective deposition (ASD)
ASD is about limiting the deposition to a specific part of the surface. Most of the work on ASD reported in the past two decades include a patterning step to define where on a substrate deposition needs to occur. As discussed in our review paper published in 2014, many area selective atomic layer deposition (ALD) approaches comprise a patterning step in which the surface is locally de-activated or activated.1
Instead of including a patterning step, some applications require the deposition on already-nanostructured surfaces. This could mean that patterning has been performed earlier on in the process flow, or that for example the aim is to deposit selectively on nanoparticles or so. Three different categories of area selective ALD that fall into this description were discussed during the workshop: (1) ASD on a device structure aimed at self-aligned fabrication, (2) topographically-selective ALD, and (3) facet-selective ALD. These categories are illustrated in Figure 1.
Figure 1 Three new categories of ASD that were discussed during the ASD2017 workshop.
Currently, one of the main motivations for area selective ALD is to tackle the challenge of aligning features in multilayered device structures with nanometer accuracy (described by edge placement errors, EPE). ASD has the potential to eliminate this bottleneck by enabling self-aligned fabrication, which was described by Gert Leusink from TEL as a change of the patterning paradigm (see Figure 2). In-self-aligned fabrication, the starting point is a partially-processed device structure that has already been patterned in an earlier process step. The aim is to selectively deposit on a specific material, without depositing on the other materials present in the device structure. In this way, the material is applied in a bottom-up fashion without performing additional lithography and etching steps, thereby eliminating the alignment challenge. ASD for self-aligned fabrication requires deposition processes that are selective to the surface of specific materials.
Figure 2 Slide from Gert Leusink (TEL) on the change of the patterning paradigm towards self-aligned and self-assembly processing. This can considered as a shift to bottom-up fabrication.
A second category of ASD that was discussed during the workshop was referred to as topographically-selective deposition.2 Stacey Bent from Stanford University described an approach for ASD relying on deactivation of the horizontal surfaces of a trench structure using an anisotropic CFx implantation step. In this implantation step an interfacial layer was formed with a hydrophobic character. Subsequently, area selective ALD of Pt was obtained at the unaffected sidewalls of the trench.
Figure 3 Slide from Stacey Bent on topographically-selective ALD.2
Rong Chen from Huazhong University of Science and Technology introduced a new concept that she referred to as facet-selective ALD. By exploiting differences in Ce(thd)4 precursor adsorption on different facets of a Pt nanoparticle, epitaxial growth of CeO2 on the Pt(111) facets was obtained. This new category of ASD can enable innovative opportunities in the field of catalysis, by providing new avenues for controlled synthesis of catalysts. The first study on facet-selective ALD was recently published.3
Figure 4 Slide from Rong Chen on facet-selective ALD.3
Improving the selectivity during the deposition
One of the main conclusions I drew during the workshop is that the community seems to be moving to adding steps during the deposition to improve the selectivity, instead of just influencing the deposition prior to the deposition. I also mentioned this as a key insight during the discussion session at the end of the workshop. Conventionally, ASD often relies on deactivation of a part of the surface by applying self-assembled monolayers or resist films prior to the deposition.1 In contrast, a large part of the presentations and posters during the workshop were about dosing various species during the deposition to improve the selectivity. This comes in two classes: (i) by re-applying deactivation molecules or (ii) by adding etching steps.
Let me first mention some examples of the first category. In the first presentation, Stacey Bent talked about the regeneration of the protective SAMs. To counteract the damage that the SAM undergoes during the deposition, dodecanethiol SAM molecules were dosed in vapor-phase after every 150 ALD cycles.4 The enhanced selectivity enabled three times thicker ZnO to be grown selectively on SiO2 without coating Cu. In addition, there were two examples of using so-called inhibitor molecules to influence the selectivity of the deposition process. John Abelson from University of Illinois at Urbana-Champaign discussed the use of inhibitors during area selective CVD of Cu. He showed that when VTMS is used as an inhibitor during Cu CVD, the nucleation on SiO2 substrates is sensibly suppressed.5 Secondly, inspired by John Abelson’s work, as well as by related work by Yanguas-Gil et al.,6 Alfredo Mameli from our group at the TU/e introduced a new approach for area selective ALD based on the use of three-step (ABC-type) ALD cycles. In this approach, the first step (A) involves the exposure of the surface to inhibitor molecules that selectively adsorb on those surfaces on which no growth should occur. Subsequently, these inhibitor molecules should block the precursor adsorption in step B. Finally, in step C, the inhibitor molecules are removed during the co-reactant pulse. Area selective ALD of SiO2 was demonstrated using an ALD process consisting of acetylacetone (Hacac) inhibitor, BDEAS precursor, and O2 plasma pulses. This process was demonstrated to result in area selective deposition of SiO2 on for example GeO2, SiNx or WOx, while it led to nucleation delays on Al2O3, TiO2, and HfO2.
There were even more examples of combinations of ASD and selective etching. I counted 8 mentions of ideas based on etching throughout the workshop. To me this seems to be connected to the fact that it is generally very difficult to make ASD processes sufficiently selective. There are always defects and impurities present on the surface on which growth can initiate. On the other hand, there is a toolbox of etching processes available that allows for etching with a high selectivity.
The main presentation on this topic was presented by Rémy Gassiloud from CEA, LETI, which focused on a supercycle approach in which ALD and selective etching were alternated.7 This supercycle approach is illustrated in Figure 5. The starting point was the plasma-assisted ALD process of Ta2O5 from TBTDET precursor and O2 plasma. This process shows a nucleation delay when performed on Si or SiO2 substrates. The nucleation delay was exploited for area selective ALD of Ta2O5 on TiN in the presence of Si as the non-growth area. An NF3 plasma was employed after every eight ALD cycles to remove the deposited Ta2O5 from the Si surface. Similar work performed in our TU/e group was shown in a poster by Sonali Chopra and Martijn Vos for the case of area selective ALD of Ru. Here, an O2 plasma etching step was included in a supercycle to improve the selectivity of Ru ALD.
Figure 5 Slide illustrating the approach of improving the selectivity of area selective ALD by combining it with atomic layer etching (ALE). I will present this slide during the tutorial of ALD 2017 (Denver, July 15-18)
Also the presentation of Younghee Lee from the University of Colorado at Boulder on atomic layer etching (ALE) was extremely relevant for this topic. When combining area selective ALD with etching, it essential that the etching step is based on self-limiting surface reactions. Otherwise, the main hallmark of ALD, its ability to conformally deposit on a nanostructured sample, is lost. The developments in the field of ALE are therefore very relevant for the ASD community. Younghee discussed several new chemistries for isotropic ALE based on the use of HF-pyridine as the co-reactant.8,9 His results on the selectivity of the different ALE processes give insight into for which ASD process the selectivity can potentially be improved.10
As mentioned by Tahsin Faraz during the discussions session, this focus on novel combinations with selective etching processes raises the question whether it is essential to aim for a high selectivity in ASD. In practice, applications will most likely include selective etching steps anyway to ensure a sufficiently high selectivity.
Development of plasma-assisted ASD processes
In the ALD community, thermal ALD processes are generally used as a starting point for developing an ASD process, because plasmas can for example destructively interact with a SAM, while plasma-assisted ALD processes typically readily nucleate on most surfaces without a nucleation delay. Considering these limitations, it is interesting to point out that there were three presentations on plasma-assisted ASD processes. As already mentioned, Alfredo Mameli presented results on area selective ALD of SiO2 for which an O2 plasma was used as the co-reactant. One of the benefits of re-applying the inhibitor molecules every cycle is that this area selective ALD approach is compatible with the use of ozone or plasmas. Ekaterina Filatova from Tyndall presented results of theoretical calculations aimed at plasma-enhanced ALD and CVD of Si-based materials. It was found that SiH4 plasma can lead to selective deposition on a Si surface, in presence of a carbon doped oxide (CDO) C-Si surface. Furthermore as already mentioned, Rémy Gassiloud talked about making the plasma-assisted ALD process of Ta2O5 selective by combining ALD with selective etching.
Moreover, plasmas can also be used for selective etching or for selective functionalization of the surface. As noted above, we have used O2 plasma etching steps in our group to improve the selectivity of Ru ALD by including the O2 plasma together with Ru ALD cycles in a supercycle. Ivan Zyulkov from IMEC employed a plasma treatment prior to Ru ALD to influence the selectivity by selective functionalization. A H2 plasma treatment transformed a SiCN surface into a hydrophilic surface, while the amorphous carbon non-growth area kept its hydrophobic character. Subsequently, area selective growth of Ru was obtained on the SiCN regions. Taken together, the broadening of ASD to plasma-assisted methods extends the set of materials that can be deposited in an area selective manner, and provides new approaches for achieving ASD.
I will discuss some of the new development described above at ALD 2017 (Denver, July 15-18) during the tutorial on “Approaches, challenges, and opportunities for area selective ALD”.
(1) Mackus, A. J. M.; Bol, A. A.; Kessels, W. M. M. The Use of Atomic Layer Deposition in Advanced Nanopatterning. Nanoscale 2014, 6, 10941–10960.
(2) Kim, W. H.; Minaye Hashemi, F. S.; Mackus, A. J. M.; Singh, J.; Kim, Y.; Bobb-Semple, D.; Fan, Y.; Kaufman-Osborn, T.; Godet, L.; Bent, S. F. A Process for Topographically Selective Deposition on 3D Nanostructures by Ion Implantation. ACS Nano 2016, 10 (4), 4451–4458.
(3) Cao, K.; Shi, L.; Gong, M.; Cai, J.; Liu, X.; Chu, S.; Lang, Y.; Shan, B.; Chen, R. Nanofence Stabilized Platinum Nanoparticles Catalyst via Facet-Selective Atomic Layer Deposition. Small 2017, 1700648.
(4) Minaye Hashemi, F. S.; Bent, S. F. Sequential Regeneration of Self-Assembled Monolayers for Highly Selective Atomic Layer Deposition. Adv. Mater. Interfaces 2016, 3 (21).
(5) Babar, S.; Mohimi, E.; Trinh, B.; Girolami, G. S.; Abelson, J. R. Surface-Selective Chemical Vapor Deposition of Copper Films through the Use of a Molecular Inhibitor. ECS J. Solid State Sci. Technol. 2015, 4 (7), N60–N63.
(6) Yanguas-Gil, A.; Libera, J. A.; Elam, J. W. Modulation of the Growth per Cycle in Atomic Layer Deposition Using Reversible Surface Functionalization. Chem. Mater. 2013, 25, 4849.
(7) Vallat, R.; Gassilloud, R.; Eychenne, B.; Vallée, C. Selective Deposition of Ta 2 O 5 by Adding Plasma Etching Super-Cycles in Plasma Enhanced Atomic Layer Deposition Steps. J. Vac. Sci. Technol. A Vacuum, Surfaces, Film. 2017, 35 (1), 01B104.
(8) Lee, Y.; George, S. M. Atomic Layer Etching of Al2O3 Using Sequential , Self-Limiting Thermal Reactions with Sn(acac)2 and Hydrogen Fluoride. ACS Nano 2015, 9 (2), 2061.
(9) George, S. M.; Lee, Y. Prospects for Thermal Atomic Layer Etching Using Sequential, Self-Limiting Fluorination and Ligand-Exchange Reactions. ACS Nano 2016, 10 (5), 4889–4894.
(10) Lee, Y.; Huffman, C.; George, S. M. Selectivity in Thermal Atomic Layer Etching Using Sequential, Self-Limiting Fluorination and Ligand-Exchange Reactions. Chem. Mater. 2016, 28 (21), 7657–7665.