Last year we finished a project called “Taking plasma ALD to the next level” (funded by the Netherlands Organization for Scientific Research (NWO), Applied and Engineering Sciences (AES) section). This project was devoted to obtaining a better understanding of plasma ALD such that open scientific and technological questions could be addressed resulting in new plasma ALD processes, a wider range of ALD materials, and novel and improved applications of ALD. The dissertation by the PhD student Karsten Arts working on the project can be downloaded here.
With his work, Karsten continued on work by several other PhD students (Saurabh Karwal, 2020, Tahsin Faraz, 2019 and Harald Profijt, 2011) that all tried to get a better understanding of the role of the ionic species in the plasma during plasma ALD. This body of work was however not just about understanding the role of ions, but we also tried (and succeeded) to enlarge the available parameter space and open up more possibilities for film preparation by ALD. The research led, for example, to the implementation of external substrate biasing in commercial plasma ALD tools as supplied by various equipment suppliers as I blogged about before. We demonstrated that energetic ions can be used to modify the properties of a large range of materials (an overview will be given below), often in a beneficial way (but not always). This can basically be done by just turning the “knob” of the ion energy without touching any of the other ALD and plasma conditions. We also discussed that ion energy control can play a very important role in area-selective deposition (ASD) and topographically-selective processing and the ion energy control by RF biasing also motivated us to work together with a company to develop a commercial power supply for tailored waveform biasing. By the way, another motivation for this development was a project on plasma-based atomic layer etching (ALE) which is funded by the same funding organization (NWO-AES). Coming back to the work of Karsten Arts, he demonstrated that not only highly energetic ions matter during plasma ALD, also fairly low energy ions can have a significant effect on ALD film growth and the material properties. See his blog here including a presentation.
So what to report on in this blog post?
First and foremost, we present a new animation that summarizes the role of ions during plasma ALD in a concise and fairly simple way. This animation is based on the research that we carried out over the years.
Next, we give an overview of our related publications. For the reader’s convenience, we list them in chronological order and briefly summarize their main message and significance to the field of ALD.
We will also give an up-to-date table with the materials that have been prepared by plasma ALD using substrate biasing and this includes also reports in the literature by other researchers. Some of those publications are listed too. Finally, an updated list of original equipment manufacturers (OEMs) is given that commercially offer plasma ALD tools with substrate biasing.
The animation: Plasma ALD with various plasma scenarios
This animation first introduces a typical plasma ALD cycle. Such a cycle is very similar to a typical thermal ALD cycle but with the main difference that the co-reactant step (step 3) involves a plasma. The precursor dosing in step 1 is similar for plasma and thermal ALD as well as the fact that the precursor and co-reactant steps are separated by purge steps (step 2 and 4). The four steps form the cycle which is repeated as many times as is required to reach the film thickness desired. Note that for ALD the surface reactions during the precursor and co-reactant step need to be self-limiting and that the dosing should be sufficiently long to reach “saturation” conditions. Also the purge steps need to be sufficiently long to avoid CVD-like conditions (also known as “parasitic CVD”).
Next the animation shows four plasma scenarios. These different scenarios can be obtained by changing the plasma conditions, at least if the plasma ALD system allows you to do so and when the specific plasma source can operate in that parameter space. Besides hardware design-related aspects, important experimental “knobs” are the pressure during the plasma step, the power applied to the plasma source, and the application of a substrate bias when operating the plasma.
- Radicals – virtually no ions. Under these conditions no or hardly any ions make it to the substrate. This can be due to the plasma ALD system configuration or due to the operation at relatively high plasma pressures. As there is no ion component, the surface reactions are completely ruled by plasma radicals. Typically, these are very “soft” conditions.
- Radicals + low energy ions. These are typical conditions for plasma ALD systems. Ions are arriving at the substrate, but they have a fairly low ion energy, let’s say < 35 eV. This is typically the case for remote plasma ALD systems or for parallel plate capacitively coupled plasmas (CCPs) with the substrate placed on the grounded electrode. The surface chemistry is still ruled by the radical species but the ions can lead to enhanced surface reactions, such as enhanced ligand or impurity removal or reactions benefiting from surface diffusion.
- Radicals + high energy ions. These are less ordinary conditions which can only be generated under special circumstances. A straightforward method is by applying a substrate bias by connecting an additional power supply to the substrate stage (the plasma ALD system must be prepared for that). Also under these conditions the surface chemistry is mostly determined by the plasma radicals but now the ion energy is so high that ions can penetrate into the subsurface region of the material and lead to material modification, such as material densification, impurity removal or crystallization.
- Radicals + too high energy ions. This is similar to the previous scenario but now the ion energies are that high such that the material modification leads to deteriorated film properties and hence to material damage. Obviously, the separation between the two scenarios with high energy ions is very material and application dependent. Besides direct damage to the material being prepared by ALD, also the underlying substrate or other films/structures on the substrate can be damaged and consequently device damage can occur.
The typical energy ranges for the four scenarios are very material dependent. Note that other plasma conditions such as the ion flux play a very important role too and the product of ion energy and flux can be a critical parameter as well for the material properties. Furthermore, the effects described do not necessarily always correspond to the general description given above. To keep the animation relatively simple to understand, these nuances have been left out of the description of the four scenarios.
Overview of our publications on the role of ions during plasma-enhanced ALD
Over the years, we have published several papers on the role of ions during plasma-enhanced ALD and on how to precisely control the energy of the ions impacting the films during plasma-enhanced ALD. In chronological order:
First paper published in which the ion flux and ion energy during plasma-enhanced ALD is reported, in this particular case for remote plasma ALD.
First full review paper about plasma-enhanced ALD, also addressing the role of ions during the process.
First report on RF substrate biasing during plasma-enhanced ALD, including measurements of the ion flux energy distribution. It is shown that the crystallinity of TiO2 can be controlled by the ion energy.
Follow-up paper on RF substrate biasing during plasma-enhanced ALD, now reporting the effect of ion energy on more metal oxides.
Another follow-up paper extending the use of substrate biasing also to metal nitrides and demonstrating the anisotropic flux of the ions in 3D structures.
Publication that shows the drastic decrease in resistivity of HfNx when applying substrate biasing during plasma-enhanced ALD. Also shows interesting results on trenches.
A paper looking at RF substrate biasing more from the plasma perspective. Includes extensive results on the ion flux energy distribution and introduces the ion energy dose parameter.
A follow-up on the first review paper about plasma-enhanced ALD from 2011. Again the role of ions and ion bombardment during plasma-enhanced ALD is addressed.
In this work we draw the conclusion that the impinging ions during the plasma-enhanced ALD of HfNx lead to a very low electrical resistivity films by suppressing the oxygen incorporation and in-grain nano-porosity in the films.
This paper revealed the surprising result that even low energy ions (<30 eV) have a significant impact on the material properties and film growth by plasma-enhanced ALD.
A novelty: an area-selective ALD process based on plasma-enhanced ALD with substrate biasing.
This publication shows that with tailored waveform biasing a much narrower ion flux energy distribution can be achieved than with RF substrate biasing. This is very promising for plasma-enhanced ALD.
Comparing results for plasma-enhanced ALD of SiO2 with RF substrate biasing for two different reactor configurations: Oxford Instruments FlexAL and Sentech Silayo.
This publication shows that low ion energies (<20 eV) have a surprisingly big influence on the growth-per-cycle of TiO2 and the crystallinity of the films.
A follow-up on the earlier work on HfNx in which experiments and first principles calculations reveal insights into the O incorporation and how this can be mitigated upon the impingement of energetic ions on the surface.
The new plasma source as employed in the Oxford Instruments AtomFab investigated in detail, includes ion flux and ion energy measurements.
Now comparing results for plasma-enhanced ALD of HfO2 when deposited with the Oxford Instruments FlexAL and the Sentech Silayo, both with RF substrate biasing.
A review paper that presents a brief overview of the results on the role of ions during plasma-enhanced ALD putting the work also in a wider perspective.
Plasma-enhanced ALD with substrate biasing: material overview and commercial equipment
The materials shown below have been prepared by plasma-enhanced ALD with substrate biasing.
* To be published
| Al2O3 | Profijt et al. (TU/e); Kim (Stanford) |
| Co3O4 | Profijt et al. (TU/e) |
| HfO2 | Faraz et al. (TU/e); Beladiya et al. (Jena & TU/e); Kim et al. (Hanyang) |
| SiO2 | Faraz et al. (TU/e); Beladiya et al. (Jena & TU/e); Park et al. (Hanyang) |
| TiO2 | Profijt et al. (TU/e); Faraz et al. (TU/e); Ratzsch et al. (Jena) |
| AlN | Legallais et al. (Grenoble) |
| HfNx | Faraz et al. (TU/e); Karwal et al. (TU/e) |
| SiNx | Faraz et al. (TU/e) |
| TaCyNx | Peeters et al. (TU/e)* |
| Ta2O5 | Chaker et al. (Grenoble); Yeghoyan et al. (Grenoble) |
| TiNx | Faraz et al. (TU/e); Merkx et al. (TU/e) |
| TiMoxNy | Chowdhury et al. (Lehigh) |
Other notable publications about plasma-enhanced ALD with substrate biasing:
Commercial ALD systems for plasma-enhanced ALD with substrate biasing
| Oxford Instruments | FlexAL; PlasmaPro ASP |
| Sentech | Fiji |
| Veeco | Silayo |
To conclude, the above shows that quite some progress has been made in the last decade with respect to gaining a better understanding of the role of ions during plasma ALD as well as to enlarging the available parameter space of the material and process properties during plasma ALD. We are confident that (precise) ion flux and energy control will play an important role in the future of plasma ALD! Stay tuned!
2 Comments