Key findings:
- ALD/Database continued to grow in 2020 by 323 new entries reaching 4527 entries in total.
- In 2020, we saw the introduction of:
- 123 new ALD processes,
- 20 new precursors,
- 17 new materials.
- Trending: most new ALD process go beyond AB-style processes, e.g. for ternary materials.
- Catch-up effect seen in first year (2019) subsided, i.e. relatively fewer old papers added.
Status of the ALD/Database in 2020
The beginning of the new year is a good time to briefly look back on the ALD/Database in 2020 just as we also did last year. The ALD/Database received 323 new entries in 2020 which included articles published in 2020 but also works published in previous years that were only added to the database in 2020. Overall, it is good to see that the field of ALD is still going strong despite the unusual circumstances that we had due to COVID-19 which significantly impacted research and work in the labs for most of us.
How to contribute to and use the ALD/Database?
You can contribute to the ALD/Database by submitting papers which only requires the DOI of the manuscript: To add a new ALD process, click on the circular menu button in the bottom right and click “add process”. To add a new reference to an existing process, navigate to the ALD process via the material and then press on the “add” at the end of the list of references. We will then check for duplicates, notation of the chemicals, and content of the submitted paper and add it to the database.
We encourage the use of the ALD/Database in your presentations and papers. You can download an image of the periodic table (as shown above) for use under the creative commons agreement from the website itself. See the round menu button (bottom right corner) on the website for the download button; you can cite the ALD/Database using its DOI:10.6100/ALDDatabase. Also let us know if you are using the ALD/Database in your presentation (send us a screenshot/picture)! This is of course fully optional but it allows us to quantify its usage better. Of course it is very good for our motivation as well!
Looking closer at the total number of 323 new entries over 2020, this is lower than last year’s number of 1293 (a non-sustainable 3.5 entries per day). This slower growth was also to be expected: The ALD/Database went live in 2019 and in the first year of its existence there was basically a community driven effort to catch up on the “older” literature; this effect has subsided to some degree by now. To verify if this is indeed an important effect having an impact on the number of submissions (e.g. not only an effect of COVID-19), an insightful metric is the time between publication in a journal and submission to the ALD/Database, abbreviated here as TbPS. The average TbPS was calculated for the entries submitted in 2020 and the same was done for the entries submitted in 2019. A subsiding catch-up effect should result in a shorter TbPS for 2020 than for 2019. This is indeed the case with a TbPS of 2.9 years for 2020 and 5.1 years for 2019. It will be interesting to monitor this metric in the coming years and see whether this trend continues.
In the calculation of the TbPS metric, we excluded the initial dataset present in the ALD/Database before it went live. These initial entries were not submitted by our users and contained a relatively high number of old papers. Including those works would have skewed the TbPS metric of 2019 appearing as if a lot of “old” papers were added that year. This initial data set was largely based on the references in several ALD review papers (see also last year’s blogpost). The use of these review papers ensured that the coverage of the ALD/Database up to the point of their publication is quite complete. Several of the review papers that we used were published just around 2010 so the coverage of the ALD/Database is expected to be fairly good up to that point. From 2010 onwards we rely on community-contributed entries and the coverage is therefore expected to improve each year as more and more works are added to the ALD/Database.
All things considered, the ALD/Database now contains 4527 entries to ALD papers on December 31 2020. The figure below shows a breakdown of the number of papers published in each year for a snapshot of the database taken in December 2019 and December 2020. The number of new publications in 2020 – i.e. the last bar in the bar chart – appears to be in-line with the previous years so a similar number of papers were published in 2020 as in 2019 (and added to the ALD/Database). The artefact in the data around 2010 – caused to some degree by the usage of review papers to initially fill the database as mentioned before – has faded a little over 2020. As can be seen, more entries have been submitted this year for the 2010 – 2015 period compared to e.g. 2005 – 2010. Although it is not clear if this artefact will go away completely but it should become less prominent in the coming years when more entries are submitted. Obviously, this depends also on your contribution!
Strategies for preparing doped and ternary materials by ALD
Strategies to synthesize complex materials such as doped and ternary materials by ALD include so-called multi-step approaches, co-dosing approaches, and super-cycle schemes as illustrated in the figure below.
In the co-dosing approach, not all of the reactants are dosed separately but at least two reactants are dosed simultaneously in the “co-dosing”. An example of this approach is the ALD process for ZnOxSy where both H2O and H2S are dosed simultaneously controlling the composition by the mixing ratio of the gas-phase reactants. In a multistep approach, a conventional ALD process consisting of two half-cycle (AB-type) is extended by adding one or more steps (ABC-, ABCD-type, etc.). These additional steps can introduce an additional constituent to the material, resulting in e.g. deposition of a ternary instead of a binary material. Another application of a multistep scheme is to modify the properties of a material via the additional step. For example in the low-temperature Pt ALD process using PtMe3CpMe and O2-plasma the regular ALD process results in PtOxat low temperatures. Metallic Pt can be deposited at low temperatures using a multistep approach by adding a H2 plasma as a third step to reduce the PtOx to metallic Pt. The supercycle scheme combines two or more ALD processes in a “larger” super-cycle which itself can be repeated. Such a super-cycle consists of N cycles of the “first” ALD process and M cycles of a “second” ALD process. For example, to synthesize Al doped MoS2, a supercycle could consist of N ALD cycles of MoS2 followed by 1 ALD cycle of AlSx to introduce the Al dopant. The composition of the film, here the concentration of the Al, can be varied simply by choosing an appropriate value for N.
The multi-step scheme is often used to extend the operating window of ALD processes or enable new capabilities. The co-dosing and supercycle approach are both frequently used to deposit multi-element materials such as doped and ternary materials. Both approaches have their own advantages and drawbacks. For example, the co-dosing approach could lead to a better mixing of the constituents than the supercycle approach. On the other hand, a possible drawback of the co-dosing approach is that the kinetics of the chemistry (depending on the partial pressures of the reactants and their reactivity) can play a role in determining the composition of the material deposited in the co-dosing step. Such effects do in principle not occur in the super-cycle approach. To which degree any of these effects play a role and to which extent they impact the properties of the prepared film varies from case-to-case.
Trending in ALD research: more complex processes and materials
In 2020, 17 new materials were added to the ALD/Database. These are materials that were not reported in papers in 2019 or earlier. The table below lists these materials with a direct link to the related paper.
| Al:MoS2 | 10.1021/acsanm.0c02167 | MnOxNy | 10.1021/acsaelm.0c00224 |
| AlOxFy | 10.1116/1.5135014 | MoOxNy | 10.1116/1.5130606 |
| CaMoO4 | 10.1116/6.0000327 | NaPxOyNz | 10.1021/acsami.0c03578 |
| CoPx | 10.1002/anie.202002280 | NaxCoO2 | 10.1116/6.0000166 |
| CsNbxOy | 10.1116/6.0000589 | NbSixOy | 10.1088/1361-6528/ab6fd6 |
| FeSixOy | 10.1116/6.0000212 | PbHfxOy | 10.1111/jace.17521 |
| LaNiO3 | 10.1038/s41467-020-16654-2 | PbHfxTiyOz | 10.1111/jace.17521 |
| LiF-CFx | 10.1016/j.jpowsour.2019.227373 | SnOxSy | 10.1016/j.ceramint.2019.10.254 |
| LiNixOy | 10.3390/en13092345 |
Of the 17 new ALD materials reported in 2020, no less than 14 of them consist of three or more reactants (shown in bold in the table) and only three processes use two reactants. As can be seen from the table, most of the materials fall in the category of ternary, multi-element, or doped materials. This clearly illustrates that ALD is no longer ruled by the preparation of “simple” binary compounds and has shifted towards the preparation of more multi-element materials. Typical ALD schemes to synthesize these multi-element compounds include co-dosing, multistep, and supercycle approaches as highlighted in the panel. In fact, this trend is already going on for quite some time as shown below. Although it varies from year-to-year, there seems to be an upwards trend in the fraction of ALD processes making use of three or more reactants. Furthermore, it is interesting to note that most the new entries in 2020 reporting on a new material with “only” two reactants involved the use of a new (precursor) molecule. Obviously it makes sense to keep the ALD process relatively uncomplicated when trying out a new precursor.
In 2020, a total of 123 new ALD processes were added to the ALD/Database. In these cases, either a new material was prepared or a new reactant (or a combination thereof) was used that was not reported in the ALD/Database so far. In this list, several trends can be identified such as the focus on 2D layered transition-metal dichalcogenides (with 22 entries) and complex oxides (with 19 entries). But also the more “classic” high-k dielectrics (13 entries) and pure metals (with 8 entries) synthesized using new reactants are well represented. In total 20 processes made use of a precursor molecule not reported before in the ALD/Database. These are summarized below.
| Al(iBu)3 | GaMe3(CH3OCH2CH2NHtBu) | InMe3(MeO(CH2)2NHtBu) | Pb(dbda) | WH2Cp2 |
| Ce(iPr2AMD)3 | [Ga(NMe2)3]2 | In(triaz)3 | Ru(CpEt)(dmopd) | Y(EtCp)2(iPr2AMD) |
| CsOtBu | Hf(BH4)4 | MoCl4O | Ru(eb)(chd) | Zn(DMP)2 |
| Ga(CpMe5) | Hf[η5:η1-Cp(CH2)(SiMe2)NMe](NMe2)2 | Na(thd) | N(Si2H5)3 | [EtZn(damp)]2 |
Conclusion
Reviewing the year 2020, we saw that the adoption of the ALD/Database in the community kept increasing. We still receive a lot of encouraging feedback the user. This is also reflected in the number of visitors, we went from ~9000 visits last year to ~14000 visits this year. On average also nearly 1 ALD process was entered in the ALD/Database per day. In 2021 we are planning to continue on the same path, we will keep on working on the website (experience) and we will continue reviewing new contributions.
If you want to help out, you can do so by submitting new and missing processes. We also really appreciate suggestions for potential improvements and reports of bugs or mistakes. You can so through the ALD/Database (see menu at the bottom right) or by contacting us directly.
To conclude, the ALD/Database team would like to wish you a happy and healthy new year and of course lots of fun with ALD in 2021.
Very nice summary! One minor edit/comment: [Ga(NMe2)3]2 is not a new precursor for the database. We published ALD of GaN using this 2020 (https://doi.org/10.1039/D0TC02085K) and while this is the first entry for GaN using this precursor in the database, the precursor can be found in several entries but then written as Ga2(NMe2)6.