Combination of Multiple Operando and In-Situ Characterization Techniques in a Single Cluster System for Atomic Layer Deposition: Unraveling the Early Stages of Growth of Ultrathin Al₂O₃ Films on Metallic Ti Substrates

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper presents a novel setup that consists of a home-made ALD reactor that can grow various materials and operate in different modes. The setup is part of a larger system that has different instruments to characterize the materials in real time, or using in-situ techniques (e.g., XPS), during the ALD cycles. This system enables the study and understanding of the complex ALD mechanism, particularly in the initial cycles. The paper uses a surface science approach to investigate the Al2O3 thermal-ALD process using trimethylaluminum (TMA) and water on different substrates (SiOx/Si and Ti/SiOx/Si). This approach can reveal the role of the substrate in the heterodeposition (below 10 nm) and can provide information to optimize the ALD process by adjusting its properties that affect the film/substrate interface.

First of all, I would like to congratulate the authors for their clear and systematic study of this complex ALD process and in general for the development of the complete home-made system described, involving both the reactor and the cluster tool. The authors have highlighted how combining multiple characterization techniques, using a surface science approach, can improve our understanding of the fundamentals of the ALD reaction mechanisms.

In my opinion, this paper can be accepted for publication in Inorganics after answering the following comments:

1) page 3: "Due to the characteristic high pressures of ALD processes, especially when a carrier gas is used, these operando XPS experiments are typically limited to synchrotron facilities."

Synchrotron sources exhibit big advantages considering the higher flux and resolution among other. However, in the last years there has been a huge development and spread of lab-based NAP-XPS systems. Do the authors know about the use of lab-based NAP-XPS systems for ALD-related research? Could they comment about it?

2) In the same page 2, they also write: "A more typical scenario is the in-situ approach, where the film is transferred under controlled conditions, i.e., high or ultra-high vacuum conditions, from the ALD reactor to the analysis chamber, thus preventing film/surface modification or the deposit of contaminants, e.g., adventitious carbon, after exposure to atmosphere."

The authors should discuss or mention here the problems that this direct connection between the ALD chamber and the XPS system may cause regarding the contamination observed in the XPS measurements. According to the authors, the transfer time from the ALD to the XPS system is only 15 minutes, but, considering the cross-contamination observed, would it not be better to use the available UHV suitcase to transfer the sample to an independent XPS instrument?

3) In page 5, the authors comment: "Although the GPC converge to the expected value above 60 cycles,"  And, one can see a clear deviation of the instantaneous growth rate of the Al 2p peak in Fig. 1c. How do you justify the convergence of the GPC despite this significant deviation? Also, the Al 2p data for the 150th cycle should be displayed in Fig. 1c as well.

4) The XPS survey spectra obtained before, during  and after the ALD process should be provided or at least commented. Based on your comments for the Ti/SiOx/Si case, the XPS chamber seems to be contaminated. Can the authors estimate the amount of TiC?

5) The authors should define in the text or in the Figure 2 caption which m/Z values correspond to water and CH4. Why did they use arbitrary units for the pressure in Figure 2?

6) On page 6, the authors claim that water could be trapped in cold spots in the reactor even though it is heated up to 120°C. This water could affect the subsequent experiments, as they suggest for the higher GPC observed for the 200°C experiment, which was performed right after the RT one. How did the authors “clean” the reactor to avoid this? Did they just leave the system at 120°C and wait? How did they ensure that there were no more “virtual leaks” when starting another experiment? Did they just dose N2 pulses? A proper cleaning before each experiment (for both RT and 200°C) should be performed, especially when dosing water.

7) Regarding the thermocouple of the ALD sample stage, the authors write: “The temperature is monitored through a K-type thermocouple in close proximity to the sample.” Have they performed a test to compare the reading with the actual temperature on the sample surface? Can they comment on that?

 

Other minor issues to check:

• Format reference 4 in the References section to match the style of the other references. 
• Reorder the references in page 2: (substrate, temperature, pressure, dose and purging time, etc.) [27,28].
• Page 3, "or synchrotron radiation [37,38]" maybe it would be convenient to include this one too: Semicond. Sci. Technol. 27 (2012) 074010.
• You should choose more distinct colors for the Ti 2+ and Ti 3+ contributions in Fig. 3a to improve the contrast and the visibility. 
• page 12, 200 ºC

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper describes an interesting vacuum cluster for studying ALD reaction mechanism. The system is thoroughly described and its operation is demonstrated by in studying the well known TMA-H2O ALD process for Al2O3. ALD increases constantly in importance especially in semiconductor industry but also in energy technologies. All methods that add our understanding on ALD chemistry are welcome and encouraged. There are however some issues related to the present design and work.

Sampling to the QMS involves a major design default. A membrane is used to limit pressure in the QMS chamber. It is not said at which temperature the membrane is (please add this) but if it as room temperature, it will condense low vapor pressure compounds. If it is at elevated temperature, it serves as substrate and dominates the signal measured by QMS: from the ALD chamber only a small fraction of gases, precursors and byproducts are pumped to the QMS chamber, but likely all byproducts forming on the membrane surfaces reach the QMS. The same applies to also all other surfaces in between ALD chamber and QMS. The authors should do simple calculations to understand  the ratio of byproducts formed in the ALD reactor vs. those formed on surfaces in between the ALD reactor and QMS.

One could have avoided the need of the membrane by choosing much smaller orifice than the 500 um aperture.Even this would have still left the contributions from other surfaces in between the ALD reactor and QMS.

Related to the above, was any Al-containing ion detected with QMS? This must be commented and discussed in the paper.

Fig. 1 has been analysed and discussed in detail with respect of the GPC. What remains to be commented is the very fast thickness increase in the very beginning, probably during the very first TMA pulse, when a thickness increment over 3 Å takes place. This is likely due to the fact that the process was run at room temperature which is exceptionally low temperature. Why it was used? 

MS data can not be reported as H2O and CH4. One must give the m/z for the followed species. One also must commented what other ions were measured and whether they were observed. There are lots of good examples in literature how to perform QMS study on ALD processes.

The observation of H2O during the TMA pulse is strange because the two should react aggressively and thus should not coexist. Anyhow, explaining this by a virtual leak from cold spots lead to a question why these were not eliminated.

Aluminum carbonate does NOT have a bond Al-C-O! Carbonate is an oxoanion that bonds through the oxygen atoms.

Homo- and heterodeposition are strange terms. Nucleation period and steady-state growth are widely used terms instead.

In Fig. 8, the connection between preparation chamber and ALD reactor is shadowed by the ellipsometer. Could this be clarified?

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have checked the authors' response to reiverew contents and found them mostly satisfactory. Some things were not properly understood however, as explained below.

With the concern of the membrane condensing low vapor pressure compounds I meant in the first place that this would prevent their detection with QMS. By stating that 'we have not noticed condensation problems' the authors seem to refer visible condensates and clogging, I presume. Or have they observed any metal ion containing species with their QMS?

This "the QMS is relatively far away from the sample, thus measuring not the
local atmosphere at the substrate surface, but a combination of the whole reactor" is of no problem at all, as long as the surfaces are at the same temperature as the substrate and, importantly, are fully saturated with the precursors so that the same ALD chemistry is happening all around.  What causes errors to the QMS results are (i) surfaces of different temperature, and (ii) depletion of precursor so that the chemistry is not saturative on all surfaces. And that is exactly what seems to be happening based on this text addiiton: "No signal from Al-containing species was detected, possibly indicating that all the TMA is consumed at the sample surface and reactor walls (see below). The increase in TMA dose does not lead to a higher GPC as measured by ellipsometry, pointing out the saturation of the process at the sample surface."
The process is saturative on the sample but not on all surfaces between the sample and QMS, and hence the results are not fully representative of the chemistry. I would conclude that the current setup is a good process and tool monitoring tool, but not a tool for detailed mechanistic studies. Exactly the same manner as in the cited NAP-XPS setup, for example.

This comment and the cited solution "...a more precise design would be that presented by M. Leskelä and coworkers in Langmuir 2000, 16, 8, 4034–4039, where the main characterization technique is residual gas analysis, and the gas entrance to the QMS is done via a capillary just above the sample." is a step to the right direction but even the short capillary has too much surface area.  In Helsinki they replaced the capillary to an orifice soon after.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf