Vertical and Lateral Manipulation of Single Cs atoms on the Semiconductor InAs(111)A

Ligthart et al. present a study on single Cs atom manipulation on semiconducting InSb(111)A surface including different manipulation modes. The paper is well written, clear for the reader and its strength lays in detailed description of lateral atom manipulation of adatoms on a semiconductor. I can recommend publication of this work after addressing the points listed below.

1) In the introduction authors state that the use of semiconductors as a substrate for creating artificial electronic states allows for higher energy resolution in STS. Do authors mean to say that use of a substrate directly influences STS resolution? Addressing the physical reason for sharper energy states obtained for atom assembled quantum dots on semiconducting substrates would be beneficial for clarifying the above statement.
2) Further in the introduction authors states that atom manipulation is more demanding on III-V semiconductor due to its polar and covalent character and I am missing a link. Can the authors elaborate on how exactly this character influences atom manipulation?
3) As a part of a motivation a part addressing the use of Cs atoms is missing. InAs(111)A substrate has native In adatoms which were previously used by e.g. Fölsch et al. [PRL 103, 096104 (2009) or Nature Nanotech 9, 505–508 (2014)]. What is the benefit of using Cs?
4) The vertical manipulation part of the paper is similar to the approach already described previously for In adatoms on the same substrate by Fölsch et al. [PRL 103, 096104 (2009)] but for with much less details. It would be nice to see a similar evaluation for Cs atom manipulation presented here especially that the lateral manipulation description is more detailed.
5) The authors state that the success rate for vertical manipulation “will vary somewhat” for different tip apexes based on data obtained for only single tip apex. What is the tip apex termination in this particular case? How does manipulation compare between bare PtIr, In terminated and Cs terminated tips?
6) The difference for data obtained for push and pull traces is not clear (Fig.3d). Are they obtained in different crystallographic directions or with different tip terminations? Is there any influence for manipulation based on local chemical environment (vacancies, atomic clusters)?
7) In Fig.4b and Fig.4c we can see beautiful structures made by atom manipulation but their electronic states signatures are missing. It is important to compare how similar or different they are versus previous reports on In/InAs(111)A [PRL 115, 076803 (2015)] and Cs/InSb(110) [Science 380, 1048-1052 (2023)] for validation of this platform.
8) In the last part the authors present stability of assembled structures (5 atoms) up to 44 K and individual Cs atoms up to 77 K. What is the reason for this difference in diffusion barrier? Does this effect scale with structure size?

1. Clearly written paper on a subject of particular topical interest.

2. Meets SciPost acceptance criterion: "Open a new pathway in an existing or a new research direction, with clear potential for multipronged follow-up work."

3. Provides new protocols for single atom manipulation on an important substrate/platform .

1. Some claims re. manipulation mechanisms -- and, as pointed out by Reviewer #1, re. the charge state of the Cs adatoms -- do not seem to be fully supported by experimental evidence. See comments in report below.

2. Not clear how the claimed indium termination of the tip is maintained throughout the experiments.

3. Although the experimental measurements are generally described in more than adequate detail, in some places greater detail on error bars/uncertainties could be provided.

This is an important and well-written paper on a topic of considerable interest to the nanoscience, surface science, and condensed matter physics communities: single atom manipulation on semiconductor surfaces. The Cs/III-V system is likely to receive renewed attention following the elegant and pioneering "quantum simulator" paper from the Radboud group (Khajetoorians et al.) published earlier this year (Science 380 1048 (2023)). Previous similarly elegant work by Folsch et al. [e.g. PRL 115, 076803 (2015)] on the assembly of quantum dot molecules on InAs(111)A has demonstrated that this particular surface is especially amenable to controlled single atom positioning with an STM tip.

As such, this paper is particularly timely and establishes protocols for manipulation of Cs atoms on InAs(111)A. I agree with the points raised by the first reviewer (i.e. Anonymous Report 1 on 2023-10-27) and would also ask for the following points to be addressed in a revised version of the manuscript:

1(a). The Introduction draws a distinction between vertical manipulation methods that transfer an atom from the tip to the substrate and vice versa, generalising that these mechanisms involve the electric field and inelastic tunnelling for surface-to-tip transfer, and vdW forces/short-range forces for tip-to-surface transfer. I think this is a rather strong statement if it is meant in a general sense. Are the authors suggesting that these mechanisms hold for the InAs(111)A substrate or do they mean that *all* vertical manipulation events, i.e. for any tip-sample combination, can be explained in this manner?

1(b). In a similar vein to 1(a) above, on p. 4 the authors argue that the jump of the adatom is "caused by inelastically tunnelling electrons which excite vibrations [20]." The reference is to a detailed analysis by F\"olsch's team that demonstrates that this is the case for In adatoms on InAs(111)A. It is not immediately obvious that the inelastic tunnelling mechanism will also dominate for the transfer of Cs adatoms. Do the authors have direct evidence for this inelastic tunnelling mechanism in their experiments on Cs/InAs(111)A?

1(c). The Introduction is a single, long paragraph. I suggest breaking this up into at least two shorter paragraphs to aid readability and clarity.

2. The final sentence of the Introduction section is intriguing: "The structures remained stable until a temperature of at least 44 K, and at 77 K no hopping of individual Cs atoms was observed." This suggests that the barrier for diffusion for Cs atoms that comprise an STM-assembled structure is lower than that for individual Cs atoms. Am I interpreting this correctly?

3. This point relates to the first reviewer's questions re. tip structure. In the second sentence of the Methods section, the authors state that an "In-coated PtIr tip was used for *all measurements*." (Emphasis mine.) How do the authors know that the tip is In-terminated, and, equally importantly, that it remains In-terminated throughout the manipulation experiments and spectroscopic probing? Is there not a finite (and perhaps rather large) probability of Cs being picked up by the tip?

4. p.2, third line from the bottom: "...reflection high energy electron diffraction (RHEED)..."

5. In Section 3.1, I suggest moving the sentence that starts "The In atoms are naturally present on the InAs(111)A surface..." so that it's the second sentence of that paragraph.

6. First sentence in main text on p.4: "...results in a sudden jump down in the current". This should be "sudden increase in the current".

7. p.5, second sentence. "Putting the adatom down is not induced by the electric field..." How do the authors rule out the influence of the electric field?

8. Fig.3 -- The mean hopping distances in Fig. 3(d) are quoted to three significant figures (i.e. down to the single pm level). What's the standard error on these mean values?

9. For Figure 4 (and in line with Reviewer 1's comment re. error bars), what was the total number of manipulation events used to generate the histogram? (My apologies if this information is in the paper and I missed it.) Is there a bias polarity dependence?

10. Caption to Fig. 5, final sentence. I assume this should be "900 mV and 30 pA".

See above for numbered list.

Ligthart et al. present a study of a novel platform for artificial lattices consisting of individual Cs atoms on InAs(111)A. They report on different successful manipulation modes with the STM tip and present examples of small artificially constructed objects, including densely packed structures of Cs atoms. The topic is timely, the paper is well written and this is indeed an interesting development, so I would clearly recommend publication of this work after the authors commented on my questions below. Most importantly, I would like to see a little more data backing the claim that this platform is interesting for studies of artificial lattices.

- The authors present a nice statistical evaluation of the manipulation's success rate in Fig. 4a (please add statistical error bars to Fig. 4a though). It would be interesting if they could provide similar statistics on the vertical manipulation mode.
- On page 4, they state that stable tips does not change significantly after vertical manipulation, however, the tip in Fig. 2b-d becomes unstable after dropping the Cs atom. Do they have other examples for vertical manipulation with no clear tip changes?
- In general, it would be interesting if the authors could comment on how they prepare the tip apex. Can you fix an unstable tip directly on the InAs surface? Are there band bending effects for different tip apexes?
- In Fig. 3, it is shown that both pulling and pushing mode can be achieved. I am puzzled though what the difference is: does it mainly depend on the precise tip apex? Or is there a dependence on (e.g.) directionality of the lateral manipulation? Please clarify.
- The authors write that semiconductors "allow for higher energy resolution in scanning tunnelling spectroscopy". I know what they want to express, but please rephrase this because it sounds as if the actual energy resolution improved, which is not true. Please also add one or two sentences about the physical reason for this improved "effective" resolution and the advantage of their platform over other experimental platforms for artificial lattices. This would highlight the work's quality and relevance throughout the paper.
- They write: "Given the low ionization energy of Cs, and its behaviour on InSb(110), the Cs atoms on InAs(111)A are charged". I don't quite understand if this is a speculation or a backed claim? Please clarify.
- Now that the authors showed that Cs can be moved, it would be important to provide more data showing that this platform actually qualifies for artificial lattice studies. For instance, is there any (sharp?) resonance peak around the structures presented in Fig. 4 verifying the confinement of charges?