From the journal Digital Discovery Peer review history

22-Feb-2024

Dear Ms Rahmanian:

Manuscript ID: DD-ART-12-2023-000257
TITLE: Autonomous millimeter scale high throughput battery research system

Thank you for your submission to Digital Discovery, published by the Royal Society of Chemistry. I sent your manuscript to reviewers and I have now received their reports which are copied below.

I have carefully evaluated your manuscript and the reviewers’ reports, and the reports indicate that major revisions are necessary.

Please submit a revised manuscript which addresses all of the reviewers’ comments. Further peer review of your revised manuscript may be needed. When you submit your revised manuscript please include a point by point response to the reviewers’ comments and highlight the changes you have made. Full details of the files you need to submit are listed at the end of this email.

Digital Discovery strongly encourages authors of research articles to include an ‘Author contributions’ section in their manuscript, for publication in the final article. This should appear immediately above the ‘Conflict of interest’ and ‘Acknowledgement’ sections. I strongly recommend you use CRediT (the Contributor Roles Taxonomy, https://credit.niso.org/) for standardised contribution descriptions. All authors should have agreed to their individual contributions ahead of submission and these should accurately reflect contributions to the work. Please refer to our general author guidelines https://www.rsc.org/journals-books-databases/author-and-reviewer-hub/authors-information/responsibilities/ for more information.

Please submit your revised manuscript as soon as possible using this link:

*** PLEASE NOTE: This is a two-step process. After clicking on the link, you will be directed to a webpage to confirm. ***

https://mc.manuscriptcentral.com/dd?link_removed

(This link goes straight to your account, without the need to log on to the system. For your account security you should not share this link with others.)

Alternatively, you can login to your account (https://mc.manuscriptcentral.com/dd) where you will need your case-sensitive USER ID and password.

You should submit your revised manuscript as soon as possible; please note you will receive a series of automatic reminders. If your revisions will take a significant length of time, please contact me. If I do not hear from you, I may withdraw your manuscript from consideration and you will have to resubmit. Any resubmission will receive a new submission date.

The Royal Society of Chemistry requires all submitting authors to provide their ORCID iD when they submit a revised manuscript. This is quick and easy to do as part of the revised manuscript submission process.   We will publish this information with the article, and you may choose to have your ORCID record updated automatically with details of the publication.

Please also encourage your co-authors to sign up for their own ORCID account and associate it with their account on our manuscript submission system. For further information see: https://www.rsc.org/journals-books-databases/journal-authors-reviewers/processes-policies/#attribution-id

I look forward to receiving your revised manuscript.

Yours sincerely,
Professor Jason Hein
Associate Editor, Digital Discovery

************

Reviewer 1

The authors present a tour de force in workflow automation for battery interface studies. MISCHBARES encompasses end-to-end workflow integration with a high degree of automation. There are elegant choices, for example using screen printing for parallel electrode synthesis, and advanced techniques, such as high resolution XPS with peak deconvolution to elucidate the chemistry of the cathode-electrolyte interface. This is an excellent fit for the journal and contribution to the community. I recommend its publication after some minor issues are addressed.

My most substantial recommendation concerns uncertainty in parameters from XPS fitting. Especially since the workflow is presented as an enabler of autonomous research, a brief discussion of uncertainty quantification in the XPS parameters, and/or in other aspects of the data analysis, would be helpful.

The analysis of the Fe XRF distribution is nice. Was the “background Fe” signal indicated in Fig. S2 subtracted from the Fe signal at the electrode spots? The text notes the use of this Fe signal to normalize the current in the CV, so I presume that’s why the current in Fig 3 has units of A/g. Please clarify in the figure caption that this is normalization via XRF and state whether this is grams of Fe or of LFP.

In Fig. S2, it would be helpful to have a semiquantitative color bar. I think I understand the Al – there sit he same amount of Al everywhere but its signal is attenuated in the LFP spots. The Fe signal also looks intuitive. Why does the spatial variation of Mn look so much like Fe? The explanation of machining contaminants from steel sounds plausible, but does this give a uniform concentration of these elements everywhere in the foil? Can the authors comment on whether a similar distribution of these elements was observed on a piece of foil prior to LFP deposition?

The SI notes a thickness of 0.008 mm for the electrodes. How was this measured?

The SI notes preparation of electrodes and drying in an open atmosphere. Then the XPS notes “to avoid contact with air, the sample environment of the NAP-XPS is directly connected to the glovebox”. The main paper states “Once the cleaned samples are dried, they are transferred for ex-situ XPS”. Perhaps in the main paper there should be a more explicit description of the sequence of steps with emphasis on transfers in and out of the glovebox. Similarly, the paper highlights the use of “near-ambient pressure Xray photoelectron spectroscopy (XPS)”, but I think the measurements were performed under UHV conditions. This isn’t a problem – it just needs to be clarified.

Reviewer 2

General comments

The authors implemented MISCHBARES, an autonomous millimetre-scale high-throughput battery research system operated by hierarchical autonomous laboratory automation and orchestration. The authors present cathode electrolyte interphase (CEI) formation in lithium-ion batteries at various potentials by correlating high-throughput electrochemistry and spectroscopy to demonstrate the usability of the presented work. The authors mention that MISCHBARES integrates automatic quality control for hardware and software to ensure high reliability through an on-the-fly fidelity assessment of each experiment. Finally, the authors note the integration with the Modular and Autonomous Data Analysis Platform (MADAP) software for Data analysis.

The presented work is exciting and describes the MISCHBARES platform as integrating different modules, pipelines, and software to facilitate autonomous millimetre-scale high-throughput battery research. The authors describe the various modules and show how to use the implemented tool in a specific case study. Despite the exciting results presented by the authors, some comments need to be addressed to improve the quality of the presented work:

- Please improve the clarity of the presented work, in particular for abstract and introduction section, facilitating a well understanding of the implemented method and helping to any reader to understand the implemented approach.

- Please add a methodology section to describe the implementation strategies, software description, libraries, components, deployment, etc.

- Please mention the license of the implemented metho

- Please clarify how the authors introduce AI control for the implemented method.

- Please improve the case of study to facilitate the clarity of the obtained results and how to use the implemented method.

- Please incorporate more details about the different data analyzis pipeline implemented in this work and does not explained in the previous work.

- Please discuss the limitations of the implemented library, the hardware and software requerimients

Specific comments

Abstract
- Please improve the abstract of the presented work because it is unclear what is presented in this work, if it is only MISCHBARES or also HELAO and MADAP, or if it is only the integration of these three previously presented works.
- I recommend improving the review of the proposed article because it is unclear what the proposed method is, how it was implemented, what the new methods are, and what was proposed by the authors in this work.

Introduction
- Please improve the redaction of the introduction; it is very complex to understand. For example, in the case of challenges description, it is unclear what the authors want to describe in the section "The challenge of multifidelity and scability". My apologies if I didn't understand the proposed section, but for me, it is unclear and cryptic at some points. The authors write an introduction describing the context, current solution, and state-of-the-art, explain the problems, and, finally, describe the proposed work, results, solved problems, and case of study with a demonstration of the proposed methodologies.

- In general, the authors claim different challenges in subsections, but understanding each of them needs to be clarified or more concise. Try to improve it, thinking about the type of reader associated with the journal.

Results and discussion

- I suggest a schematic representation of the different components included in the implemented framework and showing how each component interacts with the rest of the components.
- Please improve the redaction of the presented work. The work is fascinating and promissory, but a lousy redaction does not help to understand the work and the different possible uses.
- One of the most relevant points in the presented work is the MultiAnalyseHub module. I suggest the authors explain in more detail how all methodologies incorporated in this module are addressed. The authors could use an SI as support to include it. Also, it could be interesting to mention if the methods implemented in this module are common or if there is some previously reported work and state-of-the-art.
- The authors mention your "control protocol"; please give more details about it because it is unclear.
- Describe the content and functionalities of the different QE implemented in the proposed work.
- Please provide more justification for the incorporation of the Telegram chatbot. Is it only for notification methods, or is there an integration for decision-making?
- Please mention the differences between MultiAnalyseHub and MADAP.
- Describe the FAIR principles or add a reference about them.
- In the MultiAnalyseHub section, the authors describe this module. However, the authors incorporate some sentences like " MADAP capabilities offer a rapid assessment of experimental quality and can yield valuable scientific insights". In this work, you don't need to explain MADAP and only need to add the corresponding reference.
- Please discuss why the authors have selected a relational schema to implement the database and why they do not prefer alternatives like graph database or document database.
- Concerning the application of ML algorithms usage, at what moment do the authors recommend applying these methods? What do the authors want with ML? Patter recognition? Predictive model? Classification? The incorporation of ML methods is a bit cryptic and needs to be improved, particularly the explanation of the different tasks, what algorithms, etc.

Questions/Comments/Suggestions
- Automated quality control and data interpretation are the missing puzzle pieces towards prolonged walk-away-times in closed-loop experimentation; why?
- The implementation description of each software artefact is part of a methodology process and not a result of the work. I suggest changing the redaction to clarify it. In other words, the results usually express "what", and the methodology expresses "how".
- In Figure 1-B, why are the tables in the MER presented in different colours?
- Why do the authors use the ROI method for drop detection? Are there other alternatives? Please comment on it.
- The authors could incorporate a schematic representation of the implemented pipelines in the MultiAnalyseHub module, describing input/outputs, critical process, and optional process.
- What are the "AI control" incorporated in this work?

REVIEWER REPORT(S):
Referee: 1
Comments to the Author
The authors present a tour de force in workflow automation for battery interface studies. MISCHBARES encompasses end-to-end workflow integration with a high degree of automation. There are elegant choices, for example using screen printing for parallel electrode synthesis, and advanced techniques, such as high resolution XPS with peak deconvolution to elucidate the chemistry of the cathode-electrolyte interface. This is an excellent fit for the journal and contribution to the community. I recommend its publication after some minor issues are addressed.
We thank the reviewer for the constructive feedback and remarks. We appreciate the recognition of our work’s contribution and will carefully address the issues and consider your suggestions for revision.

- Reviewer: My most substantial recommendation concerns uncertainty in parameters from XPS fitting. Especially since the workflow is presented as an enabler of autonomous research, a brief discussion of uncertainty quantification in the XPS parameters, and/or in other aspects of the data analysis, would be helpful.
- Response: We appreciate the reviewer’s insightful recommendation regarding the uncertainty quantification, and we have taken careful steps to address this aspect in our work. Although the analysis of electrochemical tests is performed on the fly during experimentation, the evaluation of XPS data in our study was conducted post-measurement through a manual evaluation process by experts rather than on the fly. The approach was chosen to demonstrate the reliability of the outcomes of our case study within an autonomous research framework for electrochemical studies. To ensure the repeatability and reproducibility of our system, we chose XPS as our characterization technique. Here, we used the CasaXPS software for peak fitting, opting for a Shirley-type background and a peak line shape modelled as a 70% Gaussian and 30% Lorentzian convolution. This combination captures the spectral characteristics essential for accurate peak shape modelling. CasaXPS software provides two key metrics for quantitatively assessing the uncertainty and fit quality of our parameters estimates from XPS fitting: the residual standard deviation (STD), which evaluates the deviation between observed and fitted values, and the Goodness of Fit value, which assesses the model’s accuracy in capturing underlying data trends by comparing the sum of the squared residuals to the total variation in observed data. These two metrics, however, do not directly infer the physical model; they validate our parameter estimates and ensure the reliability of our analysis. We have included these metrics for every spectral region across all samples in our SI section to enhance transparency and confidence in our fitting process. Additionally, we have clarified the distinction between tests conducted in an autonomous setting and those not, and have updated the manuscript accordingly.
Regarding other aspects of data analysis: in this study, we integrated our data analysis software package into our orchestration framework. Among the various tests implemented, Cyclic Voltammetry, utilized in this study, addresses several aspects of data analysis. For instance, to find oxidation-reduction peaks, the ‘find_peaks’ module from the SciPy library is configured to identify peaks within the smoothed current signal and aims to balance sensitivity and specificity through the adjustment of parameters such as ‘width’, which filters out peak candidates that do not meet the defined span, and ‘prominence’, which establishes a minimum vertical threshold for peak distinctiveness. Although the implementation does not explicitly quantify uncertainties, it can detect notable signal features and filter out minor fluctuations that are potentially noisy. Another feature of this analysis is the calculation of the height of cathodic and anodic peak currents through a linear regression fit to approximate the background capacitive current. This approximation is achieved by selecting a window around each peak and evaluating the fit quality using the r-squared value. The peak height is then calculated as a vertical distance between the actual peak current and the intersection point of the fitted line with the voltage corresponding to the peak. Again, the method does not explicitly utilize uncertainties; the r-squared measure evaluates the quality of the fit. To clarify this data analysis aspect, we modify the SI section further.
Edits in main Manuscript: Two minor modifications to improve clarity.
Building upon our earlier studies on asynchronous web-based frameworks31, it orchestrates live visualization of electrochemical measurements, quality control and user feedback, data provenance and analysis, thus enhancing the experimental processes.

Once the cleaned samples are dried, they are transferred for ex-situ XPS analysis to characterize the synthesized CEI at different potentials and identify the formation stages of its components. The CasaXPS software61 is utilized to evaluate the outcomes.
Edits in SI section: Minor modification in figure caption and added a table with residual STD and Goodness-of-Fit for all the measured samples.


- Reviewer:The analysis of the Fe XRF distribution is nice. Was the “background Fe” signal indicated in Fig. S2 subtracted from the Fe signal at the electrode spots?
- Response: In response to the reviewer's inquiry concerning background subtraction of the Fe signal at electrode spots, we wish to clarify that after thorough evaluation, we decided not to perform such subtraction. This decision was based on our analysis, which involved defining a 10x10 grid of points at the center relative to the measurement points to assess the background signal's uniformity and intensity. Our findings showed a mean signal intensity of 16,362.09 counts per second (cps) with a standard deviation of 266.911 cps, corresponding to a variance of 71,241.7419 and a relative STD of 1.6%. Given the substantial and consistent background signal across the sample, accurately quantifying the specific reduction of the background signal attributable to the overlay of active material, proved challenging. Further analysis involved comparison with the Al signal, suggesting that the potential decrease in background signal due to the presence of active material is negligible. This conclusion was supported by the absence of Al content at the measurement points and the observation of Fe contamination on the foil, indicating that the overlay of active material does not significantly affect the overall Fe signal quantification. To ensure the integrity of our findings and avoid errors from arbitrary background signal subtraction, which we could not accurately quantify, we opted to leave the background signal unadjusted. We modified the figure caption in the SI section to improve clarity.
Edits in SI section: Added a sentence to the caption of figure.2.
Bright colors show high concentrations, while dark colors show lower concentrations. The coated points can be observed as circular spots in the elemental maps, where the presence of overlying active material reduces the background to negligible levels.


- Reviewer: The text notes the use of this Fe signal to normalize the current in the CV, so I presume that’s why the current in Fig 3 has units of A/g. Please clarify in the figure caption that this is normalization via XRF and state whether this is grams of Fe or of LFP.
- Response: We appreciate the reviewer's observation, which has prompted further clarification regarding the normalization of current in Fig.3. To clarify, the normalization process indeed involved the Fe signal from XRF analysis for quantifying the content of Fe, which serves as a proxy for the total mass of the active materials of LFP coated on the substrate. To determine the mass of the active material, we weighed the Al foil before and after coating with the material. This step enabled us to calculate the total mass of the LFP coated on the substrate. Given the distribution of the material across 121 screen-printed points, we derived a mean mass per point. However, it is important to note that the direct weight measurement does not provide the mass distribution per point. This limitation led us to use XRF measurement to obtain the Fe intensity at each point, which was then correlated to the mass of the LFP per point based on the following calculations:
- The total mass of all points combined was measured to be 3.11 mg, with the total active mass of LFP being 2.6435 mg for all 121 points, equating to an average of 0.02184 mg per point.
- The gravimetric current density was normalized by calculating the count per gram (cps/g), which is the sum of the intensity of all points (from XRF measurements) (cps) divided by the overall active material mass (g). The amount of LFP at each point was then determined by dividing the intensity of each point by the count per gram, yielding the mass in grams [cps / cps/g = g].

Edits in main Manuscript: The revised caption for Fig.3 now explicitly states that the normalization, reflected in the unit of A/g, is based on the quantification of LFP mass derived from XRF measurements.


- Reviewer: In Fig. S2, it would be helpful to have a semiquantitative color bar.
- Response: We are grateful for the reviewer's suggestion to include the semiquantitative color bar, which indeed enhances the figure's interpretability. To define the color bar ranges, we processed the raw XRF data using Python to map out the intensities at the corresponding energy levels for each element (excluding Mn), from which we determined the low and high-intensity values to establish the semiquantitative color bar range. The Mn signal estimation was derived from a detailed analysis of our Horiba XRF data report, which indicated that the Mn signal was approximately 2% of the Fe signal. This ratio was used to calibrate and adjust the color bar for Mn.
Edits in SI section: Following the reviewer’s recommendation, we have updated Fig.2 in the SI section to include these semiquantitative color bars.


- Reviewer: I think I understand the Al – there sit he same amount of Al everywhere but its signal is attenuated in the LFP spots.
- Response: Your understanding is correct. Al is uniformly present on the foil, consistent across the sample. Signal attenuation at coated spots is due to the overlay of active materials. The same effect is observed with Cr and Ni contaminants in the foil.


- Reviewer: The Fe signal also looks intuitive. Why does the spatial variation of Mn look so much like Fe?
- Response: We thank the reviewer for this remark concerning the similarity between Mn and Fe signals in our XRF results We have taken note of this observation and have identified several justifications, which we will discuss in more detail. Firstly, the presence of Mn as an impurity within the LFP matrix contributes to the observed similarity in spatial distribution patterns. As these two elements have similar ionic radii and chemical affinities, the uniform distribution highlights a homogenous incorporation of Mn impurity within the LFP structure. This uniformity indicates that regions with higher concentrations of LFP also exhibit increased signals for both Mn and Fe. However, it is important to note that there is a noticeable difference in signal intensity between Mn and Fe; the intensity of Mn signals ranges from 0.06 to 0.94 cps, which is 2% of Fe signal intensity (3 to 47 cps). This disparity indicates the lower concentration levels of Mn compared to Fe despite their similar distribution patterns. Additionally, while the overall trends for these two elements are comparable, there are differences in their spatial pattern upon closer examination. The excess Mn in certain areas of the foil, which does not correspond to an increase in Fe, may relate to variations in the manufacturing process (Below, we highlighted some slight differences in their patterns). We are open to further investigation in future studies and acknowledge the necessity to deepen our understanding of these phenomena by exploring further elemental analysis to find a more specific answer in the future.
Edits in SI section: This similarity is also illustrated in the caption of Figure S2 for further clarity.


- Reviewer: The explanation of machining contaminants from steel sounds plausible, but does this give a uniform concentration of these elements everywhere in the foil?
- Response: To address this query, we conducted a detailed analysis across the defined 10x10 grid of points of the foil (as shown in the grid on page 3). The intensity measurements for Ni and Cr contaminants were plotted in a histogram, which shows a slight degree of variability in the elemental distribution but largely follows a Gaussian distribution, which indicates a general uniformity within the measured area. Our observation did not identify any significant anomalies or regions of unexpectedly high intensity for Cr and Ni, supporting the homogenous distribution of these elements across the tested areas. While the results suggest uniformity, we can only definitively conclude that this uniformity extends to the entire roll of aluminum foil with further investigation. We remain open to expanding our analysis in future studies to further explore this uniformity across larger sections of the foil rolls. Below, the distribution plot for Cr and Ni in the background (non-coated areas) supports our findings within the examined sample.


- Reviewer: Can the authors comment on whether a similar distribution of these elements was observed on a piece of foil prior to LFP deposition?
- Response: We thank the reviewer for raising this point. It is important to clarify that we have yet to investigate the elemental distribution of the foil before the deposition process. As our study's objective was to replicate known research on the oxidation-reduction of lithium with an autonomous system, we did not directly assess the contribution of these elements to the experimental outcomes in our initial analysis regarding the foil's pre-deposition state. However, the observations from subsequent experiments, including cyclic voltammetry (CV) and XPS, indicate the absence of observable effects or contributions from these elements to our results. This observation leads us to believe that the distribution of such contaminants, if present before deposition, did not impact our study's findings. However, we acknowledge the value of understanding the material properties in their entirety and agree that examining the foil's elemental distribution prior to deposition could provide deeper insights into the overall performance. Thus, the significance of your query is well recognized, and we are keen and open to incorporating such measurement to explore this aspect in future research endeavors.


- Reviewer: The SI notes a thickness of 0.008 mm for the electrodes. How was this measured?
- Response: To determine the thickness of the electrodes, we used a digital thickness gauge, a decision driven by its simplicity for such measurements. The thickness was measured at ten distinct points across the electrode surface to ensure representativeness. The average of these measurements was then calculated to determine the thickness.
Edits in SI section: We have added this description for further clarity in our SI.


- Reviewer: The SI notes preparation of electrodes and drying in an open atmosphere.
- Response: It is correct. The electrode preparation was carried out in an open atmosphere due to the air stability and processability of the active materials. This compatibility allowed us to use a water-based slurry. After the coating process, the electrodes were dried under vacuum conditions to ensure the removal of any residual moisture. The electrodes were transferred into a glovebox following this drying phase for further processing and analysis.


- Reviewer: Then the XPS notes “to avoid contact with air, the sample environment of the NAP-XPS is directly connected to the glovebox”.
- Response: Yes, the statement is correct. Our XPS system is integrated into the glovebox, which allows us to directly transfer our samples after electrochemical measurements, performed using our autonomous system, into the XPS chamber without exposing them to air. This direct connection is essential for handling air-sensitive materials, so that their integrity is maintained during transfer. The sample environment of our XPS is an extension of our glovebox. It resembles a load-lock mechanism, which can be isolated and purged with an inert atmosphere and evacuated to achieve the desired low-pressure conditions necessary for XPS analysis. We have included a pictorial representation below to provide a more precise explanation of our laboratory setup. Further details on the setup orientation are elaborated in the “From materials discovery to system optimization by integrating combinatorial electrochemistry and data science” paper.

- Reviewer: The main paper states “Once the cleaned samples are dried, they are transferred for ex-situ XPS”. Perhaps in the main paper there should be a more explicit description of the sequence of steps with emphasis on transfers in and out of the glovebox.
- Response: We thank the reviewer for this suggestion. To clarify, as indicated in our response to a previous query, once the electrode preparation is complete, the samples are transferred to a glovebox and remain within a controlled environment for all subsequent processes.
Edits in main Manuscript: To address the reviewer’s recommendation and provide a more explicit explanation of the sequence of steps, we have made the following revision to the main Manuscript:
Our novel approach to cathode electrode preparation involved screen-printing onto the substrate in an open atmosphere due to the stability of the active material and slurry components. This method minimizes material usage and precisely defines the active material area. After preparation, the electrodes are transferred to the glovebox for further processing and analysis. For each experiment, the SDC head is positioned at the predefined measurement spot, where it dispenses a droplet of electrolyte, and ensures electrical connection before performing CV. Upon completing a series of experiments, it is necessary to remove the excess LiPF6 salt residue, a byproduct of electrolyte evaporation, from the electrodes. This is achieved by depositing a droplet of PC on each measurement spot, allowing it to soak for three minutes to dissolve the salt, and then aspirating it using the SDC head. This cleaning process is repeated three times. Once the cleaned samples are dried, they are transferred within the glovebox directly into the XPS’s sample environment for ex-situ analysis, preventing air exposure. The analysis aims to characterize the synthesized CEI at different potentials and identify the formation stages of its components. The CasaXPS software61 is utilized to evaluate the outcomes.
SI section: Additionally, an explanation of this process is also provided in the SI section.


- Reviewer: Similarly, the paper highlights the use of “near-ambient pressure Xray photoelectron spectroscopy (XPS)”, but I think the measurements were performed under UHV conditions. This isn’t a problem – it just needs to be clarified.
- Response: We acknowledge the discrepancy and appreciate the opportunity to clarify. Indeed, the XPS measurements were performed under ultra-high vacuum conditions and not near ambient pressure, as the term "NAP-XPS" might suggest. This terminology was used to highlight the capability of our instrument. However, for this specific set of measurements, we opted for UHV conditions to mitigate signal intensity loss often associated with near-ambient pressure. We appreciate your pointing out the need for more transparent communication.
Edits in main Manuscript: To address your comment, we have revised the manuscript text slightly.
Edits in SI section: Additionally, a slight modification is added to the SI section.



Referee: 2
Comments to the Author
General comments

The authors implemented MISCHBARES, an autonomous millimetre-scale high-throughput battery research system operated by hierarchical autonomous laboratory automation and orchestration. The authors present cathode electrolyte interphase (CEI) formation in lithium-ion batteries at various potentials by correlating high-throughput electrochemistry and spectroscopy to demonstrate the usability of the presented work. The authors mention that MISCHBARES integrates automatic quality control for hardware and software to ensure high reliability through an on-the-fly fidelity assessment of each experiment. Finally, the authors note the integration with the Modular and Autonomous Data Analysis Platform (MADAP) software for Data analysis.
The presented work is exciting and describes the MISCHBARES platform as integrating different modules, pipelines, and software to facilitate autonomous millimeter-scale high-throughput battery research. The authors describe the various modules and show how to use the implemented tool in a specific case study. Despite the exciting results presented by the authors, some comments need to be addressed to improve the quality of the presented work:
Response: We thank the reviewer for these encouraging comments. We acknowledge the points raised and commit to clarifying and enhancing the presentation of our case study and addressing all concerns to improve the Manuscript's quality.

- Reviewer: 1. General comments: Please improve the clarity of the presented work, in particular for abstract and introduction section, facilitating a well understanding of the implemented method and helping to any reader to understand the implemented approach.
- Response: We thank the reviewer for the broad guidance and invaluable input. We have added detailed explanations in the subsequent sections to clarify the intentions behind the implementation of Auto-MISCHABRES more comprehensively. We are confident that the revised version of the Manuscript, enriched by the reviewer's insight, is now more scientifically sound and precise. We rest assured that the reviewer will find an answer to all the posed queries addressed in this updated version.

- Reviewer: 2. General comments: Please add a methodology section to describe the implementation strategies, software description, libraries, components, deployment, etc.
- Response: We are grateful for the reviewer’s feedback. In response, we have revised the Manuscript to include a comprehensive methodology section, now clearly distinguished from the Results section. This addition details the unique and valuable strategies implemented in our work and elucidates the design philosophy of Auto-MISCHBARES, its Hub components, the software developments, and the adapted and integrated libraries and dependencies. For clarity and transparency, the code of this framework was and is open-source. A detailed technical description and operational guidelines have been appended to the SI section (details can be referred to the response of Q13). We additionally invite the scientific community to review the work and its methodologies, which are also explained and available in a dedicated video on YouTube under https://www.youtube.com/watch?v=dMZlWIy7i5s&ab_channel=FuzhanR

- Reviewer: 3. General comments: Please mention the license of the implemented methods
- Response: We acknowledge the significance of disclosing the licenses of the software used in our work and appreciate the reviewer’s emphasis on this aspect. Our implementation predominantly utilizes open-source libraries and packages, such as HELAO (https://github.com/helgestein/helao-pub?tab=MIT-1-ov-file#readme), MADAP (https://github.com/fuzhanrahmanian/MADAP?tab=MIT-1-ov-file#readme), FastAPI (https://github.com/tiangolo/fastapi/tree/master?tab=MIT-1-ov-file#readme), Flask (https://github.com/pallets/flask?tab=BSD-3-Clause-1-ov-file#readme), pgAdmin (https://www.pgadmin.org/licence/)
The connectors to manufacturer instrumentation, essential for our framework, are provided by the manufacturers and were purchased together with the instruments. While the APIs, SDKs, and DLLs are integrated into our proposed Python framework, access to these is restricted to those with licenses from the manufacturers, and the code is fully usable for license holders, provided they configure the paths to their APIs, SDKs, and DLLs accordingly. The equipment which has commercial license are for instance Syringe Pump (https://www.hamiltoncompany.com/laboratory-products/microlab-600/syringe-pump), Autolab PGSTAT302N (https://www.metrohm.com/de_de/products/a/ut30/aut302n_s.html), stepper motor (https://www.owis.eu/en/) and casaXPS (http://www.casaxps.com/).

Edits in main Manuscript: To address this important aspect comprehensively, we have added a section titled "Technological Aspect" in the SI section, which describes all modules, libraries, and devices utilized in our study. Our developed framework is also made available under the MIT license (https://github.com/fuzhanrahmanian/MISCHBARES?tab=MIT-1-ov-file#readme) to emphasize our commitment to open-source principles.


- Reviewer: 4. General comments: Please clarify how the authors introduce AI control for the implemented method.
- Response: We appreciate the reviewer's attention to terminology and the opportunity to clarify and improve our communications. In our initial abstract, we referred to the term "AI control" based on our previously published work, HELAO (https://doi.org/10.1002/admi.202101987), which uses AI-informed decision-making to predict subsequent experiments (https://github.com/helgestein/helao-pub/blob/master/testing/al_parallel.py). As we expanded and advanced this framework further in the current paper, we initially used this term to demonstrate its broader application. To enhance clarity and directly address the reviewer’s concern, we acknowledge that AI-guided experimentation is not the focus of this study. Thus, we have revised the abstract to remove references that may conflate this work with our prior publications and emphasize instead the novel contributions belonging solely to the proposed study. This revision highlights the advancements in reproducibility and reliability achievable through autonomous workflows. Here, we instead introduce the term "AI enablers" to refer to the Utilization of AI technologies, including data processing and analytics, statistics, computer vision, and machine learning, in the Development and application of tools for our autonomous workflow, Auto-MISCHABRES. This workflow integrates various AI-driven tools for tasks such as drop detection, peak detection, data extraction, and more, showcasing our methodological innovations in enhancing workflow efficiency and precision.
Edits in main Manuscript: Changing the terminology to highlight the contributions of this study solely and distinguish it from earlier publications.


- Reviewer: 5. General comments: Please improve the case of study to facilitate the clarity of the obtained results and how to use the implemented method.
- Response: Recognizing the importance of clear guidelines for the successful integration and dissemination of newly created software within the scientific community, we thank the reviewer for the opportunity to provide a higher degree of clarification about how to use the software. To this end, we have revised and updated the description of the experimental procedure section in the main Manuscript to enhance the credibility and comprehensibility of the proposed case study and its results (refer to our detailed response in Q13 for more information). Additionally, we have expanded the SI and included an extra paragraph with detailed operational guidelines that describe an in-depth technical explanation of the proposed software’s functionalities and how it performs its tasks.
Highlighted modifications in SI section: Please refer to the Operational guidlines in the SI.


- Reviewer: 6. General comments: Please incorporate more details about the different data analysis pipeline implemented in this work and does not explained in the previous work.
- Response: We thank the reviewer's insightful comment highlighting the need for a clearer distinction between our previous work and the additional expansions and new implementation introduced in this study. After carefully addressing the reviewer's specific question (see below), we wish to emphasize the additions and enhancements to our revised manuscript's methodology section. Notably, in the MultiAnalyticHub subsection, we detail all the analysis and statistical tools utilized in our current implementation (refer to the response to Q13 for more information). This section now explicitly differentiates between the methodologies previously described and those newly implemented, which are elucidated in detail. Additionally, we have included an additional figure (a sequential diagram) to better illustrate the pipeline and its interactions with different components. This complements the original Figure 1 from our initial manuscript and offers a clearer understanding of the enhancements made (please see the response to Q12 for more details). In the SI section, following the reviewer's suggestions in the specific questions part, we provide an in-depth description of our voltammetry class and its analysis tools, outlining the unique insights that can be drawn from this analysis package. This discussion strictly focuses on the analysis tools implemented for this work and deliberately omits descriptions of previous methodologies (we invite the reviewer to review our responses to Q14 for detailed modifications to this section). We are confident that these revisions distinguish between our study's various pipelines and methods and clarify the advancements made.


- Reviewer: 7. General comments: Please discuss the limitations of the implemented library, the hardware and software requirements
- Response: We value the change to enhance the transparency of our work by discussing not only the advantages but also the limitations of our framework. The primary constraint of our current setup is its hardware requirements for replicating the exact laboratory environment. Although Auto-MISCHBARES, which builds upon HELAO, is designed to be agnostic and adaptable for web communication across any device and enable configuration for diverse laboratory settings, however, this specific setup, including the camera and related equipment, is designed and engineered for use in our laboratory. Nevertheless, we provide templates within our initial framework (https://github.com/helgestein/helao-pub/blob/master/server/template_server.py) that allow users to integrate and incorporate their own devices and leverage our proposed framework for their research. A limitation regarding software is the licensing restriction of specific devices, which prevents us from releasing them as open-source, and these must be acquired and purchased directly from the manufacturers.
Edits in main Manuscript: For enhanced clarity and improved transparency, we have included a discussion on these limitations in the last section of our Manuscript.
" It should be noted that the proposed electrochemical experimentation setup is bound by the availability of the measurement devices and their licensed software. However, through the modularity and agnosticism of the platform, \gls{mischbares} can be expanded with minimal effort following the provided templates to include any laboratory device needed for any specific experimental scenario."


- Reviewer: 8. Specific comments, Abstract: Please improve the abstract of the presented work because it is unclear what is presented in this work, if it is only MISCHBARES or also HELAO and MADAP, or if it is only the integration of these three previously presented works.
- Response: Thank you for the valuable feedback, which has allowed us to clarify the contributions and structure of our work more effectively. Initially, we would like to elucidate our research with greater clarity. The proposed workflow integrates different tools, including hardware elements (laboratory instruments), an orchestration framework, automated quality control assessments during experimentation, on-the-fly data analysis, and database and data management tools, applied within the practical application of electrochemistry to demonstrate reliable experimentation. To design and execute multiple batches of experiments, the researcher only needs to access the implemented web-based user interface and select one or multiple electrochemical tests from the provided list, and the asynchronous orchestration will manage their execution (SI - Figure 1). A live demonstration illustrating the cohesive and interconnected operation of these components is available from the 10’29’’ until the end of the following video: https://www.youtube.com/watch?v=dMZlWIy7i5s&t=655s&ab_channel=FuzhanR
Auto-MISCHABRES = {hardware elements, orchestration framework (expanded HELAO), automated quality control, data analysis (expanded MADAP), database and management system, active learning (part of initial HELAO), web-UI}
HELAO (hierarchical autonomous laboratory automation and orchestration) and MADAP (Modular and Autonomous Data Analysis Platform) were two elements of this framework (as they were initially introduced and published in separate works), although their functionalities have been extended for the current study, which we will elaborate more on that.
Our initial HELAO platform, introduced previously at https://doi.org/10.1002/admi.202101987, established an asynchronous web-server communication for autonomous experimentation, including groundwork for active learning (AL) applications for experimental optimization, as demonstrated in that publication. Although the AL part of HELAO was not the main focus of this study, it offers Auto-MISCHBARES’s broader capability for optimization endeavors. The objective of this work is to demonstrate how autonomous electrochemical processes could achieve reliability and reproducibility. In other words, when processed through these workflows, experimental data can produce invaluable outcomes and generate insights. Thus, Auto-MISCHBARES represents a more advanced frame around HELAO. For instance, this framework is now equipped with unit testing, logging, complete documentation, and scalable templates, supporting various laboratory instruments, and ensuring the framework's agnosticism. Additionally, implementing a user-friendly web interface makes the framework more accessible and reliable for conducting electrochemical research.
The MADAP package, initially designed and implemented for EIS and Arrhenius analysis, as published at https://doi.org/10.1038/s41597-023-01936-3 and documented at https://fuzhanrahmanian.github.io/MADAP/index.html has now been expanded to include voltammetry tests, such as Chronopotentiometry (CP), Cyclic Voltammetry (CV), and Chronoamperometry (CA). In this paper, we selected CV analysis for our defined case study and intended to demonstrate the package's expansive applicability and robust performance across a diverse electrochemical spectrum.
To provide a more transparent abstract and enhance its clarity, we excluded the explicit references to HELAO and MADAP, as these are just the corollary to Auto-MISCHBARES and concentrated directly on the platform contributions. The method section will explain what we have added to HELAO and MADAP and how Auto-MISCHBARES integrates and advances upon these components. For this part, we will take your further recommendations in the following questions into account and directly address your concerns.


- Reviewer: 9. Specific comments, Abstract: I recommend improving the review of the proposed article because it is unclear what the proposed method is, how it was implemented, what the new methods are, and what was proposed by the authors in this work.
- Response: We thank the reviewer for the remark, which has guided us to clarify our contributions and methodology further. Our revised abstract (in the previous question) incorporates the comments in the present question. Nonetheless, we would like to provide additional details below to address your concerns comprehensively.
Proposed method clarification: Our study introduces 'Auto-MISCHBARES', an integrated platform designed to advance electrochemical research. This integration is achieved through various tools and enablers, such as web-server communication, control assessment, data analysis, and database management, which together enable the orchestration of autonomous experimental processes via a web-based user interface.
• Extract from the abstract:
… Herein, we present Auto-MISCHBARES operated with an asynchronous web-based orchestration framework that integrates modular research instrumentation designed for autonomous electrochemical experimentation. The platform allows researchers to define a range of experiments with granular parameter control, start the process, and receive a live visualization of measurements through a web-based user interface.
Implementation overview: To clarify, we would like to highlight the novel implementations utilized in this study.
- Automated Quality Control:
o Implementation of drop detection using an installed camera to ensure material sufficiency before each experiment.
o Mandatory wiping of the SDC head to prevent salt formation and crystallization between experiments.
o Contact Detection for automated lowering of the SDC head onto the substrate.
o Progress and error reporting to the user over a Telegram Chatbot for real-time updates.

- Data Analysis:
o Expansion of previously published “MADAP” package with automated analysis of voltammetry tests (novel addition)

- Database:
o Development of a Relational Database for secure experimental records storage and guarantee stable data lineage from devices and measurements in line with FAIR guidelines.

- Web Client-Server Communication
o Expansion of previously published "HELAO" orchestration framework, with logging, unit testing, modularization, and documentation.
o Expansion of the asynchronous framework to include live visualization, real-time data analysis, on-the-fly data storage and quality assessment.
o Development of a web-based user interface for the comprehensive configuration of the experiment and the granular control of the experimental parameters. The user interface was built using JavaScript, CSS, and HTML.
The Method section of the revised Manuscript provides a detailed explanation of these additions and innovations, with highlights mentioned in the abstract for brevity and clarity.
• Extract from the abstract:
… The platform allows researchers to define a range of experiments with granular parameter control, start the process, and receive a live visualization of measurements through a web-based user interface.
… Auto-MISCHBARES integrates automatic quality control for both hardware and software using AI enablers to ensure high reliability through an on-the-fly fidelity assessment of each experiment. In the presented case study, voltammetry measurements are handled through a modular platform capable of performing fully automated analysis, while data lineage is provided through relational data storage in adherence with \gls{fair} guidelines, all in real-time.

Novelty of methods:
To the best of our knowledge, designing autonomous workflows for non-aqueous battery systems is inherently complex due to issues such as material leakage through miniaturized systems, evaporation problems in glove boxes, loss of electrical connections, salt formation, and changes in the active area. As a result, all autonomous platforms have a low degree of reliability, and none of them have built-in quality control mechanisms. Additionally, for non-aqueous systems, the experimental setup must be inside a glove box and not in contact with air, an environment that makes even manual work by humans challenging. Thus, our Implementation of quality control is the key to ensuring reliability and near-full autonomy. These challenges are well-known within the community, as mentioned by https://pubs.acs.org/doi/full/10.1021/acs.accounts.2c00220?casa_token=A2QAfgyG3iUAAAAA%3AXBXBZMZdl8pI0z8_xSbxdrcI14mqfht6rzTJd8axgMaobqy1OB08akw7MWVrhh7UVMWJCyvJ7BlaMdA in 2022. Indeed, other studies have attempted to address quality control problems, but Auto-MISCHBARES is pioneering this integration in an autonomous experimental loop. Our on-the-fly automated data analysis and management, although it has been discussed in previous acceleration efforts, is designed to be coupled with ex-situ XPS analysis to provide reliable outcomes not just at the electrochemical level but across characterization technique, which makes our work unique in the battery community.
• Extract from Abstract:
… We believe quality control, complex data analysis, and management to be the missing puzzle pieces toward more complex workflow automation.
… enabling reproducible and versatile workflows for the discovery of new materials, especially for batteries.
Specific contributions:
A detailed description of the various implementations undergone for the study can be found in the revised Method section of the Manuscript at each corresponding hub. To provide a comprehensive response, we will outline the implementations inherent to the codebase of Auto-MSICHABRES below:
- DMS:
o Utilization Utilization of pgAdmin for relational database design.
o Implementation of tables with PostgreSQL and automatic table creation derived from DB schema.
o Design of database connection
o Implementation of asynchronous data storage and retrieval before, during, and after experiments using the psycopg2 library.

- Web UI
o Development of a Flask-based web framework for interaction between the Auto-MISCHBARES orchestration and the user interface.
o Development of a user interface in JavaScript, CSS, and HTML to allow user access and granular experiment definition. Experiment metadata can be defined, saved, and imported. Parameters such as experiment type and order, measurement values, substrate size and grid generation, contact point location, and material identification can be configured. The web UI can be used to launch and monitor the orchestration.
o Implementation of a customizable live visualization mechanism using Bokeh and WebSocket. The user can select variables of interest for the x and y axes to display during measurement.

- Quality Control
o Development of a Python class to integrate with asynchronous orchestration for drop detection using a USB-enabled camera. The algorithm utilizes the OpenCV library and includes a live feedback mechanism and user alerts in case of material depletion.
o Implementation of an asynchronous movement sequence to ensure optimal contact between the SDC head and substrate, prevent excessive pressure on the active material, and minimize material disruption. Walk-away user feedback was also integrated at this stage.
o Development of a mandatory movement sequence to reach a wiping pad at the beginning of every experiment was orchestrated.

- Data Analysis
o Expansion of MADAP to include analysis of cyclic voltammetry data by outputting and plotting the following information using statistics and ML, such as cycle identification, forward-, half-, and reverse-peak potentials and currents, peak-to-peak separation, overpotential, diffusion coefficient, Tafel region parameters.

We trust this response clarifies our methods and contributions and addresses the concerns raised.


- Reviewer: 10. Specific comments, Introduction: Please improve the redaction of the introduction; it is very complex to understand. For example, in the case of challenges description, it is unclear what the authors want to describe in the section "The challenge of multifidelity and scalability". My apologies if I didn't understand the proposed section, but for me, it is unclear and cryptic at some points. The authors write an introduction describing the context, current solution, and state-of-the-art, explain the problems, and, finally, describe the proposed work, results, solved problems, and case of study with a demonstration of the proposed methodologies.
- Response: We greatly appreciate the reviewer's feedback and the opportunity to clarify the message and focus of the mentioned paragraph. In response to the comments, we have made the following revisions for improved readability of the introduction and its coherence. First, we have revised the title of the section to specifically address the "Challenges of Multifidelity", recognizing that scalability, while inherently related to multifidelity, may have contributed to the confusion. By narrowing our focus, we aim to eliminate ambiguity. Scalability, although a valid concern with the increasing number of laboratory interfaces, was considered not entirely relevant for the detailed discussion in this paragraph. The concept of multifidelity here is intended to highlight both the quantity and diversity of equipment necessary to ensure the reliability and consistency of experimental results.
Our revised paragraph now concentrates on the challenges associated with non-aqueous battery systems, which historically are more prone to error and present difficulties in automation. For enhanced clarity, we have removed numerical details from one example by Dave et al.; although this study is relevant to battery studies, the inclusion of aqueous systems in their research may have caused misunderstanding. Similarly, challenges inherited from cost, although relevant and pertinent to the design of MAPs, have been removed to enhance the narrative and intuitiveness. The revised section thus describes the multifidelity challenges that arise when striving for reliable and scientifically solid discoveries. These challenges require the need for a multitude of measurements, testing, and characterization techniques to be applied to one sample. Each step of this delicate process exposes the samples to different environments, and scientific statements about the consistency of the results can just be reached when all steps achieve (multi-)fidelity to each other.
Edits in main Manuscript: Please refer to the "The challenge of multifidelity" section in the revised Manuscript.
Additionally, we have improved the clarity and transparency across other subsections of the introduction, as suggested by the reviewer. Our aim to clearly delineate the context of each section, identify existing gaps and challenges, and describe how our proposed work addresses each issue. We also outline our methods for resolving these problems and introduce our case study as a proof of concept.
Edits in main Manuscript: Some slight modifications for the second challenge regarding databases and data management.
Edits in main Manuscript: The following paragraph describes how AI tools can automate human-intensive tasks necessary to get reliable and reproducible experimental results. For clarity, the title has been changed, and some slight modifications have been made in the section.
Edits in main Manuscript: In the final paragraph, we have made additional modifications to clearly illustrate how our proposed work addresses the identified challenge. We additionally improve the presentation of our case study to clearly demonstrate the applicability of our work as a proof of concept.


- Reviewer: 11. Specific comments, Introduction: In general, the authors claim different challenges in subsections, but understanding each of them needs to be clarified or more concise. Try to improve it, thinking about the type of reader associated with the journal.
- Response: We appreciate the reviewer for emphasizing the importance of clarity in the introduction's subsections. We are committed to conveying clear and concise research. In response, we have revised the subsections of the introduction to be more aligned with this objective.
To improve readability, we have restructured the introduction (modifications are shown in the above question), which outlines each of the three challenges we address, beginning with titles that encapsulate the core issues; first is the challenge of multifidelity, which arises from conducting a variety of tests on single samples. Addressing this strengthens the outcome and reliability of experimental results. Second is the challenge of robust data management and database design. This focuses on implementing data pipelines following FAIR guidelines to ensure experimental data is trackable and results are accessible at any stage. Finally, the challenge of integrating AI enablers for reliable experimentation aims at automating simple and repetitive tasks to safeguard critical tasks such as salt removal and material sufficiency. Although these tasks may appear straightforward, their accurate execution is essential for preventing discrepancies, such as varying measurement areas or the oversight of failed experiments in a fully autonomous mode of operations. Introducing AI enablers and quality control assists in automating these tasks that are often overlooked in MAP design, all in an environment where human actions are necessary but might be error-prone and inconsistent. Each subsection in the introduction now also delineates how our study sets out to solve these challenges, and we demonstrate its efficacy through the case study of CEI formation at various potentials. The XPS characterization brings forward additional validation. Through this structured approach, we aim to ensure that readers with backgrounds in chemistry, information technology, and interdisciplinary studies can assess the validity of our efforts and, by extension, the quality of Auto-MISCHABRES.


- Reviewer: 12. Specific comments - Results and discussion: I suggest a schematic representation of the different components included in the implemented framework and showing how each component interacts with the rest of the components.
- Response: We are grateful for the reviewer’s valuable suggestion to schematize the interactions of the different components of our Auto-MISCHBARES framework. While Figure 1 in our manuscript showcases the sequence of experimental steps, we recognize and agree with the importance and need for a more detailed schematic representation. Herein, we have included an additional figure in the manuscript (a sequence diagram) that illustrates an experimental run from inception to completion. This diagram highlights the roles of all involved components and Hubs and how they are orchestrated throughout the process.
Edits in main Manuscript: We have incorporated a sequence diagram into our revised manuscript, complete with a detailed caption. This diagram elucidates how each component of our Auto-MSICHABRES pipeline interacts within a single experimental sequence. We believe this enhancement will significantly increase the transparency and understanding of our implemented framework.

- Reviewer: 13. Specific comments - Results and discussion: Please improve the redaction of the presented work. The work is fascinating and promissory, but a lousy redaction does not help to understand the work and the different possible uses.
- Response: We thank the reviewer for the opportunity to improve the presentation of our proposed work. We are dedicated to delivering a clear and structured exposition of our efforts. Following the input of the question from the Questions/Comments/Suggestions part (No.24), we have decided to restructure the Manuscript. We now include a methodology section that outlines all the efforts undertaken to create and develop Auto-MISCHBARES, distinguishing them clearly from the results of our study.
Edits in main Manuscript: In the methodology section, we now clearly list the major components introduced in the construction and development of Auto-MISCHBARES, namely the four major Hubs: Device, Server, Data, and MultiAnalytic. This introductory paragraph briefly defines each Hub to provide an overview before elaborating on the exact technological implementations of each and diving deeper into them.
Edits in main Manuscript: In the revised Manuscript, the first detailed explanation is now devoted to the DeviceHub, reversing the order from the previous version. This section describes the hardware components of our measurement setup, their connections, and functions within the system. A clear comprehension of the DeviceHub facilitates the explanation of the ServerHub, which justifies its new placement in the main Manuscript.
Edits in main Manuscript: The second detailed explanation is reserved for the ServerHub, which expands upon our previously published HELAO work. The main addition we convey in this section is the design of a Web UI, providing both a technological explanation and a detailed implementation together with a sequence diagram. Additionally, a comprehensive technology stack and operational guidelines are further elaborated in the SI.
Edits and highlighted modifications in SI section: Two Sections, technology stack and operational guidelines are added.
Edits in main Manuscript: The Manuscript’s third detailed section is reserved for the MultiAnalyticHub, which covers all software-driven operations of quality control (QC) and analysis. This section outlines a detailed structure of various QC mechanisms, including drop detection, contact detection, chatbot integration, and material wiping, with further details on the selection process for the ROI in drop detection. Additionally, this section elaborates more on how the previously published MADAP platform was integrated and expanded to meet our case study’s specific requirements. We put particular emphasis on the real-time capabilities and the enhanced electrochemical insights that are now possible with Auto-MISCHABRES.
Edits in main Manuscript: The fourth part of this Hub provides a detailed explanation of the DataHub, where we outline the general structure of the proposed database tables and its interactions with other hubs. Additionally, we also highlight the attributes of our selected relational database proposal.
Edits in main Manuscript: To conclude the presentation of the four key Hubs, we emphasize their roles, functions, and interrelationships to provide a clearer vision and understanding of the proposed Auto-MISCHABRES framework.
Edits in main Manuscript: We have additionally updated the description of the experimental procedure to enhance clarity on the software usage and deliver more substantial credibility of the proposed case study and its results.


- Reviewer: 14. Specific comments - Results and discussion: One of the most relevant points in the presented work is the MultiAnalyseHub module. I suggest the authors explain in more detail how all methodologies incorporated in this module are addressed. The authors could use an SI as support to include it. Also, it could be interesting to mention if the methods implemented in this module are common or if there is some previously reported work and state-of-the-art.
- Response: We are grateful for the reviewer’s insightful comment. Indeed, the MultiAnalyticHub is essential to Auto-MISCHABRES and plays an important role in developing this framework. In response to the suggestion, we have enriched our revised section in the main Manuscript to include a more detailed discussion on the quality control elements and the expansion of the previously published MADAP package for electrochemically focused analyses (https://github.com/fuzhanrahmanian/MADAP/tree/master/madap/echem/voltammetry). Additionally, we have elaborated on this expansion of MADAP in the SI as recommended by the reviewer. After carefully considering the reviewer’s feedback, we opted to incorporate a sequential diagram illustrating the interactions among the individual Hubs. This diagram is designed to demystify the complexities of asynchronous orchestrated experimentation and underline the comprehensive functionality that the MultiAnalyticHub contributes during the automation (Please refer to the response of Q12). All the quality control steps, and the on-the-fly analysis are performed and executed in parallel with the ongoing experimentation. For additional insight, we refer to our updated section on the MultiAnalyticHub, the newly added sequential diagram, and further elaboration on the technology stack, operational guidelines, and MADAP analysis in the SI section.
Highlighted modifications in main Manuscript: We invite the reviewer to refer to the previous question for the applied changes in the MultiAnalyticHub and the added sequential diagram.
Highlighted modifications in SI: The technology stack and operational guidelines sections are illustrated in the previous question, and we invite the reviewer to refer to the response to Q13. However, in this section, we have also added a comprehensive explanation of cyclic voltammetry analysis in the MADAD for a better clarification.


- Reviewer: 15. Specific comments - Results and discussion: The authors mention your "control protocol"; please give more details about it because it is unclear.
- Response: We appreciate the reviewer’s request for more information on the "control protocol". This term refers to the series of quality control mechanisms and measures, implemented within the MultiAnalyticHub. Our protocol consists of procedures for drop detection to monitor material depletion, mandatory movement to the wiping pad, and automated detection and verification of the SDC’s head contact with the substrate. A detailed explanation of each step can be found under the "MultiAnalyticHub" subsection in the main Manuscript and the "Technology stack" section in the revised Supplementary information (please refer to the outlined response for Q13 for more details).


- Reviewer: 16. Specific comments - Results and discussion: Describe the content and functionalities of the different QE implemented in the proposed work.
- Response: We thank the reviewer for this question and would like to elaborate more on the detailed explanation of the QC methods utilized in the MultiAnalyticHub, as outlined in our Manuscript. Even though we trust that the explanation provided in the revised Manuscript is clear and detailed, we remain committed to ensuring clarity in our response.

The quality control methods implemented in the MultiAnalyticHub are:
• Drop Detection: Implemented to monitor material presence within the pump and SDC loop. This step is implemented using a video capture to identify droplets from the head of SDC during the flushing process. The absence of droplets triggers an automatic halt of the experiment.

• Mandatory Wiping Step: Regardless of experimental design, the SDC head is programmed to always travel to the wiping pad between different measurements to remove residual material after flushing, thus preventing salt formation and buildup. This step is coordinated through orchestration API endpoints.

• Contact Detection: An asynchronous API call is used for a controlled head lowering toward the working electrode (WE) to automatically establish electrical contact. During this step, fail-safes are in place to prevent excessive pressure or connection failure that blocks the experiment. If necessary, additional material is also pumped to secure this connection. This is another demonstration of the asynchronous capability of Auto-MISCHBARES.
We want to emphasize that for transparency, our code is open source (https://github.com/fuzhanrahmanian/MISCHBARES) and can be reviewed by anyone at any time. The relevant parts of the code are the following:
• Drop Detection: Found within mischbares/quality_control/drop_detection.py with asynchronous execution from mischbares/action/hamilton_action.py

• Mandatory Wiping Step: The wiping pad’s position and the wiping movement via motor, as part of the mandatory design of sequence of experiment (SOE), are defined in mischbares/procedures/sequential_procedures.py , integral to perfom_sequential_experiment function.

• Contact Detection: The logic for head movement and failsafe mechanisms implementations are detailed in /mischbares/action/lang_action.py, within the move_down function, reserved to move down just on the WE as its target, alongside asynchronous potentiostat integration. Alternative movements are handled via moveRel and moveAbs.


- Reviewer: 17. Specific comments - Results and discussion: Please provide more justification for the incorporation of the Telegram chatbot. Is it only for notification methods, or is there an integration for decision-making?
- Response: We appreciate the reviewer’s query and would like to provide a concise and clear explanation. The Telegram bot is an "ease of life" integration. It is designed and used to keep researchers updated and informed about the progress of experiments, especially during extended periods when the researcher is not overseeing the lengthy experiment. It notifies users at the end of every experiment and updates them on the number of completed experiments and their duration. Additionally, the bot alerts users about the absence of pumpable material, allowing them to replenish supplies as needed. It also notifies and reports if electrical contact is not established, giving an opportunity to manually adjust the head and establish contact since the experiment is temporarily put "on hold". All parameters, including time and threshold settings, can be configured and adjusted to meet the researcher’s specific needs. It is important to note that the chatbot does not, and never claims to, implement any direct interaction or decision-making capabilities and has just a purely informational role.


- Reviewer: 18. Specific comments - Results and discussion: Please mention the differences between MultiAnalyseHub and MADAP.
- Response: We are grateful for the reviewer’s comment. MADAP (https://github.com/fuzhanrahmanian/MADAP/tree/master), designed for the analysis of the experimental measurement data and constitutes an integral component of the MultiAnalyticHub of Auto-MISCHBARES framework (https://github.com/fuzhanrahmanian/MISCHBARES/blob/main/mischbares/driver/analysis_driver.py). The MutliAnalyticHub encompasses the statistical methods required for two main purposes: analysis of raw data and experimental quality control. Within this Hub, MADAP specialized in the analytical segment, incorporated as a Python package, and further expanded to include voltammetry capabilities to enhance the utilization of experimental data. For a better clarification, we provide a structural summary of our framework:
Auto-MISCHBARES is structured around four Hubs = {Device, Server, MultiAnalytic, Data}
Within this structure, the MultiAnalyticHub = {Quality Control, Analysis}, where Analysis segments uses the MADAP package.
For additional context and detailed explanations concerning this Hub, we encourage the reviewer to consult our responses to questions Q12, Q13, and Q14, where have previously elaborated on these aspects.


- Reviewer: 19. Specific comments - Results and discussion: Describe the FAIR principles or add a reference about them.
- Response: We value the reviewer’s emphasis and observation of the critical role of the FAIR principles in data management. Referencing original work is indeed a staple in the scientific community, and we ensure to credit the source and properly cite the foundational work. To the best of our knowledge, the FAIR principles were introduced by Wilkinson et al. in "The FAIR Guiding Principles for scientific data management and stewardship" published in Nature - Scientific Data on 15th of March 2016 (https://www.nature.com/articles/sdata201618). These principles, integral to our study, were and are underscored in the abstract, highlighted as part of one of the challenges in the introduction, and form a core design principle that our database implementation adheres to, as described in our methods sections. This citation appears as reference number 36 in our Manuscript.
Highlighted in main Manuscript: For further clarification, we have consistently emphasized these principles throughout the main manuscript. Please refer to Abstract, Introduction section, Figure 1 caption, and Design and methodology section.


- Reviewer: 20. Specific comments - Results and discussion: In the MultiAnalyseHub section, the authors describe this module. However, the authors incorporate some sentences like " MADAP capabilities offer a rapid assessment of experimental quality and can yield valuable scientific insights". In this work, you don't need to explain MADAP and only need to add the corresponding reference.
- Response: We thank the reviewer for highlighting the specific mention of MADAP in our Manuscript. We acknowledge that a detailed explanation of MADAP’s capabilities is not the primary focus of this work. However, it is important to note that we aim to underline that MADAP is an essential component of the MultiAnalyticHub in the Auto-MISCHBARES framework and contributes unique qualities that make it a perfect fit for our use case. From our point of view, the “quality” needs to be mentioned, especially since Auto-MSICHABRES expands MADAP to include functionality for detecting landmarks in cyclic voltammetry curves. This non-trivial expansion supports our framework. The reference to the “rapid assessment” is intended to spotlight not just MADAP’s inherent quality but also the advantages that Auto-MISCHABRES poses to the experimentalist. This allows for immediate and at-a-glance review of experimental outcomes and enables early judgment calls during lengthy sequences. For instance, instead of waiting to complete 100 experiments over two weeks, Auto-MISCHBARES (leveraging MADAP) allows researchers to stay informed at predefined intervals and have the possibility of early evaluation. We have clarified and adjusted this part in the Manuscript to better convey the significant advantage and integration of MADAP within our framework (more information can be found in the response of Q13).


- Reviewer: 21. Specific comments - Results and discussion: Please discuss why the authors have selected a relational schema to implement the database and why they do not prefer alternatives like graph database or document database.
- Response: We appreciate the reviewer's insightful question regarding our choice of database architecture. Indeed, in the context of diverse data storage solutions, while alternatives such as graph and document databases present their unique advantages, our selection of a relational database for managing experimental data is driven by its inherent benefits and suitability for our research context. Relational databases provide a robust framework for handling highly structured and consistent experimental data, which is the case of our study, where raw data is delivered by a single measurement instrument, i.e., the potentiostat. Additionally, the ACID (Atomicity, Consistency, Isolation, Durability) compliance of relational databases ensures the reliability of transactions, which is essential for maintaining the integrity of experimental results. Given the well-defined structure of our data and minimal emphasis on complex relationships between data points, the simplicity and efficiency of relational databases provide a pragmatic and practical solution. This choice supports frequent, efficient data access and integration within our existing systems. Despite the ability of graph databases to handle highly interconnected data and offer flexibility and fast data traversal, the drawbacks and more complex implementation do not justify their use in our specific case of data recording and analysis of electrochemical experimentation. Our database also allows for easy expansion by adding new tables with the same structure as the existing ones, maintaining consistency across recorded data. The records from the newly added measurement device can be atomized and saved in a “raw” table with the same functions that are right now implemented in Auto-MISCHBARES.
Additionally, we would like to highlight that a document database was provided and incorporated in our previous publication through the implementation of HELAO. Every step of the experimentation, along with its inputs, outputs, and metadata, is recorded and stored in a .hdf5 formatted file. As our expansion, the relational database offers the benefit of retrieving experimental data with a single ID derived from documents saved by HELAO. This also complies with FAIR principles and provides an easy-to-use solution for the future integration of active learning decision-makers. Although these applications are not the focus of our study, we want to highlight and demonstrate the capabilities of Auto-MSICHABRES in offering a robust, reliable, and well-understood foundation for data management and experimental data handling. The revised Manuscript will further elucidate the advantages of our database choice (for more information, please revisit the response to Q13).


- Reviewer: 22. Specific comments - Results and discussion: Concerning the application of ML algorithms usage, at what moment do the authors recommend applying these methods? What do the authors want with ML? Patter recognition? Predictive model? Classification? The incorporation of ML methods is a bit cryptic and needs to be improved, particularly the explanation of the different tasks, what algorithms, etc.
- Response: We thank the reviewer for the valuable feedback and the opportunity to clarify the application of machine learning (ML) algorithms within our study. Your insight helps us further elucidate our approach and Manuscript’s clarity and align with the rigorous review process standards. Our research uses ML and statistical tools within the MultiAnalyticHub to analyze experimental data. An application for optimization is omitted. For instance, one application of ML tools involves utilizing linear regression to identify the capacitive region with a CV Curve, which is then used to determine the cathodic and anodic peak currents. This application, among others, extends the capabilities of MADAP and is now described in the SI section. The utilization of these tools for quality control is explained thoroughly in the main Manuscript (Please refer to the response to Q13). While Auto-MISCHABRES builds upon HELAO, including its integrated active learning server designed for optimizing subsequent experiments, we clarify and emphasize that optimization through ML is not the focus of this study. We mention the active learning capabilities inherited from HELAO to illustrate the potential for future work and introduce the broader context of our framework's capabilities. This distinction ensures our discussion remains focused on the direct applications and contributions of the presented study without diverging into optimization. We trust this response elucidates our selective use of ML and statistical tools in the MultiAnalyticHub for designated analytical tasks. We carefully consider and underscore our deliberate integration of these methods into our research framework, affirming our commitment to contributing valuable insights to the field. For additional details in the main Manuscript and SI section, please refer to the responses to Q13 and Q14.


Questions/Comments/Suggestions
- Reviewer: 23. Automated quality control and data interpretation are the missing puzzle pieces towards prolonged walk-away-times in closed-loop experimentation; why?
- Response: Automated quality control minimizes the necessity for continuous researcher intervention in monitoring experimental processes. Additionally, real-time error correction significantly improves experimental accuracy. These factors, combined with automated data analysis, provide insightful and invaluable information throughout the experimental process and keep researchers well-informed. This enables timely decisions on whether the system is performing satisfactorily or requires early termination. In our introduction, we address the "why" by stating, "Early demonstrations of closed-loop experiments included mostly error resilient measurement and facile data analysis, but the step towards complex and interrelated experimentation necessitates more robust data quality assessment." This statement highlighted the increased susceptibility to errors without these elements. Continuing, we then assert, "… not only streamline experimental processes but also enable scientists to undertake deeper and more intricate inquiries, accelerating discovery in various scientific domains." Here, we emphasize that incorporating these elements generates deeper insights from data and enables researchers - or AI-informed decision makers, if applicable - to optimize experiments with more informed parameters, which can again optimize outcomes and enhance experimental accuracy. Eventually, we conclude our paragraph with the sentence: "By automating and digitalizing processes, these systems increase experimental accuracy and walk-away time." summarizing our elaboration on how these technologies contribute to increased walk-away times.
Edits in main Manuscript: To address the reviewer's comment directly in the revised Manuscript and clarify our stance further, we have inserted an additional sentence directly following the discussion to improve our communication.
Automated quality control and data interpretation are the missing puzzle pieces towards prolonged walk-away-times in closed loop experimentation1. These advances increase the efficiency and innovation of the research process by minimizing the need for human oversight and ensuring the generation of reliable, insightful data.


- Reviewer: 24. The implementation description of each software artefact is part of a methodology process and not a result of the work. I suggest changing the redaction to clarify it. In other words, the results usually express "what", and the methodology expresses "how".
- Response: We are grateful for the reviewer’s recommendations regarding the structural integrity of scientific papers. To adhere to these foundational principles, we have reorganized our Manuscript to clearly separate the “how” of our work and have consolidated all segments related to the design and implementation of Auto-MISCHBARES to the methodology section. This restructuring allows us to address the “what” more explicitly in the results section and to highlight the credibility of our implemented method, where we elaborate on the results of the study case and showcase what is possible and what can be achieved by our methods toward reliable autonomous experimentation. For further clarification and to appreciate the adjustments made, we invite you to review the pertinent sections in the revised Manuscript.


- Reviewer: 25. In Figure 1-B, why are the tables in the MER presented in different colours?
- Response: We appreciate the reviewer's inquiry concerning Fig.1B. We are committed to clearly depicting our proposed work. We cannot be entirely sure what the abbreviation MER refers to since the only diagram in Figure 1b features an entity relation diagram (ERD). Assuming the ERD is the diagram in question, the different color representations of the tables are there for two main purposes: a better visualization and denote functional groupings. The tables "Raw" and "Analysis" are rendered in similar hues/shades to highlight their interconnectedness and relation. The "User" and "Motor" tables are distinguished by different colors to underline their disparate roles within both the database class schema and corresponding Python models. The classification of classes, in this case, is a deliberate design choice and can be followed in detail in the open-source code under mischbares/db/*.py , where each table is associated with its specific class and the inheritance hierarchy is clearly delineated (https://github.com/fuzhanrahmanian/MISCHBARES/tree/main/mischbares/db). Within this structure, "Experiments" is the parent class of "Measurement" and "Motor" (reflecting the association of an experiment with a motor position), which in turn is a parent class of "Procedure" that populated the raw tables. "User" is a separate class and is designed for research tracking and access control.
Should the reviewer's comment pertain to the varied colors of the semantic boxes within the experimental workflow, we clarify that these colors are for visualization differentiation and frame grouping. Additionally, we have introduced a sequence diagram to the manuscript, which assigns specific colors to different Hubs and further elucidates the role of each step in the experimental process and its affiliation with a particular Hub (Please refer to the response to Q12). We trust these adjustments and explanations will address the concerns raised and enhance the understanding of Fig. 1B's design rationale.


- Reviewer: 26. Why do the authors use the ROI method for drop detection? Are there other alternatives? Please comment on it.
- Response: We appraise the reviewer’s attention to detail and commitment to thorough explanations. In response to the reviewer's inquiry about the choice of an ROI, we have clarified and elaborated on our rationale within the Manuscript. Our preference for ROI is primarily motivated by the goal of minimizing computation time by concentrating on a specific, narrowly defined region and decreasing the likelihood of false positives, which are often a consequence of detecting movement in the non-essential and irrelevant background areas captured by the camera. It is important to note that the SDC movement is motorized, ensuring that the material is introduced at a precise and consistent location. Consequently, the ROI only requires calibration during the setup phase and remains fixed and constant throughout the experimental process. An additional motivation can also be identified pertaining to the nature of the drop being transparent. This characteristic limits the effectiveness of other Computer Vision methods beyond movement detection, which is impractical for our purposes. The transparency of the drop significantly challenges the efficacy of such methods (i.e., edge detection, pattern recognition, and others), making the focused approach on movement detection within a predefined ROI not only strategic but essential for ensuring the reliability and integrity of our experimental outcomes.


- Reviewer: 27. The authors could incorporate a schematic representation of the implemented pipelines in the MultiAnalyseHub module, describing input/outputs, critical process, and optional process.
- Response: We are grateful to the reviewer for emphasizing the importance of clarity regarding our MultiAnalyticHub. In response, we have incorporated a sequential diagram in the revised manuscript to provide a schematic representation of the implemented pipelines between all Hubs, specifically the MultiAnalyticHub and its elements along their interactions, input/output dynamic, and performed procedures. Additionally, in our revision in methodology sections, we now offer a detailed description of all processes associated with the MultiAnalyticHub. Furthermore, we have enriched the SI section with three additional subsections that cover technological aspects, operational guidelines, and a comprehensive account of electrochemical analysis. To complement these textual descriptions, we have produced a YouTube video, elaborating on each process, including their inputs and outputs, in detail (https://zenodo.org/records/10445749). We also provide access to open-source code (https://github.com/fuzhanrahmanian/MISCHBARES?tab=readme-ov-file) and comprehensive documentation (https://fuzhanrahmanian.github.io/MISCHBARES/), thus enriching the portrayal of the entire pipeline. We direct the reviewer's attention to our responses to questions Q12, Q13, Q14, Q16, and Q18, where we believe we have thoroughly addressed each aspect of the introduced Hub in detail.


- Reviewer: 28. What are the "AI control" incorporated in this work?
- Response: We appreciate the reviewer's inquiry and the opportunity to clarify our terminology. We have carefully considered the feedback and responded to the last two questions about this term, both in the general and specific abstract comments. We again acknowledge that our initial use of "AI control" may have advertently suggested a focus not central to this study. Instead, this capability is inherited from our previous research and, as such, was not intended to be a central theme in the current work. We agree that the novel contribution of this work should be distinctively highlighted. Thus, in the revised version of the abstract, we have replaced "AI control" with "AI enablers" to accurately reflect our contribution's essence and eliminate any potential confusion. This change focuses on the middleware required that facilitates the practical application of AI within our proposed reliable autonomous framework, which is the novel contribution of this work. The term "AI enablers" better captures our focus on the tools and technologies that support AI applications, distinct from the predictive capabilities associated with our previous HELAO study.
Edits in main Manuscript: We have made the corresponding correction to the abstract as indicated. By changing the terminology, we aim to clearly highlight the contributions of this study solely and distinguish it from earlier publications.


Editor Comment:
Digital Discovery strongly encourages authors of research articles to include an ‘Author contributions’ section in their manuscript, for publication in the final article. This should appear immediately above the ‘Conflict of interest’ and ‘Acknowledgement’ sections. I strongly recommend you use CRediT (the Contributor Roles Taxonomy, https://credit.niso.org/) for standardised contribution descriptions. All authors should have agreed to their individual contributions ahead of submission and these should accurately reflect contributions to the work. Please refer to our general author guidelines https://www.rsc.org/journals-books-databases/author-and-reviewer-hub/authors-information/responsibilities/ for more information.
- Response: We added a conflict of interest statement to the manuscript.


FILES TO PROVIDE WITH YOUR REVISED MANUSCRIPT :
IMPORTANT: Your original files are available to you when you upload your revised manuscript. Please delete any redundant files before completing the submission. Please carefully check the spelling and format of all author names, affiliations and funding information. If your paper is accepted for publication, it is important this information is accurate to ensure your article is correctly indexed, which may affect citations and future funding evaluation. Please note that if you have selected Accepted Manuscript publication, the author list will appear as provided in the ScholarOne submission details until your Advance Article is published and this information is updated from your article.
- Response: We thank the editors for their valuable guidance. We double-checked the spelling of the authors, and we would like to inform you that an additional author has been added. His contribution was marked accordingly in the author's contribution statement.


• A point-by-point response to the comments made by the reviewer(s). (If selecting transparent peer review, please copy your full response to reviewers into the Your Response” text box provided.)
- Response: Each question was answered point by point. We will include the response in the "Your Response Section" for transparent review. Due to the inclusion of pictures, we will include a copy of the review in the additional submission files.


• Your revised manuscript with any changes clearly marked (.doc(x) or.pdf file)
- Response: We attached both the initial manuscript and the highlighted revision version in the submission portal as a .pdf file.


• Your revised manuscript as a .doc(x) file including figures, without highlighting, track changes, etc. (If providing in TeX format instead, please also provide a final PDF version including figures). Please note that we cannot proceed with publication using a .pdf file only.
- Response: We uploaded the .tex zip file of the revised manuscript and the .pdf version to the portal.


High quality images
EITHER
embedded in a doc(x) file
OR
as separate numbered Figures, Schemes or Charts in .tif, .eps or .pdf format, with a resolution of 600 dpi or greater.
- Response: High-quality images are embedded in the .pdf file and provided in numbered .pdf format in a dedicated .zip file.


AND

• A table of contents entry: graphic maximum size 8 cm wide x 4 cm high and 1-2 sentence(s) of editable text, with a maximum of 250 characters, highlighting the key findings of the work. It is recommended authors make use of the full space available for the graphic. See our Author Guidelines for more details: https://www.rsc.org/journals-books-databases/author-and-reviewer-hub/
- Response: We provide a ToC , including a figure and a short test highlighting Auto-MISCHABRES main contributions.
A short highlight: "The high-throughput Auto-MISCHBARES platform streamlines reliable autonomous experimentation across laboratory devices through scheduling, unified quality control, feedback, and advanced real-time data management, encompassing measurement, validation, and analysis."

• Your revised Electronic Supplementary Information (if any)
- Response: We attached both the initial Supplementary Information and the highlighted revision version in the submission portal as a .pdf file.

15-Mar-2024

Dear Ms Rahmanian:

Manuscript ID: DD-ART-12-2023-000257.R1
TITLE: Autonomous Millimeter Scale High Throughput Battery Research System

Thank you for submitting your revised manuscript to Digital Discovery. I am pleased to accept your manuscript for publication in its current form. I have copied any final comments from the reviewer(s) below.

You will shortly receive a separate email from us requesting you to submit a licence to publish for your article, so that we can proceed with the preparation and publication of your manuscript.

You can highlight your article and the work of your group on the back cover of Digital Discovery. If you are interested in this opportunity please contact the editorial office for more information.

Promote your research, accelerate its impact – find out more about our article promotion services here: https://rsc.li/promoteyourresearch.

If you would like us to promote your article on our LinkedIn account [https://rsc.li/Digital_showcase] please fill out this form: https://form.jotform.com/213544038469056.

We are offering all corresponding authors on publications in gold open access RSC journals who are not already members of the Royal Society of Chemistry one year’s Affiliate membership. If you would like to find out more please email membership@rsc.org, including the promo code OA100 in your message. Learn all about our member benefits at https://www.rsc.org/membership-and-community/join/#benefit

By publishing your article in Digital Discovery, you are supporting the Royal Society of Chemistry to help the chemical science community make the world a better place.

With best wishes,

Professor Jason Hein
Associate Editor, Digital Discovery

Reviewer 1

The authors have provided a comprehensive response with associated revisions, which in my estimation have resulted in a manuscript ready for publication.