Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Deep Learning Models for Colloidal Nanocrystal Synthesis

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A transformer trained on 3,500 synthesis recipes predicts colloidal nanocrystal size to within 1.39 nm and shape with 89% accuracy, and transfers to held-out compounds like PbSe.

desk verdict A large new dataset and a plausible transformer pipeline for nanocrystal synthesis, but the augmentation split-safety question needs an answer before the headline metrics can be trusted. read the letter →

arxiv 2412.10838 v1 pith:LMNZSCCU submitted 2024-12-14 cond-mat.mtrl-sci cs.AIphysics.app-ph

Kai Gu , Yingping Liang , Jiaming Su , Peihan Sun , Jia Peng , Naihua Miao , Zhimei Sun , Ying Fu
show 2 more authors
Haizheng Zhong Jun Zhang
This is my paper · ORCID
classification cond-mat.mtrl-scics.AIphysics.app-ph
keywords colloidalnanocrystalsdeeplearningtransformerTEMimagesegmentationsemi-supervisedsizepredictionshapeclassificationrecipetransfer
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that a transformer-based deep learning model can predict the final size and shape of colloidal nanocrystals directly from their synthesis recipes. Using 3,500 recipes covering 348 nanocrystal compositions, with size labels extracted automatically from TEM images by a semi-supervised segmentation model, the authors report size prediction with a mean absolute error of 1.39 nm and shape classification at 89% average accuracy. They further claim the model transfers to recipes of compounds it never saw, such as PbSe (MAE 2.65 nm when all PbSe recipes are held out). A sympathetic reader would care because a reliable recipe-to-property model would let researchers screen and optimize syntheses computationally before going to the lab, and the attention-weight analysis offers a chemical interpretation of what matters most: composition, then precursor/ligand, then solvent.

What carries the argument

The argument rides on three components. First, a semi-supervised segmentation pipeline turns TEM images into instance masks and size labels: a Cascade-RCNN detector with Normalized Gaussian Wasserstein Distance produces weak labels, morphological erosion and dilation mark uncertain boundary regions, and a U-Net style network is trained with a binary cross-entropy loss on the confident foreground/background, reaching AP50 of 82.5%. Second, a recipe dataset of 3,500 entries supplies five condition descriptors and, for every chemical, a 512-dimensional descriptor derived from DFT-optimized structures embedded by a pre-trained GNN; a reaction-intermediate data augmentation synthesizes descriptors for pairwise reactant intermediates to expand the data tenfold. Third, a transformer encoder with a learnable CLS token consumes the variable-length sequence of chemicals plus condition features, and the attention weights of the CLS token supply the interpretation of which chemicals matter.

What would settle it

Compare model predictions against independently measured particle diameters (e.g., dynamic light scattering or small-angle X-ray scattering) on a held-out set of recipes; if the MAE against those measurements is substantially larger than 1.39 nm, the bottleneck is the segmentation-derived labels. A more direct test: take the PbSe hold-out experiment and replace the TEM-segmentation sizes with scattering-based sizes as the target, and check whether the model's performance degrades.

Watch

Extended reading notes

Core claim

The central claim is that the relationship between nanocrystal synthesis parameters and the resulting nanocrystal size and shape is learnable by a transformer model in a transferable way. The model takes condition descriptors (injection temperature, reaction temperature, time, heating rate, and molar amounts) plus 512-dimensional chemical descriptors obtained by feeding DFT-optimized 3D structures through a fine-tuned graph neural network, and predicts size as a regression and shape as a classification. The paper reports a five-fold cross-validation MAE of 1.39 nm and $R^2$ of 71.9% for size, and 89% average accuracy for shape. In held-out-composition tests the model predicts PbSe size with MAE 2.65 nm even though no PbSe recipes were in the training set, and it correctly anticipates the size of a 2023-2024 literature CaF2 recipe at 9.82 nm versus the reported 9.2 nm. The paper also claims the transformer's self-attention weights align with known reaction mechanisms, such as PbO reacting with oleic acid to form precursors and selenium interacting with TOP, and yield an importance order of composition, precursor or ligand, and solvent in determining final size.

Load-bearing premise

The size labels used as ground truth are computed as equivalent-circle diameters from TEM segmentation masks with an AP50 of 82.5%, and the segmentation, projection, and aggregation errors are not propagated into the synthesis labels, so a systematic bias in measuring non-spherical or aggregated particles would make the reported 1.39 nm MAE artificially low relative to true physical size.

Editorial extensions

If this is right

  • Synthesis recipes could be screened computationally before any lab work, with size and shape predictions guiding precursor, ligand, and solvent choices.
  • The reverse-design mode predicts the needed molar amounts of reactants for a target size, with a reported MAE of 2.41 mmol, which could accelerate recipe optimization.
  • The model's transfer to held-out compounds such as PbSe, Ag2S, Ni, and Cs2NaBiCl6 suggests it can propose starting recipes for new nanocrystal compositions.
  • Automated TEM segmentation at 1.2 million nanocrystals removes the manual-measurement bottleneck, making large recipe-to-property datasets feasible for other particle systems.
  • Attention-weight interpretation offers a data-driven ranking of chemical roles, suggesting composition dominates and solvent is least important for final size.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension the authors do not pursue is closed-loop synthesis: the model could rank candidate recipes for an autonomous platform, with each experiment returning new TEM labels that retrain the model.
  • The reported MAE may be optimistic for small-N recipes because the training loss is zeroed within a +/-15% tolerance for datasets with fewer than 300 particles; re-evaluating on recipes with N>300 only would test the headline number.
  • The reaction-intermediate augmentation assumes arbitrary pairwise reactions with 1:1 stoichiometry; if real solution-phase speciation differs from DFT-optimized geometries, this synthetic data could bias predictions, so a targeted test on reactions with well-characterized intermediates (e.g., known Pb-oleate formation) would be informative.
  • Attention weights are correlational, not causal; a perturbation experiment that removes solvent descriptors and measures the change in size MAE would test whether the solvent's low ranking reflects true causal unimportance.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper develops a transformer-based model to predict nanocrystal size and shape from colloidal synthesis recipes, using a dataset of 3,500 recipes covering 348 nanocrystal compositions and labels derived from TEM images via a semi-supervised segmentation model. The authors report a size prediction MAE of 1.39 nm and shape classification accuracy of 89%, and they demonstrate transfer to held-out compositions, e.g., PbSe with MAE = 2.65 nm when all PbSe recipes are excluded. The model uses DFT-derived chemical descriptors, graph neural network embeddings, and a reaction-intermediate-based data augmentation. The paper also analyzes self-attention weights to argue that, in determining final size, nanocrystal composition is most important, followed by precursor or ligand, with solvent least important.

Significance. If the results are robust, this is a substantial advance in data-driven nanocrystal synthesis: the dataset scale (3,500 recipes) is an order of magnitude larger than previous reports, the code and data are openly available, and the transfer tests to new compositions and the attention-based chemical interpretation are useful for hypothesis generation. The paper also ships a practical segmentation pipeline for TEM images. However, several methodological choices directly affect the central quantitative claims: the ordering of data augmentation relative to the train/test split, the propagation of segmentation uncertainty into size labels, and the ±15% loss tolerance for low-count recipes. These points need to be resolved before the reported MAE and transfer performance can be taken at face value.

major comments (3)
  1. [Data augmentation; Generalizability evaluation] The manuscript does not state whether the reaction-intermediate augmentation is applied before or after the five-fold split and the composition-holdout split. If augmentation is applied to the full 3,500-recipe dataset before splitting, augmented recipes derived from held-out PbSe (or from validation-fold) parent recipes will enter the pretraining data. In that case the reported PbSe transfer MAE of 2.65 nm and the cross-validation MAE of 1.39 nm would no longer measure generalization to unseen recipes. The statement that 'the model is pretrained on both the original and augmented data. Then, the model is fine-tuned using only the original data' does not resolve this, because pretraining would still include validation-derived augmented data. Please state explicitly that augmentation is performed after the split (or inside each training fold only) and verify this in the released code.
  2. [Calculation of equivalent circle diameter] The size labels used as regression targets are computed from instance segmentation masks with AP50 = 82.5%. The paper does not propagate segmentation uncertainty (e.g., mask boundary errors for non-spherical or aggregated nanocrystals) into the size labels, nor does it report a calibration or sensitivity analysis. A systematic segmentation bias would make the reported MAE a measure of agreement with the biased labels rather than with true physical size. The authors should provide error bars on the labels or a validation against manually measured sizes for a subset of recipes.
  3. [Training synthesis model] The tolerance threshold of ±15% for datasets with fewer than 300 nanocrystals ('If the relative error between the model's predicted size and the actual size label falls within this range, their losses are set to zero') is not fully specified. It is unclear whether this tolerance is active during evaluation of the validation folds. If the reported MAE of 1.39 nm includes this zeroing, the figure is not a raw MAE and should be reported with and without the tolerance; if the tolerance applies only during training, that should be stated explicitly. The paper should also report the number and fraction of recipes affected by this rule.
minor comments (6)
  1. [Performance evaluation] The reported MAE and R2 are averages over five folds, but no per-fold values or standard deviations are given; please include these to allow stability assessment, especially given the modest dataset size and the use of augmentation.
  2. [Nanocrystal segmentation model] The loss function is labeled L while the semi-supervised loss is labeled L sim; the text then refers to L semi in the same passage. Please align the notation.
  3. [Calculation of equivalent circle diameter] There is a typo in the formula: 'l ture' should be 'l_true'.
  4. [Main text, Fig. 3d] The phrase '8 to 18 nmby varying' is missing a space; it should read '8 to 18 nm by varying'.
  5. [Transformer layers pass] The phrase 'ReLU fiction' should be 'ReLU function'.
  6. [Discussion] The sentence 'we constructed a comprising 3500 synthesis recipes' should read 'we constructed a dataset comprising 3500 synthesis recipes'.

Circularity Check

1 steps flagged · score 6.0 of 10

Reaction-intermediate augmentation is not documented as split-safe, so the holdout and CV 'predictions' may be fitted on augmented copies of the held-out recipes; otherwise the pipeline is self-contained.

  1. fitted input called prediction [Methods, 'Data augmentation' and 'Generalizability evaluation'; Results, 'generalizability tests']
    "Through this approach, each recipe can theoretically be augmented into several dozen new recipes featuring different intermediates and molar quantities. With generated augmentation recipes, the model is pretrained on both the original and augmented data ... These tests involve selecting the whole recipes for one type of nanocrystal to serve as the validation set, while using the remaining data for training."

    The pipeline description never excludes held-out recipes from augmentation. Because augmentation is applied to 'each recipe' and pretraining uses 'both the original and augmented data', any recipe later reserved for a composition holdout (e.g., PbSe) contributes augmented offspring to the pretraining set. Those offspring inherit the parent recipe's size label and nearly the same reaction conditions, so the model has already fitted PbSe sizes before the claimed 'generalization' evaluation. The reported PbSe MAE of 2.65 nm is therefore, as written, a retrieval of pretraining information rather than a prediction from recipes whose composition was unseen; the same issue would inflate the five-fold CV MAE of 1.39 nm if augmentation precedes the split.

full rationale

Aside from the augmentation-split issue, the derivation is largely self-contained: size/shape labels come from a separately trained segmentation model applied to TEM images, the synthesis model is evaluated by five-fold cross-validation and whole-composition holdouts, and external 2023-2024 recipes provide an independent (if small) check. The self-attention importance ordering is a post hoc interpretation of the trained model, not a circular derivation, and the paper does not present it as a prediction. The only author self-citation (ref. 24, a review by Zhong and co-workers) is contextual and not load-bearing. No uniqueness theorem or fitted ansatz is imported from prior work. The central risk is the reaction-intermediate augmentation: if it is applied to all recipes before splitting, the reported transfer and CV numbers are contaminated because augmented copies of validation recipes with their labels enter pretraining. Because the manuscript does not specify split-safe augmentation, the derivation chain as written contains a fitted-input-called-prediction step. If the released code shows augmentation is applied only within the training split, the circularity score would drop to 1-2; on the manuscript text alone, the transfer claim is partially circular.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central modeling pipeline rests on several domain assumptions: the accuracy of automated segmentation labels, the plausibility of hypothetical reaction intermediates, the adequacy of DFT/GNN descriptors, and the representativeness of the bibliographic dataset. None of these are independently benchmarked or have error bars propagated into the final metrics. The only hand-set free parameters that plausibly affect the reported performance are the 15% loss tolerance and the 50% augmentation probability.

free parameters (3)
  • ±15% tolerance threshold for low-count datasets = 15%
    Hand-chosen in the training loss: predictions within ±15% of the label for recipes with fewer than 300 nanocrystals have their loss set to zero. This can artificially reduce the reported MAE. (Methods, Training synthesis model)
  • Reaction completion probability in augmentation = 50% complete, 50% partial
    The augmentation randomly chooses with 50% probability whether a simulated intermediate reaction goes to completion. This is a hand-set parameter controlling the synthetic training distribution. (Methods, Data augmentation)
  • Number of shape clusters for PbSe = 7
    The shape clustering of 130,000 PbSe nanocrystals uses K-means with seven clusters (C1 to C7); the choice of K is not justified. (Fig. 2c and Methods, Unsupervised shape clustering)
assumptions (5)
  • domain assumption TEM segmentation labels (even at AP50 of 82.5%) are accurate enough to serve as ground-truth size labels for all 3,500 recipes.
    The synthesis model's supervision depends on these labels, but the paper does not propagate segmentation or projection errors into the size labels. (Methods, Nanocrystal segmentation model and Calculation of equivalent circle diameter)
  • ad hoc to paper Any pair of chemicals in a recipe can react in a 1:1 molar ratio to form a reaction intermediate.
    The data augmentation method assumes arbitrary chemical pairs produce DFT-computed intermediates, which is not grounded in chemical mechanism. (Methods, Data augmentation)
  • domain assumption DFT-optimized 3D structures and fine-tuned GNN descriptors capture chemically relevant features for synthesis prediction.
    The validity of these descriptors for predicting nanocrystal size and shape is assumed; no ablation or independent validation is provided. (Methods, Construction of chemical descriptors)
  • domain assumption Attention weights between the CLS token and chemicals reflect the influence of chemicals on nanocrystal size.
    The interpretability analysis equates attention with chemical importance, a known fragile interpretation that lacks causal validation. (Attention weight visualization)
  • domain assumption The 3,500 recipes extracted from literature and experiments are accurately transcribed and representative.
    No detailed extraction protocol, error checking, or sampling strategy is provided; errors in the recipe data would propagate to the trained model. (Main text, dataset construction)
invented entities (1)
  • Hypothetical reaction intermediates (e.g., PbO-OA)
    purpose: Data augmentation: generate new training recipes by inserting DFT-computed intermediates and adjusting molar amounts of the original chemicals.
    These intermediates are not experimentally observed; they are constructed by assuming 1:1 reactions between arbitrary chemical pairs. (Methods, Data augmentation)

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning Models for Colloidal Nanocrystal Synthesis." pith.science (2026). https://pith.science/paper/LMNZSCCU

@misc{pith2026241210838,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Models for Colloidal Nanocrystal Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LMNZSCCU}},
  note         = {Machine review of arXiv:2412.10838}
}
read the original abstract

Colloidal synthesis of nanocrystals usually includes complex chemical reactions and multi-step crystallization processes. Despite the great success in the past 30 years, it remains challenging to clarify the correlations between synthetic parameters of chemical reaction and physical properties of nanocrystals. Here, we developed a deep learning-based nanocrystal synthesis model that correlates synthetic parameters with the final size and shape of target nanocrystals, using a dataset of 3500 recipes covering 348 distinct nanocrystal compositions. The size and shape labels were obtained from transmission electron microscope images using a segmentation model trained with a semi-supervised algorithm on a dataset comprising 1.2 million nanocrystals. By applying the reaction intermediate-based data augmentation method and elaborated descriptors, the synthesis model was able to predict nanocrystal's size with a mean absolute error of 1.39 nm, while reaching an 89% average accuracy for shape classification. The synthesis model shows knowledge transfer capabilities across different nanocrystals with inputs of new recipes. With that, the influence of chemicals on the final size of nanocrystals was further evaluated, revealing the importance order of nanocrystal composition, precursor or ligand, and solvent. Overall, the deep learning-based nanocrystal synthesis model offers a powerful tool to expedite the development of high-quality nanocrystals.

Figures

Figures reproduced from arXiv: 2412.10838 by the authors.

Figure 2
Figure 2. By incorporating additional full labels from literature, the average precision was increased to 82.5%. This [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Alcohol induced surface charging of colloidal quantum dots for controllable electrophoretic deposition processing

    physics.app-ph 2025-05 conditional novelty 6.0 of 10

    Alcohol addition positively charges oleate-capped PbSe quantum dots in non-polar solvents, enabling controllable electrophoretic deposition and size-selective separation.

Reference graph

Works this paper leans on

66 extracted references · 60 canonical work pages · cited by 1 Pith paper

  1. [1]

    Garc ´ıa de Arquer, F. P. et al. Semiconductor quantum dots: Technological progress and future challenges. Science 373, eaaz8541 (2021)

    work page 2021

  2. [2]

    R., Lifshitz, E., Sargent, E

    Kagan, C. R., Lifshitz, E., Sargent, E. H. & Talapin, D. V . Building devices from colloidal quantum dots. Science 353, aac5523 (2016)

    work page 2016

  3. [3]

    Niu, Z. et al. Anisotropic phase segregation and migration of Pt in nanocrystals en route to nanoframe catalysts. Nat. Mater. 15, 1188–1194 (2016)

    work page 2016

  4. [4]

    & Wu, L.-Z

    Li, X.-B., Tung, C.-H. & Wu, L.-Z. Semiconducting quantum dots for artificial photosynthesis. Nat. Rev. Chem. 2, 160–173 (2018)

    work page 2018

  5. [5]

    Jin, D. et al. Nanoparticles for super-resolution microscopy and single-molecule tracking. Nat. Methods 15, 415–423 (2018)

    work page 2018

  6. [6]

    Alivisatos, A. P. Semiconductor clusters, nanocrystals, and quantum dots. Science 271, 933–937 (1996)

    work page 1996

  7. [7]

    Norris, D. J. & Bawendi, M. Measurement and assignment of the size-dependent optical spectrum in CdSe quantum dots. Phys. Rev. B 53, 16338 (1996)

    work page 1996

  8. [8]

    Scholes, G. D. & Rumbles, G. Excitons in nanoscale systems. Nat. Mater. 5, 683–696 (2006)

    work page 2006

Show all 66 references
  1. [9]

    & Cheon, J

    Jun, Y .-w., Choi, J.-s. & Cheon, J. Shape control of semiconductor and metal oxide nanocrystals through nonhydrolytic colloidal routes. Angew. Chem. Int. Ed. 45, 3414–3439 (2006)

    2006

  2. [10]

    Kovalenko, M. V . et al. Prospects of nanoscience with nanocrystals. ACS Nano 9, 1012–1057 (2015)

    2015

  3. [11]

    & Wood, V

    Yarema, O., Yarema, M. & Wood, V . Tuning the composition of multicomponent semiconductor nanocrys- tals: The case of I–III–VI materials. Chem. Mater. 30, 1446–1461 (2018)

    2018

  4. [12]

    Efros, A. L. & Brus, L. E. Nanocrystal quantum dots: From discovery to modern development. ACS Nano 15, 6192–6210 (2021)

    2021

  5. [13]

    Long, Z. et al. A reactivity-controlled epitaxial growth strategy for synthesizing large nanocrystals. Nat. Synth. 2, 296–304 (2023)

    2023

  6. [14]

    Li, J. J. et al. Large-scale synthesis of nearly monodisperse CdSe/CdS core/shell nanocrystals using air- stable reagents via successive ion layer adsorption and reaction. J. Am. Chem. Soc. 125, 12567–12575 (2003)

    2003

  7. [15]

    D., Dupont, D., De Nolf, K., De Roo, J

    Tessier, M. D., Dupont, D., De Nolf, K., De Roo, J. & Hens, Z. Economic and size-tunable synthesis of InP/ZnE (E = S, Se) colloidal quantum dots. Chem. Mater. 27, 4893–4898 (2015)

    2015

  8. [16]

    & Chen, J.-F

    Pu, Y ., Cai, F., Wang, D., Wang, J.-X. & Chen, J.-F. Colloidal synthesis of semiconductor quantum dots toward large-scale production: A review. Ind. Eng. Chem. Res. 57, 1790–1802 (2018)

    2018

  9. [17]

    Lee, E., Wang, C., Yurek, J. & Ma, R. A new frontier for quantum dots in displays. Inf. Disp. 34, 10–31 (2018)

    2018

  10. [18]

    Liu, M. et al. Probing intermediates of the induction period prior to nucleation and growth of semiconductor quantum dots. Nat. Commun. 8, 15467 (2017). 10

    2017

  11. [19]

    van Embden, J., Gross, S., Kittilstved, K. R. & Della Gaspera, E. Colloidal approaches to zinc oxide nanocrystals. Chem. Rev. 123, 271–326 (2023)

    2023

  12. [20]

    Thanh, N. T. K., Maclean, N. & Mahiddine, S. Mechanisms of nucleation and growth of nanoparticles in solution. Chem. Rev. 114, 7610–7630 (2014)

    2014

  13. [21]

    Akkerman, Q. A. et al. Controlling the nucleation and growth kinetics of lead halide perovskite quantum dots. Science 377, 1406–1412 (2022)

    2022

  14. [22]

    & Aspuru-Guzik, A

    Sanchez-Lengeling, B. & Aspuru-Guzik, A. Inverse molecular design using machine learning: Generative models for matter engineering. Science 361, 360–365 (2018)

    2018

  15. [23]

    & Grossman, J

    Xie, T. & Grossman, J. C. Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties. Phys. Rev. Lett. 120, 145301 (2018)

    2018

  16. [24]

    & Zhong, H.-Z

    Peng, J., Muhammad, R., Wang, S.-L. & Zhong, H.-Z. How machine learning accelerates the development of quantum dots? Chin. J. Chem. 39, 181–188 (2021)

    2021

  17. [25]

    Deng, B. et al. CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nat. Mach. Intell. 5, 1031–1041 (2023)

    2023

  18. [26]

    Szymanski, N. J. et al. An autonomous laboratory for the accelerated synthesis of novel materials. Nature 624, 86–91 (2023)

    2023

  19. [27]

    Slattery, A. et al. Automated self-optimization, intensification, and scale-up of photocatalysis in flow. Sci- ence 383, eadj1817 (2024)

    2024

  20. [28]

    & Kumacheva, E

    Abolhasani, M. & Kumacheva, E. The rise of self-driving labs in chemical and materials sciences. Nat. Synth. 2, 483–492 (2023)

    2023

  21. [29]

    Tao, H. et al. Nanoparticle synthesis assisted by machine learning. Nat. Rev. Mater. 6, 701–716 (2021)

    2021

  22. [30]

    Masson, J.-F., Biggins, J. S. & Ringe, E. Machine learning for nanoplasmonics. Nat. Nanotechnol. 18, 111–123 (2023)

    2023

  23. [31]

    V oznyy, O. et al. Machine learning accelerates discovery of optimal colloidal quantum dot synthesis. ACS Nano 13, 11122–11128 (2019)

    2019

  24. [32]

    & Santos, M

    Baum, F., Pretto, T., K ¨oche, A. & Santos, M. J. L. Machine learning tools to predict hot injection syntheses outcomes for II–VI and IV–VI quantum dots. J. Phys. Chem. C 124, 24298–24305 (2020)

    2020

  25. [33]

    Nguyen, H. A. et al. Predicting indium phosphide quantum dot properties from synthetic procedures using machine learning. Chem. Mater. 34, 6296–6311 (2022)

    2022

  26. [34]

    Pyrz, W. D. & Buttrey, D. J. Particle size determination using TEM: A discussion of image acquisition and analysis for the novice microscopist. Langmuir 24, 11350–11360 (2008)

    2008

  27. [35]

    Lee, B. et al. Statistical characterization of the morphologies of nanoparticles through machine learning based electron microscopy image analysis. ACS Nano 14, 17125–17133 (2020)

    2020

  28. [37]

    Lin, B. et al. A deep learned nanowire segmentation model using synthetic data augmentation. npj Comput. Mater. 8, 88 (2022)

    2022

  29. [39]

    & Lee, S.-I

    Lundberg, S. & Lee, S.-I. A unified approach to interpreting model predictions. Preprint atarXiv:1705.07874 (2017)

    2017 arXiv

  30. [40]

    Mourdikoudis, S. et al. Oleic acid/oleylamine ligand pair: a versatile combination in the synthesis of col- loidal nanoparticles. Nanoscale Horiz. 7, 941–1015 (2022). 11

    2022

  31. [41]

    Preske, A. et al. Size-programmed synthesis of PbSe quantum dots via secondary phosphine chalcogenides. Chem. Mater. 31, 8301–8307 (2019)

    2019

  32. [42]

    & Vasconcelos, N

    Cai, Z. & Vasconcelos, N. Cascade R-CNN: Delving into high quality object detection. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition6154–6162 (IEEE, 2018)

    2018

  33. [43]

    Wang, J., Xu, C., Yang, W. & Yu, L. A normalized gaussian wasserstein distance for tiny object detection. Preprint at arXiv:2110.13389 (2021)

    2021 arXiv

  34. [44]

    Wang, J. et al. Deep high-resolution representation learning for visual recognition. IEEE Trans. Pattern Anal. Mach. Intell. 43, 3349–3364 (2021)

    2021

  35. [46]

    Jain, A. et al. Commentary: The Materials Project: A materials genome approach to accelerating materials innovation. APL Mater. 1, (2013)

    2013

  36. [47]

    An all-electron numerical method for solving the local density functional for polyatomic molecules

    Delley, B. An all-electron numerical method for solving the local density functional for polyatomic molecules. J. Chem. Phys. 92, 508–517 (1990)

    1990

  37. [48]

    Perdew, J. P. et al. Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approx- imation for exchange and correlation. Phys. Rev. B 46, 6671–6687 (1992)

    1992

  38. [49]

    P., Burke, K

    Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996)

    1996

  39. [50]

    Lu, S., Gao, Z., He, D., Zhang, L. & Ke, G. Data-driven quantum chemical property prediction leveraging 3D conformations with Uni-Mol+. Nat. Commun. 15, 7104 (2024)

    2024

  40. [51]

    Vaswani, A. et al. Attention is all you need. In Proceedings of the 31st International Conference on Neural Information Processing Systems 6000–6010 (NIPS, 2017). 12 Acknowledgements This work was supported by National Natural Science Foundation of China (U23A20683, H.Z.) and ...

    2017

  41. [52]

    Xiao, C. et al. VISN: virus instance segmentation network for TEM images using deep attention transformer. Briefings Bioinf. 24 (2023)

    2023

  42. [53]

    & Pachitariu, M

    Stringer, C., Wang, T., Michaelos, M. & Pachitariu, M. Cellpose: a generalist algorithm for cellular segmen- tation. Nat. Methods 18, 100–106 (2021)

    2021

  43. [54]

    Reich, C., Prangemeier, T., Franc ¸ani, A. O. & Koeppl, H. An instance segmentation dataset of yeast cells in microstructures. In 2023 45th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) 1–4 (IEEE, 2023)

    2023

  44. [55]

    Caicedo, J. C. et al. Evaluation of deep learning strategies for nucleus segmentation in fluorescence images. Cytometry Part A 95, 952–965 (2019)

    2019

  45. [56]

    Mahbod, A. et al. NuInsSeg: A fully annotated dataset for nuclei instance segmentation in H&E-stained histological images. Sci. Data 11, 295 (2024)

    2024

  46. [57]

    Mahbod, A. et al. CryoNuSeg: A dataset for nuclei instance segmentation of cryosectioned H&E-stained histological images. Comput. Biol. Med. 132, 104349 (2021)

    2021

  47. [58]

    Depto, D. S. et al. Automatic segmentation of blood cells from microscopic slides: A comparative analysis. Tissue and Cell 73, 101653 (2021)

    2021

  48. [59]

    & Cole, J

    Yildirim, B. & Cole, J. M. Bayesian particle instance segmentation for electron microscopy image quantifi- cation. J. Chem. Inf. Model. 61, 1136–1149 (2021)

    2021

  49. [60]

    R ¨uhle, B., Krumrey, J. F. & Hodoroaba, V .-D. Workflow towards automated segmentation of agglomerated, non-spherical particles from electron microscopy images using artificial neural networks.Sci. Rep. 11, 4942 (2021)

    2021

  50. [61]

    Treder, K. P. et al. nNPipe: a neural network pipeline for automated analysis of morphologically diverse catalyst systems. npj Comput. Mater. 9, 18 (2023)

    2023

  51. [62]

    Liu, Y . et al. Breaking the size limitation of directly-synthesized PbS quantum dot inks toward efficient short-wavelength infrared optoelectronic applications. Angew. Chem. Int. Ed. 62, e202300396 (2023)

    2023

  52. [63]

    Zhuang, X. et al. Trivalent europium-doped CsCl quantum dots for MA-free perovskite solar cells with inherent bandgap through lattice strain compensation. Adv. Mater. 35, 2302393 (2023)

    2023

  53. [64]

    Zhu, H. et al. Long-term stability in α-CsPbI3 quantum dots via a nonvacuum Ag+ doping strategy. J. Phys. Chem. C 127, 18727–18735 (2023)

    2023

  54. [65]

    Paik, T. et al. Shape-controlled synthesis and self-assembly of highly uniform upconverting calcium fluoride nanocrystals. Inorg. Chem. Front. 11, 278–285 (2024)

    2024

  55. [66]

    Lian, W. et al. Near-infrared nanophosphors based on CuInSe 2 quantum dots with near-unity photolumines- cence quantum yield for micro-LEDs applications. Adv. Mater. 36, 2311011 (2024)

    2024

  56. [67]

    Han, L. et al. Ternary Ag 3AuS2 nanocrystals for thin-film solar cells. Inorg. Chem. 63, 19382–19389 (2024)

    2024

  57. [68]

    Sun, X. et al. Diffusion-mediated synthesis of high-quality organic–inorganic hybrid perovskite nanocrys- tals. Nat. Synth. (2024)

    2024

  58. [69]

    Zheng, W. et al. Ultra stable X-Ray imaging through a mutually reinforcing strategy between perovskite nanocrystal-polymethyltrifluoropropylsiloxane. Adv. Funct. Mater., 2418944 (2024). 24

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.