pith. sign in

arxiv: 2606.12083 · v1 · pith:SJJKVXYSnew · submitted 2026-06-10 · ❄️ cond-mat.mtrl-sci · physics.app-ph· physics.comp-ph· physics.optics

Multilayer Screening of Double and Conventional Perovskite Solar Cells Using SCAPS-1D and Machine Learning: Optimization of ETL, HTL, and Absorber for High-Efficiency Architectures

Neda Nasiri , Seyed Mahdi Mastoor , Amirhosein Ahmadkhan Kordbacheh This is my paper

Pith reviewed 2026-06-27 08:53 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci physics.app-phphysics.comp-phphysics.optics
keywords perovskite solar cellsmachine learningSCAPS-1Ddouble perovskiteslead-freedevice screeningpower conversion efficiencySHAP analysis
0
0 comments X

The pith

Machine learning trained on SCAPS-1D data identifies lead-free double perovskite solar cell stacks with 28.85 percent efficiency.

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

The paper combines device simulations with machine learning to explore 125 different layer combinations for perovskite solar cells, using five options each for the electron transport layer, absorber, and hole transport layer. A model trained on a subset of these structures predicts the power conversion efficiency of the rest, with cross-validation confirming it ranks candidates reliably. The authors then run full simulations on the top predictions and report several structures exceeding 28 percent efficiency, many based on lead-free double perovskites. SHAP analysis further shows that the hole transport layer band gap, absorber band gap, and electron transport layer electron affinity drive most of the performance differences. This workflow is presented as a way to narrow the search space for better solar cell designs without testing every possibility.

Core claim

The central claim is that a machine learning model, trained on SCAPS-1D simulations of a representative subset, can predict and rank the power conversion efficiencies across the full set of 125 multilayer architectures, thereby identifying verified high-performance devices such as FTO/TiO₂/Cs₂AgBiBr₆/NiO/Ag at 28.85 percent PCE and FTO/SnO₂/Cs₂AgInBr₆/NiO/Ag at 28.62 percent PCE, with eight of the top eleven structures using the Cs₂AgInBr₆ absorber.

What carries the argument

A machine learning regressor trained on SCAPS-1D drift-diffusion outputs that predicts power conversion efficiency from material descriptors, with SHAP values used to rank the influence of HTL band gap, absorber band gap, and ETL electron affinity.

If this is right

  • Several specific device stacks reach PCE values that exceed a closely related literature architecture by roughly 4 percent absolute.
  • The HTL band gap, absorber band gap, and ETL electron affinity are the three descriptors with greatest impact on predicted efficiency.
  • Lead-free double perovskites, especially Cs₂AgInBr₆, dominate the highest-ranked configurations.
  • The combined simulation-plus-ML approach can be applied to screen architectures in other multilayer optoelectronic systems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The emphasis on band gap and affinity descriptors points to a practical route for material chemists to engineer new absorbers or transport layers without running full device simulations for every candidate.
  • If the top predicted stacks prove stable under real operating conditions, they could reduce reliance on lead-containing perovskites in commercial solar cells.
  • Adding experimental stability or degradation data as additional ML targets would tighten the link between simulation rankings and long-term device viability.

Load-bearing premise

The SCAPS-1D drift-diffusion model accurately captures real-device behavior for the chosen lead-free double perovskites, including all important interfacial recombination and charge transport losses.

What would settle it

Fabricating the FTO/TiO₂/Cs₂AgBiBr₆/NiO/Ag device in a lab and measuring a power conversion efficiency substantially below 28 percent would show that the simulation or ML ranking does not match experiment.

Figures

Figures reproduced from arXiv: 2606.12083 by Amirhosein Ahmadkhan Kordbacheh, Neda Nasiri, Seyed Mahdi Mastoor.

Figure 1
Figure 1. Figure 1: Schematic illustration of the multilayer perovskite solar-cell architecture used in this study, showing the [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Workflow of the proposed SCAPS-1D and machine-learning framework, showing dataset construction [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Predicted versus actual power conversion efficiency (PCE) for the out-of-sample Leave-One-Group-Out [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: SHAP summary plot showing the global feature importance ranking for PCE. Features are ranked by [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Current density–voltage (J-V) characteristics of Cs [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Current density–voltage (J-V) characteristics of perovskite solar cells with the configurations (a) [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Current density–voltage (J-V) characteristic of the FTO/ZnSe/CH [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
read the original abstract

The combinatorial design space of multilayer perovskite solar cells is vast, yet exhaustive experimental or computational screening of all possible material combinations remains impractical. Here, we integrate SCAPS-1D device simulations with machine learning to systematically explore 125 device architectures constructed from five electron transport layers (ETL), five absorbers (including lead-free double perovskites), and five hole transport layers (HTL). A representative subset of configurations is used to train a machine learning (ML) model, which predicts the power conversion efficiency (PCE) of the remaining unexplored structures. Leave-One-Group-Out cross-validation yields a Spearman rank correlation, demonstrating reliable ranking capability. SHAP (SHapley Additive exPlanations) analysis reveals that the HTL band gap, absorber band gap, and ETL electron affinity are the most influential descriptors, providing physical insights into interfacial recombination and charge extraction. The machine learning model identifies several high-performance configurations that are subsequently verified by full SCAPS-1D simulations. Among them, the device FTO/TiO$_2$/Cs$_2$AgBiBr$_6$/NiO/Ag achieves a PCE of 28.85%, and the ML-suggested structure FTO/SnO$_2$/Cs$_2$AgInBr$_6$/NiO/Ag exhibits 28.62%, outperforming a closely related literature architecture by approximately 4% absolute. Notably, eight of the top-11 structures employ the lead-free double perovskite Cs$_2$AgInBr$_6$. This work demonstrates that a physics-based, data-driven workflow combining SCAPS-1D, ML, and SHAP can accelerate the discovery of high-efficiency, environmentally friendly perovskite solar cells while providing transparent design rules. The approach is generalizable to other multilayer optoelectronic systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper integrates SCAPS-1D drift-diffusion simulations with machine learning to screen 125 multilayer perovskite solar cell architectures (5 ETL × 5 absorbers including lead-free double perovskites × 5 HTL). A representative subset trains an ML model whose PCE predictions for the remainder are ranked via leave-one-group-out cross-validation (Spearman correlation); top candidates are re-simulated in SCAPS-1D, yielding headline values of 28.85 % (FTO/TiO₂/Cs₂AgBiBr₆/NiO/Ag) and 28.62 % (FTO/SnO₂/Cs₂AgInBr₆/NiO/Ag). SHAP analysis identifies HTL band gap, absorber band gap, and ETL electron affinity as dominant descriptors.

Significance. If the ML implementation and SCAPS-1D parameter set are fully documented and the model generalizes within the simulated space, the workflow supplies a reproducible, physics-informed route to rank lead-free double-perovskite stacks and extracts interpretable design rules. The explicit enumeration of 125 structures and the verification step constitute a self-consistent computational result, though its impact remains bounded by the absence of experimental benchmarks.

major comments (2)
  1. [Abstract / Methods] Abstract and Methods: the manuscript states that an ML model was trained on a representative subset and evaluated with leave-one-group-out cross-validation yielding Spearman correlation, yet supplies no model architecture, training-set cardinality, feature list, hyper-parameters, or quantitative regression metrics (MAE, RMSE). These omissions are load-bearing for the central claim that the ML step reliably identifies high-performance architectures outside the training subset.
  2. [Results] Results: all reported PCE values, including the 28.85 % and 28.62 % figures, are generated by the identical SCAPS-1D drift-diffusion model used to create the training data; the ML step therefore performs interpolation within the same physics model rather than providing an independent prediction. The text should explicitly frame the headline efficiencies as model-consistent rankings rather than external validations.
minor comments (1)
  1. [Abstract] The abstract claims the ML-suggested structure outperforms a literature architecture by ~4 % absolute; the specific reference device and its simulated PCE should be stated for direct comparison.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the scope and limitations of our computational workflow. We address each major point below and will revise the manuscript to improve transparency and framing.

read point-by-point responses

  1. Referee: [Abstract / Methods] Abstract and Methods: the manuscript states that an ML model was trained on a representative subset and evaluated with leave-one-group-out cross-validation yielding Spearman correlation, yet supplies no model architecture, training-set cardinality, feature list, hyper-parameters, or quantitative regression metrics (MAE, RMSE). These omissions are load-bearing for the central claim that the ML step reliably identifies high-performance architectures outside the training subset.

    Authors: We agree that these implementation details are essential for reproducibility. In the revised manuscript we will add: (i) the ML algorithm and architecture, (ii) the exact cardinality of the training subset, (iii) the complete list of input features, (iv) all hyper-parameters, and (v) quantitative regression metrics (MAE, RMSE) alongside the reported Spearman correlation. These additions will be placed in a dedicated subsection of Methods and referenced in the Abstract. revision: yes

  2. Referee: [Results] Results: all reported PCE values, including the 28.85 % and 28.62 % figures, are generated by the identical SCAPS-1D drift-diffusion model used to create the training data; the ML step therefore performs interpolation within the same physics model rather than providing an independent prediction. The text should explicitly frame the headline efficiencies as model-consistent rankings rather than external validations.

    Authors: The referee is correct: every PCE value originates from the same SCAPS-1D parameter set. The manuscript already notes that ML-identified candidates are “subsequently verified by full SCAPS-1D simulations,” but we will revise the Abstract, Results, and Discussion to state explicitly that the workflow yields model-consistent rankings within the simulated physics rather than independent experimental validation. We will also adjust the phrasing around “outperforming a literature architecture” to reflect that the comparison is likewise within the same modeling framework. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The workflow consists of running SCAPS-1D drift-diffusion simulations on a representative subset of the 125 architectures to generate training data, training an ML model on that subset to rank the remainder, selecting top candidates via ML, and then re-running full SCAPS-1D on those candidates to obtain the reported PCE values (28.85 % and 28.62 %). The final efficiency numbers are therefore direct outputs of the physics simulator applied to the selected structures, not ML interpolations or fitted parameters renamed as predictions. No self-definitional step, no load-bearing self-citation, and no uniqueness theorem imported from prior author work appears in the described chain; the ML component functions only as a ranking filter whose outputs are independently re-evaluated inside the same simulator. The derivation is therefore self-contained within the simulation framework and does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated in the provided text. The workflow implicitly rests on the fidelity of SCAPS-1D and on the representativeness of the training subset.

axioms (1)
  • domain assumption SCAPS-1D drift-diffusion equations accurately represent real multilayer perovskite device physics
    The entire screening and verification pipeline depends on this untested premise.

pith-pipeline@v0.9.1-grok · 5909 in / 1515 out tokens · 28035 ms · 2026-06-27T08:53:23.982043+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

60 extracted references

  1. [1]

    Role of solar energy in achieving global net-zero targets: Policy and technological perspective

    Asadullah Muhammad Hossain Saad and A Myat. “Role of solar energy in achieving global net-zero targets: Policy and technological perspective”. In:Am. J. Energy Nat. Resour.4.1 (Jan. 2025), pp. 1–8

    2025

  2. [2]

    From lab to market: Strategies for stabilizing and scaling perovskite solar cells via printing technologies

    Xin Li et al. “From lab to market: Strategies for stabilizing and scaling perovskite solar cells via printing technologies”. en. In:Energy Environ. Mater.9.1 (Jan. 2026)

    2026

  3. [3]

    Advanced interface engineering for perovskite solar cells: The way to ensure efficiency and stability

    Dong Hoe Kim and Nam-Gyu Park. “Advanced interface engineering for perovskite solar cells: The way to ensure efficiency and stability”. en. In:Acc. Mater. Res.6.9 (Sept. 2025), pp. 1147–1157

    2025

  4. [4]

    Lead-free perovskites for solar cell applications: recent progress, ongoing challenges, and strategic approaches

    Imtiaz Ahmed, Kamal Prakash, and Shaikh M Mobin. “Lead-free perovskites for solar cell applications: recent progress, ongoing challenges, and strategic approaches”. en. In:Chem. Commun. (Camb.)61.37 (May 2025), pp. 6691–6721

    2025

  5. [5]

    Investigating the optoelectronic properties and photovoltaic performance of Na2AuGaBr6 based double perovskite solar cells via numerical simulation and AI techniques

    Bipul Chandra Biswas et al. “Investigating the optoelectronic properties and photovoltaic performance of Na2AuGaBr6 based double perovskite solar cells via numerical simulation and AI techniques”. en. In:Sci. Rep. 16.1 (Feb. 2026)

    2026

  6. [6]

    Inorganic lead-free double perovskites for eco-friendly energy conversion technology

    Kailash Rangar et al. “Inorganic lead-free double perovskites for eco-friendly energy conversion technology”. en. In:J. Mater. Sci.: Mater. Electron.36.35 (Dec. 2025)

    2025

  7. [7]

    Exploring Sm2NiMnO6 as a lead-free absorber for perovskite solar cells: Insights from theo- retical and experimental approaches

    T Sangavi et al. “Exploring Sm2NiMnO6 as a lead-free absorber for perovskite solar cells: Insights from theo- retical and experimental approaches”. en. In:Sol. Energy Mater. Sol. Cells283.113456 (May 2025), p. 113456

    2025

  8. [8]

    Performance boost in novel CBAI/SNMO double perovskite solar cells via GO-induced back surface field optimization

    Md Faruk Hossain et al. “Performance boost in novel CBAI/SNMO double perovskite solar cells via GO-induced back surface field optimization”. en. In:Phys. Chem. Chem. Phys.28.19 (May 2026), pp. 11795–11815

    2026

  9. [9]

    High performance double perovskites of Cs 2InAgBr6 and Cs 2InAgCl6 structural electronic optical and thermoelectric properties for next generation photovoltaics

    M Fatmi et al. “High performance double perovskites of Cs 2InAgBr6 and Cs 2InAgCl6 structural electronic optical and thermoelectric properties for next generation photovoltaics”. en. In:Sci. Rep.15.1 (July 2025), p. 20851

    2025

  10. [10]

    Comparative First-Principles Study of Lead-Free Cs 2AgB′Br6 (B′ = Bi, Sb, In) Double Per- ovskites for Photovoltaic Applications

    Moses Irungu. “Comparative First-Principles Study of Lead-Free Cs 2AgB′Br6 (B′ = Bi, Sb, In) Double Per- ovskites for Photovoltaic Applications”. In:International Journal of Emerging Research in Engineering and Technology12 (Dec. 2025), p. 100

    2025

  11. [11]

    Janmoni Borah and Smriti Baruah. “Design and optimization of BaSnO3/SW-CNT hybrid electron extraction architecture enabling extended spectral response in nontoxic Cs 2AgBiBr6 double perovskite photovoltaic cells: Insights from SCAPS-1D”. en. In:Physica B Condens. Matter724.418178 (Feb. 2026), p. 418178

    2026

  12. [12]

    Pinning bromide ion with ionic liquid in lead-free cs 2AgBiBr6 double perovskite solar cells

    Jiangning Li et al. “Pinning bromide ion with ionic liquid in lead-free cs 2AgBiBr6 double perovskite solar cells”. en. In:Adv. Funct. Mater.32.25 (June 2022), p. 2112991

    2022

  13. [13]

    Improved charge extraction and atmospheric stability of all-inorganic Cs 2AgBiBr6 perovskite solar cells by MoS2 nanoflakes

    Beili Pang et al. “Improved charge extraction and atmospheric stability of all-inorganic Cs 2AgBiBr6 perovskite solar cells by MoS2 nanoflakes”. en. In:Sol. Energy Mater. Sol. Cells246.111932 (Oct. 2022), p. 111932

    2022

  14. [14]

    Boosting the stability and efficiency of Cs 2AgBiBr6 perovskite solar cells via Zn doping

    Yingjun Ou et al. “Boosting the stability and efficiency of Cs 2AgBiBr6 perovskite solar cells via Zn doping”. en. In:Opt. Mater. (Amst.)129.112452 (July 2022), p. 112452

    2022

  15. [15]

    Predictive modelling of perovskite solar cell structure using machine learning: Pro- cess optimization and device performance

    Deepak Kumar Singh et al. “Predictive modelling of perovskite solar cell structure using machine learning: Pro- cess optimization and device performance”. In:2026 International Conference on Electric Power and Renewable Energy (EPREC). Durg, India: IEEE, Jan. 2026, pp. 1–6

    2026

  16. [16]

    Metal chalcogenide nanostructures: A bifunctional platform to resolve the efficiency-stability trade-off in commercial perovskite solar cells

    Alberto Boretti. “Metal chalcogenide nanostructures: A bifunctional platform to resolve the efficiency-stability trade-off in commercial perovskite solar cells”. en. In:Next Mater.10.101580 (Jan. 2026), p. 101580

    2026

  17. [17]

    Grain size influences activation energy and migration pathways in MAPbBr 3 perovskite solar cells

    Lucie McGovern et al. “Grain size influences activation energy and migration pathways in MAPbBr 3 perovskite solar cells”. en. In:J. Phys. Chem. Lett.12.9 (Mar. 2021), pp. 2423–2428

    2021

  18. [18]

    Simulation and optimization of MAPbI 2 Br and MAPbBr 3 perovskite solar cells achieving efficiencies up to 27.9%

    Yaolan Chen et al. “Simulation and optimization of MAPbI 2 Br and MAPbBr 3 perovskite solar cells achieving efficiencies up to 27.9%”. In:Phys. Scr.101.16 (Apr. 2026), p. 165507. 13

    2026

  19. [19]

    Energy band engineering and cubic crystallinity: achieving 36% efficiency in lead-free MASnBr3 perovskite solar cells via SCAPS-guided optimiza- tion

    Mohammad Mirdoraghi, Maryam Shakiba, and Marzieh Khademalrasool. “Energy band engineering and cubic crystallinity: achieving 36% efficiency in lead-free MASnBr3 perovskite solar cells via SCAPS-guided optimiza- tion”. en. In:Opt. Quantum Electron.57.6 (June 2025)

    2025

  20. [20]

    Next-generation lead-free solar cells with MASnBr3/ZnSnN2 dual absorbers for high efficiency

    Md Mehedi Hasan et al. “Next-generation lead-free solar cells with MASnBr3/ZnSnN2 dual absorbers for high efficiency”. In:Front. Mater.12.1652733 (Aug. 2025)

    2025

  21. [21]

    Study on the feasibility of high PCE C 2 N solar cell incorporating a novel HT material

    Pratap Kumar Dakua et al. “Study on the feasibility of high PCE C 2 N solar cell incorporating a novel HT material”. en. In:ChemistrySelect10.21 (June 2025)

    2025

  22. [22]

    Optimization of Pb-free highly efficient Cs 2AgInBr6 double perovskite solar cells: A numerical investigation using SCAPS

    Mahir Abrar, Ishrat Jahan Biswas, and Deidra Hodges. “Optimization of Pb-free highly efficient Cs 2AgInBr6 double perovskite solar cells: A numerical investigation using SCAPS”. en. In:J. Electron. Mater.54.6 (June 2025), pp. 4357–4365

    2025

  23. [23]

    High-efficiency lead-free MASnI3 perovskite solar cells using ZnSe ETL and NiO HTL: Optimization and comparative study

    Deepak Kumar Singh et al. “High-efficiency lead-free MASnI3 perovskite solar cells using ZnSe ETL and NiO HTL: Optimization and comparative study”. en. In:J. Phys. Chem. Solids211.113507 (Apr. 2026), p. 113507

    2026

  24. [24]

    Numerical simulation and optimization of FTO/TiO2/CZTS/CuO/Au solar cell using SCAPS-1D

    Lofty A Lotfy et al. “Numerical simulation and optimization of FTO/TiO2/CZTS/CuO/Au solar cell using SCAPS-1D”. en. In:Sci. Rep.15.1 (July 2025), p. 28022

    2025

  25. [25]

    Optimized lead-free perovskite solar cell with SnS 2 ETL and Spiro-OMeTAD HTL for ¿ 33% efficiency

    Srinivash Roula et al. “Optimized lead-free perovskite solar cell with SnS 2 ETL and Spiro-OMeTAD HTL for ¿ 33% efficiency”. en. In:Discov Electron2.1 (Sept. 2025)

    2025

  26. [26]

    Rabeya Khan et al. “Numerical simulation and performance enhancement of CsBi3I10-based heterojunction solar cell with various semiconductor layers (CZTS, CZTGS, Al0.8Ga0.2Sb, GaAs) along with machine learning- based analysis”. en. In:Sol. Energy295.113539 (July 2025), p. 113539

    2025

  27. [27]

    Kamil Monga et al. “Ag doped ZnSe as electron transport layer with enhanced electron mobility and better band alignment for efficient lead free - all inorganic CuBiSC l2 perovskite solar cells: a simulation”. In:Phys. Scr.100.8 (Aug. 2025), p. 085948

    2025

  28. [28]

    Performance enhancement of HTL free perovskite solar cells through ETL and back contact engineering

    Amina Shafique, Uzma Amin, and Ahmed Abu-Siada. “Performance enhancement of HTL free perovskite solar cells through ETL and back contact engineering”. en. In:J. Solgel Sci. Technol.117.3 (Feb. 2026)

    2026

  29. [29]

    Probing high-efficiency Cs0.05(FA0.77MA0.23)0.95Pb(I0.77Br0.23)3-based perovskite so- lar cells through first principles computations and SCAPS-1D simulation

    Okba Saidani et al. “Probing high-efficiency Cs0.05(FA0.77MA0.23)0.95Pb(I0.77Br0.23)3-based perovskite so- lar cells through first principles computations and SCAPS-1D simulation”. en. In:RSC Adv.15.10 (Mar. 2025), pp. 7342–7353

    2025

  30. [30]

    A chain entanglement gelled SnO 2 electron transport layer for enhanced perovskite solar cell performance and effective lead capture

    Yuchen Zhou et al. “A chain entanglement gelled SnO 2 electron transport layer for enhanced perovskite solar cell performance and effective lead capture”. en. In:Adv. Mater.37.8 (Feb. 2025), e2416932

    2025

  31. [31]

    Performance optimization of Cs 2 PtBr 6 -based perovskite solar cells through ETL and HTL engineering: a SCAPS-1D simulation study

    Md Sourav Talukder et al. “Performance optimization of Cs 2 PtBr 6 -based perovskite solar cells through ETL and HTL engineering: a SCAPS-1D simulation study”. In:Phys. Scr.100.12 (Dec. 2025), p. 125970

    2025

  32. [32]

    Scalable machine learn- ing models for predicting quantum transport in disordered 2D hexagonal materials

    Seyed Mahdi Mastoor, Ayda Baghervand, and Amirhossein Ahmadkhan Kordbacheh. “Scalable machine learn- ing models for predicting quantum transport in disordered 2D hexagonal materials”. en. In:Comput. Mater. Sci.266.114561 (Feb. 2026), p. 114561

    2026

  33. [33]

    Artificial intelligence in materials science and engineering: Current landscape, key chal- lenges, and future trajectories

    Iman Peivaste et al. “Artificial intelligence in materials science and engineering: Current landscape, key chal- lenges, and future trajectories”. en. In:Compos. Struct.372.119419 (Nov. 2025), p. 119419

    2025

  34. [34]

    A comprehensive review of machine learning applications in perovskite solar cells: Materials discovery, device performance, process optimization and systems integration

    Ling Mao and Changying Xiang. “A comprehensive review of machine learning applications in perovskite solar cells: Materials discovery, device performance, process optimization and systems integration”. en. In:Mater. Today Energy47.101742 (Jan. 2025), p. 101742

    2025

  35. [35]

    Predicting high-performance perovskite solar cells using AI-based machine learning models

    Shafidah Shafian et al. “Predicting high-performance perovskite solar cells using AI-based machine learning models”. en. In:Materials Today Sustainability31.101176 (Sept. 2025), p. 101176

    2025

  36. [36]

    Effect of defects on high efficient perovskite solar cells

    Sara Taheri et al. “Effect of defects on high efficient perovskite solar cells”. en. In:Opt. Mater. (Amst.) 111.110601 (Jan. 2021), p. 110601

    2021

  37. [37]

    Design and defect study of Cs 2AgBiBr6 double perovskite solar cell using suitable charge transport layers

    Hend I Alkhammash et al. “Design and defect study of Cs 2AgBiBr6 double perovskite solar cell using suitable charge transport layers”. In:Semicond. Sci. Technol.38.1 (Jan. 2023), p. 015005. 14

    2023

  38. [38]

    An in-depth investigation of the combined optoelectronic and photovoltaic properties of lead-free cs 2AgBiBr6 double perovskite solar cells using DFT and SCAPS-1D frameworks

    M Shihab Uddin et al. “An in-depth investigation of the combined optoelectronic and photovoltaic properties of lead-free cs 2AgBiBr6 double perovskite solar cells using DFT and SCAPS-1D frameworks”. en. In:Adv. Electron. Mater.10.5 (May 2024)

    2024

  39. [39]

    Simulation, optimization, and machine learning strategies for CH 3NH3PbBr3 perovskite solar cells

    Safikur Rahman Fahim et al. “Simulation, optimization, and machine learning strategies for CH 3NH3PbBr3 perovskite solar cells”. en. In:Next Energy10.100491 (Jan. 2026), p. 100491

    2026

  40. [40]

    Predictability and interrelations of spectral indicators for PV performance in multiple latitudes and climates

    M A Sevillano-Bendez´ u et al. “Predictability and interrelations of spectral indicators for PV performance in multiple latitudes and climates”. en. In:Sol. Energy259 (July 2023), pp. 174–187

    2023

  41. [41]

    Thin single crystal perovskite solar cells to harvest below-bandgap light absorption

    Zhaolai Chen et al. “Thin single crystal perovskite solar cells to harvest below-bandgap light absorption”. en. In:Nat. Commun.8.1 (Dec. 2017), p. 1890

    2017

  42. [42]

    Theoretical analysis of the optimum energy band gap of semiconductors for fabrication of solar cells for applications in higher latitudes locations

    T Zdanowicz, T Rodziewicz, and M Zabkowska-Waclawek. “Theoretical analysis of the optimum energy band gap of semiconductors for fabrication of solar cells for applications in higher latitudes locations”. en. In:Sol. Energy Mater. Sol. Cells87.1-4 (May 2005), pp. 757–769

    2005

  43. [43]

    Entropy and disorder enable charge separation in organic solar cells

    Samantha N Hood and Ivan Kassal. “Entropy and disorder enable charge separation in organic solar cells”. en. In:J. Phys. Chem. Lett.7.22 (Nov. 2016), pp. 4495–4500

    2016

  44. [44]

    Andreas Klein. “Energy band alignment at interfaces of semiconducting oxides: A review of experimental determination using photoelectron spectroscopy and comparison with theoretical predictions by the electron affinity rule, charge neutrality levels, and the common anion rule”. en. In:Thin Solid Films520.10 (Mar. 2012), pp. 3721–3728

    2012

  45. [45]

    Resolving electron and hole transport properties in semiconductor materials by con- stant light-induced magneto transport

    Artem Musiienko et al. “Resolving electron and hole transport properties in semiconductor materials by con- stant light-induced magneto transport”. en. In:Nat. Commun.15.1 (Jan. 2024), p. 316

    2024

  46. [46]

    A numerical study on the relationship between the doping and performance in P3HT:PCBM organic bulk heterojunction solar cells

    Hossein Movla, Afshin Shahalizad, and Asghar Asgari. “A numerical study on the relationship between the doping and performance in P3HT:PCBM organic bulk heterojunction solar cells”. en. In:Sci. Rep.13.1 (Feb. 2023), p. 2031

    2023

  47. [47]

    Tree-based ensemble models, algorithms and performance measures for classification

    John Tsiligaridis. “Tree-based ensemble models, algorithms and performance measures for classification”. In: Adv. Sci. Technol. Eng. Syst. J.8.6 (Nov. 2023), pp. 19–25

    2023

  48. [48]

    Interpretable ensemble learning for materials property prediction with classical interatomic potentials

    Xinyu Jiang et al. “Interpretable ensemble learning for materials property prediction with classical interatomic potentials”. en. In:Npj Comput. Mater.11.1 (Oct. 2025)

    2025

  49. [49]

    Performance optimization and machine learning-guided parameter sensitivity analysis of lead-free KGeCl3 perovskite solar cells

    Tanzir Ahamed et al. “Performance optimization and machine learning-guided parameter sensitivity analysis of lead-free KGeCl3 perovskite solar cells”. en. In:RSC Adv.16.10 (Feb. 2026), pp. 8985–9011

    2026

  50. [50]

    Machine learning-driven optimization of transport layer parameters in CsSn 0.5 Ge 0.5 I 3 perovskite solar cells

    Baseerat Bibi et al. “Machine learning-driven optimization of transport layer parameters in CsSn 0.5 Ge 0.5 I 3 perovskite solar cells”. In:IEEE Access13 (2025), pp. 185416–185432

    2025

  51. [51]

    Machine learning for perovskite solar cells: a comprehensive review on opportunities and challenges for materials scientists

    V´ ıctor de la Asunci´ on-Nadal et al. “Machine learning for perovskite solar cells: a comprehensive review on opportunities and challenges for materials scientists”. en. In:EES Solar1.6 (2025), pp. 927–957

    2025

  52. [52]

    Tabular data: Deep learning is not all you need

    Ravid Shwartz-Ziv and Amitai Armon. “Tabular data: Deep learning is not all you need”. en. In:Inf. Fusion 81 (May 2022), pp. 84–90

    2022

  53. [53]

    Band gap tuning of perovskite solar cells for enhancing the efficiency and stability: issues and prospects

    Md Helal Miah et al. “Band gap tuning of perovskite solar cells for enhancing the efficiency and stability: issues and prospects”. en. In:RSC Adv.14.23 (May 2024), pp. 15876–15906

    2024

  54. [54]

    Numerical simulations for the performance optimization of SnO2/Cs2AgInBr6/CuO lead-free perovskite solar cells

    Leila Bechane, Hani Benguesmia, and Tiouiri Hadda. “Numerical simulations for the performance optimization of SnO2/Cs2AgInBr6/CuO lead-free perovskite solar cells”. In:Eng. Technol. Appl. Sci. Res.15.6 (Dec. 2025), pp. 28706–28709

    2025

  55. [55]

    All-Inorganic Cs 2 AgInBr 6 / RbGeI 3 PSCs incorporating a p + -MoS 2 TRL

    Limei Han et al. “All-Inorganic Cs 2 AgInBr 6 / RbGeI 3 PSCs incorporating a p + -MoS 2 TRL”. In:J. Phys. Conf. Ser.3171.1 (Jan. 2026), p. 012011

    2026

  56. [56]

    Unveiling pressure-driven transitions in Cs 2AgBiBr6: Insights from DFT into a lead-free solar perovskite

    Sagita Gupta et al. “Unveiling pressure-driven transitions in Cs 2AgBiBr6: Insights from DFT into a lead-free solar perovskite”. In:East Eur. J. Phys.1 (Mar. 2026), pp. 363–372. 15

    2026

  57. [57]

    Methylammonium lead halide perovskites: Electronic structure and optical properties for tandem solar cells

    Fatima Oubihi et al. “Methylammonium lead halide perovskites: Electronic structure and optical properties for tandem solar cells”. In:E3S Web Conf.704 (2026), p. 01005

    2026

  58. [58]

    Redox-active NiOx-catalyzed Li+ capture-extraction strategy for tBP-free Spiro-OMeTAD enables exceptional damp-heat stability in perovskite solar cells

    Yun Seop Shin et al. “Redox-active NiOx-catalyzed Li+ capture-extraction strategy for tBP-free Spiro-OMeTAD enables exceptional damp-heat stability in perovskite solar cells”. en. In:Adv. Sci. (Weinh.)13.25 (May 2026), e21825

    2026

  59. [59]

    Device-level simulation of lead-free perovskite solar cells (MASnI 3, MASnBr 3, and MABiI3): A comparative study using SCAPS-1D

    M A A Fahad et al. “Device-level simulation of lead-free perovskite solar cells (MASnI 3, MASnBr 3, and MABiI3): A comparative study using SCAPS-1D”. en. In:Next Energy12.100645 (July 2026), p. 100645

    2026

  60. [60]

    Performance evaluation of metal-doped X-TiO2 electron transport layers in perovskite solar cell devices: a review

    Syed M Hasnain. “Performance evaluation of metal-doped X-TiO2 electron transport layers in perovskite solar cell devices: a review”. en. In:Bull. Mater. Sci. (India)49.1 (Feb. 2026). 16

    2026