Accurate band gap prediction based on an interpretable Δ-machine learning

Zhang, Lingyao; Su, Tianhao; Li, Musen; Jia, Fanhao; Hu, Shunbo; Zhang, Peihong; Ren, Wei

doi:10.1016/j.mtcomm.2022.104630

Cited by 7 publications

(5 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It should be noted that the method still requires the same number of samples to be computed for numerically cheaper and more costly properties. The Δ-ML method has been used for the prediction of various quantum chemical properties such as potential energy surfaces, 20 band gaps, 21 and excited state energies. 22 In the MFML method, also termed combination technique quantum machine learning (CQML), 23 it is possible to combine submodels that utilize a few training samples of the highest fidelity while using more samples from the cheaper fidelities to achieve the accuracy of a certain target fidelity.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multifidelity Machine Learning for Molecular Excitation Energies

Vinod,

Maity,

Zaspel

et al. 2023

J. Chem. Theory Comput.

View full text Add to dashboard Cite

The accurate but fast calculation of molecular excited states is still a very challenging topic. For many applications, detailed knowledge of the energy funnel in larger molecular aggregates is of key importance, requiring highly accurate excitation energies. To this end, machine learning techniques can be a very useful tool, though the cost of generating highly accurate training data sets still remains a severe challenge. To overcome this hurdle, this work proposes the use of multifidelity machine learning where very little training data from high accuracies is combined with cheaper and less accurate data to achieve the accuracy of the costlier level. In the present study, the approach is employed to predict vertical excitation energies to the first excited state for three molecules of increasing size, namely, benzene, naphthalene, and anthracene. The energies are trained and tested for conformations stemming from classical molecular dynamics and density functional based tight-binding simulations. It can be shown that the multifidelity machine learning model can achieve the same accuracy as a machine learning model built only on high-cost training data while expending a much lower computational effort to generate the data. The numerical gain observed in these benchmark test calculations was over a factor of 30 but certainly can be much higher for high-accuracy data.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Multifidelity Machine Learning for Molecular Excitation Energies

Vinod,

Maity,

Zaspel

et al. 2023

J. Chem. Theory Comput.

View full text Add to dashboard Cite

show abstract

“…Compared with the experimental band gap, the calculated band gap has a reasonable expected decrease due to the underestimate based on the PBE function. 46 In addition, the isosurface plots of the wave functions of VBM and CBM also demonstrate that the charge density is predominantly distributed on the PbBr 6 octahedra (Figure 2c,d). More importantly, the isosurface plots of the wave functions of CBM and VBM also exhibit an electronic 3D feature.…”

Section: ■ Results and Discussionmentioning

confidence: 83%

“…As shown in Figure S5, experimental band gaps of ( R -PyEA)Pb 2 Br 6 and ( S -PyEA)Pb 2 Br 6 perovskites are about 2.98 eV (416 nm) and 2.95 eV (420 nm), respectively. Compared with the experimental band gap, the calculated band gap has a reasonable expected decrease due to the underestimate based on the PBE function . In addition, the isosurface plots of the wave functions of VBM and CBM also demonstrate that the charge density is predominantly distributed on the PbBr 6 octahedra (Figure c,d).…”

Section: Results and Discussionmentioning

confidence: 99%

Wafer-Scale Patterning Integration of Chiral 3D Perovskite Single Crystals toward High-Performance Full-Stokes Polarimeter

Bai,

Wang,

et al. 2024

J. Am. Chem. Soc.

View full text Add to dashboard Cite

Chiral three-dimensional (3D) perovskites exhibit exceptional optoelectronic characteristics and inherent chiroptical activity, which may overcome the limitations of low-dimensional chiral optoelectronic devices and achieve superior performance. The integrated chip of high-performance arbitrary polarized light detection is one of the aims of chiral optoelectronic devices and may be achieved by chiral 3D perovskites. Herein, we first fabricate the wafer-scale integrated full-Stokes polarimeter by the synergy of unprecedented chiral 3D perovskites (R/S-PyEA)Pb 2 Br 6 and onestep capillary-bridge assembly technology. Compared with the chiral low-dimensional perovskites, chiral 3D perovskites present smaller exciton binding energies of 57.3 meV and excellent circular dichroism (CD) absorption properties, yielding excellent circularly polarized light (CPL) photodetectors with an ultrahigh responsivity of 86.7 A W −1 , an unprecedented detectivity exceeding 4.84 × 10 13 Jones, a high anisotropy factor of 0.42, and high-fidelity CPL imaging with 256 pixels. Moreover, the anisotropic crystal structure also enables chiral 3D perovskites to have a large linear-polarization response with a polarized ratio of 1.52. The combination of linear-polarization and circular-polarization discrimination capabilities guarantees the achievement of a full-Stokes polarimeter. Our study provides new research insights for the large-scale patterning wafer integration of high-performance chiroptical devices.

show abstract

“…A notable application of this approach is the sure independence screening and sparsifying operator (SISSO) method [ 22 ], which has been used successfully in band gap prediction models. For instance, Zhang et al trained on high-throughput calculations of two-dimensional semiconductors and utilized complex descriptors identified by the SISSO algorithm, and they achieved high accuracy in predicting HSE band gaps with a coefficient of determination ( R 2 ) of 0.96 [ 23 ]. Ma et al proposed a physically interpretable three-dimensional descriptor to obtain the Γ-point gap of twist bilayer graphene at arbitrary twist angles and different interlayer spacings, demonstrating high accuracy as evidenced by a 99% Pearson coefficient [ 24 ].…”

Section: Introductionmentioning

confidence: 99%

Feature-Assisted Machine Learning for Predicting Band Gaps of Binary Semiconductors

Huo,

Zhang,

et al. 2024

Nanomaterials

View full text Add to dashboard Cite

The band gap is a key parameter in semiconductor materials that is essential for advancing optoelectronic device development. Accurately predicting band gaps of materials at low cost is a significant challenge in materials science. Although many machine learning (ML) models for band gap prediction already exist, they often suffer from low interpretability and lack theoretical support from a physical perspective. In this study, we address these challenges by using a combination of traditional ML algorithms and the ‘white-box’ sure independence screening and sparsifying operator (SISSO) approach. Specifically, we enhance the interpretability and accuracy of band gap predictions for binary semiconductors by integrating the importance rankings of support vector regression (SVR), random forests (RF), and gradient boosting decision trees (GBDT) with SISSO models. Our model uses only the intrinsic features of the constituent elements and their band gaps calculated using the Perdew–Burke–Ernzerhof method, significantly reducing computational demands. We have applied our model to predict the band gaps of 1208 theoretically stable binary compounds. Importantly, the model highlights the critical role of electronegativity in determining material band gaps. This insight not only enriches our understanding of the physical principles underlying band gap prediction but also underscores the potential of our approach in guiding the synthesis of new and valuable semiconductor materials.

show abstract

Accurate band gap prediction based on an interpretable Δ-machine learning

Cited by 7 publications

References 38 publications

Multifidelity Machine Learning for Molecular Excitation Energies

Multifidelity Machine Learning for Molecular Excitation Energies

Wafer-Scale Patterning Integration of Chiral 3D Perovskite Single Crystals toward High-Performance Full-Stokes Polarimeter

Feature-Assisted Machine Learning for Predicting Band Gaps of Binary Semiconductors

Contact Info

Product

Resources

About