Title: Generative design of stable semiconductor materials using deep learning and density functional theory

Abstract

Semiconductor device technology has greatly developed in complexity since discovering the bipolar transistor. In this work, we developed a computational pipeline to discover stable semiconductors by combining generative adversarial networks (GAN), classifiers, and high-throughput first-principles calculations. We used CubicGAN, a GAN-based algorithm for generating cubic materials and developed a classifier to screen the semiconductors and studied their stability using first principles. We found 12 stable AA$${}^{\prime}$$${}^{\prime}$MH_{6}semiconductors in the F-43m space group including BaNaRhH_{6}, BaSrZnH_{6}, BaCsAlH_{6}, SrTlIrH_{6}, KNaNiH_{6}, NaYRuH_{6}, CsKSiH_{6}, CaScMnH_{6}, YZnMnH_{6}, NaZrMnH_{6}, AgZrMnH_{6}, and ScZnMnH_{6}. Previous research reported that five AA$${}^{\prime}$$${}^{\prime}$IrH6 semiconductors with the same space group were synthesized. Our research shows that AA$${}^{\prime}$$${}^{\prime}$MnH_{6}and NaYRuH_{6}semiconductors have considerably different properties compared to the rest of the AA$${}^{\prime}$$${}^{\prime}$MH_{6}semiconductors. Based on the accurate hybrid functional calculations, AA$${}^{\prime}$$${}^{\prime}$MH_{6}semiconductors are found to be wide-bandgap semiconductors. Moreover, BaSrZnH_{6}and KNaNiH_{6}are direct-bandgap semiconductors, whereas others exhibit indirect bandgaps.

The characterization of normal mode (CNM) procedure coupled with an adiabatic connection scheme (ACS) between local and normal vibrational modes, both being a part of the Local Vibrational Mode theory developed in our group, can identify spectral changes as structural fingerprints that monitor symmetry alterations, such as those caused by Jahn-Teller (JT) distortions. Employing the PBE0/Def2-TZVP level of theory, we investigated in this proof-of-concept study the hexaaquachromium cation case,$$\mathrm {[Cr{(OH_2)}_6]^{3+}}$$${\left[\mathrm{Cr}{\left({\mathrm{OH}}_{2}\right)}_{6}\right]}^{3+}$/$$\mathrm {[Cr{(OH_2)}_6]^{2+}}$$${\left[\mathrm{Cr}{\left({\mathrm{OH}}_{2}\right)}_{6}\right]}^{2+}$, as a commonly known example for a JT distortion, followed by the more difficult ferrous and ferric hexacyanide anion case,$$\mathrm {[Fe{(CN)}_6]^{4-}}$$${\left[\mathrm{Fe}{\left(\mathrm{CN}\right)}_{6}\right]}^{4-}$/$$\mathrm {[Fe{(CN)}_6]^{3-}}$$${\left[\mathrm{Fe}{\left(\mathrm{CN}\right)}_{6}\right]}^{3-}$. We found that in both cases CNM of the characteristic normal vibrational modes reflects delocalization consistent with high symmetry and ACS confirms symmetry breaking, as evidenced by the separation of axial and equatorial group frequencies. As underlined by the Cremer-Kraka criterion for covalent bonding, from$$\mathrm {[Cr{(OH_2)}_6]^{3+}}$$${\left[\mathrm{Cr}{\left({\mathrm{OH}}_{2}\right)}_{6}\right]}^{3+}$to$$\mathrm {[Cr{(OH_2)}_6]^{2+}}$$${\left[\mathrm{Cr}{\left({\mathrm{OH}}_{2}\right)}_{6}\right]}^{2+}$there is an increase in axial covalency whereas the equatorial bonds shift toward electrostatic character. From$$\mathrm {[Fe{(CN)}_6]^{4-}}$$${\left[\mathrm{Fe}{\left(\mathrm{CN}\right)}_{6}\right]}^{4-}$to$$\mathrm {[Fe{(CN)}_6]^{3-}}$$${\left[\mathrm{Fe}{\left(\mathrm{CN}\right)}_{6}\right]}^{3-}$we observed an increase in covalency without altering the bond nature. Distinct$$\pi $$$\pi $back-donation disparity could be confirmed by comparison with the isolated CN$$^-$$${}^{-}$system. In summary, our study positions the CNM/ACS protocol as a robust tool for investigating less-explored JT distortions, paving the way for future applications.

Graphical abstract

The adiabatic connection scheme relates local to normal modes, with symmetry breaking giving rise to axial and equatorial group local frequencies

Deary, J; Scheck, M; Schwengner, R; O’Donnell, D; Bemmerer, D; Beyer, R; Hensel, Th; Junghans, A R; Kögler, T; Müller, S E; et al(
, The European Physical Journal A)

Abstract

The electricE1 and magneticM1 dipole responses of the$$N=Z$$$N=Z$nucleus$$^{24}$$${}^{24}$Mg were investigated in an inelastic photon scattering experiment. The 13.0 MeV electrons, which were used to produce the unpolarised bremsstrahlung in the entrance channel of the$$^{24}$$${}^{24}$Mg($$\gamma ,\gamma ^{\prime }$$$\gamma ,{\gamma}^{\prime}$) reaction, were delivered by the ELBE accelerator of the Helmholtz-Zentrum Dresden-Rossendorf. The collimated bremsstrahlung photons excited one$$J^{\pi }=1^-$$${J}^{\pi}={1}^{-}$, four$$J^{\pi }=1^+$$${J}^{\pi}={1}^{+}$, and six$$J^{\pi }=2^+$$${J}^{\pi}={2}^{+}$states in$$^{24}$$${}^{24}$Mg. De-excitation$$\gamma $$$\gamma $rays were detected using the four high-purity germanium detectors of the$$\gamma $$$\gamma $ELBE setup, which is dedicated to nuclear resonance fluorescence experiments. In the energy region up to 13.0 MeV a total$$B(M1)\uparrow = 2.7(3)~\mu _N^2$$$B\left(M1\right)\uparrow =2.7\left(3\right)\phantom{\rule{0ex}{0ex}}{\mu}_{N}^{2}$is observed, but this$$N=Z$$$N=Z$nucleus exhibits only marginalE1 strength of less than$$\sum B(E1)\uparrow \le 0.61 \times 10^{-3}$$$\sum B\left(E1\right)\uparrow \le 0.61\times {10}^{-3}$ e$$^2 \, $$${}^{2}\phantom{\rule{0ex}{0ex}}$fm$$^2$$${}^{2}$. The$$B(\varPi 1, 1^{\pi }_i \rightarrow 2^+_1)/B(\varPi 1, 1^{\pi }_i \rightarrow 0^+_{gs})$$$B(\Pi 1,{1}_{i}^{\pi}\to {2}_{1}^{+})/B(\Pi 1,{1}_{i}^{\pi}\to {0}_{\mathrm{gs}}^{+})$branching ratios in combination with the expected results from the Alaga rules demonstrate thatKis a good approximative quantum number for$$^{24}$$${}^{24}$Mg. The use of the known$$\rho ^2(E0, 0^+_2 \rightarrow 0^+_{gs})$$${\rho}^{2}(E0,{0}_{2}^{+}\to {0}_{\mathrm{gs}}^{+})$strength and the measured$$B(M1, 1^+ \rightarrow 0^+_2)/B(M1, 1^+ \rightarrow 0^+_{gs})$$$B(M1,{1}^{+}\to {0}_{2}^{+})/B(M1,{1}^{+}\to {0}_{\mathrm{gs}}^{+})$branching ratio of the 10.712 MeV$$1^+$$${1}^{+}$level allows, in a two-state mixing model, an extraction of the difference$$\varDelta \beta _2^2$$$\Delta {\beta}_{2}^{2}$between the prolate ground-state structure and shape-coexisting superdeformed structure built upon the 6432-keV$$0^+_2$$${0}_{2}^{+}$level.

Acharya, S; Adamová, D; Aglieri_Rinella, G; Agnello, M; Agrawal, N; Ahammed, Z; Ahmad, S; Ahn, S U; Ahuja, I; Akindinov, A; et al(
, The European Physical Journal C)

Abstract

The elliptic flow$$(v_2)$$$\left({v}_{2}\right)$of$${\textrm{D}}^{0}$$${\text{D}}^{0}$mesons from beauty-hadron decays (non-prompt$${\textrm{D}}^{0})$$${\text{D}}^{0})$was measured in midcentral (30–50%) Pb–Pb collisions at a centre-of-mass energy per nucleon pair$$\sqrt{s_{\textrm{NN}}} = 5.02$$$\sqrt{{s}_{\text{NN}}}=5.02$ TeV with the ALICE detector at the LHC. The$${\textrm{D}}^{0}$$${\text{D}}^{0}$mesons were reconstructed at midrapidity$$(|y|<0.8)$$$\left(\right|y|<0.8)$from their hadronic decay$$\mathrm {D^0 \rightarrow K^-\uppi ^+}$$${D}^{0}\to {K}^{-}{\pi}^{+}$, in the transverse momentum interval$$2< p_{\textrm{T}} < 12$$$2<{p}_{\text{T}}<12$ GeV/c. The result indicates a positive$$v_2$$${v}_{2}$for non-prompt$${{\textrm{D}}^{0}}$$${\text{D}}^{0}$mesons with a significance of 2.7$$\sigma $$$\sigma $. The non-prompt$${{\textrm{D}}^{0}}$$${\text{D}}^{0}$-meson$$v_2$$${v}_{2}$is lower than that of prompt non-strange D mesons with 3.2$$\sigma $$$\sigma $significance in$$2< p_\textrm{T} < 8~\textrm{GeV}/c$$$2<{p}_{\text{T}}<8\phantom{\rule{0ex}{0ex}}\text{GeV}/c$, and compatible with the$$v_2$$${v}_{2}$of beauty-decay electrons. Theoretical calculations of beauty-quark transport in a hydrodynamically expanding medium describe the measurement within uncertainties.

A well-known open problem of Meir and Moser asks if the squares of sidelength 1/nfor$$n\ge 2$$$n\ge 2$can be packed perfectly into a rectangle of area$$\sum _{n=2}^\infty n^{-2}=\pi ^2/6-1$$${\sum}_{n=2}^{\infty}{n}^{-2}={\pi}^{2}/6-1$. In this paper we show that for any$$1/2$1/2<t<1$, and any$$n_0$$${n}_{0}$that is sufficiently large depending on t, the squares of sidelength$$n^{-t}$$${n}^{-t}$for$$n\ge n_0$$$n\ge {n}_{0}$can be packed perfectly into a square of area$$\sum _{n=n_0}^\infty n^{-2t}$$${\sum}_{n={n}_{0}}^{\infty}{n}^{-2t}$. This was previously known (if one packs a rectangle instead of a square) for$$1/2$1/2<t\le 2/3$(in which case one can take$$n_0=1$$${n}_{0}=1$).

Bauer, Martin; Le Brigant, Alice; Lu, Yuxiu; Maor, Cy(
, Calculus of Variations and Partial Differential Equations)

Abstract

We introduce a family of Finsler metrics, called the$$L^p$$${L}^{p}$-Fisher–Rao metrics$$F_p$$${F}_{p}$, for$$p\in (1,\infty )$$$p\in (1,\infty )$, which generalizes the classical Fisher–Rao metric$$F_2$$${F}_{2}$, both on the space of densities$${\text {Dens}}_+(M)$$${\text{Dens}}_{+}\left(M\right)$and probability densities$${\text {Prob}}(M)$$$\text{Prob}\left(M\right)$. We then study their relations to the Amari–C̆encov$$\alpha $$$\alpha $-connections$$\nabla ^{(\alpha )}$$${\nabla}^{\left(\alpha \right)}$from information geometry: on$${\text {Dens}}_+(M)$$${\text{Dens}}_{+}\left(M\right)$, the geodesic equations of$$F_p$$${F}_{p}$and$$\nabla ^{(\alpha )}$$${\nabla}^{\left(\alpha \right)}$coincide, for$$p = 2/(1-\alpha )$$$p=2/(1-\alpha )$. Both are pullbacks of canonical constructions on$$L^p(M)$$${L}^{p}\left(M\right)$, in which geodesics are simply straight lines. In particular, this gives a new variational interpretation of$$\alpha $$$\alpha $-geodesics as being energy minimizing curves. On$${\text {Prob}}(M)$$$\text{Prob}\left(M\right)$, the$$F_p$$${F}_{p}$and$$\nabla ^{(\alpha )}$$${\nabla}^{\left(\alpha \right)}$geodesics can still be thought as pullbacks of natural operations on the unit sphere in$$L^p(M)$$${L}^{p}\left(M\right)$, but in this case they no longer coincide unless$$p=2$$$p=2$. Using this transformation, we solve the geodesic equation of the$$\alpha $$$\alpha $-connection by showing that the geodesic are pullbacks of projections of straight lines onto the unit sphere, and they always cease to exists after finite time when they leave the positive part of the sphere. This unveils the geometric structure of solutions to the generalized Proudman–Johnson equations, and generalizes them to higher dimensions. In addition, we calculate the associate tensors of$$F_p$$${F}_{p}$, and study their relation to$$\nabla ^{(\alpha )}$$${\nabla}^{\left(\alpha \right)}$.

Siriwardane, Edirisuriya M. Dilanga, Zhao, Yong, Perera, Indika, and Hu, Jianjun. Generative design of stable semiconductor materials using deep learning and density functional theory. npj Computational Materials 8.1 Web. doi:10.1038/s41524-022-00850-3.

Siriwardane, Edirisuriya M. Dilanga, Zhao, Yong, Perera, Indika, & Hu, Jianjun. Generative design of stable semiconductor materials using deep learning and density functional theory. npj Computational Materials, 8 (1). https://doi.org/10.1038/s41524-022-00850-3

Siriwardane, Edirisuriya M. Dilanga, Zhao, Yong, Perera, Indika, and Hu, Jianjun.
"Generative design of stable semiconductor materials using deep learning and density functional theory". npj Computational Materials 8 (1). Country unknown/Code not available: Nature Publishing Group. https://doi.org/10.1038/s41524-022-00850-3.https://par.nsf.gov/biblio/10381640.

@article{osti_10381640,
place = {Country unknown/Code not available},
title = {Generative design of stable semiconductor materials using deep learning and density functional theory},
url = {https://par.nsf.gov/biblio/10381640},
DOI = {10.1038/s41524-022-00850-3},
abstractNote = {Abstract Semiconductor device technology has greatly developed in complexity since discovering the bipolar transistor. In this work, we developed a computational pipeline to discover stable semiconductors by combining generative adversarial networks (GAN), classifiers, and high-throughput first-principles calculations. We used CubicGAN, a GAN-based algorithm for generating cubic materials and developed a classifier to screen the semiconductors and studied their stability using first principles. We found 12 stable AA$${}^{\prime}$$′MH6semiconductors in the F-43m space group including BaNaRhH6, BaSrZnH6, BaCsAlH6, SrTlIrH6, KNaNiH6, NaYRuH6, CsKSiH6, CaScMnH6, YZnMnH6, NaZrMnH6, AgZrMnH6, and ScZnMnH6. Previous research reported that five AA$${}^{\prime}$$′IrH6 semiconductors with the same space group were synthesized. Our research shows that AA$${}^{\prime}$$′MnH6and NaYRuH6semiconductors have considerably different properties compared to the rest of the AA$${}^{\prime}$$′MH6semiconductors. Based on the accurate hybrid functional calculations, AA$${}^{\prime}$$′MH6semiconductors are found to be wide-bandgap semiconductors. Moreover, BaSrZnH6and KNaNiH6are direct-bandgap semiconductors, whereas others exhibit indirect bandgaps.},
journal = {npj Computational Materials},
volume = {8},
number = {1},
publisher = {Nature Publishing Group},
author = {Siriwardane, Edirisuriya M. Dilanga and Zhao, Yong and Perera, Indika and Hu, Jianjun},
}

Warning: Leaving National Science Foundation Website

You are now leaving the National Science Foundation website to go to a non-government website.

Website:

NSF takes no responsibility for and exercises no control over the views expressed or the accuracy of
the information contained on this site. Also be aware that NSF's privacy policy does not apply to this site.