Machine learning for discrimination of phase-change chalcogenide glasses

Abstract

Chalcogenides, despite their versatile functionality, share a notably similar local structure in their amorphous states. Particularly in electronic phase-change memory applications, distinguishing these glasses from neighboring compositions that do not possess memory capabilities is inherently difficult when employing traditional analytical methods. This has led to a dilemma in materials design since an atomistic view of the arrangement in the amorphous state is the key to understanding and optimizing the functionality of these glasses. To tackle this challenge, we present a machine learning (ML) approach to separate electronic phase-change materials (ePCMs) from other chalcogenides, based upon subtle differences in the short-range order inside the glassy phase. Leveraging the established structure–property relations in chalcogenide glasses, we select suitable features to train accurate machine learning models, even with a modestly sized dataset. The trained model accurately discerns the critical transition point between glass compositions suitable for use as ePCMs and those that are not, particularly for both GeTe–GeSe and Sb₂Te₃–Sb₂Se₃ materials, in line with experiments. Furthermore, by extracting the physical knowledge that the ML model has offered, we pinpoint three pivotal structural features of amorphous chalcogenides, that is, the bond angle, packing efficiency, and the length of the fourth bond, which provide a map for materials design with the ability to “predict” and “explain”. All three of the above features point to the smaller Peierls-like distortion and more well-defined octahedral clusters in amorphous ePCMs than non-ePCMs. Our study delves into the mechanisms shaping these structural attributes in amorphous ePCMs, yielding valuable insights for the AI-powered discovery of novel materials.

Keywords

amorphous materials descriptors / chalcogenide glass / first-principles methods / machine learning / phase-change memory

References

[1] Zhu L, Zhou J, Sun Z. Materials data toward machine learning: advances and challenges. J Phys Chem Lett. 2022; 13(18): 3965-3977.

[2] de Pablo JJ, Jackson NE, Webb MA, et al. New frontiers for the materials genome initiative. npj Comput Mater. 2019; 5(1): 41.

[3] Butler KT, Davies DW, Cartwright H, Isayev O, Walsh A. Machine learning for molecular and materials science. Nature. 2018; 559(7715): 547-555.

[4] Bartók AP, De S, Poelking C, et al. Machine learning unifies the modeling of materials and molecules. Sci Adv. 2017; 3(12): e1701816.

[5] Zhu S, Yu T, Xu T, et al. Intelligent computing: the latest advances, challenges, and future. Intell Comput. 2023; 2: 0006.

[6] Chen C, Zuo Y, Ye W, Li X, Ong SP. Learning properties of ordered and disordered materials from multi-fidelity data. Nat Comput Sci. 2021; 1(1): 46-53.

[7] Chen D, Bai Y, Ament S, et al. Automating crystal-structure phase mapping by combining deep learning with constraint reasoning. Nat Mach Intell. 2021; 3(9): 812-822.

[8] Yin J, Pei Z, Gao MC. Neural network-based order parameter for phase transitions and its applications in high-entropy alloys. Nat Comput Sci. 2021; 1(10): 686-693.

[9] Goodall REA, Parackal AS, Faber FA, Armiento R, Lee AA. Rapid discovery of stable materials by coordinate-free coarse graining. Sci Adv. 2022; 8(30): eabn4117.

[10] Rao Z, Tung PY, Xie R, et al. Machine learning-enabled high-entropy alloy discovery. Science. 2022; 378(6615): 78-85.

[11] Xian RP, Stimper V, Zacharias M, et al. A machine learning route between band mapping and band structure. Nat Comput Sci. 2022; 3(1): 101-114.

[12] Allen AEA, Tkatchenko A. Machine learning of material properties: predictive and interpretable multilinear models. Sci Adv. 2022; 8(18): eabm7185.

[13] Raabe D, Mianroodi JR, Neugebauer J. Accelerating the design of compositionally complex materials via physics-informed artificial intelligence. Nat Comput Sci. 2023; 3(3): 198-209.

[14] Jiang B, Yu Y, Cui J, et al. High-entropy-stabilized chalcogenides with high thermoelectric performance. Science. 2021; 371(6531): 830-834.

[15] Pan J, Li Q, Feng Y, et al. Parallel interrogation of the chalcogenide-based micro-ring sensor array for photoacoustic tomography. Nat Commun. 2023; 14(1): 3250.

[16] Lanza M, Sebastian A, Lu WD, et al. Memristive technologies for data storage, computation, encryption, and radio-frequency communication. Science. 2022; 376(6597): eabj9979.

[17] Lu Y, Stegmaier M, Nukala P, et al. Mixed-mode operation of hybrid phase-change nanophotonic circuits. Nano Lett. 2017; 17(1): 150-155.

[18] Moshwan R, Yang L, Zou J, Chen Z-G. Eco-friendly SnTe thermoelectric materials: progress and future challenges. Adv Funct Mater. 2017; 27(43): 1703278.

[19] Zhao L-D, Tan G, Hao S, et al. Ultrahigh power factor and thermoelectric performance in hole-doped single-crystal SnSe. Science. 2016; 351(6269): 141-144.

[20] Noe P, Verdy A, d'Acapito F, et al. Toward ultimate nonvolatile resistive memories: the mechanism behind ovonic threshold switching revealed. Sci Adv. 2020; 6(9): eaay2830.

[21] Wuttig M, Deringer VL, Gonze X, Bichara C, Raty JY. Incipient metals: functional materials with a unique bonding mechanism. Adv Mater. 2018; 30(51): e1803777.

[22] Zhu M, Cojocaru-Miredin O, Mio AM, et al. Unique bond breaking in crystalline phase change materials and the quest for metavalent bonding. Adv Mater. 2018; 30(18): e1706735.

[23] Wuttig M, Schon CF, Lotfering J, Golub P, Gatti C, Raty JY. Revisiting the nature of chemical bonding in chalcogenides to explain and design their properties. Adv Mater. 2023; 35(20): e2208485.

[24] Kooi BJ, Wuttig M. Chalcogenides by design: functionality through metavalent bonding and confinement. Adv Mater. 2020; 32(21): e1908302.

[25] Raty JY, Bichara C, Schon CF, Gatti C, Wuttig M. Tailoring chemical bonds to design unconventional glasses. Proc Natl Acad Sci U S A. 2024; 121(2): e2316498121.

[26] Schon CF, van Bergerem S, Mattes C, et al. Classification of properties and their relation to chemical bonding: essential steps toward the inverse design of functional materials. Sci Adv. 2022; 8(47): eade0828.

[27] Bruns G, Merkelbach P, Schlockermann C, et al. Nanosecond switching in GeTe phase change memory cells. Appl Phys Lett. 2009; 95(4): 043108.

[28] Wang X-P, Li X-B, Chen N-K, Chen B, Rao F, Zhang S. Phase-change-memory process at the limit: a proposal for utilizing monolayer Sb₂Te₃. Adv Sci. 2021; 8(13): 2004185.

[29] Perniola L, Sousa V, Fantini A, et al. Electrical behavior of phase-change memory cells based on GeTe. IEEE Electron Device Lett. 2010; 31(5): 488-490.

[30] Mai X, Xu Q, Yang Z, et al. Sb-Se-based electrical switching device with fast transition speed and minimized performance degradation due to stable mid-gap states. Electron. 2024;e46.

[31] Delaney M, Zeimpekis I, Lawson D, Hewak DW, Muskens OL. A new family of ultralow loss reversible phase-change materials for photonic integrated circuits: Sb₂S₃ and Sb₂Se₃. Adv Funct Mater. 2020; 30(36): 2002447.

[32] Delaney M, Zeimpekis I, Du H, et al. Nonvolatile programmable silicon photonics using an ultralow-loss Sb₂Se₃ phase change material. Sci Adv. 2021; 7(25): eabg3500.

[33] Guarneri L, Jakobs S, von Hoegen A, et al. Metavalent bonding in crystalline solids: how does it collapse? Adv Mater. 2021; 33(39): e2102356.

[34] Zhang B, Zhang W, Shen ZJ, et al. Element-resolved atomic structure imaging of rocksalt GeSbTe phase-change material. Appl Phys Lett. 2016; 108(19): 191902.

[35] Wuttig M, Steimer C. Phase change materials: from material science to novel storage devices. Appl Phys A. 2007; 87(3): 411-417.

[36] Zhang W, Ronneberger I, Zalden P, et al. How fragility makes phase-change data storage robust: insights from ab initio simulations. Sci Rep. 2014; 4: 6529.

[37] Feng M, Liu S-C, Hu L, et al. Interfacial strain engineering in wide-bandgap GeS thin films for photovoltaics. J Am Chem Soc. 2021; 143(25): 9664-9671.

[38] Hosokawa S, Hari Y, Kouchi T, et al. Electronic structures and local atomic configurations in amorphous GeSe and GeTe. J Phys Condens Matter. 1998; 10(8): 1931-1950.

[39] Xu M, Gu R, Qiao C, et al. Unraveling the structural and bonding nature of antimony sesquichalcogenide glass for electronic and photonic applications. J Mater Chem C. 2021; 9(25): 8057-8065.

[40] Akola J, Jones RO, Kohara S, et al. Experimentally constrained density-functional calculations of the amorphous structure of the prototypical phase-change material Ge₂Sb₂Te₅. Phys Rev B. 2009; 80(2): 020201.

[41] Matsunaga T, Akola J, Kohara S, et al. From local structure to nanosecond recrystallization dynamics in AgInSbTe phase-change materials. Nat Mater. 2011; 10(2): 129-134.

[42] Kassem M, Benmore CJ, Tverjanovich A, et al. Glassy and liquid Sb₂S₃: insight into the structure and dynamics of a promising functional material. J Mater Chem C. 2023; 11(14): 4654-4673.

[43] Kohara S, Kato K, Kimura S, et al. Structural basis for the fast phase change of Ge₂Sb₂Te₅: ring statistics analogy between the crystal and amorphous states. Appl Phys Lett. 2006; 89(20): 201910.

[44] Choudhary K, DeCost B, Chen C, et al. Recent advances and applications of deep learning methods in materials science. npj Comput Mater. 2022; 8(1): 59.

[45] Wang H, Fu T, Du Y, et al. Scientific discovery in the age of artificial intelligence. Nature. 2023; 620(7972): 47-60.

[46] Deringer VL, Bernstein N, Csányi G, et al. Origins of structural and electronic transitions in disordered silicon. Nature. 2021; 589(7840): 59-64.

[47] Xu M, Xu M, Miao X. Deep machine learning unravels the structural origin of mid-gap states in chalcogenide glass for high-density memory integration. InfoMat. 2022; 4(6): e12315.

[48] Giri D, Williams L, Mukherjee A, Rajan K. Quantum signatures for screening metavalent solids. J Chem Phys. 2021; 154(12): 124105.

[49] Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. ACM; 2016: 785-794.

[50] Lundberg SM, Lee S-I. A Unified Approach to Interpreting Model Predictions. Curran Associates; 2017: 4768-4777.

[51] Xie T, Grossman JC. Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties. Phys Rev Lett. 2018; 120(14): 145301.

[52] Liu Y-T, Li X-B, Zheng H, et al. High-throughput screening for phase-change memory materials. Adv Funct Mater. 2021; 31(21): 2009803.

[53] Errington JR, Debenedetti PG. Relationship between structural order and the anomalies of liquid water. Nature. 2001; 409(6818): 318-321.

[54] Gabardi S, Caravati S, Sosso GC, Behler J, Bernasconi M. Microscopic origin of resistance drift in the amorphous state of the phase-change compound GeTe. Phys Rev B. 2015; 92(5): 054201.

[55] Ward L, Dunn A, Faghaninia A, et al. Matminer: an open source toolkit for materials data mining. Comput Mater Sci. 2018; 152: 60-69.

[56] Akola J, Jones RO. Binary alloys of Ge and Te: order, voids, and the eutectic composition. Phys Rev Lett. 2008; 100(20): 205502.

[57] Kalikka J, Akola J, Jones RO, Schober HR. Density functional and classical simulations of liquid and glassy selenium. Phys Rev B. 2020; 102(10): 104202.

[58] Lvd M, Hinton GE. Visualizing Data using t-SNE. J Mach Learn Res. 2008; 9(86): 2579-2605.

[59] Lundberg SM, Erion G, Chen H, et al. From local explanations to global understanding with explainable AI for trees. Nat Mach Intell. 2020; 2(1): 56-67.

[60] Littlewood PB. The crystal structure of IV-VI compounds. I. Classification and description. J Phys C Solid State Phys. 1980; 13(26): 4855-4873.

[61] Lencer D, Salinga M, Grabowski B, Hickel T, Neugebauer J, Wuttig M. A map for phase-change materials. Nat Mater. 2008; 7(12): 972-977.

[62] Raty JY, Schumacher M, Golub P, Deringer VL, Gatti C, Wuttig M. A quantum-mechanical map for bonding and properties in solids. Adv Mater. 2019; 31(3): e1806280.

[63] Lee TH, Elliott SR. Ab initio computer simulation of the early stages of crystallization: application to Ge₂Sb₂Te₅ phase-change materials. Phys Rev Lett. 2011; 107(14): 145702.

[64] Persch C, Müller MJ, Yadav A, et al. The potential of chemical bonding to design crystallization and vitrification kinetics. Nat Commun. 2021; 12(1): 4978.

[65] Gaspard J-P. Vanishing-harmonicity and phase-change materials. Phys Status Solidi Rapid Res Lett. 2021; 15(3): 2000536.

[66] Jones RO. The chemical bond in solids—revisited. J Phys Condens Matter. 2022; 34(34): 343001.

[67] Li S, Li M, Chen L, et al. Ultra-stable, endurable, and flexible Sb(2)Te(x)Se(3 −  x) phase change devices for memory application and wearable electronics. ACS Appl Mater Interfaces. 2022; 14(40): 45600-45610.

[68] Gaspard J-P. Structure of covalently bonded materials: from the Peierls distortion to phase-change materials. CR Phys. 2016; 17(3-4): 389-405.

[69] Raty J-Y, Wuttig M. The interplay between Peierls distortions and metavalent bonding in IV-VI compounds: comparing GeTe with related monochalcogenides. J Phys D Appl Phys. 2020; 53(23): 234002.

[70] Raty JY, Godlevsky V, Ghosez P, Bichara C, Gaspard JP, Chelikowsky JR. Evidence of a reentrant Peierls distortion in liquid GeTe. Phys Rev Lett. 2000; 85(9): 1950-1953.

[71] Fujita T, Chen Y, Kono Y, et al. Pressure-induced reversal of Peierls-like distortions elicits the polyamorphic transition in GeTe and GeSe. Nat Commun. 2023; 14(1): 7851.

[72] Akola J, Jones RO. Structural phase transitions on the nanoscale: the crucial pattern in the phase-change materials Ge₂Sb₂Te₅ and GeTe. Phys Rev B. 2007; 76(23): 235201.

[73] Xu M, Cheng YQ, Wang L, et al. Pressure tunes electrical resistivity by four orders of magnitude in amorphous Ge₂Sb₂Te₅ phase-change memory alloy. Proc Natl Acad Sci U S A. 2012; 109(18): E1055-E1062.

[74] Sun Z, Zhou J, Blomqvist A, Johansson B, Ahuja R. Formation of large voids in the amorphous phase-change memory Ge₂Sb₂Te₅ alloy. Phys Rev Lett. 2009; 102(7): 075504.

[75] Song Z, Song S, Zhu M, et al. From octahedral structure motif to sub-nanosecond phase transitions in phase change materials for data storage. Sci China Inf Sci. 2018; 61(8): 081302.

[76] Hegedüs J, Elliott SR. Microscopic origin of the fast crystallization ability of Ge-Sb-Te phase-change memory materials. Nat Mater. 2008; 7(5): 399-405.

[77] Los JH, Kühne TD, Gabardi S, Bernasconi M. First-principles study of the amorphous In₃SbTe₂ phase change compound. Phys Rev B. 2013; 88(17): 174203.

[78] Kresse G, Furthmüller J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys Rev B. 1996; 54(16): 11169-11186.

[79] Kresse G, Joubert D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys Rev B. 1999; 59(3): 1758-1775.

[80] Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Phys Rev Lett. 1996; 77(18): 3865-3868.

[81] Gu RC, Xu M, Yu R, et al. Structural features of chalcogenide glass SiTe: an ovonic threshold switching material. APL Mater. 2021; 9(8): 081101.

[82] Xu M, Li B, Xu K, et al. Stabilizing amorphous Sb by adding alien seeds for durable memory materials. Phys Chem Chem Phys. 2019; 21(8): 4494-4500.

[83] Hinton G, Roweis S. Stochastic Neighbor Embedding. MIT Press; 2002: 857-864.

[84] Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn: machine learning in python. J Mach Learn Res. 2011; 12: 2825-2830.

[85] Hjorth Larsen A, Jørgen Mortensen J, Blomqvist J, et al. The atomic simulation environment—a Python library for working with atoms. J Phys Condens Matter. 2017; 29(27): 273002.

RIGHTS & PERMISSIONS

2025 The Author(s). InfoMat published by UESTC and John Wiley & Sons Australia, Ltd.