liu.seSearch for publications in DiVA
Endre søk
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Active learning with moment tensor potentials to predict material properties: Ti0.5Al0.5N at elevated temperature
Linköpings universitet, Institutionen för fysik, kemi och biologi, Teoretisk Fysik. Linköpings universitet, Tekniska fakulteten.ORCID-id: 0000-0002-6033-1130
Linköpings universitet, Institutionen för fysik, kemi och biologi, Teoretisk Fysik. Linköpings universitet, Tekniska fakulteten.ORCID-id: 0000-0001-6373-5109
Linköpings universitet, Institutionen för fysik, kemi och biologi, Teoretisk Fysik. Linköpings universitet, Tekniska fakulteten.ORCID-id: 0000-0001-7551-4717
2024 (engelsk)Inngår i: Journal of Vacuum Science & Technology. A. Vacuum, Surfaces, and Films, ISSN 0734-2101, E-ISSN 1520-8559, Vol. 42, nr 1, artikkel-id 013412Artikkel i tidsskrift (Fagfellevurdert) Published
Abstract [en]

Transition metal nitride alloys possess exceptional properties, making them suitable for cutting applications due to their inherent hardness or as protective coatings due to corrosion resistance. However, the computational demands associated with predicting these properties using ab initio methods can often be prohibitively high at the conditions of their operation at cutting tools, that is, at high temperatures and stresses. Machine learning approaches have been introduced into the field of materials modeling to address the challenge. In this paper, we present an active learning workflow to model the properties of our benchmark alloy system cubic B1 Ti0.5Al0.5N at temperatures up to 1500 K. With a minimal requirement of prior knowledge about the alloy system for our workflow, we train a moment tensor potential (MTP) to accurately model the material's behavior over the entire temperature range and extract elastic and vibrational properties. The outstanding accuracy of MTPs with relatively little training data demonstrates that the presented approach is highly efficient and requires about two orders of magnitude less computational resources than state-of-the-art ab initio molecular dynamics.

sted, utgiver, år, opplag, sider
A V S AMER INST PHYSICS , 2024. Vol. 42, nr 1, artikkel-id 013412
HSV kategori
Identifikatorer
URN: urn:nbn:se:liu:diva-200518DOI: 10.1116/6.0003260ISI: 001136625000005OAI: oai:DiVA.org:liu-200518DiVA, id: diva2:1832654
Merknad

Funding Agencies|VINNOVA Excellence Center Functional Nanoscale Materials (FunMat-II) [2022-03071]; Knut and Alice Wallenberg Foundation (Wallenberg Scholar Grant) [KAW-2018.0194]; Swedish Government Strategic Research Area in Materials Science on Functional Materials at Linkoeping University [2009 00971]; Swedish Government Strategic Research Area in the Swedish e-Science Research Centre (SeRC); Swedish National Infrastructure for Supercomputing (SNIC); CSC IT Center for Science in Finland; National Academic Infrastructure for Supercomputing in Sweden (NAISS)

Tilgjengelig fra: 2024-01-30 Laget: 2024-01-30 Sist oppdatert: 2024-04-02
Inngår i avhandling
1. Combining ab‐initio and machine learning techniques for theoretical simulations of hard nitrides at extreme conditions
Åpne denne publikasjonen i ny fane eller vindu >>Combining ab‐initio and machine learning techniques for theoretical simulations of hard nitrides at extreme conditions
2024 (engelsk)Doktoravhandling, med artikler (Annet vitenskapelig)
Abstract [en]

In this thesis I focus on combining the high accuracy of first-principles calculations with modern machine learning methods to make large scale investigations of industrially relevant nitride systems reliable and computationally viable. I study the electronic, thermodynamic and mechanical properties of two families of compounds: Ti1−xAlxN alloys at the operational conditions of industrial cutting tools and ReNx systems at crushing pres-sures comparable to inner earth core conditions. Standard first-principles simulations of materials are usually carried out at zero temperature and pressure, and while many state-of-the-art approaches can take these effects into account, they are usually accompanied by a substantial increase in computational demand. In this thesis I therefore explore the possiblities of studying materials at extreme conditions using machine learning methods with extraordinary efficiency without loss of calculational accuracy. 

Ti1−xAlxN alloy coatings exhibit exceptional properties due to their inherent ability to spinodally decompose at elevated temperature, leading to age-hardening. Since the cubic B1 phase of Ti1−xAlxN is well-studied, available high-accuracy first-principles data served as both a benchmark and data set on which to train a machine learning interatomic potential. Using the reliable moment tensor potentials, an investigation of the accuracy and efficiency of this approach was carried out in a machine learning study. Building upon the success of this technique, implementation of a learning-on-the-fly (active learning) methodology into a workflow to determine accurate material properties with minimal prior knowledge showed great promise, while maintaining a computational demand up to two orders of magnitude lower than comparable first-principles approaches. Investigations of properties of industrially lesser desired, but sometimes present hexagonal alloy phases of Ti1−xAlxN are also included in this thesis, since knowledge and understanding of all competing phases can help guide development toward improving cutting tool lifetime and performance. Furthermore, while w-Ti1−xAlxN may not be able to compete with its cubic counterpart in terms of hardness, it shows promise for other applications due to its electronic and elastic properties. 

Metastable ReNx phases are high energy materials due to their covalent N-N and Re-N bonds, leading to exceptional mechanical and electronic properties. Just like diamond, the hardest and arguably most famous metastable mate-rial naturally occurring on earth, they are stabilized by extreme pressures and high temperatures, but can be quenched to ambient conditions. Understanding the formation and existence of these non-equilibrium compounds may hold the key to unlocking a new generation of hard materials. In this thesis, all currently known phases of ReNx compounds have been investigated, encompassing both experimentally observed and theoretically suggested structures. Investigations of the convex hulls across a broad pressure range were carried out, coupled with calculations of phonons in the proposed crystals to determine both energetic and dynamical stability. Overall, the studies included in this thesis focused mainly on investigation of the ground state of ReN2 at higher pressure, where experimental results were deviating from earlier theoretical predictions. Additional research focused on specifically exploring properties and stability of novel ReN6 at synthesis conditions using the active learning workflow to train an interatomic potential. 

sted, utgiver, år, opplag, sider
Linköping: Linköping University Electronic Press, 2024. s. 87
Serie
Linköping Studies in Science and Technology. Dissertations, ISSN 0345-7524 ; 2375
HSV kategori
Identifikatorer
urn:nbn:se:liu:diva-201992 (URN)10.3384/9789180755320 (DOI)9789180755313 (ISBN)9789180755320 (ISBN)
Disputas
2024-04-19, Planck, F-building, Campus Valla, Linköping, 10:15 (engelsk)
Opponent
Veileder
Tilgjengelig fra: 2024-04-02 Laget: 2024-04-02 Sist oppdatert: 2024-04-02bibliografisk kontrollert

Open Access i DiVA

Fulltekst mangler i DiVA

Andre lenker

Forlagets fulltekst

Søk i DiVA

Av forfatter/redaktør
Bock, FlorianTasnadi, FerencAbrikosov, Igor
Av organisasjonen
I samme tidsskrift
Journal of Vacuum Science & Technology. A. Vacuum, Surfaces, and Films

Søk utenfor DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric

doi
urn-nbn
Totalt: 86 treff
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf