COREY OSES DAVID HICKS MARCO ESTERS RICO FRIEDRICH ANDRIY SMOLYANYUK CORMAC TOHER STEFANO CURTAROLO Materials Science Duke University AFLOW: Integrated infrastructure for computational materials discovery
COREY OSESDAVID HICKS
MARCO ESTERSRICO FRIEDRICH
ANDRIY SMOLYANYUK
CORMAC TOHERSTEFANO
CURTAROLOMaterials Science
Duke University
AFLOW: Integrated infrastructure for computational materials discovery
2
AFLOW School Team
Corey OsesCormac Toher David Hicks Marco Esters
Andriy Smolyanyuk Stefano CurtaroloRico Friedrich
3
AFLOW School Zoom Logistics
Microphone should be
muted if not speaking
If you have a question, use
the Chat or Raise Hand functions
4
AFLOW School Zoom Logistics
Reactions button
Raise hand button
5
AFLOW School Zoom Logistics
Raised hand button
6
AFLOW School Zoom Logistics
• Exercises will be part of each session
• For exercises, group will be separated into breakout rooms within Zoom
• One instructor will be in each breakout room, and you can ask this instructor for assistance, share your screen with them, etc.
• Exercise materials are available in the TU Dresden VNC portal – to access the portal, you will need to install and run the TU Dresden VPN: https://tu-dresden.de/zih/dienste/service-katalog/arbeitsumgebung/zugang_datennetz/vpn#
• Follow instructions for your operating system
7
AFLOW School Zoom Logistics
• Once VPN is connected, log into VNC virtual machine using link and password sent to you by email
• Please do not use the “Log out” function, as this will shut down the VM for everyone
• Please do not shut down the VM
• To exit, simply close the browser window/tab
• Please do not change the resolution in the desktop environment
8
AFLOW School Zoom Logistics
Connect toVirtual
Machine
9
AFLOW School Zoom Logistics
Use password
sent to you by email
10
AFLOW School Zoom Logistics
Clipboard and screen
resizing
11
AFLOW School Zoom Logistics
Open clipboard to copy/paste into Portal
Use clipboard to copy/paste into Portal
12
AFLOW School Zoom Logistics
Settings
Clip to Window Scaling mode
Remote Resizing
13
AFLOW School Zoom Logistics
Terminal
Home directory
14
AFLOW School Zoom Logistics
AFLOW Exercisesdirectory
15
Thermal barrier coatings:
energy/aerospace
Superhard, wear resistant coatings:
machine tools
Biocompatible, corrosion resistant
alloys: medical implants
Photovoltaic materials:
solar energy harvesting
Superconductors and heat sinks for advanced electronics
Solid state electrolyte:
Li-ion batteries
Novel Materials Applications
16
✓78
2
◆= 3, 003
✓78
3
◆= 76, 076
✓78
4
◆= 1, 426, 425
✓78
5
◆= 21, 111, 090
✓78
6
◆= 256, 851, 595
Challenge: materials space
17
Challenge: materials space
M. J. Mehl et al., Comput. Mater. Sci. 136, S1-S828 (2017), D. Hicks et al., Comput. Mater. Sci. 161, S1-S1011 (2019)
18
Challenge: materials space
10177
M. J. Mehl et al., Comput. Mater. Sci. 136, S1-S828 (2017), D. Hicks et al., Comput. Mater. Sci. 161, S1-S1011 (2019)
• Automated computational materials discovery:
19
Automated Materials Discovery
Experimental structures Automated QM calculations
Searchable/sortable database Candidate materials
Structural prototypes
ML models
Density functional theory
http://www.aflow.org
20
AFLOW: Automatic FLOW
AFLOW
DFT-code
aflow.in
ElasticElectronic Thermal
Storage
Property calculation
Database storage
Calculation management
Input file generation
DFT-code
DFT-code
DFT-code
ERROR?
Curtarolo et al., Comput. Mater. Sci. 58, 218 (2012); Calderon et al., Comput. Mater. Sci. 108A, 233 (2015)
AFLOW source code now available under GPL 3
materials.duke.edu/AFLOW/
21
AFLOW.org: web portal
Curtarolo et al., Comput. Mater. Sci. 58, 218 (2012); Calderon et al., Comput. Mater. Sci. 108A, 233 (2015)
> 3.5 million entries
> 700 million properties
breakdown of properties
aflow.orgAFLOW
Schools and Seminars
22
AFLOW.org: seminars
• Online seminars hosted by Center for Autonomous Materials Design• Seminars take place every two weeks on Thursday afternoons• See aflow.org/seminars/ for more details
Seminar sign-up link
23
AFLOW.org: applications
Curtarolo et al., Comput. Mater. Sci. 58, 218 (2012); Calderon et al., Comput. Mater. Sci. 108A, 233 (2015)
aflow.org
AFLOW.org: search page
aflow.org/search/
Element selection
24
Number of species
Element group
selection
R. H. Taylor et al., Comput. Mater. Sci. 93, 178-192 (2014)
Logic operations
AFLOW.org: search page
aflow.org/search/
Slide to Property
Filters
25R. H. Taylor et al., Comput. Mater. Sci. 93, 178-192 (2014)
Bring property
filters onto search page
AFLOW.org: search page
aflow.org/search/ 26R. H. Taylor et al., Comput. Mater. Sci. 93, 178-192 (2014)
Property Filters
AFLOW.org: search page
aflow.org/search/
Number of species=2
27
Added filter for
Egap: electronic band bap
Selected Elements
R. H. Taylor et al., Comput. Mater. Sci. 93, 178-192 (2014)
Check box to restrict
search results to specific
value range
Select “Add” to add filter
28
AFLOW.org: search page
Exercises:1. Use the advanced search functionality to find the band gap for SiC in the
zincblende structure (space group number 216).
2. Use the advanced search functionality to find the materials containing Sn but not Pb with band gaps between 1 eV and 3 eV in the ICSD catalog of the AFLOW database. How many results are returned?
Alternative server: aflowlib.duke.edu/search/ui/
R. H. Taylor et al., Comput. Mater. Sci. 93, 178-192 (2014), F. Rose et al., Comput. Mater. Sci. 137, 362 (2017)
29
Objective 1: Stability screening
elements + arrangements
Thermodynamic stability
1 0.8 0.6 0.4 0.2 0100
50
0
≠50
≠100
≠150
≠200
≠250
Zr
10.80.60.40.20
100
50
0
≠50
≠100
≠150
≠200
≠250Cu
form
ation
enth
alpy
(meV
/ato
m)
Cu 5
Zr
Cu 2
Zr
Cu 1
0Zr 7CuZ
r 2
aflow.org
C. Oses et al., J. Chem. Inf. Model. 58(12), 2477-2490 (2018)
30
Objective 1: Stability screening
elements + arrangements
Mechanical stability
Mo6PbS8 Mo6NdS8
31
Objective 2: Design rule discovery
A
BX3
ABX3 [X = F, O] ABF3 ABO3
/ B / A
Example 2: thermal conductivity in perovskites
Example 1: superconducting critical temperature
A. van Roekeghem, J. Carrete, C. Oses, S. Curtarolo, and N. Mingo,
Phys. Rev. X 6, 041061 (2016).
O. Isayev, D. Fourches, E. N. Muratov, C. Oses, K. Rasch, A. Tropsha, and S. Curtarolo,
Chem. Mater. 27(3), 735-743 (2014).
TC " TC #
32
Objective 3: Machine learning modelscrystalstructure
Voronoitessella/onandneighborssearch
infiniteperiodicgraphconstruc/onandpropertylabeling
nodes(atoms)
decomposi/ontofragments
edges(bonds)
pathfragmentsoflengthl,l=2,3,…
circularfragments(polyhedrons)
a b c
d
4
crystalstructure
ElectronicProper1es Thermo-MechanicalProper1es
metalorinsulator?
no EBG
{EBG � R :
EBG > 0}
bandgapenergy
predic4on
bulkmodulus(VRH)predic4on
{X � R}
yes
no
classifica4onmodel
regressionmodel
regressionmodels
FIG. 2. Outline of the modeling work-flow. ML models are represented by orange diamonds. Target properties predictedby these models are highlighted in green.
structure of the material and determine the atomic con-nectivity within it. In general, atomic connectivity isnot a trivial property to determine within materials.Not only must we consider the potential bonding dis-tances among the atoms, but also whether the topologyof nearby atoms allows for bonding. Therefore, we haveemployed a computational geometry approach to parti-tion the crystal structure (Figure 1a) into atom-centeredVoronoi-Dirichlet polyhedra [59–62] (Figure 1b). Thispartitioning scheme was found to be invaluable in thetopological analysis of metal organic frameworks (MOF),molecules, and inorganic crystals [63, 64]. Connectivitybetween atoms is established by satisfying two criteria:(i) the atoms must share a Voronoi face (perpendicu-lar bisector between neighboring atoms), and (ii) theinteratomic distance must be shorter than the sum ofthe Cordero covalent radii [65] to within a 0.25 A tol-erance. Here, we consider only strong interatomic in-teractions such as covalent, ionic, and metallic bonding,ignoring van der Waals interactions. Due to the ambigu-ity within materials, the bond order (single/double/triplebond classification) is not considered. Taken together,the Voronoi centers that share a Voronoi face and arewithin the sum of their covalent radii form a three-dimensional graph defining the connectivity within thematerial.
In the final steps of the PLMF construction, the fullgraph and corresponding adjacency matrix (Figure 1c)are constructed from the total list of connections. Theadjacency matrix A of a simple graph (material) with n
vertices (atoms) is a square matrix (n ⇥ n) with entriesaij = 1 if atom i is connected to atom j, and aij = 0 oth-erwise. This adjacency matrix reflects the global topol-
ogy for a given system, including interatomic bonds andcontacts within the crystal. The full graph is partitionedinto smaller subgraphs, corresponding to individual frag-ments (Figure 1d). While there are several subgraphs toconsider in general, we restrict the length l to a maximumof three, where l is the largest number of consecutive,non-repetitive edges in the subgraph. This restrictionserves to curb the complexity of the final descriptor vec-tor. In particular, we consider two types of fragments.Path fragments are subgraphs of at most l = 3 that en-code any linear strand of up to four atoms. Only theshortest paths between atoms are considered. Circularfragments are subgraphs of l = 2 that encode the firstshell of nearest neighbor atoms. In this context, circularfragments represent coordination polyhedra, or clustersof atoms with anion/cation centers each surrounded bya set of its respective counter ion. Coordination polyhe-dra are used extensively in crystallography and mineral-ogy [66].
Property labeling. PLMFs are di↵erentiated bylocal (standard atomic) reference properties [57], whichinclude: (i) general properties: the Mendeleev group andperiod numbers, number of valence electrons (NV); (ii)measured properties [57]: atomic mass, electron a�nity(EA), thermal conductivity (�), heat capacity (C), en-thalpies of atomization (�Hat), fusion (�Hfusion), and va-porization, first three ionization potentials (IP1,2,3); and(iii) derived properties: e↵ective atomic charge (Ze↵),molar volume (Vmolar), chemical hardness (⌘) [57, 67],covalent (rcov) [65], absolute [68], and van der Waalsradii [57], electronegativity (�), and polarizability. Wealso combine pairs of properties in the form of their mul-tiplication and ratio, as well as include the property value
33
AFLOW: Installation
• AFLOW can be used on Linux, Mac, and Windows
• Windows requires additional software:– Windows Subsystem for Linux (Windows 10)– Cygwin (other Windows versions)
• For instructions and installation scripts, visit:
Curtarolo et al., Comput. Mater. Sci. 58, 218 (2012); Curtarolo et al., Comput. Mater. Sci. 58, 227 (2012)
http://aflow.org/install-aflow/
34
Outline: Day 1• Keynote address: To mix or not to mix? Synthesizability
entropy-descriptors and the controversial role of vibrations in the stability of high-entropy ceramics (S. Curtarolo)
• Session 1: DFT+AFLOW (C. Oses & D. Hicks)• VASP input files • Energy calculations and k-point convergence • Primitive and conventional cell representations • Spin-polarized calculations • Density of states and band structure calculations • AFLOW input file: aflow.in• Configuration file: aflow.rc• Managing ab-initio calculations
35
Outline: Day 2• Session 2: Symmetry (D. Hicks)• AFLOW-SYM algorithm • AFLOW-SYM output files • AFLOW-SYM command line interface • AFLOW-SYM Python wrapper • Brillouin zone and AFLOW Standard k-path • AFLOW-SYM via AFLOW-Online
• Session 3: Structure prototyping (D. Hicks)• Resolving material/structural prototypes • AFLOW Crystal Prototypes (encyclopedia) • AFLOW prototype generation
36
Outline: Day 3• Session 4: Thermodynamic stability (C. Oses)
• Identifying stable vs. unstable structures • Applications: Rare-earth free Heusler magnets • AFLOW-CHULL
• Session 5: Thermodynamics of polar materials (R. Friedrich)• Thermodynamic stability of polar materials: CCE• Application: Thermodynamic stability of ternary oxides and halides • AFLOW-CCE
• Session 6: Disorder (C. Oses)• Resolving material/structural prototypes • AFLOW-POCC: Ensemble average properties • AFLOW-POCC algorithm• AFLOW-POCC via AFLOW-Online• AFLOW-POCC command line interface
37
Outline: Day 4• Session 7: Thermomechanical properties (C. Toher)• AFLOW-AEL: Elastic properties• AFLOW-AGL: Debye model
• Session 8: AFLOW.org database and APIs (M. Esters)• AFLOW.org database structure • AFLOW REST-API • AFLUX: AFLOW Search API
• Session 9: Machine Learning (C. Toher)• PLMF representation• AFLOW-ML API
38
Outline: Day 5• Session 10: AFLOW-XtalFinder (D. Hicks)
• AFLOW-Xtalfinder algorithm• AFLOW-Xtalfinder command line interface• AFLOW-Xtalfinder Python wrapper
• Session 11: Phonons in the Harmonic Approximation (M. Esters)• Introduction to phonons• AFLOW-APL: Phonon dispersions and phonon DOS• AFLOW-APL: Thermodynamic properties from phonons• AFLOW-APL: Visualizing atomic displacements
• Session 12: The Quasi-Harmonic Approximation (A. Smolyanyuk)• Introduction to the quasi-harmonic approximation• AFLOW-QHA: Calculating and visualizing thermodynamic properties