+ All Categories
Home > Documents > Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC...

Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC...

Date post: 14-Jul-2020
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
6
Supporting Information Akola et al. 10.1073/pnas.1300908110 SI Text Detail of DFT Simulation Central in the CP2K method are two representations of the electron density: localized Gaussian and plane wave (GPW) basis sets. This representation allows for an efcient treatment of the electrostatic interactions and leads to a scheme that is formally linearly scaling as a function of the system size. The valence electronion interaction is based on the norm-conserving and separable pseudopotentials of the analytical form derived by Goedecker, Teter, and Hutter (1). We considered the following valence congurations: O (2s 2 2p 4 ), Ca (2s 2 2p 6 3s 2 3p 6 4s 2 ), and Al (3s 2 3p 1 ). For the Gaussian-based (localized) expansion of the KohnSham orbitals, we use a library of contracted molecularly optimized valence double-zeta plus polarization (m-DZVP) basis sets (2), and the complementary plane wave basis set has a cutoff of 600 Rydberg for electron density (this equals to 150 Ry for wave functions in standard plane wave schemes). The molecularly optimized m-DZVP functions result in highly accurate results with less computational cost as expe- rienced with the traditional basis sets that are tted to atomic properties. Together with the GPW basis set, this enables density functional theory (DFT) simulations of systems up to 1,000 atoms or more. Effective charges of individual atoms have been evaluated from electron density (3), and chemical bond orders between atomic pairs (and Mulliken charges) have been computed from the over- laps of the atomic orbital components (with a projected com- pleteness of 97.5%). Description of the DFTRMC Approach i ) Initial hard-sphere Monte Carlo (HSMC) simulations (re- verse Monte Carlo, RMC, simulation without experimental data) with a constrained fourfold coordination for Al were used for generating starting structures. ii ) These HSMC structures were subjected to standard Monte Carlo simulations, which were tted to the experimental X-ray, neutron diffraction, and extended X-ray absorption ne structure (EXAFS) data. Several RMC models were tested in parallel with preliminary DFT simulation, and the ones that gave acceptable agreement with the EXAFS data after DFT optimization were selected. The total energy of the RMC models (before DFT optimization) reduced by 0.48 electron volt (eV)/atom and 0.42 eV/atom for the 50CaO and 64 CaO glass, respectively, in comparison with the HSMC models (step i ). iii ) The RMC structures were optimized by DFT, and the result- ing structures were simulated at 300 K for 10 ps. During molecular dynamics (MD), the DFT-optimized structures undergo minor structural changes during the rst 5 ps, which corresponds to an energy decrease of 0.04 eV/atom for the nal optimized structures of both compositions. The energy decrease with respect to the previous RMC model (step ii ) is 0.58 eV/atom and 0.64 eV/atom for 50 mol % CaO (50CaO) and 64CaO, respectively. iv) The nal RMC-renement is performed with respect to the DFT structures. The nal energy differences between the DFT minima (base structures) and RMC are 0.09 and 0.06 eV/atom for 50CaO and 64CaO, respectively. It is obvious from the energetics alone that the effect of DFT simulations is considerable. The total energy reduces by 0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one should compare this also with the energy difference between the initial HSMC and RMC models, steps i and ii ). In terms of atomic structure, the DFT treatment affects the distributions of bond lengths and angles and coordination numbers, and there are differences in the longer length scale also (e.g., rings). Importantly, there are several atoms with unfavorable bonds in the initial RMC struc- ture, which dissociate upon DFT simulations and form new bonds with other atoms. Cavity Analysis A cavity analysis has been performed as described in ref. 4. The system is divided into a cubic mesh with a grid spacing of 0.20 Å, and the points farther from any atom at a given cutoff (here 2.3 and 2.5 Å) are selected and dened as cavity domains.Each domain is characterized by the point where the distance to all atoms is a maximum. If there are no maxima closer than the di- vacancy cutoff (here 2.0 Å), we locate the center of the largest sphere that can be placed inside the cavity. This point can be used for calculating partial pair distribution functions, including vacancyvacancy correlations. Around the cavity domains we con- struct cells analogous to the Voronoi polyhedra in amorphous phases (compare the WignerSeitz cell) and analyze their vol- ume distribution. 1. Goedecker S, Teter M, Hutter J (1996) Separable dual-space Gaussian pseudopotentials. Phys Rev B Condens Matter 54(3):17031710. 2. VandeVondele J, Hutter J (2007) Gaussian basis sets for accurate calculations on molecular systems in gas and condensed phases. J Chem Phys 127(11):114105114109. 3. Tang W, Sanville E, Henkelman G (2009) A grid-based Bader analysis algorithm without lattice bias. J Phys Condens Matter 21(8):084204084207. 4. Akola J, Jones RO (2007) Structural phase transitions on the nanoscale: The crucial pattern in the phase-change materials Ge2Sb2Te5 and GeTe. Phys Rev B 76(23): 235201-1235201-10. Akola et al. www.pnas.org/cgi/content/short/1300908110 1 of 6
Transcript
Page 1: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

Supporting InformationAkola et al. 10.1073/pnas.1300908110SI Text

Detail of DFT SimulationCentral in the CP2K method are two representations of theelectron density: localized Gaussian and plane wave (GPW) basissets. This representation allows for an efficient treatment of theelectrostatic interactions and leads to a scheme that is formallylinearly scaling as a function of the system size. The valenceelectron–ion interaction is based on the norm-conserving andseparable pseudopotentials of the analytical form derived byGoedecker, Teter, and Hutter (1). We considered the followingvalence configurations: O (2s22p4), Ca (2s22p63s23p64s2), andAl (3s23p1). For the Gaussian-based (localized) expansionof the Kohn–Sham orbitals, we use a library of contractedmolecularly optimized valence double-zeta plus polarization(m-DZVP) basis sets (2), and the complementary plane wavebasis set has a cutoff of 600 Rydberg for electron density (thisequals to 150 Ry for wave functions in standard plane waveschemes). The molecularly optimized m-DZVP functions resultin highly accurate results with less computational cost as expe-rienced with the traditional basis sets that are fitted to atomicproperties. Together with the GPW basis set, this enables densityfunctional theory (DFT) simulations of systems up to 1,000 atomsor more.Effective charges of individual atoms have been evaluated from

electron density (3), and chemical bond orders between atomicpairs (and Mulliken charges) have been computed from the over-laps of the atomic orbital components (with a projected com-pleteness of 97.5%).

Description of the DFT–RMC Approach

i) Initial hard-sphere Monte Carlo (HSMC) simulations (re-verse Monte Carlo, RMC, simulation without experimentaldata) with a constrained fourfold coordination for Al wereused for generating starting structures.

ii) These HSMC structures were subjected to standard MonteCarlo simulations, which were fitted to the experimentalX-ray, neutron diffraction, and extended X-ray absorption finestructure (EXAFS) data. Several RMC models were tested inparallel with preliminary DFT simulation, and the ones thatgave acceptable agreement with the EXAFS data after DFToptimization were selected. The total energy of the RMCmodels (before DFT optimization) reduced by 0.48 electronvolt (eV)/atom and 0.42 eV/atom for the 50CaO and 64 CaO

glass, respectively, in comparison with the HSMC models(step i).

iii) The RMC structures were optimized by DFT, and the result-ing structures were simulated at 300 K for 10 ps. Duringmolecular dynamics (MD), the DFT-optimized structuresundergo minor structural changes during the first 5 ps, whichcorresponds to an energy decrease of 0.04 eV/atom for thefinal optimized structures of both compositions. The energydecrease with respect to the previous RMC model (step ii) is0.58 eV/atom and 0.64 eV/atom for 50 mol % CaO (50CaO)and 64CaO, respectively.

iv) The final RMC-refinement is performed with respect to theDFT structures. The final energy differences between theDFT minima (base structures) and RMC are 0.09 and 0.06eV/atom for 50CaO and 64CaO, respectively.

It is obvious from the energetics alone that the effect of DFTsimulations is considerable. The total energy reduces by ∼0.6 eV/atom from the initial RMC models in materials where the cohesiveenergies are of the order of 5.8 eV/atom (one should comparethis also with the energy difference between the initial HSMCand RMC models, steps i and ii). In terms of atomic structure,the DFT treatment affects the distributions of bond lengths andangles and coordination numbers, and there are differences inthe longer length scale also (e.g., rings). Importantly, there areseveral atoms with unfavorable bonds in the initial RMC struc-ture, which dissociate upon DFT simulations and form new bondswith other atoms.

Cavity AnalysisA cavity analysis has been performed as described in ref. 4. Thesystem is divided into a cubic mesh with a grid spacing of 0.20 Å,and the points farther from any atom at a given cutoff (here 2.3and 2.5 Å) are selected and defined as “cavity domains.” Eachdomain is characterized by the point where the distance to allatoms is a maximum. If there are no maxima closer than the di-vacancy cutoff (here 2.0 Å), we locate the center of the largestsphere that can be placed inside the cavity. This point can beused for calculating partial pair distribution functions, includingvacancy–vacancy correlations. Around the cavity domains we con-struct cells analogous to the Voronoi polyhedra in amorphousphases (compare the Wigner–Seitz cell) and analyze their vol-ume distribution.

1. Goedecker S, Teter M, Hutter J (1996) Separable dual-space Gaussian pseudopotentials.Phys Rev B Condens Matter 54(3):1703–1710.

2. VandeVondele J, Hutter J (2007) Gaussian basis sets for accurate calculations onmolecular systems in gas and condensed phases. J Chem Phys 127(11):114105–114109.

3. Tang W, Sanville E, Henkelman G (2009) A grid-based Bader analysis algorithm withoutlattice bias. J Phys Condens Matter 21(8):084204–084207.

4. Akola J, Jones RO (2007) Structural phase transitions on the nanoscale: The crucialpattern in the phase-change materials Ge2Sb2Te5 and GeTe. Phys Rev B 76(23):235201-1–235201-10.

Akola et al. www.pnas.org/cgi/content/short/1300908110 1 of 6

Page 2: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

-4

-2

0

2

4

1050

k (Å-1)

k3χ(

k)

Fig. S1. The k3·χ(k) spectra for 50CaO glass (red) and 64CaO glass (blue) measured at Ca K edge.

3.1

3.0

2.9

2.8

2.7706560555045

CaO mol%

Den

sity

(g/

cm3 )

Fig. S2. Density of CaO–Al2O3 glasses measured by using a dry pycnometer.

Akola et al. www.pnas.org/cgi/content/short/1300908110 2 of 6

Page 3: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

-2

-1

0

1

2

20151050

Q (Å-1)

Sij(

Q)

Al-Al Al-O Al-Ca

O-O Ca-O Ca-Ca

-3

-2

-1

0

1

20151050

3

2

1

0

-1

-2

-320151050

2

1

0

-1

-220151050

-2

-1

0

1

20151050

-3

-2

-1

0

1

2

3

20151050

Fig. S3. Partial structure factors Sij(Q) for 50CaO glass (red) and 64CaO glass (blue) obtained by RMC–DFT simulation.

0.6

0.5

0.4

0.3

0.2

0.1

0.04.03.53.02.52.0

64CaO glass64CaO crystal

0.6

0.5

0.4

0.3

0.2

0.1

0.04.03.53.02.52.0

50CaO glass50CaO crystal

1.0

0.5

0.03.53.02.52.01.5

64CaO glass64CaO crystal

1.0

0.5

0.03.53.02.52.01.5

50CaO glass50CaO crystal

C

A B

Che

mic

al s

treng

th (b

ond

orde

r)

Al-O distance (Å)

Che

mic

al s

treng

th (b

ond

orde

r)

Ca-O distance (Å)

Che

mic

al s

treng

th (b

ond

orde

r)

Al-O distance (Å)

Che

mic

al s

treng

th (b

ond

orde

r)

Ca-O distance (Å)

D

Fig. S4. Scatter plot of bond orders (chemical strengths) of Al–O (A, C) and Ca–O (B, D) bonds as a function of distance for CaO–Al2O3 glass. (A and B) 50CaOand (C and D) 64CaO composition glass. The red points correspond to crystalline reference structures. For reference, the bond order of an ideal covalent singlebond is unity.

Akola et al. www.pnas.org/cgi/content/short/1300908110 3 of 6

Page 4: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

2.5

2.0

1.5

1.0

0.5

0.0180120600

Bond angle (degree)

Pro

babi

lity

(arb

. Uni

ts)

Al-Al-Al Al-O-Al O-Al-O O-O-O

O-Ca-O Ca-O-Al Ca-O-Ca Ca-Ca-Ca

2.0

1.5

1.0

0.5

0.0180120600

2.5

2.0

1.5

1.0

0.5

0.0180120600

1.5

1.0

0.5

0.0180120600

1.5

1.0

0.5

0.0180120600

2.5

2.0

1.5

1.0

0.5

0.0180120600

3.0

2.5

2.0

1.5

1.0

0.5

0.0180120600

1.5

1.0

0.5

0.0180120600

Fig. S5. Bond angle distribution for 50CaO glass (red) and 64CaO glass (blue) calculated from RMC–DFT model.

Fig. S6. The DFT–RMC optimized atomic configuration and cavities in the 50CaO (A) and 64CaO glass (B). The cavity volume is 4.4% and 2.7% with a cutoffdistance of 2.5 Å (magenta isosurface) for a test particle, and 13.0% and 10.0% with that of 2.3 Å (cyan isosurface) for 50CaO and 64CaO glass, respectively.

Akola et al. www.pnas.org/cgi/content/short/1300908110 4 of 6

Page 5: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

HOMO

HOMO -2

HOMO -1

HOMO -3

LUMO

LUMO +2

LUMO +1

LUMO +3

A

B

Fig. S7. Visualizations of the highest occupied single-particle Kohn–Sham (KS) states (A) and the lowest unoccupied KS states (B) in the 64CaO glass. The statesare spin-degenerate.

Akola et al. www.pnas.org/cgi/content/short/1300908110 5 of 6

Page 6: Supporting Information - PNAS · The total energy reduces by ∼0.6 eV/ atom from the initial RMC models in materials where the cohesive energies are of the order of 5.8 eV/atom (one

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5

Energy (eV)

DO

S (

eV-1

)

Fig. S8. Calculated electronic density of states (DOS) for the 64CaO glass. The impurity states [lowest unoccupied molecular orbital (LUMO), LUMO+1, andLUMO+2] are indicated by arrows.

Table S1. NAl–O, NO–Al, NCa–O, NO–Ca, and NO–O calculated from theDFT–RMC model for CaO−Al2O3 glasses

Samples NAl–O/NO–Al, 2.50 Å NCa–O/NO–Ca, 2.80 Å NO–O, 3.50 Å

50CaO 4.26/2.13 5.02/1.25 8.4264CaO 4.14/1.73 4.92/1.83 7.67

Table S2. Connectivity of AlOn–AlOn, AlOn–CaOx, and CaOx–CaOx

in CaO−Al2O3 glasses

Pair

50CaO 64CaO

Connectivity % Connectivity %

AlOn–AlOn Corner 87 Corner 88Edge 13 Edge 12Face 0 Face 0

AlOn–CaOx Corner 73 Corner 76Edge 25 Edge 23Face 2 Face 1

CaOx–CaOx Corner 77 Corner 75Edge 22 Edge 24Face 1 Face 1

Akola et al. www.pnas.org/cgi/content/short/1300908110 6 of 6


Recommended