A Map of the Inorganic Ternary Metal...
1
A B N 2 Depth of A-N binary hull Thermochemical competition w/ binaries Electronic Interaction of A and B in ternary New Stable Ternary Space Experimentally Known in ICSD Stabilizable ΔμN 2 < 1 eV/N Metastable vs. Binaries Stable Metastable vs. Elements A Map of the Inorganic Ternary Metal Nitrides Wenhao Sun 1 , Christopher Bartel 2 , Elisabetta Arca 3 , Sage Bauers 3 , Bethany Matthews 4 , Bernardo Orvañanos 1 , Janet Tate 4 , Bor-Rong Chen, 5 Laura Schelhas, 5 Michael F. Toney, 5 William Tumas 3 , Andriy Zakutayev 3 , Stephan Lany 3 , Aaron Holder 3 , Gerbrand Ceder 1 Nitrides are an exciting class of solid-state compounds, with unique structures and compelling properties for next-generation functional materials. Unfortunately, nitrides are rare in nature and difficult to synthesize. There are only ~300 unique ternary metal nitrides in the ICSD, in contrast to over 4,000 ternary metal oxides. We employ a suite of high-throughput computational materials discovery tools to broadly survey stability relationships across the inorganic ternary metal nitrides. By clustering the ternary nitrides into chemical families with distinct stability and metastability, we produce a map that visualizes both promising and risky compositions for further exploratory synthesis. We reveal the fascinating interplay between chemistry, nitride composition, and solid-state bonding in governing the stability of ternary nitrides. CONSTRUCTION OF THE MAP Multi-factor Clustering: - Formation Energies - Stable/Metastable/Unstable - Chemical group Data-Type Continuous Nominal Ordinal Multi-Feature Hierarchical Agglomeration reveals underlying relationships and chemical families across the ternary metal nitrides chemical space Formation energy stable compounds Formation energy of all compounds Metastable / Unstable Gower Metric: 1 2 1 2 M -M i M M i i G W d − = Linear combination of distance metrics with user-assigned weights ICSD Predicted SYNTHESIS OF NEW NITRIDES Data-mined ionic substitution suggests new ternary nitride structures based on rational ionic substitutions on known phases “Redox Mediated Stability in Zinc Molybdenum Nitrides Elisabetta Arca et al., JACS, (2018) Ternary M 1 -M 2 -N Spaces Previously Known Newly Predicted Portion of map Stable Ternary Systems (Blue) 189 92 281 (26%) - Stable Alkali-Metal-Nitride Systems 124 76 200 - Stable Metal-Metal-Nitride Systems 65 16 81 Metastable vs. Stable Binaries, ΔH f < 0 (Green) 98 322 420 (44%) Metastable vs. Elements, ΔH f > 0 (Red) 20 241 261 (27%) Ternary A x B y N z Phases Previously Known Newly Predicted Total Number Stable Ternary Phases 213 203 416 Metastable, ΔE Hull < 70 meV/atom 39 36 75 Metastable, ΔE Hull < 200 meV/atom 85 175 260 Metastable: Stabilizable ΔμN < +1 eV/N 3 92 95 How does chemistry drive the stability of the ternary metal nitrides? Metallicity : COHP-computed A-A, B-B, A-B and bonds Covalency : COHP-computed A-N, B-N, and N-N bonds Ionicity : Ratio of Net Atomic Charges (NAC) over the Summed Bond Order (SBO) obtained from the Density Derived Electrostatic and Chemical (DDEC) Depth of the Binary Hull (eV/atom) Stable Ternary Spaces Sc Ti Zr Y Hf Al B Ta Si V Nb Mg Ba Ca Li Mo K Rb W Cr Ga Fe Ge Re Mn Co Zn Na C Ni In Os Sn Ru Sb Pd Ir Pt Bi S Cs Rh Pb Te Se Ag Sr Au Cd Cu M-N bonds are weak M-N bonds are very strong Nitrides unlikely to form in this chemistry Binaries are too stable, Ternaries likely to decompose Intermediate Good to form ternaries nitrides with Thermochemical competition with binary nitrides Electronic Interaction of A and B in ternary Inductive effect, electropositive metal A donates electron density to B-N covalent bond, oxidizing the more electronegative metal, which can lead to nitrogen-rich nitrides. Reductive effect, nitrogen oxidation or nitrogen deficiency provides electrons to Me-Me bonds, reducing the metals and increasing metallicity. ‘Depth of the Binary Hull’ is the energy of lowest-energy binary nitride in a Me-N space, and serves as a proxy for the strength of the solid-state metal- nitrogen bond When plotting stable ternary spaces vs. depth of binary hull, a ‘volcano’ plot emerges, where ternaries are unstable against binaries with deep hulls, or do not form at all if Me-N bonding is weak in the first place. Alkali and Alkaline Earth form stable ternaries more readily than elements with similar hull depth. ΔE Hull = E A-B-N – E A-N – E B-N Stable ternary nitrides lie on the convex hull, shown on the right. The energy of a ternary nitride can be expressed relative to the hull as We explain ternary stability from 1.) stability of the competing binaries, and 2.) quantum chemical solid-state bonding in the ternary. Molecular Orbital Theory Solid-State Bonding Sr 6 Ga 5 N Sr 4 In 2 N Ca 2 InN Na 3 IrN 2 MgTa 2 N 3 KPtN N-Rich 243 compounds N-Poor 56 compounds N-Poor 35 compounds Ionicity M-M Metallicity M-N Covalency M-M Metallicity M-N Covalency N-Rich 19 compounds Ionicity Metal-Metal-N Alkali-Metal-N Reduced Oxidized Reduced Oxidized Reduced Oxidized Reduced Oxidized We conduct a large-scale data-mining analysis of chemistry, nitride composition, and solid-state bonding in governing the stability of 353 ternary metal nitrides. Cross-validated by “predicting” known nitrides. ~80% chance of recovery. Train DMSP algorithm on known Pnictides We extend the 300 nitrides to 6000 ternary nitrides DMSP offers an efficient way to probe both nitride structures and stoichiometries over a broad compositional space. Clustering on individual features gives cluster maps with limited scope. We develop a multi-feature distance metric that captures high-level stability trends and local-scale chemical relationships We evaluate phase stability of DMSP-predicted structures using tools within the Pymatgen and Materials Project infrastructure DMSP may not predict the ground-state structure if a prototype structure does not exist. However, identification of a ternary with ΔH f < 0 in an otherwise empty space highlights that space for further investigations of ground-state structures. We use reactive sputtering to synthesize 10+ new Zn- and Mg- based ternary nitrides Mg-based ternary nitrides tend to crystallize in rocksalt-derived structures, whereas Zn-based ternary nitrides form in a Wurtzite- derived structure New ternary nitrides are all semiconductors with E gap ~ 1-2 eV By changing the Zn:Mo ratio in Zn- Mo-N ternaries, we can modulate the molybdenum oxidation state from Mo 4+ to Mo 6+ , turning metallic ZnMoN 2 into a wide- bandgap Zn 3 MoN 4 semiconductor This redox-mediated behavior suggests a new class of II x -TM-N semiconductors, with tunable optoelectronic properties.
Transcript of A Map of the Inorganic Ternary Metal...
A
B
N2
Depth of A-N binary hull
Thermochemical competitionw/ binaries
Electronic Interaction of A and B in ternary
New StableTernary Space
ExperimentallyKnown in ICSD
StabilizableΔμN2 < 1 eV/N Metastable vs.
BinariesStable Metastable vs.
Elements
A Map of the Inorganic Ternary Metal NitridesWenhao Sun1, Christopher Bartel2, Elisabetta Arca3, Sage Bauers3, Bethany Matthews4, Bernardo Orvañanos1, Janet Tate4, Bor-Rong Chen,5
Laura Schelhas,5 Michael F. Toney,5 William Tumas3, Andriy Zakutayev3, Stephan Lany3, Aaron Holder3, Gerbrand Ceder1
Nitrides are an exciting class of solid-state compounds, with unique structures and compelling properties for next-generation functional materials.
Unfortunately, nitrides are rare in nature and difficult to synthesize. There are only ~300 unique ternary metal nitrides in the ICSD, in contrast to over 4,000 ternary metal oxides.
We employ a suite of high-throughput computational materials discovery tools to broadly survey stability relationships across the inorganic ternary metal nitrides.
By clustering the ternary nitrides into chemical families with distinct stability and metastability, we produce a map that visualizes both promising and risky compositions for further exploratory synthesis.
We reveal the fascinating interplay between chemistry, nitride composition, and solid-state bonding in governing the stability of ternary nitrides.
CONSTRUCTION OF THE MAP
Multi-factor Clustering: - Formation Energies- Stable/Metastable/Unstable- Chemical group
Data-TypeContinuousNominalOrdinal
Multi-Feature Hierarchical Agglomeration reveals underlying relationships and chemical families across the ternary metal nitrides chemical space
Formation energy stable compounds
Formation energy of all compounds
Metastable/ Unstable
Gower Metric:
1 2 1 2M -M
i
M M i
i
G W d− = Linear combination of distance metricswith user-assigned weights
ICSD Predicted
SYNTHESIS OF NEW NITRIDES
Data-mined ionic substitution suggests new ternary nitride structures based on rational ionic substitutions on known phases
“Redox Mediated Stability in Zinc Molybdenum NitridesElisabetta Arca et al., JACS, (2018)
Ternary M1-M2-N Spaces Previously Known Newly PredictedPortion of
map
Stable Ternary Systems (Blue) 189 92 281 (26%)- Stable Alkali-Metal-Nitride Systems 124 76 200
- Stable Metal-Metal-Nitride Systems 65 16 81
Metastable vs. Stable Binaries, ΔHf < 0 (Green) 98 322 420 (44%)
Metastable vs. Elements, ΔHf > 0 (Red) 20 241 261 (27%)
Ternary AxByNz Phases Previously Known Newly Predicted Total Number
Stable Ternary Phases 213 203 416
Metastable, ΔEHull < 70 meV/atom 39 36 75
Metastable, ΔEHull < 200 meV/atom 85 175 260
Metastable: Stabilizable ΔμN < +1 eV/N 3 92 95
How does chemistry drive the stability of the ternary metal nitrides?
Metallicity: COHP-computed A-A, B-B, A-B and bonds
Covalency: COHP-computed A-N, B-N, and N-N bonds
Ionicity: Ratio of Net Atomic Charges (NAC) over the Summed Bond Order (SBO) obtained from the Density Derived Electrostatic and Chemical (DDEC)
Depth of the Binary Hull (eV/atom)
Stab
le T
ern
ary
Spac
es
ScTi
Zr
Y
Hf
AlB
Ta
SiV
Nb
Mg
Ba
CaLi
Mo K
RbW
Cr
Ga
FeGe
Re
MnCo
Zn Na
C
Ni
InOs
Sn
RuSb
Pd
IrPt
Bi
S
Cs
RhPb
TeSe
Ag
Sr
Au
CdCu
M-N bonds are weakM-N bonds are very strong
Nitrides unlikely to form in this chemistry
Binaries are too stable,Ternaries likely to decompose
Intermediate
Good to form ternaries nitrides with
Thermochemical competition with binary nitrides
Electronic Interaction of A and B in ternary
Inductive effect, electropositive metal Adonates electron density to B-N covalent bond, oxidizing the more electronegative metal, which can lead to nitrogen-rich nitrides.
Reductive effect, nitrogen oxidation or nitrogen deficiency provides electrons to Me-Me bonds, reducing the metals and increasing metallicity.
‘Depth of the Binary Hull’ is the energy of lowest-energy binary nitride in a Me-N space, and serves as a proxy for the strength of the solid-state metal-nitrogen bond
When plotting stable ternary spaces vs. depth of binary hull, a ‘volcano’ plot emerges, where ternaries are unstable against binaries with deep hulls, or do not form at all if Me-N bonding is weak in the first place.
Alkali and Alkaline Earth form stable ternaries more readily than elements
with similar hull depth.
ΔEHull = EA-B-N – EA-N – EB-N
Stable ternary nitrides lie on the convex hull, shown on the right. The energy of a ternary nitride can be expressed relative to the hull as
We explain ternary stability from 1.) stability of the competing binaries, and 2.) quantum chemical solid-state bonding in the ternary.
Molecular Orbital Theory
Solid-State Bonding Sr6Ga5N
Sr4In2N
Ca2InN
Na3IrN2
MgTa2N3
KPtN
N-Rich243 compounds
N-Poor56 compounds
N-Poor35 compounds
Ionicity
M-MMetallicity
M-N Covalency
M-MMetallicity
M-N Covalency
N-Rich19 compounds
Ionicity
Metal-Metal-NAlkali-Metal-N
Red
uce
dO
xid
ized
Red
uce
dO
xid
ized
Reduced OxidizedReduced Oxidized
We conduct a large-scale data-mining analysis of chemistry, nitride composition, and solid-state bonding in governing the stability of 353 ternary metal nitrides.
Cross-validated by “predicting” known nitrides. ~80% chance of recovery.
Train DMSP algorithm on
known Pnictides
We extend the 300 nitrides to 6000 ternary nitrides
DMSP offers an efficient way to probe both nitride structures and stoichiometries over a broad compositional space.
Clustering on individual features gives cluster maps with limited scope.
We develop a multi-feature distance metric that captures high-level stability trends and local-scale chemical relationships
We evaluate phase stability of DMSP-predicted structures using tools within the Pymatgen and Materials Project infrastructure
DMSP may not predict the ground-state structure if a prototype structure does not exist. However, identification of a ternary with ΔHf < 0 in an otherwise empty space highlights that space for further investigations of ground-state structures.
We use reactive sputtering to synthesize 10+ new Zn- and Mg- based ternary nitrides
Mg-based ternary nitrides tend to crystallize in rocksalt-derived structures, whereas Zn-based ternary nitrides form in a Wurtzite-derived structure
New ternary nitrides are all semiconductors with Egap ~ 1-2 eV
By changing the Zn:Mo ratio in Zn-Mo-N ternaries, we can modulate the molybdenum oxidation state from Mo4+ to Mo6+, turning metallic ZnMoN2 into a wide-bandgap Zn3MoN4 semiconductor
This redox-mediated behavior suggests a new class of IIx-TM-N semiconductors, with tunable optoelectronic properties.
![nP q arXiv:1205.5252v4 [math.NT] 30 Dec 2013 · 2013-12-31 · arXiv:1205.5252v4 [math.NT] 30 Dec 2013 MINOR ARCS FOR GOLDBACH’S PROBLEM H. A. HELFGOTT Abstract. The ternary Goldbach](https://static.fdocument.org/doc/165x107/5e68df189e8aca31703bbe63/np-q-arxiv12055252v4-mathnt-30-dec-2013-2013-12-31-arxiv12055252v4-mathnt.jpg)
