Jour
nal
Name
Supplementary Information
Sabine Körbel,a,b‡ Miguel A.L. Marques,c,d and Silvana Botti∗a,d
An illustrative binary convex hull for Sn–Cl is shown in panel (a) In Tables ESI-1 to ESI-3 we list the energy above the convex hull,
of Fig. ESI-1, while the ternary phase diagram of Sn–Cl–Rb is Ehull , in eV/atom for each compound, together with the magnetic
displayed in panel (b) of Fig. ESI-1. If a compound is stable with moment µ in units of µB , the indirect and direct band gaps from
respect to decomposition, it lies by definition on the convex hull; PBE and HSE, the effective hole mass in units of the electron mass,
otherwise it lies above the hull. In the binary phase diagram we the modulus of the electric polarization |P| in µC/cm2 , and the
can observe that SnCl lies above the hull. It is therefore unstable, formation energy Ef in eV/atom, for all perovskite compounds
in this case with respect to decomposition into SnCl2 and Sn. On which are within 25 meV/per atom from the convex hull of ther-
the other hand, RbSnCl3 (here in the ilmenite crystal structure) modynamic stability.
lies on the hull of the ternary phase diagram, hence it is stable.
To some extent it is possible to find a relation between the As a side comment, the concept of the effective mass is only valid
possible value for the band-gap and the minimal difference for semiconductors, and although we obtain vanishing effective
in electronegativity between A or B and X elements, ∆ χ = hole masses for the metallic compounds, in reality of course the
min (χ(X) − χ(B), χ(X) − χ(A)). As we can see in Fig. ESI-2, band- electric conductivity of these compounds will be finite. Similarly,
gaps usually increase with ∆ χ, although values are still quite the ferroelectric polarization can only be larger than zero for ma-
scattered. The upper limit of the band gap can approximately be terials with a band gap.
described by the straight line Eg ≤ 10/3∆ χ.
Where available, data for the same composition (not necessar-
Figure ESI-3 shows the dependency of the effective hole mass
ily in the perovskite crystal structure) from databases are also
on the band gap. One may identify a trend to larger effective
included in the tables. The Materials Project database is abbrevi-
masses with larger band gaps for the compounds with direct gaps,
ated as “MP”, the Computational Materials Repository 1 as “CMR”,
whereas for the indirect band gap materials, the scattering is large
and the Open Quantum Materials Database as “OQMD”. The for-
(we can identify the direct band gap materials in the graph as
mation energies in the CMR and most of those in the OQMD are
the cases where direct and indirect band gap fall on top of each
those of the ideal cubic perovskites (space group 221). Since here
other).
we allow for structural distortions which can lower the energy,
our formation energies are often more negative (stronger cohe-
a
sion) than those of the CMR and the OQMD, whereas they usually
Institut für Festkörpertheorie und -optik, Friedrich-Schiller-Universität Jena, Max-
Wien-Platz 1, 07743 Jena, Germany, E-mail: skoerbel@uni-muenster.de are very close to those in the MP database.
b
Institut Lumière Matière, UMR5306 Université Lyon 1-CNRS, Université de Lyon, F-
69622 Villeurbanne Cedex, France
c
Institut für Physik, Martin-Luther-Universität Halle-Wittenberg, D-06099 Halle, Ger- References
many
d
European Theoretical Spectroscopy Facility
‡E-mail: silvana.botti@uni-jena.de 1 https://cmr.fysik.dtu.dk, 2015.
J
our
nal
Name,
[yea
r][
,vol
.
], 1–6 | 1
Table ESI-1 Calculated properties of ABX3 compounds: distance from the convex hull Ehull in eV/atom, magnetic moment µ per five-atom unit cell in
units of the Bohr magneton, indirect and direct band gaps from PBE and HSE06 functionals, effective hole mass m∗h in units of the electron mass,
modulus of the ferroelectric polarization |P| in the LDA, and formation energies Ef in eV/atom from this work and the MP, CMR, and OQMD databases.
New compounds are marked by a •
2| J
our
nal
Name,
[yea
r][
,vol
.
],
1–6
Table ESI-2 Continuation of Table ESI-1
J
our
nal
Name,
[yea
r][
,vol
.
], 1–6 | 3
Table ESI-3 Continuation of Table ESI-2
4| J
our
nal
Name,
[yea
r][
,vol
.
],
1–6
Table ESI-4 Continuation of Table ESI-3
Cl Sn 11
0.0 HSE indirect
HSE direct
10 10/3 ∆ χ
−0.2
Ef [eV/atom]
9
−0.4
SnCl
−0.6 Ehull 8
−0.8 7
SnCl4
SnCl2
−1.0
Eg (eV)
Cl 0.0 4
-0.3 3
SnCl4
SnCl2 -0.6 2
Ef (eV/atom)
Rb2 SnCl6
RbSnCl3 1
RbCl SnCl -0.9
0
-1.2 0.5 1.0 1.5 2.0 2.5 3.0 3.5
∆χ
-1.5
Fig. ESI-2 (Color online) band gaps of the stable perovskites vs
-1.8 electronegativity difference between cations and X anion (see text).
Rb RbSn RbSn3 Sn
Fig. ESI-1 (Color online) Panel (a): Formation energy (in units of
eV/atom) as a function of composition for the system Sn–Cl. Panel (b):
Projected formation energy (in units of eV/atom) as a function of
composition for the system Rb–Sn–Cl. The color map represents the
formation energy E f . Stable compounds (those on the convex hull) are
connected by lines. For the sake of readability, most unstable or
metastable phases are omitted.
J
our
nal
Name,
[yea
r][
,vol
.
], 1–6 | 5
20
PBE indirect
PBE direct
15
m∗h
10
0
-1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Eg (eV)
Fig. ESI-3 (Color online) Effective hole mass m∗h vs. band gap for the
stable perovskites.
6| J
our
nal
Name,
[yea
r][
,vol
.
],
1–6