these approaches ( 47 , 48 ). The MA+cation
from the added MACl initially participates in
forming the perovskite intermediate phase at
room temperature. During the subsequent an-
nealing process, the intermediate phase is con-
verted into the perovskite crystals with the
vaporization of MACl. The slow vaporization of
MACl and thus retarded crystal growth kinetics
promote the growth of highly crystalline perov-
skite crystals. Regardless of the high molar con-
centrations of the excess MACl (~40 mol %)
in the precursor solution, the final OLHP films
contain only a small amount of residual MA+
(~5 mol %) ( 49 ), and thus their bandgaps
(~1.50 to 1.53 eV) are comparable to that of
pure FAPbI 3. This approach enables the for-
mation of phase-pure and highly crystalline
FAPbI 3 -based films, and it has been broadly
used in studies that report record-efficiency
PSCs (vertical dashed line in Fig. 2D) ( 50 , 51 ).
Relatedly, the dual combination of MACl and
CsCl was also found to effectively stabilize the
perovskite phase ( 52 ). A similar approach was
developed by using gaseous methylammonium
thiocyanate (MASCN) ( 53 ). It was reported that
MASCN is incorporated onto the surface of
hexagonal FAPbI 3 crystals to assist in the con-
version of the hexagonal phase into the cubic
polymorph. For either MACl or MASCN, the
counter anions were also reported to contrib-
ute to stabilizing the FAPbI 3 perovskite phase.
Further developments may elucidate the dif-
ferent roles of the counter anions as well as
explore alternative cations or counter anions
that can impart additional functionalities.
The phase energetics of materials is gov-
erned by the Gibbs free energy of the phase,
which is in turn determined by the summa-
tionofthebulkandsurfacefreeenergies.
Therefore, the phase conversion of metastableOLHP phases can be modulated by surface-
energy manipulation. For example, it was dem-
onstrated that the metastable CsPbI 3 perovskite
phase can be stabilized by synthesizing nano-
sized crystals where the surface-energy contri-
bution becomes dominant compared with that
of its bulk counterpart ( 54 ). Such an approach
was found to be also valid for FAPbI 3 , where
nanocrystalline FAPbI 3 demonstrated a supe-
rior phase stability compared with its corre-
sponding bulk crystals ( 55 ). Surface-energy
modulation can also be achieved by adopting
surface-functionalizing oversized A cations
( 56 – 58 ). Bulky ammonium cations, such as
phenethylammonium (PEA+), butylammonium
(BA+), or isopropylammonium, are spontane-
ously expelled to the grain boundaries and
surfaces during perovskite crystallization be-
cause of their inability to incorporate into the
crystalline lattice bulk. The bulky cations are
free from steric constraints at the surface and
thus can coordinate with the surface A site by
bondingwiththesurfaceBX 64 – .Suchsurface
functionalization was suggested to modulate
the crystal surface energy to stabilize the cubic
perovskite phase. This approach allows for the
stabilization of metastable OLHP phases with-
out changing its bulk composition and inher-
ent bandgap.
The phase energetics of OLHPs were also
suggested to be dependent on the hydrogen-
bond number and bond strength between the
A cation and surrounding inorganic framework
( 5 ). This is illustrated by using the divalent
methylenediammonium (MDA2+)cationthat
canformagreaternumberofhydrogenbonds
with the BX 64 – lattice than FA+( 50 ). Compared
with MA+, lower concentrations of MDA2+are
required to stabilize the cubic phase of FAPbI 3 ,
even though the ionic radius of MDA2+is
slightly oversized compared with that of FA+.
The incorporation of MDACl 2 may generate
FA vacancy and interstitial Cl defects to main-
tain charge neutrality but, importantly, with-
out forming deep trap states. It is noteworthy
that the PSC performance and stability im-
proved regardless of the generated additional
defects, but further investigations may be nec-
essary to decouple their full consequences.
Nevertheless, incorporation of MDA2+inad-
vertently induces tensile lattice strain as a
result of its large cation radius. A direct cor-
relation between residual lattice strain and a
detrimental effect on the OLHP charge carrier
dynamics has been reported ( 59 ). To com-
pensate for the residual strain, it was fur-
ther reported that the smaller Cs+cation can
be incorporated in equimolar amounts with
MDA2+(Fig. 3B) ( 51 ). Tensile strain relaxation
by Cs+incorporation resultingly increased
the carrier lifetime owing to a reduced defect
concentration in the film. The case of MDA2+
illustrates the possibility of expanding into
higher-valency cations for use in OLHPs, althoughLeeet al.,Science 375 , eabj1186 (2022) 25 February 2022 4 of 10
Fig. 3. Stabilization of the cubic FAPbI 3 phase by A cation engineering.(A) Schematics illustrating
the metastability of the cubica-FAPbI 3 phase and its corresponding free-energy diagram (left) and
approaches to stabilize the cubica-FAPbI 3 phase by tuning the bulk or surface free energy (right). R.T.,
room temperature. (B) Schematic illustration of lattice strain engineering by using MDA2+and Cs+
cations. Adapted with permission from ( 51 ). (C) Schematics showing the free-energy diagrams at different
temperatures and refined structures of FAPbI 3 at the corresponding temperatures. At temperatures lower
than the phase-transition temperature (Tc), the FA+cation is oriented within the hexagonal crystal structure,
whereas its orientation becomes isotropic (random) in the cubic crystal structure at temperatures higher
thanTc, contributing to the entropy of the system. Adapted with permission from ( 12 ).
RESEARCH | REVIEW