1
In situ growth of Nanoparticles through
Control of Nonstoichiometry
Authors: Dragos Neagu
1
*, George Tsekouras
1,2
, David N. Miller
1
, Hervé Ménard
3
and John
T.S. Irvine
1
*
Affiliations:
1
University of St Andrews, St. Andrews, KY16 9ST, Scotland, United Kingdom.
G.T. formerly at address
1
, currently at:
2
Laboratory for High Performance Ceramics, Swiss Federal Laboratories for Materials
Science and Technology, Überlandstrasse 129, 8600 Dübendorf, Switzerland
3
Sasol Technology (UK) Ltd. St. Andrews, KY16 9ST, Scotland, United Kingdom.
*Correspondence to: dn67@st-andrews.ac.uk, jtsi@st-andrews.ac.uk
Abstract
Surfaces decorated with uniformly dispersed catalytically active nanoparticles play a key role
in many fields including renewable energy and catalysis. These structures are typically
prepared by deposition techniques, but alternatively they could be made by growing the
nanoparticles in situ directly from the (porous) backbone support. Here we demonstrate that
growing nano-size phases from perovskites can be controlled through judicious choice of
composition, particularly by tuning deviations from the ideal ABO
3
stoichiometry. This non-
stoichiometry facilitates a change in equilibrium position to make particle exsolution much
more dynamic, enabling the preparation of compositionally diverse nanoparticles (i.e.
metallic, oxides, or mixtures) and seems to afford unprecedented control over particle size,
distribution and surface anchorage. The phenomenon is also shown to be strongly influenced
2
by surface reorganisation characteristics. The concept exemplified here may serve in the
design and development of more sophisticated oxide materials with advanced functionality
across a range of possible domains of application.
Introduction
The development of tailored functional materials consisting of catalytic nanoparticles
dispersed on external surfaces or on the inner surface of porous crystals is of key importance
in many fields including catalysis, photocatalysis and energy conversion and storage (e.g.
fuel cells, electrolysis cells, batteries). Generally, both types of microstructure are produced
through deposition techniques (e.g. physical vapor deposition
1,2
or chemical impregnation
3,4
,
respectively), in which the catalysts or the catalyst precursors are attached onto the surface
during a thermal treatment. While these approaches are widely applied, they offer limited
control over the size, distribution and anchorage of the deposited species, not only during
preparation, but also during ageing, and may be time consuming and costly.
When perovskite oxides (ABO
3
) are employed as supporting frameworks it has been shown
that certain catalysts can be incorporated as cations on the B-site of the perovskite lattice
under oxidizing conditions and partly exsolved as nanoparticles upon subsequent reduction,
thus opening the possibility of in situ growth of catalysts
5
(Figure 1a). As compared to
traditional deposition techniques, this process has been shown to produce finer and better
distributed catalyst nanoparticles, is more time- and cost-effective (that is, it does not require
multiple “deposition” steps or expensive precursors), its reversibility means that catalyst
agglomeration may be avoided through re-oxidation, thus greatly enhancing the lifetime of
the catalyst
5,6
.
So far, this concept has only been demonstrated for A to B stoichiometric perovskites (A/B =
1) and only for a limited number of easily reducible, catalytically active cations (Ni
2+
, Ru
2+
,
3
Rh
4+
, Pd
4+
and Pt
4+
– see Supplementary Figure S1), because the exsolution phenomenon is
believed to be exclusively driven by the ease with which these cations reduce to metals
6–10
.
Moreover, in the majority of these systems, exsolutions occur preferentially within the bulk
rather than on the surface, rendering most of the nano-particles inaccessible for catalysis and
thus decreasing the overall effectiveness of the approach
11
. Recently we found that harder-to-
reduce cations can also be exsolved and, additionally, exsolutions emerge preferentially on
the surface when highly A-site deficient perovskites (A/B < 1) are employed (e.g. TiO
2-δ
exsolutions from La
0.4
Sr
0.4
TiO
3
)
12
. Thus, A-site deficiency could serve as a general driving
force for triggering B-site exsolution to produce a wider range of nanoparticle compositions
with superior surface distribution and coverage. Moreover, A-site deficiency (henceforth
denoted α, in e.g. A
1-
α
BO
3
) is the natural choice for designing systems that exsolve the B-site
since in these systems exsolution acts to locally revert the perovskite towards a stable,
“defect-free”, ABO
3
stoichiometry. In initially stoichiometric compositions exsolution is
typically accompanied by undesirable A-site cation containing phases (Figure 1b)
5
Here we investigate the contribution of various factors for the in situ growth of nanoparticles
from perovskites, and find that perovskite nonstoichiometry emerges as instrumental in this
phenomenon. We also illustrate the thought process which enables one to tune
nonstoichiometry and thus tailor particle growth for a palette of relevant applications.
4
a
c
b
Figure 1 | Diagrams anticipating the key role of perovskite nonstoichiometry for in situ growth of nanoparticles.
(a) Schematic representation of in situ exsolution of catalysts from a particle. In oxidizing conditions
catalytically active species typically reside on the B-site perovskite sublattice as cations (small red dots inside
the particle). Upon reduction, B-site species closer to the surface exsolve as nanoparticles (larger red circles),
effectively decorating the outer surface of the parent perovskite particle. (b) Perovskites with A-site deficiency
are anticipated to be more suitable for promoting the exsolution of B-site species as compared to cation
stoichiometric perovskites (c) When the perovskite nonstoichiometry landscape is represented in a Cartesian
plot (x and y axes corresponding to oxygen and A-site nonstoichiometry, respectively), deficient and excess
perovskites appear to be diametrically opposed with respect to the defect-free ABO
3
, thus outlining their
observed antagonistic characteristics. In this plot, representative perovskites and perovskite-like series are
outlined in grey (e.g. the Ruddlesden-Popper series A
n+1
B
n
O
3n+1
, the oxygen deficient A
n
B
n
O
3n-1
brownmillerite
type series, or the oxygen excess A
n
B
n
O
3n+2
homologous series), while the specific compositions prepared for
this study are outlined in colored boxes (red: deficient perovskites and blue: excess perovskites).
Results and discussion
Studied systems in the context of perovskite nonstoichiometry
To provide a clear, visual distinction between the different nonstoichiometry classes, we
associate them to the quadrants of a Cartesian plot having the O-site and A-site
nonstoichiometry as x and y axis, respectively, and the defect-free ABO
3
as origin
(Figure 1c). Through this representation we emphasize the diametrically opposed position of
the deficient (A/B < 1 and/or O/B < 3) and excess perovskites (A/B > 1 and/or O/B > 3) with
respect to the stoichiometric ABO
3
, which anticipates their distinct structure and thus
contrasting defect chemistries. The ideal perovskite structure (e.g. SrTiO
3
) may be visualised
as a continous 3D network of corner-sharing BO
6
octahedra, in which A-site cations occupy
A
n+1
B
n
O
3n+1
O (+)
excess
O (-)
deficient
A (-)
deficient
A
1-α
BO
3-γ
Excess perovskites
(Intergrowths)
Deficient perovskites
(Vacancies)
A
1-α
BO
3
ABO
3
A
n
B
n
O
3n+2
A
n
B
n
O
3n-1
ABO
3-γ
A (+)
excess
A
1+α
BO
3
(La
0.3
Sr
0.7
)(Ti
0.94
M
0.06
)O
3+
M
m+
=Ti
4+
, Ni
2+
(La
0.4
Sr
0.4
)(Ti
0.94
M
0.06
)O
3-
M
m+
=Mn
2+/3+
, Fe
2+/3+
, Ni
2+
, Cu
2+
(La
x
Sr
1-3x/2
)TiO
3
(La
0.8
Ce
0.1
)(Ni
0.4
Ti
0.6
)O
3
(La
0.4+x(4-m)
Sr
0.4-x(4-m)
)(Ti
1-x
M
x
)O
3
x=0.06; M
m+
= Fe
3+
, Ni
2+
5
the resulting cubo-octahedral cavities. Deficiency is achieved by creation of A, O-site
vacancies while preserving overall octahedra connectivity
13–16
.Excess, however, is
accommodated, by alternating perovksite slabs with intergrowth regions (where the excess
species reside), thus locally disrupting the continuity of the octahedra network
(Supplementary Figure S2)
17–19
.
We exemplify the exsolution phenomenon on compositions derived from the archetype
perovskite, SrTiO
3
, and make use of the exceptional ability of the perovskite lattice to
accommodate dopants of different size (through tilting and/or deformation of the octahedra)
and charge (by adopting deficient or excess stoichiometry), in order to tailor the cation
constitution and nonstoichiometry. Thus, La
3+
was substituted for Sr
2+
in SrTiO
3
, its higher
charge compensated by extra oxygen to produce the A-site stoichiometric, oxygen excess
series La
y
Sr
1-y
TiO
3+y/2
or by A-site vacancies leading to the A-site deficient series
La
x
Sr
1-3x/2
TiO
3
. We selected the member y = 0.30 for the oxygen excess series and members
with x ≤ 0.4 for the A-site deficient series to avoid the compositional domains where oxygen
defects or A-site vacancies, respectively, start ordering
14,18,20
. We then attempted to replace
6% of the Ti
4+
ions in some of these compositions with cations of interest to be exsolved,
such as Mn
2+/3+
, Fe
2+/3+
, Ni
2+
, Cu
2+
and Ce
3+/4+
. The full list is included in Figure 1c, outlining
the position of the studied composition in the perovskite nonstoichiometry space and
structural motif class.
Accommodation of cation substitution into the pervoskite structure
Where the size of the dopant and the host were similar (Supplementary Figure S3) we
observed the formation of the perovskite phase (Figure 2a) and the size of the perovskite
(pseudocubic) unit cell shifted in proportion to the size of the dopant (Figure 2b), thus
confirming substitution for the systems La
0.4
Sr
0.4
(Mn/Ni/Fe/Cu)
x
Ti
1-x
O
3-γ
(γ = (4-m)·x/2,
where m and x are the charge and stoichiometry of the dopant, respectively). Ce
4+
proved to