Variable-Temperature Multinuclear Solid-State NMR Study of Oxide Ion Dynamics in Fluorite-Type Bismuth Vanadate and Phosphate Solid Electrolytes

Article abstract of DOI:10.1021/acs.chemmater.8b05143

Ionic conducting materials are crucial for the function of many advanced devices used in a variety of applications, such as fuel cells and gas separation membranes. Many different chemical controls, such as aliovalent doping, have been attempted to stabilize ?-Bi₂O₃, a material with exceptionally high oxide-ion conductivity which is unfortunately only stable over a narrow temperature range. In this study, we employ a multinuclear, variable-temperature NMR (VT-NMR) spectroscopy approach to characterize and measure oxide-ionic motion in the V- and P-substituted bismuth oxide materials Bi_0.913V_0.087O_1.587, Bi_0.852V_0.148O_1.648, and Bi_0.852P_0.148O_1.648, previously shown to have excellent ionic conduction properties (Kuang et al., Chem. Mater. 2012, 24, 2162; Kuang et al., Angew. Chem., Int. Ed. 2012, 51, 690). Two main ¹⁷O NMR resonances are distinguished for each material, corresponding to O in the Bi-O and V-O/P-O sublattices. Using VT measurements ranging from room temperature to 923 K, the ionic motion experienced by these different sites has then been characterized, with coalescence of the two environments in the V-substituted materials clearly indicating a conduction mechanism facilitated by exchange between the two sublattices. The lack of this coalescence in the P-substituted material indicates a different mechanism, confirmed by ¹⁷O T₁ (spin-lattice relaxation) NMR experiments to be driven purely by vacancy motion in the Bi-O sublattice. ⁵¹V and ³¹P VT-NMR experiments show high rates of tetrahedral rotation even at room temperature, increasing with heating. An additional VO₄ environment appears in ¹⁷O and ⁵¹V NMR spectra of the more highly V-substituted Bi_0.852V_0.148O_1.648, which we ascribe to differently distorted VO₄ tetrahedral units that disrupt the overall ionic motion, consistent both with line width analysis of the ¹⁷O VT-NMR spectra and experimental results of Kuang et al., showing a lower oxide-ionic conductivity in this material compared to Bi_0.913V_0.087O_1.587 (Chem. Mater. 2012, 24, 2162). This study shows that solid-state NMR is particularly well suited to understanding connections between local structural features and ionic mobility and can quantify the evolution of oxide-ion dynamics with increasing temperature.

Full text of DOI:10.1021/acs.chemmater.8b05143

Subscriber access provided by Iowa State University | Library
Article
Variable-temperature multinuclear solid-state NMR study of oxide ion
dynamics in fluorite-type bismuth vanadate and phosphate solid electrolytes
Matthew T. Dunstan, David M. Halat, Matthew L. Tate, Ivana Radosavljevic Evans, and Clare P. Grey
Chem. Mater., Just Accepted Manuscript • DOI: 10.1021/acs.chemmater.8b05143 • Publication Date (Web): 29 Jan 2019
Downloaded from http://pubs.acs.org on February 6, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
Variable-temperature multinuclear solid-state NMR
study of oxide ion dynamics in fluorite-type bismuth
vanadate and phosphate solid electrolytes
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Matthew T. Dunstan,†,¶ David M. Halat,†,¶ Matthew L. Tate, Ivana Radosavljevic
‡
‡
∗,†
Evans, and Clare P. Grey
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge, CB2 1EW,
United Kingdom, and Department of Chemistry, University Science Site, Durham University,
South Road, Durham DH1 3LE, United Kingdom
E-mail: cpg27@cam.ac.uk
Abstract
Ionic-conducting materials are crucial for the function of many advanced devices used in a
variety of applications, such as fuel cells and gas separation membranes. Many different chem-
ical controls, such as aliovalent doping, have been attempted to stabilise δ-Bi O , a material
2
3
with exceptionally high oxide ion conductivity which is unfortunately only stable over a narrow
temperature range. In this study, we employ a multinuclear, variable-temperature NMR spec-
troscopy approach to characterise and measure oxide ionic motion in the V- and P-substituted
bismuth oxide materials Bi0.913V0.087O1.587, Bi0.852V0.148O1.648 and Bi0.852P0.148O1.648, previ-
ously shown to have excellent ionic conduction properties (Kuang et al., Chem. Mater. 2012,
*
†
To whom correspondence should be addressed
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge, CB2 1EW, United Kingdom
Department of Chemistry, University Science Site, Durham University, South Road, Durham DH1 3LE, United
‡
Kingdom
¶Contributed equally to this work
1
ACS Paragon Plus Environment

Chemistry of Materials
Page 2 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4, 2162; Kuang et al., Angew. Chem. Int. Ed. 2012, 51, 690). Two main 17O NMR res-
2
onances are distinguished for each material, corresponding to O in the Bi–O and V–O/P–O
sublattices. Using variable-temperature (VT) measurements ranging from room temperature
to 923 K, the ionic motion experienced by these different sites has then been characterised,
with coalescence of the two environments in the V-substituted materials clearly indicating a
conduction mechanism facilitated by exchange between the two sublattices. The lack of this
coalescence in the P-substituted material indicates a different mechanism, confirmed by 17O
T1 (spin-lattice relaxation) NMR experiments to be driven purely by vacancy motion in the Bi–
O sublattice. 51V and 31P VT-NMR experiments show high rates of tetrahedral rotation even
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
at room temperature, increasing with heating. An additional VO environment appears in 17O
4
and 51V NMR spectra of the more highly V-substituted Bi0.852V0.148O1.648, which we ascribe
to differently distorted VO tetrahedral units that disrupt the overall ionic motion, consistent
4
17
both with linewidth analysis of the O VT-NMR spectra and experimental results of Kuang et
al. showing a lower oxide ionic conductivity in this material compared to Bi0.913V0.087O1.587
(
Chem. Mater. 2012, 24, 2162). This study shows solid-state NMR is particularly well suited
to understanding connections between local structural features and ionic mobility, and can
quantify the evolution of oxide-ion dynamics with increasing temperature.
Introduction
Anionic-conducting materials, especially oxide ion conductors, have been the focus of continued
research efforts over the past decades due to their potential use in renewable energy applications,
such as oxygen-permeable membranes and solid oxide fuel cells (SOFCs).1–3 A crucial parameter
holding back their widespread implementation is the undesirably high temperature at which they
show sufficiently high ionic conduction, which even for currently used, best-in-class materials
4
(such as yttria-stabilised zirconia, YSZ) is >923 K. There is still very much a need to discover
and develop novel ionic conducting materials that operate at lower temperatures and to characterise
the connection between structure and conductivity in order to move towards rational design of
2
ACS Paragon Plus Environment

Page 3 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
optimal materials.
Bismuth oxides and various substituted derivatives have been mined as a rich compositional
space within which to find new fast ion conductors, sparked by the extremely high oxygen ionic
conductivity values of the cubic δ phase of Bi O , which is not stable at lower temperatures due
2
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
to transformation to a monoclinic polymorph below 1003 K.5,6 Of the families of materials that
3+ 7–12
have been explored, including phases resulting from isovalent doping (rare earth Ln )
as
well as aliovalent doping (such as pentavalent P, V, Nb, and/or Ta),13–22 vanadium-substituted
bismuth oxides have shown the most promising values of ionic conductivity at relatively lower
temperatures.23,24 These materials often exhibit complex superstructures; understanding the role
the V dopant plays in enabling the superionic behaviour of the Bi–O lattice is correspondingly
complex, but is nonetheless necessary in elucidating the best way to optimise these materials for
device applications.
In the previous work of Kuang et al., two novel bismuth-rich phases in the Bi1−xV O
sys-
1.5+x
x
tem (x = 0.087, 0.095) were found to have high oxide ion conductivity between 573–773 K, com-
23
petitive with the best-performing Bi O -based materials currently known. Subsequent studies
2
3
involving ab initio molecular dynamics (AIMD) simulations found that for these materials, as well
as a related phase with x = 0.148 (also known as β-Bi V O ), distinct structural features, such
4
6
8
89
as variably-coordinated VO polyhedra, played key roles in the ionic migration pathways through
x
the material, in particular facilitating oxygen exchange between the polyhedra and the Bi–O sub-
lattice.24 The structure of the higher-substituted material Bi0.852V0.148O1.648 is shown in Figure 1,
highlighting the layers containing Bi and O interspersed with layers containing VO polyhedra.
4
This phase can be thought of as having a monoclinic, ordered fluorite superstructure where V5+
replaces some of the Bi3+ sites in δ-Bi O , with concomitant reordering of oxygen vacancies to
2
3
preserve fourfold coordination around the V atoms. Darriet et al. reported a very similar structure
in the same monoclinic C2/m space group for the P-substituted analogue Bi0
P
O1.648 (also
.852 0.148
25
reported as Bi P O ), with layers of PO tetrahedra rather than VO . At lower doping levels,
4
6 8 89
4
4
such as in the x = 0.087 (Bi0.913V0.087O1.587) phase, a long-range ordered pseudo-cubic 3 × 3 ×
3
ACS Paragon Plus Environment

Chemistry of Materials
Page 4 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
fluorite superstructure is instead adopted. In the more highly substituted, monoclinic phases, the
ordering of vacancies and dopants leads to a disruption of the underlying cubic symmetry of the
Bi O sublattice, which by comparison is essentially maintained in Bi0.913V0.087O1.587
.
2
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1: Crystallographic structure of Bi0.852V
along the [110] direction, as reported by Kuang et al. Bismuth atoms are shown in purple, oxygen
atoms in red, and vanadium atoms (and VO tetrahedral units) in brown. The monoclinic (space
group C2/m) structure comprises a characteristic stacking of two layers containing VO tetrahedra
O
(also known as β-Bi V O ), shown
0.148 1.648 46 89
8
2
4
¯
4
4
(
Bi V O ) and one bismuth oxide rock salt layer (Bi O ) containing only OBi units. All VO
14 31 18 27 4 4
4
groups are isolated.
In this work, we report a comprehensive experimental study of the oxide-ion dynamics of both
Bi0.913V0.087O1.587 and Bi0.852V0.148O1.648 using multinuclear variable-temperature solid state nu-
clear magnetic resonance (VT-NMR), and compare the conduction mechanisms of these phases
1
7
to the P-substituted analogue Bi0
P
O1.648. O VT-NMR experiments have previously been
.852 0.148
shown to be particularly well-suited to the elucidation of ion conduction mechanisms in related ma-
2
6
terials such as the Aurivillius-type layered α-Bi V O , γ-Bi V Ti O
and the fluorite-
4
2
11
4
1.7 0.3 10.85
2
7
related Bi Mo O . These VT-NMR experiments can distinguish amongst crystallographic
26
10 69
4
ACS Paragon Plus Environment

Page 5 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
oxygen environments present in the phases as well as their corresponding motional processes
and rates. In this study, we have used 17O, 51V and 31P NMR to characterise the mechanisms
of, and activation energy values for, ionic motion in Bi0.913V0.087O1.587, Bi0.852V0.148O1.648, and
Bi0
P
O1.648. We find that oxide-ion exchange between the Bi–O and V–O sublattices occurs
.852 0.148
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
with increasing temperature for both Bi0.913V0.087O1.587 and Bi0.852V0.148O1.648, whereas no evi-
dence of exchange between the sublattices is observed for Bi0 O1.648. We also find that the
P
.852 0.148
concentration of dopant is critical for maximising the rate of conduction, with higher dopant con-
centrations leading to the formation of VO polyhedra that do not participate in ionic conduction
x
pathways and instead hinder the overall movement of oxide ions through the structure. Further-
more, the use of dopant atoms that are unable to support variable O coordination (e.g., P) hinders
exchange between the two sublattices; this finding helps to explain previous reports of oxide-ion
conductivity about half an order of magnitude lower for the P-substituted phase relative to the
24,28
β-Bi V O material over the same temperature range.
4
6
8
89
Experimental Methods
17
Synthesis and O-enrichment
Samples of Bi0.913V0.087O1.587, Bi0.852V0.148O1.648 and Bi
P
O1.648 were synthesised ac-
0.852 0.148
cording to the previous literature.23,24,28 Polycrystalline samples were synthesised by a conven-
tional solid state route using Bi O (99.9%, Sigma-Aldrich), V O (99.99%, Sigma-Aldrich) and
2
3
2
5
(NH )H PO (≥98%, Sigma-Aldrich). Stoichiometric quantities of the starting materials were
4
2
4
ground under isopropanol and heated at 700 °C, 750 °C, 800 °C and finally at 825 °C (for the
vanadates) and 850 °C (for the phosphate), for 12 h at each temperature (using heating and cool-
ing rates of 5 °C/min) with intermediate grinding. Sample purity was confirmed by laboratory
powder X-ray diffraction (XRD) carried out with a Bruker D8 Advance diffractometer using Cu
Ka radiation, via Pawley and Rietveld fitting implemented in Topas Academic software.29–31 17O-
enriched samples were obtained by heating as-synthesised powders to 900 °C under an atmosphere
5
ACS Paragon Plus Environment

Chemistry of Materials
Page 6 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
17
of 70% O (Cambridge Isotope Laboratories, USA; used as received) in a Pt crucible inside a
2
sealed quartz tube for 24 h. Samples were cooled slowly from the enrichment temperature to
17
17
maximise O uptake. Sample purity after O-enrichment was confirmed using room-temperature
X-ray powder diffraction (XRD) data collected on a Panalytical X’Pert Pro diffractometer using
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Cu K radiation (data shown in Supporting Information); the reflections could be indexed to a cu-
α
bic space group for Bi0.852V0.148O1.648, and a monoclinic space group for Bi0.852 0.148
P
O1.648 and
Bi0.852V0.148O1.648
.
Solid-state NMR spectroscopy
1
7
51
31
Solid-state O, V and P magic-angle spinning (MAS) NMR spectra were obtained at 16.4 T
on a Bruker Avance III 700 MHz spectrometer operating at a Larmor frequency of 94.95 MHz
1
7
51
31
( O), 184.05 MHz ( V) and 283.39 MHz ( P); initial room-temperature measurements were
performed using a double resonance 1.3 mm Bruker probe, at a MAS rate of 60 kHz, and using
a rotor-synchronised spin-echo pulse sequence with a π/2 excitation pulse length of 1.4 µs (for
17O), with a recycle delay of 50 ms. Variable-temperature measurements from room temperature
to 973 K were performed at the same field using a triple resonance 7 mm Bruker laser probe.32,33
Temperature calibration of the probe was carefully performed by measuring the 79Br resonance
of KBr over the entire temperature range,34 with KBr typically included in the same rotor as the
sample as an internal temperature reference. The temperatures given in the text correspond to
actual sample temperatures with an estimated accuracy of ±10, 20 and 30 K in the 293–473 K,
473–673 K and 673–873 K temperature ranges, respectively. Samples were packed in 4 mm boron
nitride (BN) inserts within 7 mm zirconia rotors, which were subsequently spun at 4 kHz for
variable-temperature experiments. One-dimensional experiments were performed with a rotor-
synchronised spin-echo pulse sequence, using a 9 µs, 4 µs, or 6.5 µs π/2 excitation pulse, with
recycle delays of 50 ms, 0.5 s, or 30 s (for 17O, 51V, and 31P, respectively). Additional 17
O
measurements, including relaxation experiments, were obtained at 11.7 T on a Bruker Avance III
5
00 MHz spectrometer operating at a Larmor frequency of 67.83 MHz, using a double resonance
6
ACS Paragon Plus Environment

Page 7 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
Bi
4
mm Bruker probe at a MAS rate of 12.5 kHz, with an excitation pulse of 2.2 µs (for O sites)
V
or 5.0 µs (for O sites). Variable-temperature spin-lattice (T ) relaxation experiments at this field
1
were performed using a standard saturation recovery pulse sequence, with a maximal delay of 2.4 s
Bi
V
(
for O sites) or 0.24 s (for O sites). Temperature calibration of this probe was also performed by
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
79
17
measuring the Br resonance of KBr, with estimated accuracy of ±5 K. Finally, O measurements
were performed at 4.7 T on a Bruker Avance III 200 MHz spectrometer operating at a Larmor
frequency of 27.14 MHz, using a double resonance 1.3 mm Bruker probe at a MAS rate of 40 kHz,
with an excitation pulse of 0.5 µs, and a recycle delay of 50 ms; these experiments were performed
in order to extract quadrupolar parameters from the field-dependent second-order quadrupolar shift
17
(
see Supporting Information for further details). O chemical shifts were externally referenced to
51
35,36
H O at 0 ppm, V chemical shifts were externally referenced to NH VO at −570 ppm,
and
2
4
3
31
P chemical shifts were externally referenced to H PO (aq) at 0 ppm. NMR data were processed
3
4
with TopSpin 3.2.
Results
17O NMR
1
7
Room-temperature O MAS NMR spectra of all three materials (Figure 2) were initially obtained
with fast MAS (60 kHz) in order to achieve higher resolution. For each spectrum, a number of
different oxygen environments can be distinguished with resonances ranging from ∼130–630 ppm.
Bi
The common feature at 220 ppm is assigned to oxygen within the Bi–O sublattice (O ), consistent
with this chemical environment occurring in all three structures. Similar 17O chemical shifts of
3
7
oxygen environments coordinated to Bi have been previously reported, such as in Bi O
(195
2
3
ppm), as well as the so-called ‘BIMEVOX’ systems26 (250–270 ppm), Bi WO
38
(242 ppm), and
2
6
3
9
17
BiM PO with M = Cd, Zn (200–250 ppm). In these previous O NMR studies of Bi-containing
2
6
oxides, the local O environment corresponds to OBi , with a shift range of 190–270 ppm; we
4
are not aware of 17O shifts reported for OBi sites. The feature at 620 ppm present in both V-
3
7
ACS Paragon Plus Environment

Chemistry of Materials
Page 8 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
V
substituted samples is attributed to O in VO tetrahedra (O ), consistent with the established shift
4
V
26,40
range of O environments (600–1000 ppm).
Finally, the resonance at 130 ppm, present only
P
in spectra of the P-substituted sample, is assigned to O in PO tetrahedra (O ), with a shift that
4
compares favourably to recent 17O NMR studies of a range of solid inorganic acid phosphates
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
80–150 ppm).41,42 A minor peak observed at 140 ppm in both V-substituted samples is tentatively
assigned to bismuth silicate (estimated at ∼1 wt.%), which forms by reaction with the quartz vessel
17
17
during the O-enrichment procedure; similar O shifts have been reported for other silicates such
as glasses and apatite oxide-ion conductors.43,44
OV
Bi0.913V0.087O1.587
OBi
#
Bi0.852V0.148O1.648
#
OP
Bi0.852P0.148O1.648
600 400 200
δ O (ppm)
17
Figure 2: Room temperature 17O MAS NMR spectra of Bi0
and Bi0.852V0.148O1.648 obtained at 16.4 T at a spinning rate of 60 kHz, with a recycle delay of 50
P
O1.648, Bi0.913V0.087O1.587
.852 0.148
ms. All three samples show a common feature at ∼220 ppm, assigned to O in the Bi–O sublattice
Bi
V
P
(O ), with additional resonances arising from VO and PO environments (O and O ) at ∼620
4
4
ppm and ∼130 ppm, respectively. The # symbol denotes the peak at 140 ppm assigned to a minor
(
∼1 wt. %) bismuth silicate impurity phase.
17
Representative O NMR spectra obtained upon heating all three materials to 823 K are col-
8
ACS Paragon Plus Environment

Page 9 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
17
lated in Figure 3; the full datasets are shown in the Supporting Information (SI). The O spectra
for both V-substituted samples show qualitatively similar behaviour with increasing temperature,
V
Bi
and in particular both the O and O resonances significantly narrow in linewidth up to 573 K.
Above 573 K, both resonances disappear for both samples, before reappearing as a single coa-
lesced resonance at 823 K; the shift of this coalesced feature is 332 ppm for Bi0.913V0.087O1.587
and 391 ppm for Bi0.852V0.148O1.648. This behaviour is indicative of exchange between the two
different types of sites occurring on the so-called NMR timescale, where the rate of motion ap-
proaches the frequency separation between the two spectral features. Interpolation using the full
set of spectra (see SI) allows us to estimate the temperature at which the maximum broadening
and initial coalescence is observed, where the rate of ionic motion is directly proportional to the
frequency separation. On this basis, we calculate sublattice exchange rates of 87 kHz at 723 K
for pseudo-cubic Bi0.913V0.087O1.587, and 86 kHz at 748 K for monoclinic Bi0.852V0.148O1.648. At
comparable temperatures, the motional rate is somewhat slower in the more highly V-substituted,
crystallographically lower-symmetry material.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
For each phase, the chemical shift of the fully coalesced peak also gives information as to the
2−
Bi
V
residence time of mobile O ions in O environments vs. O environments. The coalesced shift
Bi
V
of 332 ppm for the Bi0.913V0.087O1.587 phase corresponds to a ratio of 70:30 (O :O ), while the
Bi
V
shift of 391 ppm for Bi0.852V0.148O1.648 gives a ratio of 52:48 (O :O ). This result is intuitively
reasonable given the higher concentration of V in Bi0.852V0.148O1.648. Furthermore, the expected
Bi
V
ratios of O :O sites calculated from the nominal stoichiometry values, assuming isolated VO
4
tetrahedra, (i.e., 79:21 for Bi0.852V0.148O1.648 and 64:36 for Bi0.913V0.087O1.587) are significantly
larger than the actual ratios calculated from the coalescence data. This result suggests the VT-
NMR data are sensitive to the presence of oxygen vacancies in the Bi–O sublattice and/or over-
coordination of the VO units. For Bi0.852V0.148O1.648, the relative number and/or residence time of
4
V
O sites is 33% larger than that expected solely from stoichiometry, whereas it is 43% larger than
expected for Bi0.913V0.087O1.587, implying that the latter phase has a higher average V coordination.
Kuang et al. also found from AIMD simulations at 1373 K that the coordination number of V sites
9
ACS Paragon Plus Environment

Chemistry of Materials
Page 10 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
varied between 4.0 and 4.3 for both V-substituted phases, with the larger average coordination
number observed for the lower V-substituted material, in excellent agreement with the present
results.23,24
17
For Bi0
P
O1.648, the temperature-dependent behaviour of the O spectra differs signif-
.852 0.148
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
P
Bi
icantly from that of the V-substituted samples. With increasing temperature, both the O and O
resonances exhibit linewidth narrowing that is indicative of increased rates of local oxide-ionic
motion, and this narrowing allows for observation of an additional resonance for each environment
above ∼473 K. However, even at 873 K, the two groups of resonances remain distinct, showing
no sign of the coalescence that is seen in Bi0.913V0.087O1.587 and Bi0.852V0.148O1.648. This result
is indicative of ionic motion that remains local (e.g., PO rotations or librations) or is confined to
4
distinct sublattices, with minimal long-range exchange between the Bi–O and P–O sublattices on
the NMR timescale. Moreover, the appearance of spinning sideband intensity arising from the OP
Bi
environment (but not from O ) above 673 K is suggestive of increasing rates of local ionic motion,
and further supports the idea that the motion remains distinct between the two sublattices.
Closer examination of the isotropic peaks (i.e., the central transition, or CT, centreband) corre-
V
17
sponding to the O environments in the O spectra of the V-substituted samples shows that, for
Bi0.913V0.087O1.587, the CT centreband can be fit to a single Voigt function, corresponding to a sin-
V
gle type of O environment. On the other hand, the fit to the CT centreband for Bi0.852V0.148O1.648
requires two such functions with relative intensity varying with temperature (Figure 4), indicating
V
the presence of two distinct types of O environments in the more highly V-substituted material.
V
The simple two-peak fits to the O features are justified, as the quadrupole coupling constants
V
(
C ’s) of the O sites are below 0.9 MHz, and second-order quadrupolar effects are too small to
Q
account for the observed lineshapes, including the broad linewidths (see SI). The relatively broad
V
Bi
room-temperature linewidths (for the O sites for the V-doped samples, and the O sites in gen-
eral) are likely to arise due to J-coupling to the abundant quadrupolar nuclei 51V (I = 7/2) and
209Bi (I = 9/2).
V
The two types of O environments display different temperature-dependent behaviours and
10
ACS Paragon Plus Environment

Page 11 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
Bi0.913V0.087O1.587
OV
OBi
823 K
673 K
5
5
73 K
23 K
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
473 K
5
73 K
3
2
73 K
93 K
OV
OBi
293 K
600 400 200
Bi0.852V0.148O1.648
680 600
δ 17O (ppm)
OV
260 180
OBi
8
6
73 K
73 K
x5
x5
573 K
5
4
23 K
73 K
573 K
3
2
73 K
93 K
OV
OBi
2
93 K
6
00 400 200
680 600
δ 17O (ppm)
260 180
Bi0.852P0.148O1.648
OP
OBi
573 K
8
73 K
*
*
*
*
*
5
23 K
673 K
*
*
*
*
*
4
3
73 K
73 K
5
73 K
OBi
OP
293 K
2
93 K
250 200 150 100
150 110
δ 17O (ppm)
230 180
1
7
Figure 3: Variable-temperature O MAS NMR spectra of Bi0.913V0.087O1.587, Bi0.852V0.148O1.648
O1.648 obtained at 16.4 T at a spinning rate of 4 kHz, with a recycle delay of 50
ms. The high-temperature O spectra for both V-substituted samples show first the broadening
and disappearance (at 673 K) and then the coalescence of the two distinct O features above this
temperature, indicative of increasingly fast motion between the O and O environments with
and Bi0
P
.852 0.148
1
7
17
Bi
V
increasing temperature. In contrast to the V-substituted samples, for Bi
P
O1.648 no coales-
0.852 0.148
1
7
cence of the two O resonances is observed w 1i t1h increasing temperatures, indicating a minimal
ACS Paragon Plus Environment
rate of O hopping between distinct environments over this temperature range. Spinning sidebands
are denoted by asterisks.

Chemistry of Materials
Page 12 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Bi0.852V0.148O1.648
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
expt
ꢀ
t
“620”
(O )
V1
“630”
V2
(O )
5
23 K
4
4
3
73 K
23 K
73 K
293 K
6
60 640 620 600 580
17
δ O (ppm)
Figure 4: Variable-temperature 17O MAS NMR spectra of Bi0.852V0.148O1.648, focusing on the
isotropic peaks of the O environments, measured at 16.4 T at a spinning rate of 4 kHz from
room temperature to 523 K. The O resonance in the Bi0.852V0.148O1.648 sample comprises two
distinct Voigt functions, with relative intensity changing with increasing temperature (compared to
the single peak seen in Bi0.913V0.087O1.587). Experimental spectra are shown in blue, summed fits
in red, and the individual fitted lineshapes in purple and green respectively.
V
V
12
ACS Paragon Plus Environment

Page 13 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
(a)
V1
Bi0.913V0.087O1.587 (O )
2
0
V1
Bi0.852V0.148O1.618 (O )
V2
18
16
Bi0.852V0.148O1.618 (O )
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
14
12
10
8
3
00
350
400
450
500
550
T (K)
(b)
P1
Bi0.852P0.148O1.618 (O )
9
P2
Bi0.852P0.148O1.618 (O )
7
5
3
5
00
600
700
800
900
T (K)
Figure 5: 17O MAS NMR CT full width at half maximum (FWHM) linewidths of the O and O
environments in the three phases as a function of temperature. (a) In both V-substituted samples,
V
P
V1
one resonance at ∼620 ppm (O ) narrows with increasing temperature, indicative of increasing
local tetrahedral motion, until a minimum linewidth is attained; further motional increases lead
to a regime in which CT linewidth is insensitive to motion. For the Bi0.852V0.148O1.648 sample,
V2
a second resonance at ∼630 ppm (O ) is observed with a relatively temperature-invariant CT
linewidth, indicating the presence of VO tetrahedra that do not undergo motion on the spectral
4
P
P1
P2
timescale. (b) Two distinct O resonances at ∼135 ppm (O ) and ∼145 ppm (O ) are resolved
for Bi at temperatures >473 K. The linewidth of both resonances narrows with
P O
.852 0.148 1.648
0
increasing temperature (albeit with an initial broadening for the resonance at 145 ppm), reaching a
minimum value at ∼650 K.
13
ACS Paragon Plus Environment

Chemistry of Materials
Page 14 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
thus provide insights into the underlying oxide-ion motional mechanisms that distinguish the V-
substituted samples. Both V-substituted compounds share a common OV1 environment centred at
∼620 ppm at room temperature, which narrows as a function of increasing temperature, reach-
ing a minimum CT linewidth between 473 and 523 K (Figure 5a). This environment therefore
undergoes thermally-activated ionic motion with a motional rate eventually exceeding the range
to which the CT linewidth is sensitive, with no further narrowing above 523 K until coalescence
occurs (see SI). By contrast, the linewidth of the second, higher-frequency OV2 peak (at ∼630
ppm at room temperature), which is only seen for Bi0.852V0.148O1.648, retains a relatively constant
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
V
linewidth over the same temperature range. The observation of two distinct O environments im-
plies that full local isotropic O exchange is not occurring at these temperatures. Similar narrowing
Bi
behaviour is seen in the CT linewidth for the O environment in both V-substituted samples (see
SI), also suggesting ionic motion on the Bi–O sublattice. However, the extent of narrowing is
greater for Bi0.913V0.087O1.587 than for Bi0.852V0.148O1.648, again implying faster motion, a smaller
Bi
distribution of local O chemical environments, or both.
17
P
For the Bi0
P
O1.648 sample, the O resonances corresponding to the O environment(s)
.852 0.148
also display different behaviours across distinct temperature ranges. From room temperature to
373 K, what appears to be a single resonance first broadens with increasing temperature, before
narrowing at a slightly higher chemical shift (see SI). This unusual behaviour could be due to co-
∼
P
Bi
alescence of multiple O resonances (some of which are hidden under the O resonance at room
temperature) due to local O rotations within isolated PO units. Above 523 K, another feature ap-
4
P
pears at 145 ppm, indicating the presence of an additional O chemical environment (Figure 3 and
Figure S3). Both features undergo linewidth narrowing with increasing temperature (see Figure 5b
and the SI) and move to higher chemical shifts. As described later (see Discussion), the observed
P
resonances correspond to subtly different local O environments that persist at high temperatures,
V
a phenomenon which may also occur for the O environments in Bi0.852V0.148O1.648 but cannot be
Bi
directly observed because of coalescence with the O feature.
14
ACS Paragon Plus Environment

Page 15 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
51V NMR
The room-temperature 51V MAS NMR spectra for the more highly V-substituted monoclinic
sample, Bi0.852V0.148O1.648, shown in Figure 6 (right), reveal three major distinct features with
isotropic shifts at −487, −493 and −508 ppm, with broad spinning sideband manifolds arising from
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
1
the corresponding satellite transitions or STs ( V is a quadrupolar nucleus with I = 7/2). Room-
5
1
temperature V NMR spectra taken at higher spinning speed (10 kHz) more easily allow the three
distinct environments to be observed (shown in the SI), but unfortunately we could not acquire
VT-NMR spectra at very high temperatures at this relatively fast MAS rate. The relatively small
difference in chemical shift of these V environments is most likely due to small differences in
the local structure surrounding the VO tetrahedra. The resonances lie within the known shift
4
range for 4- and 5-coordinate vanadium sites,26 with the higher-frequency resonance at −487 ppm
likely to be more 5-coordinate in nature, i.e., to experience 5-coordinate bonding environments a
greater percentage of the time, as suggested by previous AIMD simulations also showing average
V coordination numbers up to 4.3.23,24
5
1
With increasing temperature, the V spinning sideband manifolds of Bi0.852V0.148O1.648 first
broaden, before reappearing and narrowing up to 823 K. We assign these characteristic changes
to increasing rates of ionic motion on the order of the MAS rate (i.e., ∼4 kHz), as seen in similar
studies.27,45–47 In particular, the relevant time scale for this process is influenced by the strengths of
51
anisotropic spin interactions that give rise to the sidebands, primarily quadrupolar coupling for V
but to a lesser extent chemical shift anisotropy (CSA).27 The three features remain distinct, with
a gradual change to higher frequencies of −456, −472 and −505 ppm, at 673 K. The large change
in the shift of the first site in particular is consistent with an increase in coordination number
consistent with the formation of VO5,48 while the latter two sites remain closer to 4-coordinate.
The fact that all three signals persist at high temperature implies their local structural environments
remain distinct, even with very fast rates of nearby O2– ion exchange.
The corresponding VT spectra of the lower V-substituted, pseudo-cubic phase Bi0.913V0.087O1.587
however, show evidence for only one type of V environment centred at −478 ppm (Figure 6, left)
,
15
ACS Paragon Plus Environment

Chemistry of Materials
Page 16 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Bi0.913V0.087O1.587
Bi0.852V0.148O1.648
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
*
6
73 K
73 K
*
** ** * ** *
*
*
*
*
* ** * * *
***** ***
*
* * ***
*
*
*
5
*
*
*
*
*
*
*
*
** * * ** *
*
*
** ** ** *
4
73 K
93 K
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
2
***
*
* *** **
*
*** ** *
-
300 -500 -700 -440 -470 -500 -530
-300 -500 -700 -460 -480 -500 -520
δ 51V (ppm)
δ 51V (ppm)
Figure 6:
Bi0.852V0.148O1.648 obtained at 16.4 T at a spinning rate of 4 kHz, with a recycle delay of 0.5 s. At
Variable-temperature 51V MAS NMR spectra of Bi0.913V0.087O1.587 and
5
1
room temperature, the isotropic resonance in the V NMR spectrum of Bi0.913V0.087O1.587 is al-
ready broadened, indicating significantly fast motion involving the VO environment with respect
4
to the spinning frequency, whereas the corresponding spectrum of Bi0.852V0.148O1.648 indicates a
relatively slower motional rate. Asterisks denote spinning sidebands (for Bi0.913V0.087O1.587 only
shown at highest temperature for clarity). For Bi0.852V0.148O1.648, dashed lines mark the three
isotropic resonances; some spinning sidebands appear as shoulders next to, or nearly on top of, the
isotropic resonances.
16
ACS Paragon Plus Environment

Page 17 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
with a much broader CT linewidth at room temperature and no observed ST spinning sideband
intensity, consistent with a distribution of similar VO polyhedra. Upon heating, the CT linewidth
4
narrows markedly, accompanied by the appearance and then linewidth narrowing of spinning
sidebands; interestingly, this corresponds to the higher-temperature behaviour of the V sites in
the highly V-substituted sample (Bi0.852V0.148O1.648). As with Bi0.852V0.148O1.648, a similar shift
change to lower frequency (−458 ppm) occurs, corresponding to an average increase in V coordi-
nation. We can conclude that the motion probed by the single V environment, even at room temper-
ature, is already of a similar rate to the MAS spinning frequency (4 kHz) and therefore much faster
than that experienced by the equivalent sites in Bi0.852V0.148O1.648. Additionally, there is no evi-
dence of additional V environments as in the lower V-substituted Bi0.913V0.087O1.587. These results
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
51
17
from V NMR are in good agreement with the O NMR data, in which at moderate temperatures
up to coalescence, only one average feature in the V–O region is observed for Bi0.913V0.087O1.587
,
but two such peaks are present for Bi0.852V0.148O1.648
.
31P NMR
3
1
The room temperature P MAS NMR spectra of Bi
P O1.648 comprise a single broad reso-
.852 0.148
0
nance that narrows with increasing temperature (Figure 7), allowing us to distinguish four different
local environments. We assign these linewidth changes to the influence of nearby oxygen motion
3
1
that enters the fast motional regime at high temperature and thus no longer broadens the P reso-
1
7
nances, a result that concurs with the narrowing linewidths seen in the corresponding O spectra.
At high temperatures, the four distinct 31P resonances reflect small changes in the number and
arrangement of nearby Bi and P atoms, in good agreement with the four reported crystallographic
P sites in the monoclinic Bi0
P
O1.648 structure.25 This same resolution is not achieved in
.852 0.148
5
1
51
the V spectra, which is ascribed to some second-order quadrupolar line broadening of V (I =
7
/2) that persists even at high temperatures, as well as most likely increased disorder of the VOx
polyhedra as compared to PO4.
17
ACS Paragon Plus Environment

Chemistry of Materials
Page 18 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
73 K
573 K
4
73 K
93 K
2
4
2 0 -2 -4 -6 -8
δ P (ppm)
31
3
1
Figure 7: Variable-temperature P MAS NMR spectra of Bi
P
O1.648 obtained at 16.4 T at
0
.852 0.148
31
a spinning rate of 4 kHz, with a recycle delay of 30 s. At room temperature, the P NMR spectrum
51
shows a single highly broadened feature similar to the V NMR spectra of Bi0.913V0.087O1.587. A
narrowing isotropic linewidth is seen with increasing temperature that reveals four distinct 31
resonances.
P
18
ACS Paragon Plus Environment

Page 19 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
17
O T measurements
1
Measurement of the spin-lattice (T ) relaxation times as a function of temperature can be used to
1
8
gain insight into ionic motion correlation rates on the order of the Larmor frequency (∼10 Hz),
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a motional regime well above that accessible through CT linewidth measurements, which are only
4
5
sensitive to motion on the order of the MAS rate (∼10 −10 Hz). (We note that T measurements
1
may also probe a range of motional rates several orders of magnitude above and below the Larmor
frequency, depending on the specific relaxation mechanism(s) and the temperature of the system.)
By measuring the 17O T values for both the O and O sites as a function of temperature, the
Bi
V
1
relative contributions to ionic motion in these materials can be determined. Results from 17O
T1 measurements for Bi0.913V0.087O1.587, Bi0.852V0.148O1.648 and Bi0.852P0.148O1.648 are shown in
Figure 8.
1
7
Bi
For all three materials, the O T of the O environment decreases with increased temperature,
1
suggestive of changes in ionic motion within the Bi–O sublattice on the MHz timescale. The mo-
tion involving the Bi–O sublattice as probed by the T relaxation measurements is ascribed to the
1
1
7
local movement of oxygen vacancies, which have previously been measured with O NMR relax-
ometry in other oxide-ion conductors, e.g., Y-doped ceria49 and Bi WO . The relevant motions
38
2
6
in these systems contributing to temperature-dependent changes in the T values are typically con-
1
sidered to comprise short-range displacements such as nearest neighbour and/or nearby vacancy
jumps,50 which may be distinct from motions influencing bulk ionic conductivity.
V
On the other hand, the T values for the O environments in both V-substituted materials have
1
V
a weaker dependence on temperature. While the earlier linewidth analysis of the O environment
clearly probes increasing local VO rotational or librational motion, the lack of any change to
4
the T relaxation rates would suggest that this motion is not in the sensitive regime centred about
1
6
10
V
the Larmor frequency (i.e., ∼ 10 to ∼ 10 Hz). We note that the T values for the O site
are surprisingly short; the previous 17O NMR study of Bi Mo O by Holmes et al.27 also
1
2
6
10 69
found comparably short T times (∼10 ms at and above room temperature) for MoO tetrahedral
1
4
Bi
V
units. At higher temperatures the T values for the O and O environments appear to converge,
1
19
ACS Paragon Plus Environment

Chemistry of Materials
Page 20 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
consistent with earlier observed coalescence of these two sites and the accompanying averaging of
17
their apparent O quadrupolar parameters. This supports the inference that quadrupolar relaxation
mechanism leads to the observed changes in T1.
P
The O environment in Bi
P
O1.648 appears to have a short, temperature-invariant T1
0
.852 0.148
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
value of ∼1 ms, as shown in the SI, although signal from this site could not be deconvoluted at
17
Bi
several temperatures in the O relaxation measurements due to overlap with the O resonance.
T (K)
500
400
350
V
Bi0.913V0.087O1.587 (O env.)
V
2
Bi0.852V0.148O1.618 (O env.)
1
0
0
Bi
Bi0.913V0.087O1.587 (O env.)
Bi
Bi0.852V0.148O1.618 (O env.)
Bi
Bi0.852P0.148O1.618 (O env.)
1
1
2
.00 2.25 2.50 2.75 3.00
-1
3
-1
T (10 K )
Figure 8: Inverse 17O relaxation rates (T ) as a function of inverse temperature for
−1
1
Bi
V
Bi0.913V0.087O1.587, Bi0.852V0.148O1.648 and Bi
P
O
, for both the O and O environ-
0.852 0.148
1.648
V1
V2
ments. In the case of Bi0.852
V
0.148O1.648, the two distinct O and O environments (at ∼620 and
−
1
∼
630 ppm) have similar T1 values and the features have been analysed together. In the case of
P
Bi
P
O1.648, T values for the O environment were consistently too short to reliably mea-
sure (< 1 ms) and/or the O features were not resolved at certain temperatures due to overlap with
0
.852 0.148
1
P
Bi
the O resonance.
Activation energies
Bi
If it is assumed that the temperature dependence of the spin-lattice (T ) relaxation of the O
1
17
site(s) is driven by transient, motion-induced changes in the apparent O quadrupolar coupling,
20
ACS Paragon Plus Environment

Page 21 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
activation energies for oxide ion and/or vacancy mobility can be extracted. In this analysis, inverse
−1
correlation times τc are first calculated, which can be derived from fits to the T data using the
1
following equation:51,52
!
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ꢀ
ꢁ
2
2π2
25
η
1 =
·C · 1+
·
τc
+
4τc
,
(1)
2
Q
Q
2
2
2 2
T1
3
1+ω τ
1+4ω τ
0 c
0
c
1
7
where ω is the Larmor frequency of O (67.83 MHz at 11.7 T), and values of the proportionality
0
ꢂ
ꢃ
2
η
2
Q
constant C · 1+
, which depend on the estimated quadrupole coupling constant C and
Q
Q
3
Bi
quadrupolar asymmetry parameter η of the O environments, have been obtained by acquiring
Q
room-temperature spectra at multiple magnetic fields and determining the field dependence of the
second-order quadrupolar shift (further details in Supplementary Information).
−1
−1
If we then take the τc values to be equal to the average jump rates τ for the relevant
motional processes, we can fit the linear parts of each curve to extract an activation energy EA
and pre-exponential factor τi for the ionic motion, using the Arrhenius relationship:52
−1
ꢂ
ꢃ
EA
−1
−1
τ
= τi ·exp −
(2)
k T
B
Given that exchange at higher temperatures leads to an apparent average C value between the
Q
V
Bi
O and O sites, we have restricted this fitting to temperatures ≤ 400 K to obtain an accurate
value for motion within the Bi–O sublattice solely. The results of this fitting are shown in Figure 9
and Table 1. All samples have similar E values within error, consistent with a common oxide ion
A
motion present within the Bi–O sublattice.
A similar Arrhenius relationship can also be applied to the temperature-dependent 17O CT
V
P
linewidth(s) for the O and O environments plotted in Figure 5, under the assumption that this
variation is related to changes in the rate of oxide-ion motion. In this analysis, the logarithm of the
inverse CT linewidth is plotted against inverse temperature, which gives a linear relationship when
the relevant motional rate is on the order of the MAS rate, i.e., typically when the rate is between
1
00 Hz and 10 kHz.45–47,53 The results of the linewidth fitting are shown in Figure 10 and Table 1.
21
ACS Paragon Plus Environment

Chemistry of Materials
Page 22 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
V
P
17
Bi
In contrast to that of the O and O environments, the O CT linewidths of the O features
display more complicated behaviour with new environments appearing and disappearing at differ-
ent temperatures; therefore, E values could not be reliably derived in this way (see SI).
A
T (K)
400
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
500
350
Bi
Bi0.913V0.087O1.587 (O env.)
Bi0.852V0.148O1.618 (O env.)
Bi0.852P0.148O1.618 (O env.)
1
6.5
5.5
Bi
Bi
1
14.5
1
3.5
2.5
1
2
.00 2.25 - 2.50 2.75 3.00
1 3 -1
T (10 K )
−1
Figure 9: Arrhenius fitting of the inverse correlation times τ which have been extracted from the
1
7
−1
Bi
inverse O relaxation rates T of the O environments; the E derived from the low temperature
A
1
region corresponds to local motion within the Bi–O sublattice.
22
ACS Paragon Plus Environment

Page 23 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
T (K)
500
8
00 700 600
400
P1
Bi0.852P0.148O1.618 (O env.)
Bi0.852P0.148O1.618 (O env.)
-
1.0
P2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
-
1.5
2.0
-
-2.5
V1
Bi0.913V0.087O1.587 (O env.)
V1
-3.0
Bi0.852V0.148O1.618 (O env.)
1
.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8
-1
3
-1
T (10 K )
1
7
V
P
Figure 10: Arrhenius fitting of inverse O CT linewidths of the O and O environments. The
P1
P2
O and O environments correspond to the peaks at ∼135 ppm and ∼145 ppm, respectively, and
V1
V2
the O environment corresponds to the peak at ∼620 ppm (the other O site has a temperature-
invariant linewidth). The E derived in this way corresponds to oxide-ion motion within the VO
A
4
and PO polyhedral units, such as rotations or librations, on the spectral timescale (typically on the
4
order of ∼1 kHz).
17
Table 1: Activation energies E of oxide ion motion for the materials studied, determined from O
A
1
7
relaxation (T ) measurements and analysis of the O CT linewidths. Activation energies derived
1
1
7
P
from the O T values for the O environment(s) in Bi0.852P0.148O1.648 are not reported, as the T1
values were consistently too small to reliably measure (< 1 ms), or the spectral features were not
resolved at certain temperatures due to overlap with the O resonance.
1
Bi
Sample
Method
Environment
EA (eV)
Bi
Bi0.913V0.087O1.587
Bi0.852V0.148O1.648
T1
T1
T1
O
0.16 ± 0.02
0.16 ± 0.02
0.11 ± 0.02
0.07 ± 0.02
0.06 ± 0.02
0.04 ± 0.02
0.04 ± 0.02
Bi
O
Bi
Bi0
P
O1.648
O
.852 0.148
V1
Bi0.913V0.087O1.587 CT linewidth (620 ppm)
Bi0.852V0.148O1.648 CT linewidth (620 ppm)
O
O
O
O
V1
P1
Bi0
Bi0
P
P
O1.648 CT linewidth (135 ppm)
O1.648 CT linewidth (145 ppm)
.852 0.148
P2
.852 0.148
23
ACS Paragon Plus Environment

Chemistry of Materials
Page 24 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discussion
The previously published structural and AIMD studies of these systems23,24 suggested that long-
range, isotropic oxide-ion motion in Bi0.852V0.148O1.648 and Bi0.913V0.087O1.587 takes place through
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
exchange between the Bi–O and VO sublattices, facilitated by the variable coordination number of
4
17
V. The O VT MAS NMR spectra clearly support this mechanism, as the coalesced resonances ob-
served at high temperatures are directly indicative of a motionally averaged O environment rapidly
Bi
V
sampling both O and O sites, albeit occurring at a slower rate than the local motional processes
17
within the Bi–O sublattice. Moreover, the coalesced O shift lying closer to that of the position
V
of the O resonance than would be expected purely from stoichiometry implies the presence of
over-coordinated V sites, with the average O environment residing more frequently near V. These
51
results agree with the V resonances measured between −450 and −510 ppm (Figure 6), which
are consistent with both 4- and 5-coordinate vanadium environments observed previously,48,54,55
51
as well as a change towards more highly-coordinated V sites (higher frequency V shifts) at higher
51
temperatures; we do however note that the isotropic V shifts are difficult to fully resolve at the
slow spinning speeds used in the VT-NMR experiments.
For Bi0
P
O1.648, the E values found for motion within the PO units agree relatively
.852 0.148 4
A
17
well with the small activation barriers for tetrahedral rotation previously found using O NMR in
tetraoxoanion-containing materials,56 indicating that the relatively straightforward CT linewidth
analysis nonetheless leads to reasonable activation energy values. Moreover, the E values are
A
identical within error to those obtained for motion involving the VO units in the V-substituted
4
materials, suggesting that similar motional processes operate in both types of dopant sublattices.
This study also provides insight as to the different local oxide-ion conduction mechanisms op-
erating in the different phases. The previously measured differences in bulk ionic conductivity for
Bi0.913V0.087O1.587 and Bi0.852V0.148O1.648 (σ ≈ 10−2 and 10 S cm at 673 K, respectively)
−3
−1
V
now become apparent in terms of variations in the O environments present in these materials.
V
For the more highly V-substituted, monoclinic Bi0.852V0.148O1.648, two distinct types of O en-
vironments are observed (OV1 and O ), with different rates of associated O motion as revealed
V2
24
ACS Paragon Plus Environment

Page 25 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
by the differences in the temperature dependence of their 17O CT linewidths. By comparison,
only a single OV1 environment is resolved for Bi0.913V0.087O1.587. We note that the OV1 site
in Bi0.852V0.148O1.648 and Bi0.913V0.087O1.587 has a very similar chemical shift (∼620 ppm) and
temperature-dependent linewidth behaviour in both samples, which we ascribe to O in a more
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
highly-coordinated local VO environment with a faster rate of local motion and/or sublattice ex-
4
change. The appearance of the second OV2 resonance in Bi0.852V0.148O1.648 with a temperature-
invariant 17O CT linewidth is indicative of O sites that do not participate to the same degree in
exchange and instead may hinder long-range oxygen motion. Structurally, the presence of these
additional OV2 environment(s) is a consequence of higher V-doping and is consistent with shorter
distances between neighbouring VO tetrahedra.20 Interestingly, two O sites also appear in the
P
4
17O spectra of Bi0
P
O1.648 (O and O ), and in both the P- and V-substituted samples we
P1
P2
.852 0.148
observe a ∼1:3 intensity ratio of the two sites at higher temperatures (see Figure 4 and SI). We
note the Bi0.852V0.148O1.648 and Bi
P
O1.648 phases are structurally analogous, and possess
0.852 0.148
25
four crystallographically distinct P/V sites; in the reported crystal structures, one of the P/V sites
appears to possess different local tetrahedral distortions, which may give rise to the distinct oxygen
environments with the ∼1:3 intensity ratio we observe.
17
The O VT-NMR spectra of Bi
P
O1.648 (Figure 3) indicate conduction processes that
0
.852 0.148
differ from the sublattice exchange evident from the two-site coalescence observed for the V-
Bi
substituted samples. Focussing on the O region of the spectra, three distinct regimes are active
at different temperatures. From room temperature to 473 K, only one broad OBi resonance is
observed, consistent with slow ionic motion and/or a large chemical shift distribution that prevents
the resolution of individual sites. From 523–673 K two distinct environments at ∼220 and ∼200
ppm are resolved, which are tentatively ascribed to O near, and not near, an oxide ion vacancy in
the Bi–O sublattice; this assignment is consistent with the oxygen vacancies observed in this phase
previously.25 (We note a two-peak fit is justified as second-order quadrupolar coupling lineshapes
are not expected, with further details given in the SI.) At these temperatures, the higher-frequency
resonance at ∼220 ppm displays a narrowing linewidth with increasing temperature. We attribute
25
ACS Paragon Plus Environment

Chemistry of Materials
Page 26 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
this behaviour to the vacancy-adjacent oxygen atoms undergoing hopping into vacancy sites. By
contrast, the isolated oxygens in fully occupied regions of the lattice are immobile, and therefore
the corresponding spectral feature (at ∼200 ppm) exhibits an invariant linewidth with increasing
temperature. (Two OBi resonances at ∼220 ppm and ∼200 ppm are also observed in the 17
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
spectra of both V-substituted samples, especially in the spectra collected at 573 K, showing that O
sites adjacent to vacancies can be distinguished in all the studied phases.)
Bi
Finally, at temperatures >723 K, the two O environments in the P-substituted material coa-
lesce to a single peak due to sufficiently rapid oxide-ion motion within the entire Bi–O sublattice,
i.e., on the NMR timescale, no O sites are isolated from vacancies. Over the same temperature
range, both the 17O and 31P NMR spectra reveal an increasing number of different OP environ-
ments that can be resolved due to motional narrowing; however, this motion must be restricted
to localised tetrahedral rotations because coalescence between the Bi–O and PO sublattices is
4
17
not observed. In contrast, the same behaviour cannot be observed in O NMR spectra of the V-
Bi
V
51
substituted samples due to the coalescence of the O and O resonances; however, the V NMR
spectra of Bi0.852V0.148O1.648 still indicate the presence of V sites with differing coordination num-
bers even at high temperature.
The variable coordination of the dopant ion plays a crucial role in understanding the differ-
ing ionic conductivities of these fluorite-type materials. Experimentally, the Bi0
P
O1.648
.852 0.148
exhibits the lowest conductivity, followed by Bi0.852V0.148O1.648 and Bi0.913V0.087O1.587.23,24,28
Although dopants clearly aid in stabilising the cubic δ-Bi O structure, they also tend to disrupt
2
3
three-dimensional isotropic conduction within the Bi–O sublattice. However, to a limited degree
cooperative exchange between the dopant and Bi–O sublattices can offset this effect. The differ-
ence in the conductivity of pseudo-cubic Bi0.913V0.087O1.587 and monoclinic Bi0.852V0.148O1.648
highlights this first point, with the maintenance of the long-range 3D isotropic structure in the
former case contributing to its exceptional ionic conduction. In the case of Bi0
P
O1.648,
.852 0.148
although increasing rates of oxide-ion motion clearly do occur within the Bi–O sublattice as we
have shown, the inability of phosphorus to adopt variable O coordination hinders significant sub-
26
ACS Paragon Plus Environment

Page 27 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
lattice exchange and leads to overall lower bulk conductivity. Similar observations have been made
57
58
regarding the Bi Re O
and Bi Nb O
1.5+x
phases, which utilise dopant atoms that adopt
2
8
2
49
1-x
x
variable O coordination, but also undergo doping-induced symmetry lowering of the crystal struc-
ture from cubic to tetragonal; this in turn leads to correspondingly lower oxide-ion conductivity
than in the vanadate materials.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Seen in the context of previous studies on oxygen ion conductors, and Bi-based materials in
particular, our results further consolidate the state of mechanistic understanding of the conduction
processes, while also adding local-level insights regarding structure and dynamics. Firstly, con-
27
duction in the V-substituted materials appears to operate in a similar fashion to Bi Mo O ,
2
6
10 69
with rapid localised rotation measured for dopant polyhedra at low temperatures preceding the on-
set of long-range, isotropic conduction via cooperative exchange with the Bi-containing sublattice
at higher temperatures. However, in the Bi Mo O system the onset of sublattice exchange
2
6
10 69
appears correlated with and possibly driven by a monoclinic-triclinic phase transition at 583 K,
in contrast to the temperature-invariant structures of the vanadates and phosphates studied here.
17
Also, the previous work on the molybdates as well as other solid-state O NMR results indicating
rapid polyhedral rotation at low temperatures (e.g., E = 0.07 eV in KMnO 56) give us confidence
A
4
that the very small activation energies for VO and PO rotation do indeed represent local rota-
4
4
tional or librational motion. Finally, as discussed in the prior study of Kim and Grey on V- and
26
Ti-substituted Bi O materials, distorted polyhedra are critical in facilitating ionic mobility. In
2
3
that work, of the three local V environments (symmetric tetrahedral, symmetric pentahedral and
distorted tetrahedral) present in α-Bi V O and observed by 51V NMR, the most distorted en-
4
2
11
vironment was associated with the highest oxide-ion mobility, as probed by variable-temperature
51
17
V{ O} TRAPDOR NMR experiments. This mirrors our results which suggest that distorted and
overcoordinated VO polyhedra contribute to higher oxygen conductivity for Bi0.913V0.087O1.587
,
4
51
while the presence of additional, more symmetric V environments in Bi0.852V0.148O1.648 are re-
sponsible for the slightly lower oxygen conductivity of this phase.
27
ACS Paragon Plus Environment

Chemistry of Materials
Page 28 of 37
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Conclusions
In this work, a multinuclear VT MAS NMR approach has been applied to elucidate the influence
of V and P substitution on the oxide-ion conductivity of promising stabilised bismuth oxide (δ-
Bi O ) phases. We have shown that variable-temperature solid-state NMR is particularly well-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
suited to understanding connections between local structural features and ionic mobility, and can
quantify the evolution of ion dynamics with temperature.
Initially, room temperature 17O MAS NMR measurements have been performed to charac-
terise the local O environments present in each material; we are able to distinguish between bis-
Bi
V
P
muth oxide-like (O ) environments and other sites corresponding to VO (O ) and PO (O )
4
4
polyhedral units. Next, upon heating the V-substituted samples, the resonances arising from these
chemically distinct sublattices first sharpen, then disappear, and finally reappear as a single coa-
lesced (averaged) peak at high temperature, conclusively showing oxide-ion exchange between the
two sublattices. In contrast, 17O NMR spectra of the phosphate phase do not evidence two-site
coalescence within the studied temperature range (up to 873 K); we conclude this material exhibits
a different and less effective oxide-ion conduction mechanism, as supported by lower values of
ionic conductivity measured previously.23,24,28
17
Arrhenius-based analyses of both the O T relaxation rates and central transition (CT) linewidth
1
data allow us to determine activation energies for thermally-activated motional processes within
the two distinct sublattices (Bi–O and VO /PO ); the much lower activation energy for local VO
4
4
4
rotations or librations as compared to those of PO helps to explain the differences in bulk con-
4
V
ductivity between the materials. Furthermore, two different O environments are observed for the
higher V-substituted phase Bi0.852V0.148O1.648. We argue that with increased V doping, symmetry
lowering from pseudo-cubic to monoclinic leads to two different types of local VO environments
4
with differing rates of local ionic motion, which in turn disrupts the three-dimensional oxide-ion
conduction network. These findings highlight the crucial interplay between dopant concentration
and the long-range ionic conduction pathways, and also point out the influence of the coordina-
tion flexibility of the dopant. These two factors (dopant concentration and coordination) directly
28
ACS Paragon Plus Environment

Page 29 of 37
Chemistry of Materials
1
2
3
4
5
6
7
8
9
affect the available oxide-ion mobility mechanisms in Bi O -based materials and thus determine
2
3
the overall functional performance.
In summary, this study highlights the potential for multinuclear variable-temperature solid-
state NMR, with its specificity for different nuclei as well as sensitivity to local dynamics, to
provide unique insights into the underlying mechanisms driving bulk ionic conduction in superi-
onic conductors. In turn, these mechanisms suggest design rules that can be utilised in future work
to optimise the next generation of ionic conducting materials with the ultimate goal of rational
materials design.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Acknowledgement
M.T.D. is supported by a Junior Research Fellowship from Clare College, Cambridge, and a STFC
Early Career Award. D.M.H. acknowledges funding from the Cambridge Commonwealth Trusts,
and is grateful for support from NECCES, an Energy Frontier Research Center funded by the U.S.
Department of Energy, Office of Science, Office of Basic Energy Sciences under Award No. DE-
SC0012583. I.R.E. acknowledges the Royal Society and the Leverhulme Trust for the award of a
Senior Research Fellowship.
Supporting Information
1
7
XRD characterisation, complete VT-NMR datasets, details of O quadrupolar parameter fitting,
1
7
further O NMR saturation recovery and CT data.
Data
All supporting data for this work can be found on https://www.repository.cam.ac.uk.
29
ACS Paragon Plus Environment