EES Batteries

View Article Online

PAPER View Journal | View Issue

Probing the electrochemical behaviour of lithium

This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

Cite this: EES Batteries, 2025, 1, 527

imide as an electrolyte for solid-state batteries†
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Jeremy P. Lowen, a Teresa Insinna, b Tharigopala V. Beatriceveena, a

Mark P. Stockham, a,d Bo Dong, a Sarah J. Day,c Clare P. Grey, b

Emma Kendrick, d Peter R. Slater, a Paul A. Anderson a and
Joshua W. Makepeace *a

All-solid-state batteries utilising a Li-metal anode have long promised to be the next-generation of high-
performance energy storage device, with a step-change in energy density, cycling stability and cell safety
touted as potential advantages compared to conventional Li-ion battery cells. A key to enabling this
technology is the development of solid-state electrolytes with the elusive combination of high ionic con-
ductivity, wide electrochemical stability and the ability to form a conductive and stable interface with Li
metal. Presently, oxide and sulfide-based materials, particularly garnet and argyrodite-type structures,
have proved most promising for this application. However, these still suffer from a number of challenges,
including resistive lithium metal interfaces, poor lithium dendrite suppression (at high current density) and
low voltage stability. Here we report the first application of lithium imide, an antifluorite-structured material,
as a solid electrolyte in a Li-metal battery. Low-temperature synthesis of lithium imide produces promising
Li-ion conductivity, reaching >1 mS cm−1 at 30 °C using a modest post-synthetic mechanochemical treat-
ment, as well as displaying at least 5 V stability vs. Li+/Li. In situ electrochemical operation of lithium imide
with Li-metal electrodes reveals an apparent 1000-fold increase in its measured conductivity, whilst appear-
ing to remain an electronic insulator. It is postulated that stoichiometry variation at the grain boundary may
Received 26th March 2025, contribute to this conductivity improvement. Furthermore, the material is shown to possess impressive resis-
Accepted 31st March 2025
tance to hard shorting under high current density conditions (70 mA cm−2) as well as the ability to operate
DOI: 10.1039/d5eb00058k in Li-metal battery cells. These results not only highlight the promising performance of lithium imide, but
rsc.li/EESBatteries also its potential to be the basis for a new family of antifluorite based solid electrolytes.

Broader context
Solid electrolyte materials offer a promising solution to the safety and performance challenges posed by conventional liquid electrolyte batteries, particularly
the issue of metal dendrite formation. However, achieving all of the demanding requirements for solid electrolytes—high ionic conductivity, low electronic
conductivity, good processability, and excellent stability under operating conditions—remains a significant challenge. In this study, we introduce lithium
imide, a relatively unexplored ionic conductor with promising properties for solid-state battery applications. The material demonstrates excellent compatibil-
ity with lithium metal, high ionic conductivity, and a low-temperature synthesis route compared to other leading electrolytes. Interestingly, the material’s
conductivity increases dramatically upon cycling with lithium, reaching liquid-like levels. This unexpected behaviour is hypothesized to result from compo-
sitional changes at the grain boundaries, forming more disordered structures. Given previous exploration of lithium imide materials in the context of hydro-
gen storage, this study lays the groundwork for a novel family of solid electrolytes based on lithium imide.

The growing demand for batteries with high power and greater
energy density is driving the push for adopting lithium metal
a
School of Chemistry, University of Birmingham, Edgbaston, B15 2TT, UK. anodes in the next generation of Li-ion battery technology.1,2
E-mail: j.w.makepeace@bham.ac.uk This transition is however hindered by a number of challenges
b
Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge, CB2 including lithium dendrite propagation, the formation of
1EW, UK
c
‘dead’ lithium, and chemical incompatibility with current
I11 Beamline, Diamond Light Source, Didcot, OX11 0QX, UK
d
School of Materials and Metallurgy, University of Birmingham, B15 2TT, UK
liquid electrolytes, which leads to uncontrolled solid electro-
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/ lyte interphase (SEI) growth during cycling.1–3 A key strategy
10.1039/d5eb00058k towards resolving these issues is to replace organic liquid elec-

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 527
 View Article Online

Paper EES Batteries

trolytes with a solid ceramic or polymer alternative forming a that Li+ diffuses via an interstitialcy-type mechanism, intrinsically
so-called all solid-state battery (ASSB).4–6 Employing a solid- linked with the rotation of N–H bonds within the structure.37,38
state electrolyte (SSE) could offer chemical compatibility with Although moisture sensitive, Li2NH can reportedly be formed in
lithium metal, provide a mechanical barrier to dendrite as little as 10 minutes at 210 °C and is composed of largely abun-
growth, enable the use of future high-voltage cathodes, and dant and lightweight elements, making it more suitable for low
address the safety concerns associated with current liquid cost and high energy density systems as compared to other arche-
electrolytes.4,7 Finding a suitable SSE material is not, however, typal SSEs.39 Herein we demonstrate the significant potential of
straightforward; the performance of the electrolyte must meet Li2NH as a SSE, with a new highest recorded bulk ionic conduc-
a number of strict requirements including: tivity for the material (>1 mS cm−1 at RT), wide electrochemical
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

• High ionic conductivity of at least 0.1 mS cm−1 at room stability window (≥5 V) and excellent high current density oper-
temperature.7,8 ational capability (at up to 70 mA cm−2). We demonstrate that
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

• Electronically insulating.7,8 in situ operation of a Li2NH SSE using a Li-metal symmetric cell
• A wide electrochemical stability window comparable to or induces a metastable further increase in the conductivity
greater than that of a conventional liquid electrolyte (up to (>10 mS cm−1) without bulk structural changes to the material,
∼4.2 V vs. Li+/Li) with a reduction potential close to that of indicating possible activation of fast grain boundary diffusion.
lithium metal (0 V vs. Li+/Li).7–9 Operation of hybrid solid-state batteries utilising a Li2NH SSE
• Low resistance interfaces with both lithium metal and the and Li-metal anode is also achieved with two separate cathode
relevant cathode. materials (LiFePO4 and TiS2). This material therefore represents a
• Lightweight, cheap, and easily accessible through com- promising new system for achieving high-performance solid-state
mercially viable synthesis routes. Li-metal batteries.
To date, numerous ceramic material types (ranging from
oxides to sulfides) with a variety of structures (e.g. Garnet,
LISICON, Argyrodite) have been researched for this appli- Structure and synthesis
cation, though the majority have yet to reach the market.10–15
Even the most extensively researched materials, where Lithium imide is widely accepted to adopt a cubic antifluorite-
impressively high ionic conductivities have been recorded, still based structure, although there are differing reports as to
have significant challenges to their practical application. whether at room temperature this is a simple antifluorite cell
Oxide-based materials, such as lithium garnets, are often with disordered N–H orientations (Fm3 ˉm symmetry) or a larger
mechanically hard and form surface lithium carbonates, con- ˉ
superstructure (e.g. Fd3m symmetry, Fig. 1a) with ordered dis-
sequently exhibiting poor wettability with lithium metal and placement of Li-ions into octahedral holes (Fig. 1b) and tetra-
highly resistive interfaces.16,17 Sulfide-based materials, whilst hedral coordination of the lithium vacancy by N–H groups
softer, display a low upper-voltage stability and are extremely (Fig. 1c).40–43 In essence, this structure represents an ordered
moisture sensitive.18 Fluorite-structured oxides and fluorides Frenkel defect variation on the classic antifluorite structure.
have long been researched in the context of solid oxide fuel Given the propensity for Li2NH to form antifluorite-structured
cells and solid-state fluoride batteries, due to their high Fm3ˉm symmetry solid solutions with other N–H based
ambient temperature anionic conductivity.19–23 Despite this, materials, it is unlikely that there are two room temperature
comparatively little attention has been paid to corresponding polymorphs of Li2NH. Instead, experimental reports of the dis-
lithium antifluorite materials for Li-ion solid electrolyte appli- ordered structure for stoichiometric Li2NH are much more likely
cation, aside from lithium nitride chloride systems and recent off-stoichiometry due to amide (NH2−), nitride (N3−) or hydride
phosphorus-doped lithium sulphide.24–26 (H−) impurities associated with the synthesis method.31,44–46
Lithium imide (Li2NH), a nitrogen-based complex metal Multiple molecular dynamics and density functional theory
hydride with an antifluorite-type structure, has previously been simulations investigating the structure of Li2NH support this
explored for its readily-reversible hydrogen storage reaction assertion, indicating that stoichiometric Li2NH should take the
and impressive catalytic activity for ammonia ordered superstructure at room temperature.37,47–50 It should be
decomposition.27–30 A key characteristic driving the perform- noted that previous reports of the Li-ion conductivity of Li2NH
ance of this material in these applications is its high reported have almost exclusively been from samples with this disordered
lithium-ion conductivity (10−5–10−4 S cm−1), yet only one study structure and not therefore stoichiometric Li2NH. The synthesis
has investigated the electrochemical characteristics of the of Li2NH was achieved using the solid-state reaction of Li3N
material.31–34 This may relate to initial suggestions of a narrow with LiNH2 first reported by Hu and Ruckenstein.39 The powder
operational voltage range, though these were not experi- X-ray diffraction pattern of Li2NH is displayed in Fig. 1d. The
mentally verified. Indeed, subsequent studies of lithium imide peak at 6.9° corresponds to the (111) reflection characteristic of
point towards a promising set of properties: it is thermally the 2a × 2a × 2a antifluorite superstructure of stoichiometric
stable up to 600 °C, has an apparently wide electrochemical Li2NH. Rietveld analysis was performed using a previously
stability window and 7Li NMR data suggest the main charge reported Fd3 ˉm structure to fit the Li2NH phase.43
carrier to be Li+, signifying that Li2NH is likely electronically The inset Raman spectrum displays a broad peak at
insulating.35,36 Molecular dynamics simulations have indicated approximately 3180 cm−1 corresponding to the linear imide

528 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Fig. 1 (a) Fd3̄m structure of stoichiometric Li2NH with tetrahedral Li ions shown in red, interstitial Li ions in orange, N in blue and H in grey. (b)
Octahedral hole occupied by interstitial Li-ions (c) tetrahedral coordination of local N–H bonds towards vacancy left by Li displacement. (d) Powder
XRD pattern including experimental data and Rietveld fit of Li2NH with inset Raman spectra. (e) Nyquist impedance spectra of Au|Li2NH|Au cells at
30 °C with different pellet preparation strategies. (f ) Arrhenius plots of each Au|Li2NH|Au cell. (g) Nyquist impedance spectra of a Li|Li2NH|Li cell
before and after simple heat treatment. Cold-pressed imide was used. A schematic of proposed interface stabilisation is shown.

stretch and is consistent with previously reported this measurement is displayed in Fig. 1e and the full Nyquist
spectra.34,45,51 Two minor peaks at 3240 cm−1 and 3275 cm−1 plot with equivalent circuit model is found in ESI Fig. 1†
are likely indicative of very minor levels of residual amide ions (results of fitting found in ESI Table 1†). One semi-circle may
left in the solid solution.45 Compared to other solid electrolyte be observed which is assigned to the impedance associated
preparations, the synthesis conditions for Li2NH are very mild. with the combined bulk/grain boundary transport, followed by
For example, typical garnet-type oxides often require multiple a spike observed at high frequency, which is characteristic of
firings at temperatures >900 °C to complete their synthesis, Li2NH–Au interface charging. This blocking behaviour is a
whilst in this synthesis the sample is calcined in a single step good indication that Li2NH is a purely ionic conductor. The
at only 325 °C.52 conductivity was calculated using the total bulk/grain bound-
ary contribution and was found to be 0.054 mS cm−1. This is
of the same order as the previous literature for this phase.34 A
Ionic and electronic conductivity potential detriment to the total conductivity of this system is
likely the porosity of the cold-pressed pellet which had a rela-
The ionic conductivity of as-prepared, cold pressed Li2NH was tive density of 76.5%. The electronic conductivity and transfer-
determined at 30 °C via electrochemical impedance spec- ence numbers of Li2NH were determined using DC polaris-
troscopy (EIS) on an Au|Li2NH|Au cell. The Nyquist plot for ation on the same Au|Li2NH|Au held at 40 °C (ESI Fig. 2†).

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 529
 View Article Online

Paper EES Batteries

The electronic conductivity was calculated to be 1.2 nS cm−1 posed mechanical interface formation process is displayed as
whilst ti was found to be 0.998, confirming that all charge a schematic in Fig. 1g.
transfer may be attributed to ionic movement and that Li2NH
is an electronic insulator.
Given that the ionic conductivity for cold-pressed Li2NH Electrochemical stability
does not meet the threshold for SSE application we have
pursued further pellet preparation strategies to improve this The electrochemical stability of Li2NH was assessed via cyclic
property. Fig. 1e displays Nyquist plots of these other prep- voltammetry (CV) between −0.5 V and 5 V vs. Li/Li+ on a Li|
arations at 30 °C. Full Nyquist plots with equivalent circuit Li2NH|Steel cell where the Li-metal electrode was melded to
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

fitting can also be found in ESI Fig. 1† (fitting results in ESI the Li2NH pellet via the method used above. No extra pressure
was applied to the cell beyond the spacer and spring used in
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Table 1†). Similar to cold-pressed Li2NH, one semi-circle and

spike can be observed for each preparation. A pre-treatment of fabrication. The cyclic voltammograms for this measurement
Li2NH powder through low-energy ball milling before cold are displayed in Fig. 2a. Anodic and cathodic currents corres-
pressing was found to improve the conductivity to 0.34 mS ponding to lithium stripping ( positive current, Li → Li+ + e−)
cm−1, whilst retaining a similar pellet relative density of 77%. and lithium plating (negative current, Li+ + e− → Li) can be
This improvement in conductivity is likely due to improved observed near 0 V. No currents corresponding to electrolyte
grain boundary diffusion, potentially through inducing dis- decomposition were observed over 16 cycles, indicating that
order from milling. Similar effects from milling have been Li2NH is electrochemically stable across the measured voltage
reported for fluorite-based materials, however, to our knowl- range, or at least forms a stable electrode–electrolyte inter-
edge this is the first report of this behaviour for an antifluorite phase. However, it can be observed that the currents corres-
material.53 Hot pressing Li2NH at 325 °C for just one hour has ponding to lithium stripping/plating increase with increasing
an even greater effect than milling, improving the conductivity cycle number. This has previously been ascribed to an improv-
to 1 mS cm−1 (Fig. 1e inset), the highest recorded conductivity ing interfacial morphology between the SSE and Li-metal
for this material at room temperature. These improvements during cycling, however, it is also possible that the ionic con-
are likely due to morphological effects as well as a gain in
pellet relative density to 85% during hot pressing. Arrhenius
plots for the conductivity of all three pellet preparations across
a range of temperatures are shown in Fig. 1f. The activation
energy for lithium diffusion was calculated to be 0.50(2) eV for
cold pressed Li2NH, 0.52(2) eV for milled + cold pressed and
0.44(1) eV for hot pressed Li2NH. These values are of the order
of previously reported values for Li2NH.32–34 The lower acti-
vation energy for hot pressed Li2NH is likely a reflection of
improved grain boundary diffusion and the lower porosity of
the pellet. Given the conductivity gains with these relatively
simple physical treatments, it is likely that further optimi-
sation of the sample morphology will yield additional
improvements.

Li-metal interface
Fig. 1g displays Nyquist plots for a Li|Li2NH|Li cell at 30 °C
before and after heat treatment. Upon initial construction of
the cell a large interfacial impedance is observed (∼3308 Ω
cm2) between the Li2NH and Li-metal. However, a simple heat
treatment of just 60 °C for 6–8 hours largely eliminates this
interfacial resistance (∼47 Ω cm2). Equivalent circuit fitting
results of these spectra may be found in ESI Fig. 3 and ESI
Table 2.† The conductivity of the cell post heat-treatment is
found to be in good agreement with Au|Li2NH|Au data at
0.059 mS cm−1. Post thermal treatment X-ray diffraction ana-
lysis (ESI Fig. 4†) displays that no crystalline impurities are
Fig. 2 (a) Cyclic voltammograms of a Li|Li2NH|Steel cell at 40 °C. (b)
formed during this procedure, indicating that this process is Integrated current vs. cycle number for stripping and plating. (c) DC
likely to be either a mechanical wetting of the interface or a polarisation taken (c) before and after (d) cycling. (e) Post-CV (16 com-
surface-confined reaction which improves conductivity. A pro- pleted cycles) powder XRD pattern of material after cell disassembly.

530 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

ductivity of the cell is improving through another mechanism, It is possible that the discrepancy in lithium stripped and
allowing more lithium to diffuse each cycle.54 plated reflects a change in the lithium content at the surface
The rate of increase in plating/stripping currents with cycle or grain boundaries of Li2NH particles. Fluorite-type anionic
number follows two regimes, as displayed in Fig. 2b. Initially conductors have been found to have increased ionic mobility
(cycles 1–7), the proportion of lithium stripped/plated through the grain boundaries due to stoichiometry
increases slowly, followed by a rapid increase (cycles 7–16). In variation.53,55,56 Furthermore, recent ab initio simulations of a
this second regime, the ratio of stripping peak area to plating Li2NH surface catalysing ammonia decomposition with sub-
peak area appears to stabilise (ESI Fig. 5†). By analysing the sequent formation of a non-stoichiometric imide-amide par-
difference in the total amount of lithium stripped and plated ticle surface results in a highly disordered, quasi-liquid
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

it was found that 5.2% more lithium is plated than stripped surface, where fast-ionic diffusion is extremely plausible.57
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

(calculated as a percentage of the number of moles of lithium While these simulations were at 500 °C, it is conceivable that
in Li2NH). This discrepancy may be explained either by the for- similar particle surface stoichiometry variation under electro-
mation of dead lithium as a result of loss of contact at the Li– chemical conditions might induce an analogous effect, result-
Li2NH interface or may be representative of a chemical change ing in ‘ionic highways’ along the grain boundaries of Li2NH
in the cell. A loss of lithium could for instance indicate the for- particles.
mation of an imide-amide phase (Li2−xNH1+x) which are well
known and exist over a wide stoichiometry range where up to
two-thirds of imide ions can be replaced by NH2− within the Lithium stripping and plating
Li2NH structure.31
It is clear that there is a mechanical and/or chemical inter- To assess the long-term stability of Li2NH in contact with a
action between lithium metal and Li2NH. To probe this inter- lithium metal anode under battery operating conditions and
action a DC polarisation experiment using a voltage of 0.5 V further understand the interaction between the two materials,
was undertaken both before and immediately (within lithium stripping and plating experiments (S&P) under con-
2 minutes) after the CV experiment, with data displayed in stant current conditions were conducted on a Li|Li2NH|Li cell.
Fig. 2c and d. The plateau currents imply an electronic conduc- Fig. 3a displays the time-dependent voltage profile of the cell
tivity pre-CV and post-CV of 0.45 nS cm−1 and 54 nS cm−1, at 40 °C cycled for 200 cycles at 5, 10, 20, 40, 60 and 80 μA
respectively (see ESI Fig. 6†), indicating that the cell is still cm−2 and then for 25 cycles at increasing current densities up
electronically insulating with a small increase, possibly reflect- to 10 mA cm−2. An inset image displays a zoomed-in view of
ing lithium penetration reducing the effective thickness of the several voltage profiles at a current density of 10 μA cm−2 and
pellet. This is also reflected in the calculated transference 80 μA cm−2. Under these conditions Li2NH is shown to
numbers pre- and post-CV, where ti is found to be 0.981 and undergo over 750 hours of cycling (>1250 cycles) without a dra-
0.999 respectively, indicating dominant lithium-ion conduc- matic drop in current that might indicate a hard short-circuit.
tion. Post-CV there is a far greater spike in current upon initial However, during 40 μA cm−2 cycling, a slow decrease in the
polarisation (74 nA pre-CV to 1.06 mA post-CV) and a much voltage profile is observed resulting in a lower stable cycling
faster relaxation period compared to the pre-CV measurement. voltage. This voltage drop corresponds to the overall cell resis-
Fitting of both sets of data using exponential decay functions tance reducing each cycle and is also observed at higher
revealed a decrease in this current relaxation period of 4.48(6) current densities at accelerated rates. Whilst this could be the
s pre-CV to 2.28(1) s post-CV (ESI Fig. 6 and ESI Table 3†). result of slow dendrite penetration it is also consistent with
Given that this spike is typically associated with the initial observations from the CV experiments (Fig. 2) that ionic con-
movement of ionic charge carriers upon polarisation this inter- ductivity increases whilst cycling with a Li-metal electrode.
action is therefore indicated to result in an overall increase in EIS measurements (Fig. 3b–f ) were completed both before
the ionic conductivity of the cell. Cell disassembly and sub- and after S&P to understand further the effect on conductivity
sequent X-ray diffraction measurement (Fig. 2e) again indi- whilst cycling with Li-metal (Bode plots and phase angle data
cates no change in the bulk structure of Li2NH, nor formation of each EIS measurement are included in ESI Fig. 7†). Fig. 3b
of impurities detected during this process. Given the lack of and c show the Li|Li2NH|Li cell Nyquist plots before interface
apparent bulk chemical change, several possible mechanisms stabilisation and post-thermal treatment respectively. Before
for this increased conductivity are conceivable: stabilisation two diffusion mechanisms are observed, one at
1. An electrochemically-driven chemical or morphological 103–104 Hz ascribed to interfacial diffusion with Li-metal, and
process resulting in a more conductive electrode–electrolyte another at ∼107 Hz (with a phase angle of −45°) assigned to
interface. bulk diffusion. After stabilisation the interfacial component
2. An in situ reversible order–disorder phase transition of disappears. In both cases a low frequency inductive loop is
Li2NH to a simpler antifluorite cell (Fd3ˉm → Fm3 ˉm) resulting in observed. The exact cause of this remains ambiguous however
increased ionic diffusion as theorised in computational studies.50 these have been ascribed in other systems to interfacial stoi-
3. Grain boundary/particle surface stoichiometry variation chiometry variation and may reflect a similar effect at the
on a local scale undetectable via a bulk technique such as surface of the Li2NH particles.58,59 Fig. 3d displays the Nyquist
diffraction. plot of the cell immediately after cycling where a decrease in

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 531
 View Article Online

Paper EES Batteries

This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Fig. 3 (a) Lithium stripping and plating data of a Li|Li2NH|Li cell cycled at 40 °C at various increasing current densities. The positive and negative
currents are not displayed to not obscure the voltage data, a line indicating the set current density is instead used alternating between positive and
negative currents in 15-minute intervals. Nyquist plots of cell (b) before interface stabilisation, (c) after interface stabilisation, (d) immediately post-
cycling, (e) 2 days after cycling, and (f ) 1 month after cycling.

cell impedance by a factor of ∼1000 corresponding to a con- to the bulk structure nor appreciable formation of impurities
ductivity of >10 mS cm−2 is observed. The high frequency bulk (ESI Fig. 10†).
diffusion process observed before cycling is no longer present, Given the limitations of the structural changes which can
replaced by a process with a phase angle of 90°. Minimal be ascertained from average structure measurements, solid
capacitive behaviour is seen (ESI Fig. 8†), indicating that state NMR spectroscopy was employed as a means of assessing
charge is able to move freely through the cell.60 Combining local changes in the material. ESI Fig. 11† displays 7Li and 1H
these two observations points towards the change observed spectra for pristine and post-cycled Li2NH. The 7Li NMR spec-
being surface related. The conductivity calculated post-cycling trum of the pristine Li2NH consists of a sharp Lorentzian line
is on the order of the best solid-state ionic conductors known centred at 3.30 ppm with an underlying broader component
to date, however, the exact nature of this improvement is enig- centred at 3.40 ppm, suggesting two separate Li ion environ-
matic. The other possibility is the formation of soft short cir- ments, one more mobile than the other. The post-cycled
cuits resulting in mixed ionic-electronic conductivity, which sample displays a shift in the broader component to 3.03 ppm,
could result in a similar impedance spectrum to that shown in indicative of a minor change in this environment post cycling
Fig. 3d, though the post-CV DC polarisation experiments indi- and potentially minor stoichiometry variation. It should be
cated minimal evidence for this. noted that changes in signal intensity between the two
Fig. 3e and f display EIS measurements on the same cell materials remain inconclusive due to measurements having
after set periods of time at rest. Over this time the cell is been run at different magnetic fields (the cycled sample
observed to display relaxation/recovery behaviour. This is having therefore experienced greater polarisation) and on
reflected in both an observed recovery of cell resistance as well different probes (with different Q-factors). However, qualitat-
as in the phase angle data, where two days after cycling the ively we observe a drop in the 7Li signal intensity between the
high-frequency angle returns to around 45° indicating a bulk pristine and cycled material, perhaps suggesting Li loss. A
diffusion limited process. After one month, the low frequency change is also observed in the 1H NMR spectra where the pris-
diffusion process assigned to a Li2NH|Li interfacial com- tine sample shows a variety of environments associated with
ponent returns, whereby the impedance spectrum is compar- the NHx groups: one at −4.35 ppm which is likely the imide
able to that of prior to stabilisation. This implies a dynamic groups, one at −1.53 ppm which may be residual amide
and metastable interfacial process underpins the increased groups as discussed above, and some very low intensity reso-
conductivity observed. Repeat experiments performed on sep- nances at 3.97 ppm, which are in the chemical shift region of
arate cells show similar conductivity improvement and corres- saline hydrides.61 Post cycling, the main peak shifts to −
ponding recovery behaviour (ESI Fig. 9†). Synchrotron X-ray 5.13 ppm and the second peak to −2.28 ppm again indicating
diffraction of the post-cycled material shows again no change stoichiometry variation compared to the pristine sample (dis-

532 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

cussed in further detail below). Furthermore, the echo delay in

the Hahn echo pulse sequence can be increased to filter out
fast relaxing components (i.e. having a short transverse, T2,
relaxation time). Such T2-filtered 1H spectra (ESI Fig. 12†) high-
light a higher proportion of less mobile hydrogen environ-
ments in the post-cycled sample. Both amide and hydride
anions are expected to be less mobile than the free rotation of
the imide group within Li2NH, suggesting the presence of one/
both of these anions in low quantities. Whilst these data indi-
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

cate that the observed process is potentially non-stoichio-

Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

metric in its nature, the metastability demonstrated by the

observed recovery behaviour limits the insight offered by ex
situ analysis on the mechanism of the conductivity increase.
Therefore, in situ diffraction and NMR measurements were
performed to gain temporal resolution.

In situ X-ray diffraction and solid-state

7
Li NMR
An accelerated S&P experiment with in situ synchrotron diffrac-
tion was carried out to assess bulk structure behaviour during
cycling, particularly with regard to potential metastable phases
or stoichiometry variation. In this case cycling was started at
40 μA cm−2 and doubled every 3 hours up to 10 mA cm−2. The
electrochemical data for this experiment are displayed in ESI
Fig. 13† with a contour plot of the observed diffraction data at
each indicated current density displayed in Fig. 4a (example
diffraction pattern shown in ESI Fig. 14†). Again, despite the
accelerated timescale of this experiment, there is no sudden
and sustained drop in voltage to indicate a short circuit and a
similar lowering of resistance is observed. The diffraction data
display no change in the average structure of Li2NH over the
course of this experiment. Peaks corresponding to a small
Li3N component are present throughout and are thought to
originate on the outside lithium metal surface closest to the
Kapton window of the cell, where visible nitriding could be
seen. ESI Fig. 15† displays a separate cell run on a further Fig. 4 Contour plots of (a) synchrotron X-ray diffraction data and (b) 7Li
accelerated timescale (current density doubled every hour) SS non-spinning NMR spectra from Li|Li2NH|Li coin cells during lithium
stripping and plating at the indicated current densities, with (c) key
along with cycling for 18 hours at 10 mA cm−2. The lowering of
SSNMR spectra fitted with a CSA static model. A schematic of the pro-
resistance and stable cycling behaviour at 10 mA cm−2 is well posed mechanism of enhanced grain boundary Li-ion conduction
demonstrated here together with no corresponding change in through non-stoichiometric Li–N–H phase formation at grain bound-
average structure. After a rest period (approximately 6 hours) aries, and subsequent diffusion into the grains is shown in (d).
this cell was rerun on a similar programme from 40 μA cm−2
up to 70 mA cm−2. At the highest current densities, a large
polarisation in the voltage profile of each cycle is observed increase in conductivity may be due to small-scale stoichio-
along with variation in the measured diffraction signal for Li metry variation confined to the grain boundary.
metal, but it again appears that there is no hard short circuit. An in situ 7Li NMR S&P experiment was performed to
The flattened voltage profiles for the second cell which evolve further probe this hypothesis. The electrochemical data for
at 10 mA cm−2 (Fig. S15†) have been suggested to reflect the this experiment are displayed in ESI Fig. 16† with a contour
presence of soft-short circuits,62 although it is difficult to plot of the observed NMR spectra in Fig. 4b. The spectra show
determine this unambiguously. The lack of structural variation a strong signal centred close to 0 ppm from Li2NH and a
whilst cycling excludes the possibility of a bulk phase tran- smaller signal from Li metal at around 260 ppm. In this case,
sition of Li2NH from Fd3 ˉm → Fm3 ˉm. This, combined with the the cell was cycled at each current density for 3 hours then
ex situ NMR data, strengthens the hypothesis that the observed doubled, starting from 40 μA cm−2 up to 2.56 mA cm−2. The

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 533
 View Article Online

Paper EES Batteries

cell was then cycled at 320 μA cm−2 for 12 hours before a 12-hour Fig. 4d for a schematic depiction of this process). We stress
rest period. The electrochemical data can be observed to be that this variation in stoichiometry is very minor and not sig-
noisy, possibly indicating dendrite formation although this is nificant enough to alter the average structure of the material
countered by the observed recovery of the voltage profiles. No sig- and hence we do not observe this relaxation via diffraction.
nificant change in the shift of the Li2NH peak is observed whilst Additional surface sensitive Dynamic Nuclear Polarisation
current is applied, indicating that the local Li environment (DNP) NMR experiments could help confirm the local environ-
within Li2NH does not change drastically during cycling. The ment at the grain boundary.
absence of significant shifts outside the diamagnetic region As this potential stoichiometric variation occurs whilst in
(except for Li metal itself) also confirms that the sample does not contact with lithium metal, it is prudent to consider any poss-
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

become electronically conductive.63 The development of a small ible reactions between the two materials. Eqn (1) and (2) detail
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Li metal component (∼267 ppm) in addition to the original Li possible mechanisms for stoichiometric variation.
metal signal (∼263 ppm) indicates the expected formation of Li
microstructures during cycling. The deposited Li is likely to be xLi þ Li2 NH ! Li2þx NH ð1Þ
rough (the Li–Li2NH interphase is quite flat) rather than dendri-
ðx þ 1ÞLi2 NH ! Li2x NH1þx þ xLi3 N ð2Þ
tic, as the latter experiences different bulk magnetic susceptibility
effects, since it grows perpendicularly to the Li metal in the cell Eqn (1) relates to the formation of surface imide-nitride-
(and to the applied field), resulting in a larger shift (∼10 ppm) hydride through reaction with lithium, whilst eqn (2) details
from that of the bulk Li signal.64 the formation of an imide-amide phase and lithium nitride. As
Monitoring the variation in the integral of the Li metal and such, an assessment of the nature of the stoichiometry vari-
Li2NH peaks over time (ESI Fig. 17†) reveals initially a drop in ation observed was conducted via NMR measurement of two
the Li-metal signal in the first hour followed by an increase in more ex situ samples: a lithium imide-amide (Li deficient com-
the next 14 h, this increase being consistent with some for- pared to Li2NH, Li1.917NH1.083) and a lithium imide-nitride-
mation of Li microstructures at the Li–Li2NH interface. A hydride (Li2.083NH) (ESI Fig. 18†). Both of these materials also
similar drop is observed in the Li2NH signal, which then take an antifluorite-type structure and simply represent a shift
recovers as the current density is doubled from 40 μA cm−2 to in bulk stoichiometry compared to Li2NH (see ESI Fig. 19† for
80 μA cm−2 (a decrease in the cell potential is also observed at diffraction and structural data). The 7Li and 1H NMR spectra
this point). The signal then slowly decays throughout the of these samples confirm a change in the local structure com-
remaining cycling period suggesting potential Li loss from the pared to pristine Li2NH (ESI Fig. 11†). For both the imide-
Li2NH phase. The 7Li NMR spectra corresponding to pre- amide and imide-nitride-hydride, the 7Li spectra contain two
cycling, at the final cycle, and after 12 h resting at zero current resonances (sharp and broad). These are centred at 2.08 ppm
(Fig. 4c) were fitted using a chemical shift anisotropy (CSA) (sharp) and 1.56 ppm (broad) for the amide-imide and
static model: two components were fitted, one quasi-axial 3.36 ppm (sharp) and 3.03 ppm (broad) for the imide-nitride-
likely corresponding to Li in the interstitial sites (broader, in hydride (we note that the linewidth of the imide-nitride-
purple in the spectra) and one rhombic assigned to the Li in hydride is however ∼2× that of the other samples analysed
tetrahedral sites (narrow and in blue). The fits show that the here). In both samples there is a shift to lower ppm in the
broad component progressively shifts during cycling (3.40 → broad component as compared to pristine Li2NH. Observing
3.25 ppm), with a slightly more significant shift occurring the 7Li spectra in isolation, the post-cycled material appears to
during the rest period (3.25 → 3.03 ppm), while the narrow resemble the imide-nitride-hydride most closely, particularly
component remains approximately constant at ∼3.30 ppm. considering the chemical shift of the broad component is
While it is difficult to unambiguously determine the nature identical (3.03 ppm). However, the 1H spectra show two reso-
of the different local environments, the observed changes in nances centred at −5.59 and −2.74 ppm for the amide-imide
chemical shift after cycling suggest that the stoichiometry of and three resonances at −4.98, 0.67 and 3.00 ppm for the
Li2NH varies upon cycling. This may give insight into the imide-nitride-hydride. In this case the 1H spectrum of the
observed changes in conductivity and resistance recovery post-cycled sample most closely resembles the spectrum of the
behaviour discussed above. We hypothesise that under applied imide-amide: in both spectra the main component (light blue
current or potential there is a stoichiometry variation at the in ESI Fig. 18†) shifted to a more a negative ppm compared to
grain boundary leading to a disordering of the surface of the pristine Li2NH. Comparison to these fixed off-stoichiometry
Li2NH particles and a highly conductive state, similar to that samples is therefore likely to be a simplification of the overall
described for Li2NH ammonia cracking catalysts.57,65 There is picture and it is possible that formation of imide-amide and
likely some exchange of this highly conductive state with the imide-nitride-hydride phases are happening simultaneously
bulk (reflected by the change in signal before and immediately during cycling. This is potentially evidenced by the complex
after cycling). However, the change is metastable and when the range of 1H environments revealed in the T2-filtered experi-
electrochemical bias is removed the compositional gradients ments. We do however note that observation of Li loss from
lessen through diffusion further into the bulk grain, as rep- the Li2NH phase in our in situ NMR and CV experiments,
resented by the greater shift in signal after rest and resulting suggest the change in stoichiometry is on average towards an
in the recovery in resistance observed in the EIS spectra (see imide-amide type phase (Li2−xNH1+x). Furthermore, it is these

534 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

lithium imide-amide phases that have been computationally

observed to exhibit significant surface disorder.65
It is difficult to rule out the possibility that soft-short circuit-
ing might be also contributing to the enhanced conductivity
observed in the stripping and plating experiments66 and is cer-
tainly possible in the higher current density data presented.
However, the lack of voltage spikes or sudden drops normally
associated with shorting behaviour, the persistent low electronic
conductivity measured within a few minutes of the CV experi-
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

ments, and the absence of dendritic Li microstructures in the

Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

in situ NMR experiments indicates atypical behaviour for soft-

shorting. Furthermore, the system appears quite stable against
hard-shorting, which is encouraging. It may be that penetration
of Li into the pellet is at the heart of the observed stoichiometry
variation along grain boundaries, with rapid reaction of dendrites
resulting in more conductive pathways. It is clear that further
analysis of this phenomenon is required.

Li-metal battery proof-of-concept

Fig. 5 (a) Charge–discharge cycles of a Li|Li2NH|TiS2 cell at 40 °C, C/
tests 48. (b) Capacity and coulombic efficiency vs. cycle number of Li|Li2NH|
TiS2 cell. (c) Charge–discharge cycles of a Li|Li2NH|LiFePO4 cell at
The proof-of-concept performance of Li2NH in Li-metal battery 40 °C, C/35. (d) Capacity and coulombic efficiency vs. cycle number of
cells was evaluated using a Li-metal anode and two different Li|Li2NH|LiFePO4 cell.
cathode materials (TiS2 and LiFePO4). In both cases a small
amount of liquid electrolyte (10 μL, LiPF6 in EC : DMC for TiS2,
LiPF6 in EC : EMC with 2 wt% vinyl chloride for LFP) was used
to wet the solid electrolyte–cathode interface forming so-called
Summary
hybrid solid-state batteries. Fig. 5 shows charge–discharge The results reported here demonstrate the promising perform-
measurements operated at 5 mA g−1 for each of these cells ance of antifluorite Li2NH for application as a solid electrolyte
(Fig. 5a – TiS2, Fig. 5b – LFP) as well as capacity and columbic in solid-state Li-ion batteries. A high initial ionic conductivity
efficiency as a function of cycle number (Fig. 5c – TiS2, Fig. 5d – of Li2NH is further improved through modest post-synthetic
LFP). With both cathode materials multiple charge–discharge milling and hot-pressing regimes resulting in the highest
cycles are observed, demonstrating the first application of Li2NH reported conductivity for this material at 1 mS cm−1. High
as a functioning solid electrolyte in a full cell. With a TiS2 current operation at up to 70 mA cm−2 as well as wide voltage
cathode, over 40 cycles were completed with a relatively high stability (0–5 V vs. Li/Li+) is reported. An interaction with Li-
initial discharge capacity of 126 mA h g−1. Capacity fade was metal electrodes is also observed, with an apparent improve-
observed in both cells, with the TiS2 cell terminating after 42 ment in conductivity to >10 mS cm−1 whilst appearing to
cycles and the LFP cell diminishing after 18 cycles. This cell remain electronically insulating. This interaction is hypoth-
degradation is likely from the formation of an unstable CEI layer esised to occur along the grain boundary of the material and
at the Li2NH–cathode interface due to a reaction with the liquid may be as a result of stoichiometry variation at the Li2NH par-
electrolyte, and is particularly prominent at voltages above 3.5 V, ticle surface producing a highly disordered surface. The
hence the worse performance of the LFP cells. This is supported impressive performance of this material in these initial investi-
by the inferior stability of LFP cells when the 2 wt% vinyl chloride gations suggests the potential for further enhancements.
additive (included for more favourable CEI formation) is not Through the understanding of the wealth of possible compo-
used in the liquid electrolyte formulation (ESI Fig. 18†). In either sition modifications of lithium imide from the hydrogen
case, no indication of reduction in cell impedance analogous to storage literature, a new family of imide-based ionic conduc-
the S&P experiments was observed, however, it is possible that tors are open for further investigation.
this is due to the low number of cycles in these battery cycling
experiments. As this is the first demonstration of this material in
operation, we anticipate that further optimisation of cell formu-
lation both in hybrid and all-solid-state configurations will Methods
further improve performance.65 At the very least, given the
known ability of the antifluorite structure to incorporate amide, Sample handing
nitride, hydride and halides, there is significant potential for a All sample handing was performed in an argon filled glovebox
new family of high-conductivity materials to be explored.31,67,68 (MBraun, Unilab, <0.1 ppm H2O, <0.1 ppm O2).

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 535
 View Article Online

Paper EES Batteries

Synthesis Preparation of Li2NH for electrochemical measurements

Synthesis of Li2NH was achieved by the solid-state reaction of The Li2NH powder was pressed into a pellet (diameter 10 mm,
LiNH2 (Sigma Aldrich hydrogen storage grade) with Li3N thickness 1–2 mm) under an Ar atmosphere. For Au|Li2NH|Au
(Sigma Aldrich >99.5%) in stoichiometric quantities. cells, a small amount of vacuum grease was applied to the
curved side edge of each pellet before both sides were sput-
Li3 N þ LiNH2 ! 2Li2 NH
tered with gold using an Agar auto sputter coater placed inside
The reagents were weighed out in order to form 1 g of an argon filled glove bag. Sputtered pellets were returned to a
Li2NH (0.3973 g LiNH2 and 0.6027 g Li3N). Samples were first glovebox and the gold and grease were removed from the sides
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

hand-ground using an agate pestle and mortar for 2 minutes to leave gold electrodes only on the faces of each pellet. For Li|
before being transferred into a Si3N4 milling jar (volume Li2NH|Li cells, lithium foil (Pi-KEM, 0.25 mm thickness) was
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

45 ml) prefilled with 20 g of 5 mm diameter Si3N4 milling cut into 10 mm circles (0.0099 g) using a hole punch and any
balls. The jar was sealed and transferred from the glovebox to surface oxidation was removed using a spatula. The Li foil was
a Fritsch Pulverisette 7 Premium Line planetary micro mill. then applied to both sides of the pellet and light pressure was
The mixture was milled at a rate of 150 rpm for 1 hour. Once applied by hand. For Li|Li2NH|Steel cells the Li metal was only
returned to the glovebox the resultant mixture was transferred applied to one side of the pellet. In all cases, each cell was con-
into a quartz tube and fitted with a Young’s tap T-piece con- structed into coin cells for all electrochemical measurements.
nected by an Ultra-Torr fitting. This reaction vessel was then Coin cell assembly took place by placing each pellet with
clamped into a tube furnace (Lenton Furnaces, LTF 12/25/250 attached electrodes in a CR2032 cell cap (0.9150 g). A stainless-
fitted with a Eurotherm 3216P1 controller) and gas lines were steel spacer (0.5 mm thickness, 0.7422 g) and spring (0.25 mm
attached either side of the Young’s tap. Argon gas was then thickness, 0.1983 g) were placed on top followed by a CR2032
allowed to flow through the tap. The furnace was ramped to cell cap with an attached O-ring (0.8853 g). The cell was then
325 °C at a rate of 2 °C min−1 and held at that temperature for sealed using a Hohsen Corp coin cell crimper. Stabilisation of
12 hours. The sample of Li1.917NH1.083 was synthesised by the Li–Li2NH interface was conducted on already constructed
varying the Li3N to LiNH2 ratio to give the appropriate coin cells by placing cells in an oven (Memmert, UN55) at
stoichiometry. 60–80 °C for 6–8 hours.
For the lithium imide-nitride-hydride sample (Li2.083NH),
lithium nitride hydride was synthesised by the reaction of Electrochemical impedance spectroscopy
lithium nitride with lithium hydride (Sigma Aldrich, 98%)
Variable temperature electrochemical impedance spectroscopy
according to the following reaction:
measurements were taken in a Carbolite VST 12/400 tube
Li3 N þ LiH ! Li4 NH: furnace heated to a maximum of 110 °C, with the impedance
The powder mixture was milled as above, pressed into a measurements (100 mV perturbation) being taken by a Hewlett
pellet and then heated in a microwave reactor (CEM Discover) Packard 4192A LF between frequencies of 5 Hz to 13 MHz.
for five rounds of 1-minute heating at 300 W under argon flow. Measurements on Li|Li2NH|Li cells were taken using a
The synthesised lithium nitride hydride was then mixed with Solartron 1260 impedance analyser (100 mV perturbation)
lithium imide, milled as above and heated to 540 °C under from 1 mHz to 10 MHz. Impedance data were analysed using
flowing argon for 12 hours (2 °C min−1 ramp rate): Zview software by Scribner. In general, resistance values for
Li2NH were taken as the high intercept of the first semi-circle.
0:083  Li4 NH þ 1:917  Li2 NH ! 2Li2:083 NH: Where data were fitted to equivalent circuit models, resistance
of the material was taken as the combined resistance of R1
X-ray diffraction and R2 (see ESI Fig. 1 and ESI Table 1†). Area specific resis-
Laboratory powder X-ray diffraction (XRD) measurements were tance was calculated by normalising Li|Li2NH|Li data to the
carried out using a Stoe Stadi-P instrument (Mo Kα1 source) surface area of the pellet (0.7854 cm2) and dividing the calcu-
with samples sealed in capillaries made from Cole-Parmer lated resistance by 2 to account for the two Li electrodes. For
polyimide tubing (0.7 mm internal diameter). Rietveld analysis Fig. 3d (ESI Fig. 7†) the resistance was taken as the high inter-
of XRD patterns was performed using TOPAS Academic cept of the assumed semi-circle (approximately 61 Ω cm).
software.69
DC polarisation
Raman spectroscopy
Electronic conductivity of Li2NH was evaluated using DC polar-
Raman measurements were taken using Renishaw InVia
isation experiments. Different voltages (0.5 V & 0.75 V) were
Raman microscope. Samples were loaded in a borosilicate
applied for a period of 5 hours on an Au|Li2NH|Au cell held at
glass capillary (0.7 mm internal diameter) to an approximate
40 °C using a Biologic VMP3. The electronic conductivity was
height of 5 mm and sealed with vacuum grease. A laser wave-
calculated using the following equation:
length of 532 nm was used to analyse the samples. Spectra
were taken in the region of 3000 cm−1 to 3400 cm−1 using a L If
σe ¼  ð3Þ
laser power of 0.5% and a typical exposure time of 10 seconds. A V

536 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

where L is the thickness of the pellet, A is the surface area of Li2NH spectra were acquired on a 16.4 T (ωH = 700 MHz)
the pellet, V is the applied potential, and If refers to the Bruker Avance III spectrometer. In both cases a Bruker 1.3 mm
average current of the last half hour of the relaxation period. magic angle spinning (MAS) probe was used with a MAS fre-
Transference numbers were calculated from the voltage vs. quency of 40 kHz. The spectra were externally referenced to
time plots using the following equations: glycine (spun at 20 kHz) at 8.00 ppm (δ 1H) and Li2CO3 at
0.00 ppm (δ 7Li). In all cases, a rotor-synchronised Hahn-echo
If
te ¼ ð4Þ pulse sequence (90°–τ–180°–τ–acquire) was used for quantitat-
It
ive measurements. The recycle delay adopted was at least 5 ×
ti ¼ 1  te ð5Þ T1, the T1 having been measured via a saturation recovery
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

pulse sequence ((sat)n–τ–90°–acquire). For 7Li NMR, pulse

where te and ti are the electronic and ionic transference
lengths of 1.0 μs (at 11.7 T) and 2.05 μs (at 16.4 T) and recycle
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

numbers respectively and It is the initial current upon

delays of 1.36 s ( pristine), 10 s (cycled), 2.65 s (Li1.917NH1.083)
polarisation.
were set; for 1H NMR, pulse lengths of 0.87 μs (at 11.7 T) and
Cyclic voltammetry 2.15 μs (at 16.4 T) and recycle delays of 28.9 s ( pristine), 23.4 s
(cycled) and 26.5 s (Li1.917NH1.083) were set. All spectra were
Cyclic voltammetry was conducted on Li|Li2NH|Steel cells scaled according to the mass of the sample and number of
held at 40 °C using a Biologic VMP3 across a range of voltages residuals recorded.
from −0.5 V up to 5 V vs. Li/Li+ at a scan rate of 1 mV s−1. In situ NMR experiments were conducted on a 7.05 T (ωH =
Electronic conductivity was evaluated both before and immedi- 300 MHz) Bruker Avance NMR spectrometer equipped with an
ately after the CV experiment using the Hebb–Wagner in situ NMR probe (NMR Service GmbH) with automatic
method.71 The same cell had a 0.5 V potential applied for tuning and matching capabilities and built-in highly shielded
5 hours and the current response measured. The electronic electrochemistry connections. Additional radiofrequency low-
conductivity was calculated as per eqn (3). Transference pass filters were used on the connection to the potentiostat to
numbers were calculated as per eqn (4) and (5). prevent interference. A 12 mm inner diameter solenoid coil
was used and the cell (made of polyether ether ketone, PEEK)
Lithium stripping and plating
was oriented so that the lithium chips and the electrolyte
Lithium stripping and plating experiments were conducted on pellet were parallel to the main magnetic field.70 During
symmetrical Li|Li2NH|Li coin cells held at 40 °C using a cycling, NMR spectra were recorded continuously using a one
Biologic VMP3. These symmetric cells were cycled under con- pulse sequence with a pulse length of 5.6 μs and a recycle
stant current conditions at various current densities for delay of 1 s, quantitative for Li metal and providing enough
15 minutes plus a 10 s rest per half cycle making each full signal and time resolution for the solid electrolyte. The time
cycle approximately 30 minutes long. resolution was of ∼1 min per spectrum. The spectra were exter-
nally referenced to LiCl (aq) at 0.00 ppm.
Synchrotron X-ray powder diffraction
All spectra were recorded and processed using Bruker
Where appropriate, synchrotron measurements were under- Topspin 2.1, 3.6.2 and 3.6.4, fitted using DMfit software using
taken on the I11 beamline at Diamond Light Source, with the Chemical Shift Anisotropy (CSA) MAS/static model and
powder samples sealed in 0.5 mm borosilicate capillaries. analysed and plotted using home-written MATLAB scripts.71
Lithium stripping and plating experiments were conducted on
modified Li|Li2NH|Li coin cells with a 3 mm diameter Kapton
window to allow for beam penetration. These cells were cycled Charge–discharge measurements
under identical conditions to the stripping and plating experi- Charge–discharge measurements were undertaken using a
ments except on an accelerated timescale and using an Ivium Biologic VMP3 cell tester with the cells held at 40 °C, cells
Octostat potentiostat. Diffraction data (λ = 0.824042 Å) were were rested for at least 24 h prior to cycling. To assemble Li|
collected on each of the two cycling cells every 100 s using Li2NH|LiFePO4 cells, Li metal was first applied to one side of a
position sensitive detectors (PSD), with 10 s total collection cold pressed pellet (10 mm diameter, 1.2 mm thickness,
time per scan. The cells were aligned in the beam relative to excess capacity = 38.2 mA h) before being clamped on a hot
the diffraction pattern of a 0.5 mm borosilicate capillary plate and held at 60 °C for 1 hour followed by 180 °C for
measurement of lithium imide. 2 hours. Once cooled, the pellet would be placed in a CR2032
cell cap and 10 μL of 1M LiPF6 in EC : DMC with 2 wt% vinyl
Solid state NMR chloride (1 : 1, Sigma Aldrich) was added dropwise to the
Solid-state Nuclear Magnetic Resonance (NMR) experiments surface of the pellet. Commercially purchased LiFePO4 elec-
were performed on the pristine Li2NH powder as synthesized, trode sheets (Pi-KEM) with an active loading of 7.44 mg cm−2
after cycling (ex situ) and in situ during electrochemical (cut into 12 mm disks) were placed onto the stack with a steel
cycling. Ex situ solid state NMR spectra on the pristine spacer (0.5 mm thickness) and spring (0.25 mm thickness) fol-
material and Li1.917NH1.083 were acquired on a 11.7 T (ωH = lowed by a CR2032 cell cap with an attached O-ring. The cell
500 MHz) Bruker Avance III spectrometer, while the cycled was then sealed using a Hohsen Corp coin cell crimper.

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 537
 View Article Online

Paper EES Batteries

For Li|Li2NH|TiS2 cells, a cathode slurry was prepared by acknowledged for the provision of beamtime (CY35016), and
mixing 80 wt% TiS2 (99.9%, Sigma Aldrich, ball milled at 200 Stephen Thompson is acknowledged for support during this
pm for 1 hour) as an active material, 8 wt% polyvinylidene flu- experiment. Ezri McGarrigle is acknowledged for assistance
oride (PVDF 5130, Solvay) as a binder and 12 wt% carbon with the schematic diagram (Fig. 4d). Josh Deakin and Louise
black (TimCal, C65) as a conducting additive with N-methyl-2- Male are acknowledged for support with the laboratory powder
pyrrolidone (NMP) as the solvent in a THINKY mixer at 2000 diffraction and Raman equipment.
rpm for 15 minutes. The slurry was coated uniformly on alu-
minium foil using the doctor blade coating technique and
dried in a vacuum oven at 120 °C for 24 h. The electrodes were
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

cut into 12 mm disks for further use. The active material References
loading of the cathode was maintained at 3 mg cm−2. A cold-
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

pressed Li2NH pellet of 10 mm diameter and 1.2 mm thick- 1 D. Lin, Y. Liu and Y. Cui, Reviving the lithium metal anode
ness was used as the SSE. Li metal was used as an anode and for high-energy batteries, Nat. Nanotechnol., 2017, 12(3),
was melded on one side of the pellet as described above 194–206.
(37.9 mA h excess capacity). 10 µL of 1 M LiPF6 in EC/EMC 2 W. Xu, J. Wang, F. Ding, X. Chen, E. Nasybulin, Y. Zhang,
(1 : 1) liquid electrolyte was again added while assembling the et al., Lithium metal anodes for rechargeable batteries,
coin cells. The cells were discharged and charged at a current Energy Environ. Sci., 2014, 7(2), 513–537.
density of 5 mA g−1, cycling between 1.5 V and 3.2 V. 3 P. Bai, J. Li, F. R. Brushett and M. Z. Bazant, Transition of
lithium growth mechanisms in liquid electrolytes, Energy
Environ. Sci., 2016, 9(10), 3221–3229.
Author contributions 4 T. Famprikis, P. Canepa, J. A. Dawson, M. S. Islam and
C. Masquelier, Fundamentals of inorganic solid-state elec-
JWM and JPL conceived the study and analysed the data. JPL trolytes for batteries, Nat. Mater., 2019, 18(12), 1278–1291.
conducted the experiments. PAA, PRS and EMK assisted with 5 J. W. Fergus, Ceramic and polymeric solid electrolytes for
electrochemical data interpretation and experiment design. lithium-ion batteries, J. Power Sources, 2010, 195(15), 4554–
TVB collected and analysed the TiS2 cell data. TI and CPG 4569.
designed experiments and analysed data from the solid-state 6 A. Manthiram, X. Yu and S. Wang, Lithium battery chem-
NMR spectroscopy. SD assisted with collection of synchrotron istries enabled by solid-state electrolytes, Nat. Rev. Mater.,
X-ray diffraction data. MPS and BD assisted with the electro- 2017, 2(4), 1–16.
chemical experiments and associated analysis. JWM and JPL 7 Z. Zhang, Y. Shao, B. Lotsch, Y. S. Hu, H. Li, J. Janek, et al.,
prepared the manuscript with input from all authors. New horizons for inorganic solid state ion conductors,
Energy Environ. Sci., 2018, 11(8), 1945–1976.
8 J. B. Goodenough and Y. Kim, Challenges for rechargeable
Data availability Li batteries, Chem. Mater., 2010, 22(3), 587–603.
9 T. Zhang, B. Fuchs, M. Secchiaroli, M. Wohlfahrt-Mehrens
ESI† is available for this paper. Raw data supporting the
and S. Dsoke, Electrochemical behavior and stability of a
figures presented in the paper can be accessed through the
commercial activated carbon in various organic electrolyte
University of Birmingham e-data repository (https://doi.org/
combinations containing Li-salts, Electrochim. Acta, 2016,
10.25500/edata.bham.00001212). Correspondence and
218, 163–173.
requests for materials should be addressed to Dr Joshua
10 V. Thangadurai, S. Narayanan and D. Pinzaru, Garnet-type
Makepeace ( j.w.makepeace@bham.ac.uk).
solid-state fast Li ion conductors for Li batteries: Critical
review, Chem. Soc. Rev., 2014, 43(13), 4714–4727.
11 C. Wang, K. Fu, S. P. Kammampata, D. W. McOwen,
Conflicts of interest A. J. Samson, L. Zhang, et al., Garnet-Type Solid-State
The authors declare no competing interests. Electrolytes: Materials, Interfaces, and Batteries, Chem.
Rev., 2020, 120(10), 4257–4300.
12 R. Murugan, V. Thangadurai and W. Weppner, Fast lithium
Acknowledgements ion conduction in garnet-type Li7La3Zr2O12, Angew. Chem.,
Int. Ed., 2007, 46(41), 7778–7781.
JWM acknowledges a Future Leaders Fellowship (MR/S03403X/ 13 N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama,
1). JPL and MPS acknowledge the University of Birmingham R. Kanno, M. Yonemura, et al., A lithium superionic con-
for studentship funding. PRS acknowledges the EPSRC grant ductor, Nat. Mater., 2011, 10(9), 682–686.
(GENESIS project: EP/R024006/1), and the Faraday Institution 14 S. Boulineau, M. Courty, J. M. Tarascon and V. Viallet,
CATMAT grant (FIRG057 & FIRG063) for funding. T. I. and Mechanochemical synthesis of Li-argyrodite Li6PS5X (X =
C. P. G. were supported by an ERC Advanced Investigator Cl, Br, I) as sulfur-based solid electrolytes for all solid state
Grant for C. P. G. (EC H2020 835073). Diamond Light Source is batteries application, Solid State Ionics, 2012, 221, 1–5.

538 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry
 View Article Online

EES Batteries Paper

15 P. Adeli, J. D. Bazak, K. H. Park, I. Kochetkov, A. Huq, 29 J. W. Makepeace, T. J. Wood, H. M. A. Hunter, M. O. Jones

G. R. Goward, et al., Boosting Solid-State Diffusivity and and W. I. F. David, Ammonia decomposition catalysis
Conductivity in Lithium Superionic Argyrodites by Halide using non-stoichiometric lithium imide, Chem. Sci., 2015,
Substitution, Angew. Chem., Int. Ed., 2019, 58(26), 8681–8686. 6(7), 3805–3815.
16 A. Sharafi, S. Yu, M. Naguib, M. Lee, C. Ma, H. M. Meyer, 30 C. A. Brooker-Davis, J. W. Makepeace and T. J. Wood,
et al., Impact of air exposure and surface chemistry on Li- Enhancement of the Catalytic Activity of Lithium Amide
Li7La3Zr2O12 interfacial resistance, J. Mater. Chem. A, 2017, towards Ammonia Decomposition by Addition of
5(26), 13475–13487. Transition Metals, J. Ammonia Energy, 2023, 01, 46–58.
17 T. Krauskopf, H. Hartmann, W. G. Zeier and J. Janek, 31 J. W. Makepeace, J. M. Brittain, A. Sukhwani Manghnani,
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

Toward a Fundamental Understanding of the Lithium C. A. Murray, T. J. Wood and W. I. F. David, Compositional
Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

Metal Anode in Solid-State Batteries - An Electrochemo- flexibility in Li-N-H materials: implications for ammonia
Mechanical Study on the Garnet-Type Solid Electrolyte catalysis and hydrogen storage, Phys. Chem. Chem. Phys.,
Li6.25Al0.25La3Zr2O12, ACS Appl. Mater. Interfaces, 2019, 2021, 23(28), 15091–15100.
11(15), 14463–14477. 32 B. A. Boukamp and R. A. Huggins, Ionic conductivity in
18 L. R. Mangani and C. Villevieille, Mechanical: Vs. chemical lithium imide, Phys. Lett. A, 1979, 72(6), 464–466.
stability of sulphide-based solid-state batteries. Which one 33 W. Li, G. Wu, Z. Xiong, Y. P. Feng and P. Chen, Li + ionic
is the biggest challenge to tackle? Overview of solid-state conductivities and diffusion mechanisms in Li-based
batteries and hybrid solid state batteries, J. Mater. Chem. A, imides and lithium amide, Phys. Chem. Chem. Phys., 2012,
2020, 8(20), 10150–10167. 14(5), 1596–1606.
19 P. Hagenmuller, J. M. Réau, C. Lucat, S. Matar and 34 B. Paik and A. Wolczyk, Lithium Imide (Li2NH) as a Solid-
G. Villeneuve, Ionic conductivity of fluorite-type fluorides, State Electrolyte for Electrochemical Energy Storage
Solid State Ionics, 1981, 3–4(C), 341–345. Applications, J. Phys. Chem. C, 2019, 123(3), 1619–1625.
20 N. I. Sorokin and B. P. Sobolev, Nonstoichiometric fluor- 35 J. Zhang and Y. H. Hu, Promoter effect of Cl- anion on
ides-Solid electrolytes for electrochemical devices: A review, decomposition of lithium amide and lithium imide, in ACS
Crystallogr. Rep., 2007, 52(5), 842–863. National Meeting Book of Abstracts, 2011, pp. 8058–8064.
21 J. M. Reau and P. Hagenmuller, Fast ionic conductivity of 36 P. J. Haigh, R. A. Forman and R. C. Frisch, Nuclear mag-
fluorine anions with fluorite -or tysonite-type structures, netic resonance of 7Li and 1H in solid lithium imide,
Rev. Inorg. Chem., 1999, 19(1–2), 45–77. lithium amide, and lithium nitride, J. Chem. Phys., 1966,
22 M. Mogensen, D. Lybye, N. Bonanos, P. V. Hendriksen and 45(3), 812–816.
F. W. Poulsen, Factors controlling the oxide ion conduc- 37 T. Mueller and G. Ceder, Ab initio study of the low-tempera-
tivity of fluorite and perovskite structured oxides, Solid ture phases of lithium imide, Phys. Rev. B:Condens. Matter
State Ionics, 2004, 174(1–4), 279–286. Mater. Phys., 2010, 82(17), 1–7.
23 P. Molaiyan and R. Witter, Crystal phase and surface defect 38 G. Miceli, M. Ceriotti, S. Angioletti-Uberti, M. Bernasconi
driven synthesis of Pb1−xSnxF2 solid solution electrolyte for and M. Parrinello, First-principles study of the high-temp-
fluoride ion batteries, J. Electroanal. Chem., 2019, 845, 154– erature phase of Li2NH, J. Phys. Chem. C, 2011, 115(14),
159. 7076–7080.
24 C. Szczuka, B. Karasulu, M. F. Groh, F. N. Sayed, 39 Y. H. Hu and E. Ruckenstein, Ultrafast reaction between
T. J. Sherman, J. D. Bocarsly, et al., Forced Disorder in the Li3N and LiNH2 to prepare the effective hydrogen storage
Solid Solution Li3P-Li2S: A New Class of Fully Reduced material Li2NH, Ind. Eng. Chem. Res., 2006, 45(14), 4993–
Solid Electrolytes for Lithium Metal Anodes, J. Am. Chem. 4998.
Soc., 2022, 144(36), 16350–16365. 40 T. Ichikawa and S. Isobe, The structural properties of
25 W. Li, M. Li, P. H. Chien, S. Wang, C. Yu, G. King, et al., amides and imides as hydrogen storage materials, Z.
Lithium-compatible and air-stable vacancy-rich Li9N2Cl3 Kristallogr., 2008, 223(10), 660–665.
for high-areal capacity, long-cycling all-solid-state lithium 41 T. Noritake, H. Nozaki, M. Aoki, S. Towata, G. Kitahara,
metal batteries, Sci. Adv., 2023, 9(42), eadh4626. Y. Nakamori, et al., Crystal structure and charge density
26 V. Landgraf, T. Famprikis, J. de Leeuw, L. J. Bannenberg, analysis of Li2NH by synchrotron X-ray diffraction, J. Alloys
S. Ganapathy and M. Wagemaker, Li5NCl2: A Fully- Compd., 2005, 393(1–2), 264–268.
Reduced, Highly-Disordered Nitride-Halide Electrolyte for 42 K. Ohoyama, Y. Nakamori, S. I. Orimo and K. Yamada,
Solid-State Batteries with Lithium-Metal Anodes, ACS Appl. Revised crystal structure model of Li2NH by neutron
Energy Mater., 2023, 6(3), 1661–1672. powder diffraction, J. Phys. Soc. Jpn., 2005, 74(1), 483–487.
27 P. Chen, Z. Xiong, J. Luo, J. Lin and K. L. Tan, Interaction 43 M. P. Balogh, C. Y. Jones, J. F. Herbst, L. G. Hector and
of hydrogen with metal nitrides and imides, Nature, 2002, M. Kundrat, Crystal structures and phase transformation of
420, 20–22. deuterated lithium imide, Li2ND, J. Alloys Compd., 2006,
28 J. F. Herbst and L. G. Hector, Energetics of the Li amide/Li 420(1–2), 326–336.
imide hydrogen storage reaction, Phys. Rev. B:Condens. 44 W. I. F. David, M. O. Jones, D. H. Gregory, C. M. Jewell,
Matter Mater. Phys., 2005, 72(12), 2–9. S. R. Johnson, A. Walton, et al., A mechanism for non-stoi-

© 2025 The Author(s). Published by the Royal Society of Chemistry EES Batteries, 2025, 1, 527–540 | 539
 View Article Online

Paper EES Batteries

chiometry in the lithium amide/lithium imide hydrogen 57 M. Yang, U. Raucci and M. Parrinello, Reactant-Induced
storage reaction, J. Am. Chem. Soc., 2007, 129(6), 1594– Dynamics of Lithium Imide Surfaces during the Ammonia
1601. Decomposition Process, Nat. Catal., 2023, 6, 829–836.
45 J. W. Makepeace and W. I. F. David, Structural Insights into 58 A. C. Lazanas and M. I. Prodromidis, Electrochemical
the Lithium Amide-Imide Solid Solution, J. Phys. Chem. C, Impedance Spectroscopy─A Tutorial, ACS Meas. Sci. Au,
2017, 121(22), 12010–12017. 2023, 3(3), 162–193.
46 N. Tapia-Ruiz, N. Sorbie, N. Vaché, T. K. A. Hoang and 59 S. Taibl, G. Fafilek and J. Fleig, Impedance spectra of Fe-
D. H. Gregory, Rapid microwave synthesis, characterization doped SrTiO3 thin films upon bias voltage: Inductive loops as
and reactivity of lithium nitride hydride, Li4NH, Materials, a trace of ion motion, Nanoscale, 2016, 8(29), 13954–13966.
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

2013, 6(11), 5410–5426. 60 A. West, J. Irvine and D. Sinclair, Electroceramics:

Open Access Article. Published on 02 April 2025. Downloaded on 6/30/2025 10:46:59 AM.

47 N. G. Kravchenko, V. V. Zhdanov, E. B. Kaizer and Characterization by Impedance Spectroscopy, Adv. Mater.,

A. S. Poplavnoi, First Principles Investigations of the 1990, 2(3), 132–138.
Crystal Structure, Electron Spectra, and Chemical Bonding 61 K. Hayashi, P. V. Sushko, Y. Hashimoto, A. L. Shluger and
in the Low-Temperature Phase of Lithium Imide, J. Struct. H. Hosono, Hydride ions in oxide hosts hidden by hydrox-
Chem., 2018, 59(6), 1258–1264. ide ions, Nat. Commun., 2014, 5, 1–8.
48 M. L. Bonnet, M. Iannuzzi, D. Sebastiani and J. Hutter, 62 S. Menkin, J. B. Fritzke, R. Larner, C. de Leeuw, Y. Choi,
Local disorder in lithium imide from density functional A. B. Gunnarsdóttir, et al., Insights into soft short circuit-
simulation and NMR spectroscopy, J. Phys. Chem. C, 2012, based degradation of lithium metal batteries, Faraday
116(35), 18577–18583. Discuss., 2024, 248, 277–297.
49 G. Miceli, M. Ceriotti, M. Bernasconi and M. Parrinello, 63 W. D. Knight, Nuclear Magnetic Resonance Shift in Metals,
Static disorder and structural correlations in the low-temp- Phys. Rev., 1949, 76, 1259.
erature phase of lithium imide, Phys. Rev. B:Condens. 64 H. J. Chang, N. M. Trease, A. J. Ilott, D. Zeng, L. S. Du,
Matter Mater. Phys., 2011, 83(5), 2–6. A. Jerschow, et al., Investigating Li Microstructure
50 C. M. Araújo, A. Blomqvist, R. H. Scheicher, P. Chen and Formation on Li Anodes for Lithium Batteries by in Situ
6
R. Ahuja, Superionicity in the hydrogen storage material Li/7Li NMR and SEM, J. Phys. Chem. C, 2015, 119(29),
Li2NH: Molecular dynamics simulations, Phys. Rev. B: 16443–16451.
Condens. Matter Mater. Phys., 2009, 79(17), 1–4. 65 F. Mambretti, U. Raucci, M. Yang and M. Parrinello, How
51 Y. Kojima and Y. Kawai, IR characterizations of lithium Does Structural Disorder Impact Heterogeneous Catalysts?
imide and amide, J. Alloys Compd., 2005, 395(1–2), 236– The Case of Ammonia Decomposition on Non-Stoichiometric
239. Lithium Imide, ACS Catal., 2024, 14, 1252–1256.
52 S. Ramakumar, C. Deviannapoorani, L. Dhivya, 66 M. J. Counihan, K. S. Chavan, P. Barai, D. J. Powers,
L. S. Shankar and R. Murugan, Lithium garnets: Synthesis, Y. Zhang, V. Srinivasan, et al., The phantom menace of
structure, Li+ conductivity, Li+ dynamics and applications, dynamic soft-shorts in solid-state battery research, Joule,
Prog. Mater. Sci., 2017, 88, 325–411. 2024, 8, 64–90.
53 C. Rongeat, M. A. Reddy, R. Witter and M. Fichtner, 67 P. A. Anderson, P. A. Chater, D. R. Hewett and P. R. Slater,
Nanostructured fluorite-type fluorides as electrolytes for Hydrogen storage and ionic mobility in amide-halide
fluoride ion batteries, J. Phys. Chem. C, 2013, 117(10), systems, Faraday Discuss., 2011, 151, 271–284.
4943–4950. 68 A. Gotoh, H. Obayashi, R. Nagai, S. Mochizuki and
54 J. B. Grinderslev, L. N. Skov, J. G. Andreasen, S. Ghorwal, T. Kudo, Hitachi Maxell Ltd. LITHIUM COMPOUNDS,
J. Skibsted and T. R. Jensen, Methylamine Lithium United States Patent, 4411971, 1983, 10–25.
Borohydride as Electrolyte for All-Solid-State Batteries, 69 A. Coelho, TOPAS and TOPAS-Academic: An optimization
Angew. Chem., Int. Ed., 2022, 61(32), 1–7. program integrating computer algebra and crystallographic
55 Z. Wang, Z. Ye, Y. Zhang, C. Li and Y. Zeng, Fluorite-doped objects written in C++, J. Appl. Crystallogr., 2018, 51, 210–218.
Gd0.1Ce0.9O1.95 polycrystalline electrolytes with extraordi- 70 O. Pecher, J. Carretero-Gonzalez, K. J. Griffith and
narily enhanced oxide ionic conductivities controlled by C. P. Grey, Materials’ methods: NMR in battery research,
depletion degree of oxygen vacancies at grain boundaries, Chem. Mater., 2017, 29(1), 213–242.
J. Power Sources, 2020, 480, 229118. 71 D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé,
56 I. W. Chen, Grain boundary kinetics in oxide ceramics with B. Alonso, et al., Modelling one- and two-dimensional
the cubic fluorite crystal structure and its derivatives, solid-state NMR spectra, Magn. Reson. Chem., 2002, 40(1),
Interface Sci., 2000, 8(2), 147–156. 70–76.

540 | EES Batteries, 2025, 1, 527–540 © 2025 The Author(s). Published by the Royal Society of Chemistry