Energy &

Environmental
Science
View Article Online
PAPER View Journal | View Issue

Operando investigations of the solid electrolyte

Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

interphase in the lithium mediated nitrogen

Cite this: Energy Environ. Sci.,
2024, 17, 3482 reduction reaction†
Niklas H. Deissler, a J. Bjarke V. Mygind, a Katja Li, a Valerie A. Niemann, b

Peter Benedek, b Valentin Vinci,c Shaofeng Li,a Xianbiao Fu,a

Peter C. K. Vesborg, a Thomas F. Jaramillo, b Jakob Kibsgaard, a
Jakub Drnecc and Ib Chorkendorff *a

The lithium-mediated nitrogen reduction reaction (Li-NRR) represents a promising approach for
electrochemical nitrogen activation, in which the solid electrolyte interphase (SEI) layer formed on the
electrochemically plated lithium plays a key role. Herein, we used time-resolved, operando, grazing incidence
wide-angle X-ray scattering (GI WAXS) to identify SEI species and reaction intermediates in the Li-NRR,
comparing LiBF4 and LiClO4 as electrolyte salts. We demonstrated how the SEI composition influences the Li-
NRR performance by regulating proton transport to the plated Li. When LiBF4 is used as the electrolyte salt,
Received 7th December 2023, the formation of LiF and lithium ethoxide (LiEtO) is observed. Reaction intermediates such as LiH and LiNxHy
Accepted 18th April 2024 species were found and provide insight into reaction pathways towards undesired and desired products,
DOI: 10.1039/d3ee04235a respectively. Observed restructuring of the Cu (111) single crystal substrate also indicates interaction with plated
Li that could possibly influence the Li-NRR performance. Together, these experiments give molecular insight
rsc.li/ees into how to design Li-NRR systems and their SEI layers for optimal performance.

Broader context
Ammonia holds a prominent position as one of the most manufactured chemicals globally (4180 million tons annually), primarily because of its pivotal role in
creating synthetic fertilizers. Recently, there has been significant interest in the lithium-mediated nitrogen reduction reaction (Li-NRR) as a sustainable alternative to
the Haber–Bosch process, viable at ambient temperatures and pressures. In the Li-NRR process, lithium metal is electrodeposited from an organic electrolyte
containing a lithium salt. Subsequently, nitrogen is split through its reaction with the deposited lithium and intermediate lithium–nitrogen species are then
protonated by a proton donor, leading to the formation of ammonia. Improvements in selectivity towards ammonia, higher current densities and the development of
a flow-cell allowing for continuous ammonia synthesis have been achieved using LiBF4 instead of LiClO4 as electrolyte salt. Its beneficial impact has been attributed to
the solid electrolyte interphase (SEI) layer, which forms from reactions of the electrolyte with lithium. The application of operando grazing incidence wide-angle X-ray
scattering (GI WAXS) provides insight into the mostly unexplored function and composition of the SEI layer formed by LiBF4.

Introduction might prove suitable as an energy carrier to replace fossil fuels.2

Currently, ammonia is produced almost exclusively from the
Every year, more than 180 million tons of ammonia are Haber–Bosch process, which requires elevated temperatures
produced, mainly for use as fertilizer.1 In the future, ammonia and pressures (400–450 1C, 150–200 bar).3 Electrochemical
processes have been considered as alternatives for many years,
a
however, progress was hindered by many erroneous reports.4
Department of Physics, Technical University of Denmark, Kongens Lyngby,
Denmark. E-mail: [email protected]
The lithium-mediated nitrogen reduction reaction (Li-NRR) was
b
Department of Chemical Engineering, Stanford University, Stanford, CA, USA, first reported in 1930 by Fichter et al.5 and later by Tsuneto
SUNCAT Center for Interface Science and Catalysis, SLAC National Accelerator et al.6,7 in 1993. Despite these reports, there was no subsequent
Laboratory, Menlo Park, CA, USA exploration or follow-up on this discovery. In 2019 our group
c
Experimental Division, European Synchrotron Radiation Facility, Grenoble, France
established the Li-NRR as the only reliable strategy for electro-
† Electronic supplementary information (ESI) available: Detailed experimental
methods, photos of setup, electrochemical data, 1D diffractograms of all experi-
chemical ammonia production from elemental nitrogen (N2).8
ments, time traces of all species, time development of 2D detector images of all In the Li-NRR, lithium (Li) is plated on the cathode from
species, XPS data. See DOI: https://doi.org/10.1039/d3ee04235a an electrolyte consisting of an organic solvent, typically

3482 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Paper Energy & Environmental Science

tetrahydrofuran (THF), a lithium-containing salt, and a proton electrode and a LiFePO4 (LFP) electrode as ref. 24 and 25 Before
carrier, usually ethanol (EtOH). Plated Li can react with N2 to each experiment the Cu (111) single crystal was annealed in a
form intermediate Li–N-compounds, which can be protonated tube furnace for 4 1 h at 950 1C in a 1 : 3 H2 : Ar atmosphere. All
by the proton carrier thereby forming ammonia.9,10 Plated Li operando measurements were conducted at the ID31 beamline
also reacts with the electrolyte to form the solid electrolyte at the European synchrotron radiation facility (ESRF). The
interphase (SEI) layer;11 however, its composition and function experimental setup and subsequent data treatment are shown
are mostly unknown. There have been substantial efforts aim- schematically in Fig. 1. The Cu (111) single crystal electrode was
ing at a better understanding of the role of the SEI, though aligned to minimize its contribution to the recorded detector
most studies rely on ex situ techniques.12–16 In situ investiga- images. Following an initial period (425 min) of open circuit
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

tions of the SEI by means of neutron reflectometry and grazing voltage (OCV) a linear sweep voltammetry (LSV) with a
incidence wide-angle X-ray scattering (GI WAXS) have been scan rate of 20 mV s1 was conducted until Li-plating
conducted only with LiClO4 as electrolyte salt at current den- potential was reached. subsequently chronopotentiometry
sities below 1 mA cm2.17,18 (CP) at 2 mA cm2 was carried out for 65 min, followed by
Recent research driven by the growing interest in Li-NRR has another OCV measurement (420 min). Recording of GI WAXS
led to significantly increased faradaic efficiencies (FE) and measurements began when the initial OCV measurement was
current densities.12,19,20 Further, the development of flow-cell started. Detector images recorded every 5–10 s were radially
systems and the implementation of hydrogen on the anode integrated giving 1D diffractograms as shown in Fig. 1. Follow-
enables continuous ammonia synthesis without the reliance on ing the intensity of a given peak over time gives information
sacrificial agents.13,21 In early studies mainly lithium perchlo- about the time development of the corresponding species.
rate (LiClO4) was used as electrolyte salt.6–9,11,12 Recent studies Additionally, the time development of the detector image pixels
by our group using a pressurized one-compartment cell as well provides insights into the crystallinity of corresponding spe-
as a flow cell employed lithium tetrafluoroborate (LiBF4) as an cies. Experimental details can be found in the ESI† (Fig. S1).
alternative electrolyte salt, attributing SEI structure and com-
position a critical role in improved FEs and long-term
stability.13,20,22 To date, however, the SEI layer in the LiBF4- Results & discussion
based electrolyte has mainly been characterized ex situ14,20 and
in one in situ XRD study,23 and it remains to be determined how Fig. 2A shows the peaks corresponding to the Li (110) reflec-
it enables enhanced NH3 selectivity and stability compared to tion, exemplarily for all Li reflections, while the time develop-
the LiClO4-based electrolyte. ment of the maximum intensity of the peaks is shown in
In this work, we present an operando GI WAXS study Fig. 2B. Diffractograms showing different q-ranges and refer-
comparing the SEI in the Li-NRR with LiBF4 or LiClO4 as ence patterns are shown in the ESI† (Fig. S2–S28) In experi-
electrolyte salt. GI WAXS was used to characterize the crystal- ments without EtOH, the intensity of the Li (110) peak
line deposits on a Cu (111) single crystal working electrode increased continuously while current was applied. When mea-
during the LI-NRR using LiClO4 or LiBF4 with or without proton suring at OCV after the CP, the peak intensity decreased only
donor, EtOH, thereby enabling us to monitor Li0 and its slowly, indicating that the plated Li remained on the electrode
reaction with the electrolyte, EtOH, and N2. These in situ X- surface, dissolving only slowly. Accordingly, the working elec-
ray measurements have the benefit of detecting transient trode potentials shown in Fig. 2C did not increase within the
species that are not possible to observe ex situ. We found that first 20 min after the CP when no EtOH was present in the
a LiF containing SEI layer formed by LiBF4, stabilizes the plated electrolyte. Corresponding counter electrode potentials are
Li by limiting proton transport to the electrode. Intermediate shown in the ESI† (Fig. S29) as well as LSV measurements
LiNxHy species as well as LiH were detected, giving insight into and electrochemical data for all experiments (Fig. S29–S45,
reaction pathways towards both desired and undesired pro- ESI†). The observed FEs for ammonia without EtOH, 0.1 
ducts. The formation of Li2CO3, likely related to electrolyte 0.1% and 0.4  0.4% for LiClO4 and LiBF4 respectively, are
degradation was observed with LiClO4. This work provides in agreement with previously reported values for unpressurized
molecular understanding of the SEI layer to guide further batch reactors.6,7,10,26,27 The stability of the plated Li
development of Li-NRR towards higher selectivities and stabi- indicates that also the competing reduction of protons by Li
 
lities, including the rational design of artificial SEIs, which to þ þ 1
Li þ H ! Li þ H2 is limited.
date, have not been explored in this system. 2
When an electrolyte with 1 M LiClO4 and 1 vol% EtOH was
used, the Li (110) peak only appeared temporarily and was
Experimental absent throughout most of the measurement. This only tran-
sient appearance suggests that Li is plated but then converted
With each electrolyte composition, LiBF4 (1 M) or LiClO4 (1 M) by fast reactions. Correspondingly, the working electrode
in THF, both with EtOH or without EtOH (1 vol% vs. 0 vol%), potential increased almost immediately after the end of the
two synchrotron experiments were conducted. A Cu (111) single CP (Fig. 2C). The low FE of 2  1%, observed for this system,
crystal was used as working electrode, a Pt-mesh as counter suggests that the main reaction is the hydrogen-forming

This journal is © The Royal Society of Chemistry 2024 Energy Environ. Sci., 2024, 17, 3482–3492 | 3483
 View Article Online

Energy & Environmental Science Paper

Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

Fig. 1 Schematic of the experimental setup and the data processing. Synchrotron X-rays are directed in grazing incidence on a Cu (111) single crystal
working electrode used for the lithium-mediated nitrogen reduction reaction. 2D detector images, obtained from the diffracted X-rays, were integrated,
giving diffraction patterns, and used to identify SEI components. Pixels corresponding to a specific q-vector were plotted with time, where t = 0 is set to
the beginning of a chronopotentiometry (CP) measurement.

 
1 Li was still present on the electrode24 even though it could not
reduction of protons by Li Li þ Hþ ! Liþ þ H2 . Side reac-
2 be detected by WAXS. Furthermore, in all electrolytes, when Li
tions of Li with the solvent likely further decreased the FE.28,29 is detected, the intensity increase is not uniform over all
Since only a small volume (1.1 mL) of N2 saturated electrolyte corresponding pixels on the detector (S46–S53, ESI†) and thus
without flow or stirring could be used in the operando cell, a it is likely that Li does not form a fully polycrystalline phase but
shortage of N2 at the electrode can likely explain the low FE.9–11 rather forms fewer larger crystallites. Therefore, it is possible
While N2 likely would not directly change the SEI layer, it for Li to be present on the electrode in small quantities without
should be noted that Li–N species might impact its structure being detected as the crystallites may not be aligned properly
and composition. with the incoming beam to satisfy the Bragg conditions, which
When LiBF4 was used as electrolyte salt (FE = 3  2%), can explain the delayed increase in working electrode potential
lithium remained on the electrode for a prolonged time even in as shown in Fig. 2C. Furthermore, some Li might have plated as
the presence of EtOH (Fig. 2B). Nevertheless, after an initial an amorphous phase, which was undetectable by GI WAXS in
continuous increase for 25 min, the peak intensity decreased operando conditions. The difference in Li plating behavior
until the end of the CP, indicating that Li was consumed in confirms that the SEI layer formed when using LiBF4 is better
reactions. After the CP, the working electrode potential at protecting the plated Li from reacting with protons in the
remained at 3.4 V vs. LFP for 10 min, indicating that some electrolyte, as hypothesized by Li et al.20

3484 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Paper Energy & Environmental Science

Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

Fig. 2 (A) Integrated WAXS patterns showing the Li (110) reflection, measured using different electrolyte compositions. (B) Time development of the Li
(110) peak intensities when using different electrolyte compositions. (C) Working electrode potentials in the corresponding experiments. The counter
electrode was Pt-mesh and a LFP electrode was used as reference. A 1 M solution of either LiClO4 or LiBF4 in THF with either 0 vol% or 1 vol% EtOH was
used as electrolyte. During chronopotentiometry (CP), a current density of 2 mA cm2 was applied.

The observed differences in Li plating behavior point to a reached (Fig. 3A and Fig. S54–S58, ESI†). As only crystallites
substantially different SEI layer formed depending on the above a certain size can be detected with X-ray scattering there
electrolyte composition. Previous studies suggest that the SEI is a delay between the formation and the detection of a species,
layer formed by LiBF4 consists in large parts of LiF.20 Fig. 3A also dependent on cell alignment. Therefore, LiF might have
and B confirm the formation of crystalline LiF in experiments also formed prior to Li plating in other measurements but was
with LiBF4 containing electrolyte. Notably, peaks corres- not detected until after the LSV measurement. Furthermore,
ponding to the LiF were already observed in the first measure- due to the variability of the cell resistance (200–600 O) and the
ment before current was applied to the cell, which indicates resulting difference in ohmic drop (iR) correction, the actual,
that LiBF4 had already partially decomposed to LiF, possibly applied potentials during LSV may vary for different
due to the heat evolved during the dissolution of LiBF4 in THF. experiments.
Fig. 3A and B show that the intensity of the LiF peak increased Contrary to the non-uniform increase of intensity shown for
slightly during the initial OCV period. This continued decom- the Li (110) reflection in Fig. 4A, a homogenous increase of
position might be facilitated by the X-ray beam.30 With EtOH, a intensity over all corresponding detector pixels was observed
steep increase in the LiF intensity is observed already at B2 V for LiF, as exemplary shown for the (111) reflection in the LiBF4/
vs. LFP during the LSV measurement (Fig. S34, ESI†), showing 1 vol% EtOH experiment in Fig. 4B, and the ESI† (Fig. S59–S61).
that electrochemical LiF and SEI formation already occurred This homogeneous increase over all corresponding pixels
prior to Li plating. SEI formation before Li plating has pre- indicates that polycrystalline LiF composed of small crystallites
viously been confirmed to occur with LiClO4 but not yet with is deposited. The formation of LiF is most likely accompanied
LiBF4.17,18 However, this steep increase in LiF intensity was not by the formation of BF3 (LiBF4 - LiF + BF3), which can
always observed clearly before Li plating potentials were participate in various organic reactions and therefore might

This journal is © The Royal Society of Chemistry 2024 Energy Environ. Sci., 2024, 17, 3482–3492 | 3485
 View Article Online

Energy & Environmental Science Paper

Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

Fig. 3 Time development of peak intensities corresponding to observed SEI species and reaction intermediates when using LiBF4 without (A) or with
EtOH (B) in the electrolyte. Traces of low intensity peaks (LiH, LiEtO, LiNH2, Li3N) were smoothed by using Savitzky–Golay filtering.

play a critical role in the organic part of the SEI and could Continuing LiF formation might prove problematic for long-
possibly explain the better performance of LiBF4 compared to term stability as it irreversibly removes Li+ ions from the
LiPF6.12,31,32 Even though no quantification based on peak electrolyte. As shown in Fig. 3B the LiF intensity increases
intensities is possible, the high counts observed for all LiF again when the Li intensity decreases. This increase might be
peaks suggest that substantial amounts are formed. It should a result of the SEI retreating into the X-ray beam as the plated Li
be noted that at higher q-vectors splitting of LiF peaks dissolves. LiF might also form from the plated Li as it has been
appeared, as shown in the ESI† (Fig. S11 and S15). This splitting shown that LiF easily forms on Li exposed to fluoride contain-
is likely caused by the parallax effect from the formation of ing electrolytes36
multiple separate deposits of LiF on the working electrode, As shown in Fig. 3B, when EtOH was used in the electrolyte,
possibly a consequence of a non-uniform current density dis- after 25 min of applied current the intensity of peaks corres-
tribution due to the electrode configuration. ponding to Li steeply decrease in intensity for B4 min, coin-
As more Li is plated, the intensity of LiF peaks decreases ciding with an increase of intensity for LiF peaks and the
(Fig. 3A and B and Fig. S54–S56), suggesting that parts of the appearance of new peaks. Corresponding time traces shown
LiF are lifted above the X-ray beam, as Li plates below it. An in Fig. 3B have been smoothed using Savitzky–Golay filtering.
inverse trend for the development of the Li and LiF peaks was The raw data is available in the ESI† (Fig. S54), as well as plots
observed both with and without EtOH in the electrolyte, as of the overlapped time traces (Fig. S54 and S56, ESI†). This very
depicted in Fig. 4C and D. Due to its low solubility33 and high dynamic behavior in the observed species could indicate the
electrochemical stability34,35 it is unlikely that decreases in LiF formation of a crack in the protective SEI layer accompanied by
intensity were caused by a reaction or dissolution of LiF. These reactions of Li with the electrolyte, forming new species as
results show that, as previously predicted,13,20 LiF is a key depicted in Fig. 4D. Crack formation is a phenomenon well-
compound in the SEI layer, that can limit proton transport to known in Li metal batteries.37–40 Furthermore, ex situ studies of
the plated Li (Fig. 4C and D). Experiments with EtOH showed this system have shown a more porous SEI layer when EtOH is
an overall increase in LiF intensity while a current was applied. present, potentially caused by the formation of hydrogen

3486 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Paper Energy & Environmental Science

Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

Fig. 4 Time development of all detector pixels corresponding to the Li (110) reflection when using LiBF4 without EtOH in the electrolyte (A) and the LiF
(111) reflection when using LiBF4 with EtOH in the electrolyte (B). Schematic depiction of crystalline deposits on the working electrode when using LiBF4
without EtOH (C) and with EtOH (D).

bubbles when protons are reduced.14 Crack formation might decreasing intensity of LiH peaks throughout the rest of the
therefore be initiated by gas formation. As the X-ray beam is experiment.42,43 Therefore, LiH formation might be both a con-
only 6 mm wide this is likely a local occurrence and not sequence and an intermediate of parasitic hydrogen formation.
representative of the whole electrode, suggesting that deposits Nevertheless, considerable LiH formation was observed indicating
are not homogeneously distributed over the whole electrode. This that despite the temporary exposure of Li to the electrolyte, the
phenomenon is an uncontrolled process that could explain varied proton/EtOH transport to the electrode is limited, which could
results between repeat experiments (Fig. S57, ESI†). In future indicate a quick restoration of the SEI layer. This quick SEI
studies it might be advantageous to measure more spots on the reformation in fluorine containing electrolytes, related to the easy
electrode. After the steep decrease, the Li peak intensities con- formation of LiF has been attributed with improved stability of Li
tinue to decrease with a smaller slope suggesting that the metal batteries.36 Similarly in Li-NRR fast restoration of the SEI
protective SEI has partially reformed. Some of the appearing limits the direct exposure of plated Li even after the formation of
peaks can clearly be attributed to LiH (Fig. 3B and Fig. S10 and cracks in the SEI. Formation of LiH was also observed in experi-
S11, ESI†). LiH formation has been suggested previously,20,41 but ments without EtOH for both LiClO4 and LiBF4 (Fig. 3A and Fig.
here we provide experimental proof of LiH formation in the Li- S4, S7, S17, S19 S62 and S63, ESI†) which further supports the
NRR. It has been demonstrated previously that LiH can form from hypothesis that it is a reaction occurring under proton-limited
the reaction of H2 with Li in inorganic electrolyte solutions conditions. Low intensities suggest that this LiH formation is less
 
1 pronounced and might arise from the decomposition of THF on
H2 þ Li þ e ! LiH .42 Under proton-limited conditions,
þ 
2 the anode or impurities (water, alcohols) in the solvent.44,45
an electrochemical pathway (Li + e + H+ - LiH) might be Notably, no LiH was detected when LiClO4 was used with EtOH
possible. However, LiH is expected to react with protons or EtOH, (Fig. S21, S25 and S26, ESI†), further indicating that here the SEI
forming H2 (LiH + EtOH - Li+ + EtO +H2), which explains the does not limit proton transport to the electrode. It should be

This journal is © The Royal Society of Chemistry 2024 Energy Environ. Sci., 2024, 17, 3482–3492 | 3487
 View Article Online

Energy & Environmental Science Paper

noted that the LiH surface has been predicted to be active towards detected by GI WAXS is unlikely. These results could suggest that
ammonia synthesis at Li plating potentials and therefore might the formation of cracks in the SEI layer is beneficial to Li-NRR
also play a role here.41 Other appearing peaks have been attrib- performance as intermediate species were only observed after-
uted to LiEtO (q-vector = 0.775 Å1), LiNH2 (q-vector = 2.150 Å1), wards. When the SEI layer is too thick, transport of reactants
and Li3N (q-vector = 1.660 Å1). It should be noted that usually might be insufficient, but this transport limitation is alleviated
species were identified on the basis of multiple reflections, but when cracks are formed. It is more likely that with an intact SEI
due to low peak intensities for the aforementioned species only a layer only crystallites, insufficiently large for detection, are
single peak each could be used. Even though this has been formed. A crack in the SEI exposed plated Li to the electrolyte
reported by others, it has to be acknowledged that a certain leading to the formation of greater amounts of intermediate Li–N
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

degree of uncertainty is associated with the peak assignment.18 species. The simultaneous formation of LiH suggests significant
An extensive discussion of peak assignments is provided in the hydrogen evolution, indicating that the increased formation of Li–
ESI† (Fig. S64–S70). In view of the time courses of their intensities, N species was accompanied by a lower selectivity as any advanta-
it is nevertheless reasonable to attribute the peaks to reaction geous effects of the SEI were mitigated. Beneficial or not, crack
products of Li. Assuming correct peak assignment, they indicate formation and accompanying reactions show that there is
that LiEtO and Li3N formed immediately after SEI cracking, dynamic change of the deposits throughout Li-NRR experiments,
whereas generation of LiNH2 was slightly delayed. This observa- suggesting a multitude of simultaneous processes. These pro-
tion is in accordance with the proposed reaction mechanism for cesses must be understood to enhance Li-NRR performance. In
ammonia formation in which Li3N is progressively protonated none of the LiBF4 experiments LiOH was detected which is
and experimental proof of intermediate protonated Li–N in contrast to a previous study utilizing in situ XRD.23
species.10,41 However, the protonation of Li3N was assumed to The formation of LiOH is likely linked to the presence of water
 
be fast, whereas the detection of LiNH2 here suggests that at least 1
Li þ H2 O ! H2 þ LiOH . Therefore, differences in water
the final protonation step is slow and potentially rate-limiting. 2
Otherwise, the formation of sufficiently large crystallites to be content can explain the absence of crystalline LiOH in this study.

Fig. 5 (A) Integrated GI WAXS patterns with Li2CO3 reflections marked, using LiClO4 without EtOH in the electrolyte. (B) Time development of Li2CO3
reflections, when using LiClO4 and EtOH in the electrolyte. Traces were smoothed by running averages over 5 values. Schematic depiction of crystalline
deposits on the working electrode when using LiClO4 without EtOH (C) and with EtOH (D).

3488 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Paper Energy & Environmental Science

Notably, with the only exception being LiF, all observed species Fig. 5B time traces of Li2CO3 peaks, when using LiClO4 with
are only observed temporarily and disappear almost completely EtOH, are shown. Time traces were smoothed by running
before or shortly after the end of the CP. While the disappearance averages over 5 values. The raw data is available in the ESI†
of some species before the end of the CP is likely coincidental and (Fig. S77).
due to the small amounts formed, this dynamic behavior high- Li2CO3 was generated only with an electrolyte containing
lights the need for operando investigations. EtOH and LiClO4, which might suggest EtOH oxidation. Inter-
When LiClO4 is used as electrolyte salt fewer crystalline mediate species of the multi-step oxidation of EtOH such as
species were observed, as shown in Fig. 5A and B. Remarkably, acetic acid were detected previously with GC–MS,28 providing a
no LiCl was detected which is in contrast to ex situ possible pathway to Li2CO3 formation. Decomposition pro-
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

XPS measurements (Fig. S71–S73, ESI†) and previous ducts of THF previously detected by NMR,16,19,27 forming only
reports.11,15,20 Ex situ XPS spectra of the deposits for all in the presence of EtOH such as 2-ethoxy-THF and g-
electrolyte conditions is provided in the ESI† (Fig. S71–S76). butyrolactone provide other possible precursors for carbonate
The apparent absence of LiCl in these operando measurements formation. Li2CO3 could be observed during the OCV before
means that no sufficiently large crystallites are formed within any current was applied, suggesting that it can also be formed
the X-ray beam. If the deposits were not uniformly distributed it without an electrochemical step, possibly facilitated by the X-
is possible that species present on the electrode would not be ray beam. More Li2CO3 signals however, were observed during
detected. However, LiCl was not detected in any of the experi- the CP, indicating that an electrochemical process directly or
ments, even when measuring on multiple spots throughout the indirectly facilitates its formation. It should also be noted that
experiment. Possibly LiCl formed during the experiment is not all diffraction peaks are visible at the same time as the
dissolved in the electrolyte and then detected ex situ in the peaks arise from single illuminated spots on the detector (Fig.
residue. Depth profiling shown in Fig. S74 (ESI†) suggests S78–S88, ESI†). Therefore, only a small number of Li2CO3
however that this is not the case as the LiCl signal increases crystals were present at any time, explaining why only a partial
with sputtering time. Therefore, it is likely that LiCl either diffraction pattern can be observed. These results suggest that
formed amorphous deposits or its crystallites were too small to Li2CO3 is not stable on the electrode and is possibly reduced46
be detected. Furthermore, in previous GI WAXS measurements, or dissolved, especially in a proton rich environment,44,47
LiOH and Li2O were detected.17,18 Neither Li2O nor LiOH could which could explain the transient behavior shown in Fig. 5B.
be detected clearly in the experiments reported here. A broad However, the disappearance of a peak does not necessarily
increase in intensity around 2.3 Å1 (Fig. S28, ESI†), observed correspond to the disappearance of the observed species e.g., a
when using LiClO4 with EtOH, could correspond to LiOH. For crystal might move and not be aligned to fulfill the Bragg
this experiment measurements were conducted on two spots on conditions, especially when only a small number of crystals
the electrode, however the signals were only observed on one are present. With LiBF4, no Li2CO3 was detected, indicating
spot, further showing that deposits are not homogeneously that LiClO4 plays a role as an oxidizer and oxygen source in the
distributed on the electrode during the Li-NRR process. In formation of Li2CO3.

Fig. 6 Recorded 2D detector images before and after a Li-NRR experiment (A). Time development of all pixels corresponding to the Cu (111) reflection
(q = 3.012 Å1) (B). A solution of 1 M LiBF4 in THF with 1 vol% EtOH was used as electrolyte. During the chronopotentiometry (CP) a current of 2 mA cm2
was applied.

This journal is © The Royal Society of Chemistry 2024 Energy Environ. Sci., 2024, 17, 3482–3492 | 3489
 View Article Online

Energy & Environmental Science Paper

As depicted in Fig. 5C, in the absence of EtOH, plated Li is Nevertheless, dynamic changes of the deposits, such as the
stable on the surface of the electrode. Aside from LiH no formation of cracks in the SEI layer were observed. LiH was
crystalline species were detected suggesting that the SEI con- identified as another SEI compound which, in contrast to LiF is
sists largely of amorphous, likely organic species. In the not stable. Importantly, LiH is likely an intermediate towards
presence of EtOH, small amounts of Li2CO3 are also formed the major competing reaction, H2 evolution, in this system.
but the SEI layer does not limit proton transport enough for Li This report provides experimental proof for LiH formation
to be stable on the surface, as depicted in Fig. 5D. Therefore, during Li-NRR, and understanding its formation is critical to
only a small number of Li crystals is present at a given time, mitigating unwanted side reactions, and enhancing NH3 selec-
and no stable LiH is formed. The apparent absence of crystal- tivity. Aside from SEI compounds, intermediate reaction spe-
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

line species, when using LiClO4, as depicted in Fig. 5C and D is cies were observed, demonstrating the presence of
in sharp contrast to the SEI formed from LiBF4. This intermediate LiNxHy species. All observed species showed
finding provides a rationale for the reported improvement dynamic behavior over the duration of the experiments, high-
of FE by adding oxygen or water as this may lead to the lighting the multitude of processes occurring during Li-NRR.
formation of possibly crystalline LiOH and Li2O, which might Finally, Cu substrate-Li interactions were observed, motivating
be beneficial.11,48 further study into substrate effects on Li-NRR performance.
In multiple experiments, a restructuring of the Cu (111) Altogether, these findings inform the molecular design of Li-
single crystal was observed, indicated by a transition from a NRR systems to adequately passivate Li, mitigate HER, promote
single crystal refraction pattern to that characteristic of a LiNxHy intermediate formation, and consider Li interactions
polycrystalline sample as exemplarily shown in Fig. 6A for the with the cathode. This work motivates achieving a molecular
LiBF4/1 vol% EtOH experiment. This restructuring of the sur- understanding of the SEI resulting from other system enhance-
face and the associated increase in intensity outside of the ments including operating in a flow-cell with hydrogen oxida-
single crystal refraction spots does not begin before a potential tion on the anode, and inspires new approaches such as the
is applied as shown in Fig. 6B and the ESI† (Fig S55, S61–S63 design of artificial SEIs.
and S89–S98). Previously, changes to a Mo substrate in the
initial minutes of LI-NRR have been attributed to intercalation
of Li-ions in a molybdenum oxide layer.18 In this study, how- Author contributions
ever, no copper oxide is detected, and the substrate changes Conceptualization: N. H. D., J. B. V. M., K. L., J. D., I. C. Data
occur also at potentials where no oxide would be present. curation: N. H. D., J. B.V. M., K. L., V. A. N., P. B., V. V. Formal
Therefore, this effect might rather be caused by Li plating. analysis: N. H. D., J. B. V. M., K. L., V. A. N., P. B., S. L., X. F.
However, at room temperature no stable Cu–Li alloy is expected Investigation: N. H. D., J. B. V. M., K. L., V. A. N., P. B., V. V.
to form.49 Notably, the surface continues to change even after Methodology: N. H. D., J. B. V. M., K. L., V. A. N., P. B., V. V, J. D.
the experiment when no more current is applied. Diffusion of Software: N. H. D. Supervision: P. C. K. V., T. F. J., J. K., J. D., I.
Li ions into Cu electrodes has been reported for anode-free C. Writing – original draft: N. H. D. Writing – review & edidting:
solid state batteries.50,51 These Li atoms act as nucleation sites N. H. D., J. B. V. M., K. L., V. A. N., P. B., V. V., S. L., X. F., P. V.,
for further Li plating,50 which could also lead to roughening T. F. J., J. K., J. D., I. C.
and restructuring of the Cu surface, due to the high surface
mobility of Cu atoms. Determining the exact mechanism of
surface restructuring requires further investigation and is Conflicts of interest
beyond the scope of this study. A roughened surface likely
inhibits homogeneous Li electrodeposition and promotes den- The authors declare no competing interests.
drite formation. Inhomogeneous Li plating will impact the
structure of the SEI and dendrites could compromise the Acknowledgements
structure of the SEI layer. Both these processes likely impact
Li-NRR efficiency. Clearly there is an interaction of the system We thank Asger B. Moss for help with writing code. We thank
with the Cu substrate, however it remains unclear whether Suzanne Z. Andersen, Jakob B. Pedersen and Mattia Saccoccio
there is an effect on Li-NRR performance. for general scientific discussion. We gratefully acknowledge the
funding by Villum Fonden, part of the Villum Center for the
Science of Sustainable Fuels and Chemicals (V-SUSTAIN grant
Conclusion 9455), Innovationsfonden (E-ammonia grant 9067–00010B) and
the European Research Council (ERC) under the European
In this work, by using operando GI WAXS, it was shown that Union’s Horizon 2020 research and innovation programme
unlike the SEI layer formed form LiClO4, the SEI layer formed (grant agreement no. 741860). X. F. was supported under the
by LiBF4, limits the proton/EtOH transport to the electrode, MSCA European Postdoctoral Fellowships (Electro-Ammonia
which explains the previously observed higher selectivity Project 101059643). V. A. N. was supported under the National
towards ammonia. Furthermore, it was confirmed that LiF Science Foundation Graduate Research Fellowship Program
plays a crucial role in the formation of this beneficial SEI layer. under grant no. DGE-1656518 and the Camille and Henry

3490 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024
 View Article Online

Paper Energy & Environmental Science

Dreyfus Foundation. We acknowledge the European Synchro- 17 S. J. Blair, M. Doucet, J. F. Browning, K. Stone, H. Wang,
ton Radiation Facility (ESRF) for the provision of synchrotron C. Halbert, J. Avilés Acosta, J. A. Zamora Zeledón,
radiation using beamline ID 31. A. C. Nielander, A. Gallo and T. F. Jaramillo, ACS Energy
Lett., 2022, 7, 1939–1946.
18 S. J. Blair, M. Doucet, V. A. Niemann, K. H. Stone,
References M. E. Kreider, J. F. Browning, C. E. Halbert, H. Wang,
P. Benedek, E. J. McShane, A. C. Nielander, A. Gallo and
1 L. E. Apodaca, Mineral commodity summaries 2023: NITRO- T. F. Jaramillo, Energy Environ. Sci., 2023, 16(8), 3391–3406.
GEN (FIXED)—AMMONIA, National Minerals Information 19 H.-L. Du, M. Chatti, R. Y. Hodgetts, P. V. Cherepanov,
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

Center, 2023. C. K. Nguyen, K. Matuszek, D. R. MacFarlane and

2 D. R. MacFarlane, P. V. Cherepanov, J. Choi, B. H. R. A. N. Simonov, Nature, 2022, 609, 722–727.
Suryanto, R. Y. Hodgetts, J. M. Bakker, F. M. Ferrero Vallana 20 S. Li, Y. Zhou, K. Li, M. Saccoccio, R. Sažinas, S. Z. Andersen,
and A. N. Simonov, Joule, 2020, 4, 1186–1205. J. B. Pedersen, X. Fu, V. Shadravan, D. Chakraborty,
3 F. Haber and R. Le Rossignol, Z. Elektrochem. Angew. Phys. J. Kibsgaard, P. C. K. Vesborg, J. K. Nørskov and
Chem., 1913, 19, 53–72. I. Chorkendorff, Joule, 2022, 6, 2083–2101.
4 J. Choi, B. H. R. Suryanto, D. Wang, H.-L. Du, R. Y. Hodgetts, 21 N. Lazouski, M. Chung, K. Williams, M. L. Gala and
F. M. Ferrero Vallana, D. R. MacFarlane and A. N. Simonov, K. Manthiram, Nat. Catal., 2020, 3, 463–469.
Nat. Commun., 2020, 11, 5546. 22 S. Li, Y. Zhou, X. Fu, J. B. Pedersen, M. Saccoccio, S. Z.
5 F. Fichter, P. Girard and H. Erlenmeyer, Helv. Chim. Acta, Andersen, K. Enemark-Rasmussen, P. J. Kempen, C. D.
1930, 13, 1228–1236. Damsgaard, A. Xu, R. Sažinas, J. B. V. Mygind, N. H.
6 A. Tsuneto, A. Kudo and T. Sakata, Chem. Lett., 1993, 851–854. Deissler, J. Kibsgaard, P. C. K. Vesborg, J. K. Nørskov and
7 A. Tsuneto, A. Kudo and T. Sakata, Lithium-mediated electro- I. Chorkendorff, Nature, 2024, DOI: 10.1038/s41586-024-07276-5.
chemical reduction of high pressure N, to NH 3, 1994, vol. 367. 23 X. Cai, Z. Shadike, X. Cai, X. Li, L. Luo, L. An, J. Yin, G. Wei,
8 S. Z. Andersen, V. Čolić, S. Yang, J. A. Schwalbe, A. C. Nielander, F. Yang, S. Shen and J. Zhang, Energy Environ. Sci., 2023, 16,
J. M. McEnaney, K. Enemark-Rasmussen, J. G. Baker, A. R. 3063–3073.
Singh, B. A. Rohr, M. J. Statt, S. J. Blair, S. Mezzavilla, 24 E. J. McShane, P. Benedek, V. A. Niemann, S. J. Blair,
J. Kibsgaard, P. C. K. Vesborg, M. Cargnello, S. F. Bent, T. F. G. A. Kamat, A. C. Nielander, T. F. Jaramillo and
Jaramillo, I. E. L. Stephens, J. K. Nørskov and I. Chorkendorff, M. Cargnello, ACS Energy Lett., 2023, 8, 230–235.
Nature, 2019, 570, 504–508. 25 R. Tort, O. Westhead, M. Spry, B. J. V. Davies, M. P. Ryan,
9 S. Z. Andersen, M. J. Statt, V. J. Bukas, S. G. Shapel, M.-M. Titirici and I. E. L. Stephens, ACS Energy Lett., 2023, 8,
J. B. Pedersen, K. Krempl, M. Saccoccio, D. Chakraborty, 1003–1009.
J. Kibsgaard, P. C. K. Vesborg, J. Nørskov and 26 N. Lazouski, K. J. Steinberg, M. L. Gala, D. Krishnamurthy,
I. Chorkendorff, Energy Environ. Sci., 2020, 13, 4291–4300. V. Viswanathan and K. Manthiram, ACS Catal., 2022,
10 N. Lazouski, Z. J. Schiffer, K. Williams and K. Manthiram, 5197–5208.
Joule, 2019, 3, 1127–1139. 27 J. B. Valbæk Mygind, J. B. Pedersen, K. Li, N. H. Deissler,
11 K. Li, S. Z. Andersen, M. J. Statt, M. Saccoccio, V. J. Bukas, M. Saccoccio, X. Fu, S. Li, R. Sažinas, S. Z. Andersen,
K. Krempl, R. Sažinas, J. B. Pedersen, V. Shadravan, Y. Zhou, K. Enemark-Rasmussen, P. C. K. Vesborg, J. Doganli-Kibsgaard
D. Chakraborty, J. Kibsgaard, P. C. K. Vesborg, J. K. Nørskov and I. Chorkendorff, ChemSusChem, 2023, 16, e202301011.
and I. Chorkendorff, Science, 2021, 374, 1593–1597. 28 R. Sažinas, S. Z. Andersen, K. Li, M. Saccoccio, K. Krempl,
12 K. Li, S. G. Shapel, D. Hochfilzer, J. B. Pedersen, K. Krempl, J. B. Pedersen, J. Kibsgaard, P. C. K. Vesborg, D. Chakraborty
S. Z. Andersen, R. Sažinas, M. Saccoccio, S. Li, D. Chakraborty, and I. Chorkendorff, RSC Adv., 2021, 11, 31487–31498.
J. Kibsgaard, P. C. K. Vesborg, J. K. Nørskov and 29 X. Cai, C. Fu, H. Iriawan, F. Yang, A. Wu, L. Luo, S. Shen,
I. Chorkendorff, ACS Energy Lett., 2022, 7, 36–41. G. Wei, Y. Shao-Horn and J. Zhang, iScience, 2021,
13 X. Fu, J. B. Pedersen, Y. Zhou, M. Saccoccio, S. Li, R. Sažinas, 24, 103105.
K. Li, S. Z. Andersen, A. Xu, N. H. Deissler, J. B. V. Mygind, 30 H. G. Steinrück, C. Cao, M. R. Lukatskaya, C. J. Takacs,
C. Wei, J. Kibsgaard, P. C. K. Vesborg, J. K. Nørskov and G. Wan, D. G. Mackanic, Y. Tsao, J. Zhao, B. A. Helms, K. Xu,
I. Chorkendorff, Science, 2023, 379, 707–712. O. Borodin, J. F. Wishart and M. F. Toney, Angew. Chem., Int.
14 K. Steinberg, X. Yuan, C. K. Klein, N. Lazouski, M. Mecklenburg, Ed., 2020, 59, 23180–23187.
K. Manthiram and Y. Li, Nat. Energy, 2022, 8, 138–148. 31 H. Heaney, Encyclopedia of Reagents for Organic Synthesis, ed.
15 O. Westhead, M. Spry, A. Bagger, Z. Shen, H. Yadegari, S. John Wiley & Sons, Ltd, Chichester, UK, 2001, p. rb250.
Favero, R. Tort, M. Titirici, M. P. Ryan, R. Jervis, Y. Katayama, 32 N.-S. Kwon and S.-W. Ryu, Solid State Ion., 2021,
A. Aguadero, A. Regoutz, A. Grimaud and I. E. L. Stephens, 361, 115563.
J. Mater. Chem. A, 2023, 11, 12746–12758. 33 J. Jones, M. Anouti, M. Caillon-Caravanier, P. Willmann and
16 E. J. McShane, V. A. Niemann, P. Benedek, X. Fu, A. C. D. Lemordant, Fluid Phase Equilib., 2009, 285, 62–68.
Nielander, I. Chorkendorff, T. F. Jaramillo and M. Cargnello, 34 Y. Zhu, X. He and Y. Mo, ACS Appl. Mater. Interfaces, 2015, 7,
ACS Energy Lett., 2023, 4024–4032. 23685–23693.

This journal is © The Royal Society of Chemistry 2024 Energy Environ. Sci., 2024, 17, 3482–3492 | 3491
 View Article Online

Energy & Environmental Science Paper

35 W. D. Richards, L. J. Miara, Y. Wang, J. C. Kim and G. Ceder, 44 K. Krempl, J. B. Pedersen, J. Kibsgaard, P. C. K. Vesborg
Chem. Mater., 2016, 28, 266–273. and I. Chorkendorff, Electrochem. Commun., 2022, 134,
36 M. He, R. Guo, G. M. Hobold, H. Gao and B. M. Gallant, 107186.
Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 73–79. 45 G. R. Fulmer, A. J. M. Miller, N. H. Sherden, H. E. Gottlieb,
37 R. D. Deshpande and D. M. Bernardi, J. Electrochem. Soc., A. Nudelman, B. M. Stoltz, J. E. Bercaw and K. I. Goldberg,
2017, 164, A461–A474. Organometallics, 2010, 29, 2176–2179.
38 A. Ramasubramanian, V. Yurkiv, T. Foroozan, M. Ragone, 46 B. Han, Z. Zhang, Y. Zou, K. Xu, G. Xu, H. Wang,
R. Shahbazian-Yassar and F. Mashayek, ACS Appl. Energy H. Meng, Y. Deng, J. Li and M. Gu, Adv. Mater., 2021,
Mater., 2020, 3, 10560–10567. 33, 2100404.
Published on 19 April 2024. Downloaded by University of Queensland on 6/18/2025 3:26:18 PM.

39 W. Lu, L. Sun, Y. Zhao, T. Wu, L. Cong, J. Liu, Y. Liu and 47 R. Abegg, F. Auerbach and I. Koppel, Handbuch der anorga-
H. Xie, Energy Storage Mater., 2021, 34, 241–249. nischen Chemie, S. Hirzel, Leipzig, 1908, vol. 2.
40 D. E. Galvez-Aranda and J. M. Seminario, J. Electrochem. 48 M. Spry, O. Westhead, R. Tort, B. Moss, Y. Katayama, M.-
Soc., 2018, 165, A717–A730. M. Titirici, I. E. L. Stephens and A. Bagger, ACS Energy Lett.,
41 J. A. Schwalbe, M. J. Statt, C. Chosy, A. R. Singh, B. A. Rohr, 2023, 8, 1230–1235.
A. C. Nielander, S. Z. Andersen, J. M. McEnaney, J. G. Baker, 49 W. Ga˛sior, B. Onderka, Z. Moser, A. De ˛bski and T. Gancarz,
T. F. Jaramillo, J. K. Norskov and M. Cargnello, ChemElec- Calphad, 2009, 33, 215–220.
troChem, 2020, 7(7), 1542–1549. 50 Y. Huang, H. Chen and L. Nyholm, Small, 2023,
42 D. Aurbach and I. Weissman, Electrochem. Commun., 1999, 2306829.
1, 324–331. 51 D. Rehnlund, F. Lindgren, S. Böhme, T. Nordh, Y. Zou,
43 T. R. P. Gibb, Jr. and C. E. Messer, NYO-3957, 4393297, 1954, J. Pettersson, U. Bexell, M. Boman, K. Edström and
DOI: 10.2172/4393297. L. Nyholm, Energy Environ. Sci., 2017, 10, 1350–1357.

3492 | Energy Environ. Sci., 2024, 17, 3482–3492 This journal is © The Royal Society of Chemistry 2024