pubs.acs.

org/JPCC Article

Mechanism of the N2 Cleavage Promoted by Lithium vs Other Alkali

and Alkaline-Earth Metals
Luis Miguel Azofra,* José Miguel Doña-Rodríguez, Douglas R. MacFarlane, and Alexandr N. Simonov
Cite This: J. Phys. Chem. C 2025, 129, 1198−1205 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
Downloaded via GWANGJU INST SCIENCE & TECHNOLOGY on January 20, 2025 at 08:39:45 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: The lithium-mediated nitrogen reduction reaction

(Li-NRR) currently stands as the most feasible route for producing
ammonia (NH3) from atmospheric nitrogen (N2) in a sustainable
manner under close-to-ambient conditions. The key step enabling
the Li-NRR to occur under such mild conditions is the capability of
metallic lithium (Li0) to spontaneously adsorb N2 and cleave the
triple nitrogen−nitrogen bond via a redox reaction producing
lithium nitride (Li3N)�a well-known but yet to be fully
understood process. The present study theoretically explores the
distinctive behavior of Li0, Na0, K0, Be0, Mg0, and Ca0 in terms of
their N2-philicity and metal−surface interaction, also delving into
the mechanistic aspects of diffusion and the cleavage kinetics of the
N2 molecule for Li0. In this sense, Li0 exhibits greater N2
chemisorption than the other metals examined, only surpassed by Ca0 which reveals a very high N2-philicity. In addition, the
cleavage and splitting of the N�N bond is promoted by Li0 with activation barriers less than 1.0 eV for the different Li0 facets
studied in this work.

■ INTRODUCTION
Ammonia (NH3) stands as the second most-produced
ammonia-generation devices as compared to the intricate
Haber−Bosch reactors operating at extremely high pressure
chemical compound globally,1 primarily intended for the and elevated temperature. However, a short yet highly
fertilizer industry, which sustains the world’s population. The instructive period of intense research on the NRR over the
industrial synthesis of NH3 through the Haber−Bosch process2 past decade has demonstrated that practical yield rates and
has been hailed as the most significant invention of the last selectivity of electrochemical N2 conversion to NH3 are
century,3 despite the drawback of releasing hundreds of currently achievable through a redox-mediated process only, in
millions of tons of CO2 into the atmosphere, due to its energy- particular through the lithium-mediated NRR (Li-NRR).6−8
intensive conditions and the requirement for the use of H2 Over several years of intense research on the Li-NRR, very
derived from fossil fuels.4 According to estimates by Hatzell significant progress has been achieved, transitioning it from a
and co-workers,5 if current approaches to NH3 production batch step-by-step process9,10 to continuous ammonia electro-
persist, CO2 emissions for the manufacturing of fixed-nitrogen synthesis.11−13 Innovations in the electrolyte and electrode
fertilizer feedstocks may surpass 1.3 Gton per year by 2050. design have supported demonstration of close-to-practical
This represents a disconcerting scenario in the context of a ammonia production rates,14−22 which are currently being
climate action emergency and also corresponds to significant translated into practical device prototypes. An important
consumption of the limited resources of fossil fuels. recent extension of the Li-NRR was the exploration of other
Consequently, the development of alternative, renewable- mediators23,24 including the successful demonstration of the
based processes for scalable ammonia production has become calcium-mediated NRR25 and a magnesium-based electro-
a global societal challenge. This challenge requires a delicate chemical activation of N2 at a close to ambient temperature.26
balance between the inevitable need to produce this essential However, the Li-NRR still presents the most feasible pathway
commodity and the imperative to reduce its carbon footprint.
A significant body of recent research now focuses on
developing novel materials and technologies for converting Received: November 1, 2024
atmospheric nitrogen gas (N2) into NH3 under mild Revised: December 11, 2024
conditions. Within the realm of electrochemistry and Accepted: December 16, 2024
heterogeneous catalysis, the nitrogen reduction reaction Published: December 31, 2024
(NRR) has emerged as a promising approach. This is primarily
due to the potentially lower level of complexity of electrolytic

© 2024 American Chemical Society https://doi.org/10.1021/acs.jpcc.4c07438

1198 J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

toward NH3 electrosynthesis from the perspectives of

productivity, stability in operation, and possibility of being
■ RESULTS AND DISCUSSION
N2 Adsorption: The First Stage of Contact with Metal.
coupled to a sustainable anode process like H2 or H2O The adsorption capability of N2 is defined by the lone pairs of
oxidation. this molecule. These two centers of electron density, located at
In our interest to understand why lithium exhibits such the extremes of each nitrogen atom, are poorly nucleophilic
highly favorable behavior in activating N2 as compared to other and weakly basic, as demonstrated by the presence of the low
metals, we present a comprehensive density functional theory local minimum of the molecular electrostatic potential.42 The
(DFT) study that analyzes, at the mechanistic level, the binding of N2 with electropositive centers takes place primarily
behavior of a range of alkali and alkaline-earth metals, viz., Li, with metals. In this context, we seek to explore reactive
Na, K, Be, Mg, and Ca during N2 adsorption. Notwithstanding patterns related to periodicity: do all metals favor spontaneous
that the N2 activation by Li0 is a highly energetically favorable interactions with N2? Are there metals with greater or lesser
process, this fundamental step of the Li-NRR still presents N2-philicity? Is there a pattern that categorizes them?
experimental challenges, as for example demonstrated in a In our recent study, analyzing the competition between N2
recent in situ nuclear magnetic resonance study.27 In this with protons and NOx for their reduction on surfaces of
sense, we also shed light on the diffusion of N2 and its different transition metals,43 we did not discuss the free
cleavage/splitting in/on Li0. Our approach is supported by the energies for the adsorption of N2 to form *N2. To comprehend
so-called chemical nitrogen splitting model28 and thus the the N2-philicity nature in the series of d2 to d10 d-block metals,
specific focus herein is on the first mechanistic step of the i.e., the affinity of these to interact with N2, we present the
process, viz. generation of nitrides, in particular lithium nitride results computed for flat surfaces (see ref 43 for full details).
(Li3N).29 This work supplements previous efforts, including Table 1 summarizes the binding free energies at room
seminal DFT investigations by Nørskov and co-workers9,30 and
by us,31 extending the scope of these studies toward an analysis Table 1. Metal−N2 Binding Free Energies (ΔGb) for a
of the greater variability of the metals, surface structures, and Series of d-Block Metals
types of interactions at the surface and subsurface levels.
M ECa ΔGb M ECa ΔGb

■ COMPUTATIONAL METHODS
The mechanism of N2 adsorption and splitting was studied by
Ti
Fe
d2
d6
First Row
−0.18
−0.08
V
Co
d3
d7
−0.14
0.01
DFT through the Generalized Gradient Approximation Ni d8 0.04 Cu d9 -b
(GGA). The revised Perdew−Burke−Ernzerhof (RPBE) Zn d10 0.41
functional with Pade approximation32 was used with a plane- Second Row
wave cutoff energy of 400 eV. Bulk materials were fully Nb d3 0.00 Mo d4 −0.19
optimized with very tight convergence criteria, specifically, Ru d6 −0.18 Rh d7 −0.08
energy and force convergence limits equal to 10−5 eV/atom Pd d8 0.19 Ag d9 0.26
and |0.001| eV/Å, respectively. The Brillouin zone (periodic Third Row
boundary conditions) was sampled via the Monkhorst−Pack Re d5 −0.15 Os d6 −0.17
scheme,33 increasing the number of k-points up to electronic Ir d7 0.07 Pt d8 0.27
energy differences less than |0.01| eV. Once the bulk structure Au d9 0.40
was optimized, different facets were built and reoptimized. In a
Electronic configuration. bNo *N2 minimum was located for Cu.
all cases, a vacuum width of 15 Å was imposed with the aim of
avoiding interactions between periodic slabs. For the modeling temperature for the end-on configuration of *N2 (ΔGb),
of the clean surfaces and N2 and split N adsorbed states, energy revealing a clearly differentiated behavior among early to late
and force convergence limits were set to 10−4 eV/atom and | transition metals. In general terms, elements with a more
0.01| eV/Å, respectively. Transition states (TSs) for N2 pronounced metallic character exhibit slight spontaneity in the
splitting were located using the improved DIMER method,34,35 metal−N2 interaction. However, for metals further to the right
increasing the cutoff for force convergence to |0.03| eV/Å. In in the d-block series, this interaction ceases to be exergonic,
all cases, explicit dispersion correction terms in the energy demonstrating increasingly higher values of nonspontaneity.
were included through the D3 method with the standard To complement this data, end-on binding of *N2 on selected
parameters programed by Grimme and co-workers.36,37 Free p-block metals, specifically fcc Al(111), fcc Pb(111), and hcp
energies were calculated for gases and adsorbates, as explained Bi(001), was found herein to be nonspontaneous as well (free
in Section 1 of Supporting Information. All optimization and energies of adsorption of 0.41, 0.40, and 0.31 eV, respectively)
vibrational frequency calculations were performed through the and also demonstrated the presence of small imaginary
facilities provided by the Vienna Ab Initio Simulation Package frequencies.
(VASP, version 5.4.4).38−41 In this sense, minima and first- Within this data set, the presence of a periodic behavior in
order TSs were corroborated by the presence of none or one N2-philicity is evident, indicating a linear correlation for d4 to
imaginary frequency, respectively. A series of NVT [substance d10 metals within this element series (R2 = 0.90; Figure S1,
(N), volume (V), and temperature (T); canonical ensemble] Section 2 of Supporting Information). Consequently, the
molecular dynamics (MD) simulations at the quantum level greater the metallic character, the greater the strength of the
were carried out to analyze the diffusion effects of N2 on/in metal−N2 interaction.
Li0. All ab initio MD simulations were carried out at the same If this is so for this series of d-block metals, then what about
level of theory as the optimizations with time steps of 0.5 fs. the s-block metals with a more pronounced metallic character?
Full details regarding the computational protocol can be found In the case of lithium metal, Li0, its capability to adsorb and
in Section 1 of Supporting Information. split N2 was studied for a series of low Miller index surfaces,
1199 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

namely, (001), (011), and (111) for body-centered cubic transition toward a side-on adsorption mode. It should be
(bcc) packing. The first is equivalent to the (010) and (100) noted that Ludwig et al.30 also carried out calculations on *N2
facets and the second to the (101) and (110) ones. As seen in for the case of the Li(011) surface, with their observations
Figure 1, (001) and (011) facets result in flat surfaces with being in agreement with our results. In their case, binding free
energy for *N2 was computed to be −0.48 eV at 300 K and 1
bar and in our case it was computed to be −0.46 eV under
standard conditions.
NVT MD simulations at the quantum level also support this
behavior of N2 in its interaction with the Li0 surfaces. Starting
from initial structures in which N2 is separated by 4 Å from the
nearest surface Li0 atom, Figure 3 displays the center of mass

Figure 1. Top views of the (001), (011), and (111) surfaces of bcc
Li0.

proximal surface Li0 atoms forming angles of 90° for the first
facet and 109.5 and 70.5° for the second facet. The (111) facet
is, however, a stepped surface with terrace Li0 atoms forming
angles of 120 and 60°. These models allow us to have high
structural and surface variability in order to analyze their effects
on the interactions with N2.
Thus, the side-on configuration of *N2 exhibits highly
exergonic interactions at room temperature with binding free
energies of −0.66, −0.46, and −0.85 eV for the (001), (011),
and (111) facets, respectively. These values suggest that N2 is
strongly adsorbed, if not captured, on the surface of Li0, with
the binding being highly dependent on the geometry
(optimized structures of these minima are displayed in Figure
2). On the one hand, for all side-on configurations of *N2, the

Figure 3. NVT MD simulations of N2 interaction with (001), (011),

and (111) surfaces of bcc Li0. The starting points are initial structures
in which N2 is separated by 4 Å from the nearest surface Li atom.
Purple curves show the center of mass for specific Li layers, while blue
curves represent the center of mass for N2 (both along the Z axis).
The light blue line represents the center of mass for N2 as computed
Figure 2. Top and side views for the side-on adsorption of N2 onto in each side-on *N2 minimum. Insets show selected structures,
(001), (011), and (111) facets of bcc Li0 and end-on configuration for specifically those in which the center of masses of the surface layer
(011). R(N�N) and ΔGb are corresponding N�N bond lengths and N2 coincide, i.e., snapshots in which it can be considered that N2
and binding free energies, respectively. begins to penetrate the Li0 structure.

N2 molecule interacts with surface Li0 atoms, slightly along the Z axis of selected Li0 layers and N2 over 4 ps of the
embedding itself in the crystalline structure. While the N� NVT MD simulations. In all cases, a gradual approach of N2
N distance is 1.12 Å in the gas phase, this distance is elongated toward the Li0 surface is observed, coinciding with a decrease
to 1.27, 1.28, and 1.32 Å upon adsorption onto (001), (011), in electronic energy, indicating an evolution toward more
and (111) surfaces, respectively. These values unmistakably favorable states. Thus, the N2 molecule interacts with the
indicate activation of the N2 molecule when adsorbed on Li0, surface, passes through it until being embedded, and even
with proximal Li−N distances in the range between 1.8 and 2.1 interacts between the surface and subsurface layers in the case
Å. of the (011) and (111) facets. As a result, the insertion of N2
On the other hand, the end-on configuration is only into the Li0 structure occurs on a picosecond scale. The N2
identified for the (011) facet, showing nonspontaneous center of mass of the minimum energy configuration for the
binding free energies of 0.22 eV. In the cases of (001) and side-on configuration (previously described in Figure 2),
(111), this configuration displays one imaginary frequency, indicates that N2 remains formally captured on/in Li0 over
indicating its energetic instability and its inclination to time in all cases (Figure 3).
1200 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Thus, lithium behaves as a material with a very high N2- (−0.82 and −0.85 eV, respectively), observing, in this latter
philicity, as observed experimentally. However, do other s- case, an elongated N�N bond (1.44 Å in both cases). When
block metals exhibit similar behavior? To further investigate using the sv pseudopotential, Be(110) also exhibits a slightly
this, we have carried out a comprehensive computational spontaneous binding free energy of −0.06 eV for end-on *N2.
analysis of the possible end-on and side-on interactions These results reveal a clearly differentiated reactive character
between N2 and flat surfaces for the s-block metals Na0 from dissimilar facets of the crystal (see the optimized
(bcc), K0 (bcc), Be0 (hcp), Mg0 (hcp), and Ca0 (fcc), as structures in Figure 4).
well as a stepped surface equivalent to Li(111). In all cases, For the case of Ca0, no standard pseudopotential is available,
representative slabs of the same size as those used for the restricting the analysis to sv pseudopotential with included
modeling of Li0 have been built. In addition, different semicore s states treated as valence states; unfortunately, the pv
pseudopotentials have been tested in order to rule out pseudopotential with included semicore p states treated as
contradictory results due to convergence problems common valence states has led to unfinished calculations due to
in the analysis of properties for this type of metal. For example, convergence problems. As seen in Tables 2 and S4, end-on
results given in Figure 2 for Li0 using standard pseudopotential *N2 interactions are shown to be nonspontaneous, however,
(Table 2) coincide with those obtained employing the so- Ca0 presents very intense side-on N2 capture free energies with
called sv pseudopotential (Table S4), which includes semicore highly spontaneous values that even exceed those computed
s states treated as valence states. for Li0. In these cases, N�N bond elongations are also
observed, with values between 1.3 and 1.4 Å (see optimized
Table 2. Metal−N2 Binding Free Energies, ΔGb, in eV, for a structures in Figure 4), being in consonance with the recent
Series of s-Block Metalsa report by Fu et al. on calcium-mediated NRR.25
Despite all of our efforts, the results for Mg0, whatever the
standard standard pseudopotential used, as well as after having tested different
lithium (bcc) (end-/side-) sodium (bcc) (end-/side-)
computational adjustments, have only yielded nonspontaneous
(001), (010), 0.22 −0.66 (001), (010), 0.35 0.38 *N2 interactions on all surfaces. This leaves open the question
(100) (100)
(011), (101), 0.22 −0.46 (011), (101), 0.37 0.39
of how N2 activation occurs in the case of the experimentally
(110) (110) observed magnesium-based electrochemical activation of
(111) 0.17 −0.85 (111) 0.32 0.32 dinitrogen.26
sv beryllium standard Diffusion of N2 in Lithium Bulk: Stronger Stabiliza-
potassium (bcc) (end-/side-) (hcp) (end-/side-) tion of *N2. An interesting feature of Li0 is its capability of
(001), (010), (100) 0.35 0.37 (001) 0.52 0.38 capturing N2 as suggested by our calculations at the DFT level.
(011), (101), (110) 0.34 0.35 (010), (100) 0.13 0.36 To further explore this behavior, we carried out an NVT MD
(111) 0.30 (011), (101) 0.59 simulation for the case of a (011) surface of a Li0 slab densely
(110) 0.05 covered with N2. Specifically, nine N2 molecules were
(111) −0.18 −0.82 stochastically placed at a minimum distance of 2.0 Å with
magnesium standard respect to the surface and a minimum distance of 2.0 Å
(hcp) (end-/side-) calcium (fcc) sv (end-/side-) between each N2 molecule.
(001) 0.46 1.09 (001), (010), (100) 0.28 −2.04 As seen in Figure 5, after 4 ps of NVT MD simulation, two
(010), (100) 0.40 0.96 (011), (101), (110) 0.28 −1.90 of the nine N2 molecules moved away from the surface toward
(011), (101) 0.45 0.42 (111) 0.28 −1.60 the vacuum region. Four of the remaining N2 molecules
(110) 0.37 0.39 (211) 0.26 −2.38 remained on the surface through physical adsorption via end-
(111) 0.37 0.78 on configurations. Most interestingly, three of the molecules
a
Note: standard and sv refer to standard pseudopotential and moved below the surface layer of Li atoms. The interactions of
pseudopotential with semi-core s states treated as valence states, this submersive type indicate chemisorption, being differ-
respectively. This table summarizes Table S4, in which full data are entiated, during certain periods of time in the MD simulation,
detailed. between molecules being captured on the surface and those in
subsurface Li0 layers.
As seen in Tables 2 and S4, none of the surfaces of the alkali Delving into this, Figure 6 presents the optimized structures
Na0 and K0 metals present spontaneous values of binding free of a captured N2 molecule within the subsurface Li0 layer for
energies for *N2 interactions, with ΔGb values close to 0.3−0.4 the (001), (011), and (111) facets. Remarkably, the elongation
eV. Such weak interactions cannot even be considered as of the N�N bond is more pronounced in this case as
adequate physisorption due to the long distance of N2 with compared to the surface interactions (cf. Figures 2 and 6).
respect to the surfaces. Specifically, the higher R(N�N) values suggest that the N2
For the case of Be0, we have found that, for all flat slabs, molecule is activated to a greater extent when located within
surface beryllium atoms are so proximal in distance that no subsurface Li0 atoms, which can be expected to facilitate its
side-on N2 insertions occur. Hence, these interactions are subsequent splitting into cleaved N atoms. The binding free
either very weak, like those observed in Na0 and K0, or lead to energy for *N2 in the subsurface (−0.53 eV) is slightly less
end-on interactions, in all cases with nonspontaneous binding favorable than on the surface (−0.66 eV) only for Li(001); but
free energies. However, interesting results are observed for an opposite trend is found for Li(011) and Li(111) structures,
Be(111) using both standard and sv pseudopotentials (Table which provide notably stronger binding in the bulk (−0.85 vs
S4). In this case, N2 is spontaneously adsorbed when −0.46 eV and −1.37 vs −0.85 eV, respectively). Based on these
interacting with one defective surface Be atom in the end-on data, we can conclude that N2 is likely to diffuse from the
*N2 geometry (−0.18 and −0.26 eV, respectively) and two surface to inner layers of the Li0 bulk from the (011) and
defective surface Be atoms in the side-on *N2 adsorption mode (111) facets.
1201 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 4. Top and side views for minima between N2 and s-block metals (except for lithium, which is shown in Figure 2) with spontaneous binding
free energies. All structures correspond to those optimized using pseudopotentials with semicore s states treated as valence states (the so-called sv
pseudopotentials; see Table S4). R(N�N) and ΔGb are corresponding N�N bond lengths and binding free energies, respectively.

Figure 5. NVT MD simulation of nine N2 molecules initially placed

on a (011) surface of bcc Li0 at a minimum distance of 2.0 Å. Pink,
light blue, and dark blue represent the center of mass for
nonadsorbed, physisorbed, and chemisorbed N2 molecules, while
purple indicates the center of mass for the top-most exposed layers
(first to fourth) of Li0 (in all cases along the Z axis).

Figure 7. Free energy profiles for the cleavage of N2 in the subsurface

layers of (001) (red), (011) (blue), and (111) (green) facets of bcc
Li0, and optimized structures for TSs along with the activation free
energies (ΔGac) relative to *N2 and R(N�N) values. Note that for
Li(011), two different TSs were located.

For all Li0 facets studied in this work, a similar profile is

seen: (i) spontaneous chemisorption of N2; (ii) presence of a
barrier due to activation of the N�N cleavage; and (iii)
Figure 6. Side views for *N2 captured within subsurface layers of splitting of N2 into two N adatoms showing an important drop
(001), (011), and (111) facets of bcc Li0, along with corresponding in free energy.31 Only for the case of Li(001) is the free energy
N�N distances [R(N�N)] and binding free energies (ΔGb). barrier for N2 splitting larger than the energy released during
N2 chemisorption. This is not the case for Li(011) and
N2 Splitting: How Lithium Promotes the Cleavage of Li(111), where the energy drop for N2 capture far exceeds the
the N�N Bond. Once N2 is captured inside the Li0 structure, energy required to subsequently split N2. For example, while
we were able to locate TSs that describe the cleavage of the the activation free energy for the N�N cleavage is computed
N�N bond. Figure 7 depicts the free energy profile from the as 0.78 eV for Li(001), this parameter is only 0.42 eV for
separated species (N2 gas and Li0 surface) to the N atoms Li(111). Note that these activation free energies are calculated
produced by cleavage of the N�N bond. as the difference between G(TS) and G(*N2). For Li(011),
1202 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

the reaction is even more favorable and, interestingly, exhibits

two distinct TSs with barriers of 0.31 and 0.14 eV.
■
*
ASSOCIATED CONTENT
sı Supporting Information
Our results for Li(011) are somewhat more favorable than The Supporting Information is available free of charge at
those computed by Ludwig et al.30 using the climbing image https://pubs.acs.org/doi/10.1021/acs.jpcc.4c07438.
nudged elastic band method and metadynamics calculations, in
which a free energy barrier of ca. 0.80 eV was obtained at 300 Full computational details, extended data on N 2
K and 1 bar in both cases. In this sense, examination of the TSs adsorption on transition metal slabs, structural informa-
reported by Ludwig et al. suggests that it involves a tion for the modeled s-block metals, metal−N2 binding
free energies for a series of s-block metals, and energy
configuration in which the N moieties interact with both
evolution with time (PDF)
surface and subsurface layers. In our TSs, the N moieties
interact with the subsurface Li atoms only, which might be
indicative of a more stable TS examined in our analysis. In this
regard, frequency calculations have corroborated that the
■ AUTHOR INFORMATION
Corresponding Author
saddle points located by us using the improved dimer method Luis Miguel Azofra − Instituto de Estudios Ambientales y
correspond to first-order TSs for N�N cleavage, with Recursos Naturales (iUNAT), Universidad de Las Palmas de
imaginary frequencies in the range of 350i cm−1 (see Table Gran Canaria (ULPGC), Las Palmas de Gran Canaria
S2, Section 1 of Supporting Information for further details). 35017, Spain; orcid.org/0000-0003-4974-1670;
Email: [email protected]
■ CONCLUSIONS
In summary, our DFT calculations confirm that metallic
Authors
José Miguel Doña-Rodríguez − Instituto de Estudios
lithium and calcium enable the strongest N2 adsorption of all s- Ambientales y Recursos Naturales (iUNAT), Universidad de
and d-block metals examined herein. The adsorption capability Las Palmas de Gran Canaria (ULPGC), Las Palmas de
of Li0 and other s-metals depends on the facets, which produce Gran Canaria 35017, Spain
different surface terminations. Most importantly and different Douglas R. MacFarlane − School of Chemistry, Monash
to d-block metal−N2 interactions, N2 is found to be embedded University, Clayton, Victoria 3800, Australia; orcid.org/
into the Li0 surface and subsurface rather than interacting with 0000-0001-5963-9659
the surface metal atoms only. This represents an absorption of Alexandr N. Simonov − School of Chemistry, Monash
N2 by Li0, which as far as we are aware, is a distinctive behavior University, Clayton, Victoria 3800, Australia; orcid.org/
0000-0003-3063-6539
specific of Li0 yet to be explored if similarly N2-philic Ca0
exhibits the same behavior that is highly favorable for the Complete contact information is available at:
redox-mediated NRR. Ab initio MD calculations corroborate https://pubs.acs.org/10.1021/acs.jpcc.4c07438
the N2 diffusion from vacuum into the subsurface of Li0, and
the possibility of the N2 interactions with Li0 atoms from both Notes
the surface and the subsurface layers. The free energy of N2 The authors declare the following competing financial
binding within the inner layers of the Li0 bulk is stronger than interest(s): D.R.M. and A.N.S. are shareholders in Jupiter
Ionics Pty Ltd, a start-up company focused on commercializing
the energy of chemisorption on the surface for the Li(011) and
ammonia production technology.
Li(111) structures. Once N2 is captured on/in Li0, the N�N
bond undergoes significant elongation, which produces
activated structures for the subsequent splitting of the
molecule. The activation barriers for the N�N cleavage
■ ACKNOWLEDGMENTS
L.M.A. is a Ramón y Cajal fellow (ref RYC2021-030994-I).
within bcc Li0 structures were computed to fall within the L.M.A. and J-MDR are grateful to MCIN/AEI and
range between 0.15 and 0.80 eV relative to *N2, with the most NextGenerationEU/PRTR for the financial support through
facile reaction predicted for Li(111) and especially for Li(011). project ref. PID2022-143294OB-I00. A.N.S. and D.R.M. are
grateful for funding from the Australian Research Council
Our efforts add to the other DFT studies in the area,9,30,31,44
through Linkage Project LP210301321 and Future Fellowship
and we believe that our findings will help toward a better FT200100317 (A.N.S.). The authors are also grateful to the
understanding of this distinctive chemistry as well as the design KAUST Supercomputer Laboratory (KSL), KSA, for providing
of future materials45,46 for this process of great societal and the computational resources (Shaheen II and Shaheen III).
economic importance.
Finally, the cleavage of the N�N bond into two *N species
is hypothesized to be followed by a charge transfer from Li0 to ■ REFERENCES
(1) Giddey, S.; Badwal, S. P. S.; Munnings, C.; Dolan, M. Ammonia
produce a nitrogenated layer, which can be protonated to as a Renewable Energy Transportation Media. ACS Sustain. Chem.
release NH3 and regenerate the Li+ mediator. Theoretical Eng. 2017, 5, 10231−10239.
analysis of these steps presents specific challenges, including (2) Haber, F.; Le Rossignol, R. Ü ber Die Technische Darstellung
but not limited to the preferred binding sites for hydro- von Ammoniak Aus Den Elementen. Z. Elektrochem. Angew. Phys.
genation, structural modifications during the formation of NHx Chem. 1913, 19, 53−72.
(3) Smil, V. Detonator of the Population Explosion. Nature 1999,
species, as well as an in-depth analysis of the energy cost of
400, 415.
hydrogenation on perfect or defective (N-vacancies) Li3N (4) MacFarlane, D. R.; Cherepanov, P. V.; Choi, J.; Suryanto, B. H.
surfaces, among others. These mechanistic aspects are the R.; Hodgetts, R. Y.; Bakker, J. M.; Ferrero Vallana, F. M.; Simonov, A.
focus of our ongoing work. N. A Roadmap to the Ammonia Economy. Joule 2020, 4, 1186−1205.

1203 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

(5) Lim, J.; Fernández, C. A.; Lee, S. W.; Hatzell, M. C. Ammonia (21) Nguyen, N.-T.; O’Dell, L. A.; Dinh, K. N.; Hodgetts, R. Y.;
and Nitric Acid Demands for Fertilizer Use in 2050. ACS Energy Lett. Nguyen, C. K.; Banerjee, K.; Truong, D. T. H.; Bakker, J. M.; McKay,
2021, 6, 3676−3685. A.; MacFarlane, D. R.; et al. Nitrogen Electroreduction to Ammonia
(6) Andersen, S. Z.; Č olić, V.; Yang, S.; Schwalbe, J. A.; Nielander, with Phosphonium Proton Shuttles: Mass-Transport vs. Electrode
A. C.; McEnaney, J. M.; Enemark-Rasmussen, K.; Baker, J. G.; Singh, Surface Chemistry Effects. Chem. 2024, 10, 3622.
A. R.; Rohr, B. A.; et al. A Rigorous Electrochemical Ammonia (22) Cherepanov, P. V.; Krebsz, M.; Hodgetts, R. Y.; Simonov, A.
Synthesis Protocol with Quantitative Isotope Measurements. Nature N.; MacFarlane, D. R. Understanding the Factors Determining the
2019, 570, 504−508. Faradaic Efficiency and Rate of the Lithium Redox-Mediated N2
(7) Suryanto, B. H. R.; Du, H.-L.; Wang, D.; Chen, J.; Simonov, A. Reduction to Ammonia. J. Phys. Chem. C 2021, 125, 11402−11410.
N.; MacFarlane, D. R. Challenges and Prospects in the Catalysis of (23) Tort, R.; Bagger, A.; Westhead, O.; Kondo, Y.; Khobnya, A.;
Electroreduction of Nitrogen to Ammonia. Nat. Catal. 2019, 2, 290− Winiwarter, A.; Davies, B. J. V.; Walsh, A.; Katayama, Y.; Yamada, Y.;
296. et al. Searching for the Rules of Electrochemical Nitrogen Fixation.
(8) Choi, J.; Suryanto, B. H. R.; Wang, D.; Du, H.-L.; Hodgetts, R. ACS Catal. 2023, 13, 14513−14522.
Y.; Ferrero Vallana, F. M.; MacFarlane, D. R.; Simonov, A. N. (24) Tort, R.; Bagger, A.; Westhead, O.; Kondo, Y.; Khobnya, A.;
Identification and Elimination of False Positives in Electrochemical Winiwarter, A.; Davies, B. J. V.; Walsh, A.; Katayama, Y.; Yamada, Y.;
Nitrogen Reduction Studies. Nat. Commun. 2020, 11, 5546. et al. Correction to “Searching for the Rules of Electrochemical
(9) McEnaney, J. M.; Singh, A. R.; Schwalbe, J. A.; Kibsgaard, J.; Lin, Nitrogen Fixation. ACS Catal. 2024, 14, 3169−3170.
J. C.; Cargnello, M.; Jaramillo, T. F.; Nørskov, J. K. Ammonia (25) Fu, X.; Niemann, V. A.; Zhou, Y.; Li, S.; Zhang, K.; Pedersen, J.
Synthesis from N2 and H2O Using a Lithium Cycling Electrification B.; Saccoccio, M.; Andersen, S. Z.; Enemark-Rasmussen, K.; Benedek,
Strategy at Atmospheric Pressure. Energy Environ. Sci. 2017, 10, P.; et al. Calcium-Mediated Nitrogen Reduction for Electrochemical
1621−1630. Ammonia Synthesis. Nat. Mater. 2024, 23, 101−107.
(10) Kim, K.; Cho, H.; Jeon, S. H.; Lee, S. J.; Yoo, C.-Y.; Kim, J.-N.; (26) Krebsz, M.; Hodgetts, R.; Johnston, S.; Nguyen, C. K.; Hora,
Choi, J. W.; Yoon, H. C.; Han, J.-I. Lithium-Mediated Ammonia Y.; MacFarlane, D. R.; Simonov, A. N. Reduction of Dinitrogen to
Electro-Synthesis: Effect of CsClO4 on Lithium Plating Efficiency and Ammonium through a Magnesium-Based Electrochemical Process at
Ammonia Synthesis. J. Electrochem. Soc. 2018, 165, F1027. Close-to-Ambient Temperature. Energy Environ. Sci. 2024, 17, 4481−
(11) Lazouski, N.; Chung, M.; Williams, K.; Gala, M. L.; Manthiram, 4487.
K. Non-Aqueous Gas Diffusion Electrodes for Rapid Ammonia (27) Luo, R.; Gunnarsdóttir, A. B.; Zao, E. W. Direct in Situ NMR
Synthesis from Nitrogen and Water-Splitting-Derived Hydrogen. Nat. Observation of Lithium Plating, Corrosion, Nitridation and
Catal. 2020, 3, 463−469.
Protonolysis for Ammonia Synthesis. ChemRxiv 2024,
(12) Fu, X.; Pedersen, J. B.; Zhou, Y.; Saccoccio, M.; Li, S.; Sažinas,
DOI: 10.26434/chemrxiv-2024-cpf4j.
R.; Li, K.; Andersen, S. Z.; Xu, A.; Deissler, N. H.; et al. Continuous-
(28) Cai, X.; Fu, C.; Iriawan, H.; Yang, F.; Wu, A.; Luo, L.; Shen, S.;
Flow Electrosynthesis of Ammonia by Nitrogen Reduction and
Wei, G.; Shao-Horn, Y.; Zhang, J. Lithium-Mediated Electrochemical
Hydrogen Oxidation. Science 2023, 379, 707−712.
Nitrogen Reduction: Mechanistic Insights to Enhance Performance.
(13) Li, S.; Zhou, Y.; Fu, X.; Pedersen, J. B.; Saccoccio, M.;
Andersen, S. Z.; Enemark-Rasmussen, K.; Kempen, P. J.; Damsgaard, iScience 2021, 24, 103105.
(29) Tsuneto, A.; Kudo, A.; Sakata, T. Lithium-Mediated Electro-
C. D.; Xu, A.; et al. Long-Term Continuous Ammonia Electrosyn-
thesis. Nature 2024, 629, 92−97. chemical Reduction of High Pressure N2 to NH3. J. Electroanal. Chem.
(14) Gao, L.-F.; Cao, Y.; Wang, C.; Yu, X.-W.; Li, W.-B.; Zhou, Y.; 1994, 367, 183−188.
Wang, B.; Yao, Y.-F.; Wu, C.-P.; Luo, W.-J.; et al. Domino Effect: (30) Ludwig, T.; Singh, A. R.; Nørskov, J. K. Subsurface Nitrogen
Gold Electrocatalyzing Lithium Reduction to Accelerate Nitrogen Dissociation Kinetics in Lithium Metal from Metadynamics. J. Phys.
Fixation. Angew. Chem., Int. Ed. 2021, 60, 5257−5261. Chem. C 2020, 124, 26368−26378.
(15) Suryanto, B. H. R.; Matuszek, K.; Choi, J.; Hodgetts, R. Y.; Du, (31) MacFarlane, D. R.; Simonov, A. N.; Vu, T. M.; Johnston, S.;
H.-L.; Bakker, J. M.; Kang, C. S. M.; Cherepanov, P. V.; Simonov, A. Azofra, L. M. Concluding Remarks: Sustainable Nitrogen Activation
N.; MacFarlane, D. R. Nitrogen Reduction to Ammonia at High − Are We There Yet? Faraday Discuss. 2023, 243, 557−570.
Efficiency and Rates Based on a Phosphonium Proton Shuttle. Science (32) Hammer, B.; Hansen, L. B.; Nørskov, J. K. Improved
2021, 372, 1187−1191. Adsorption Energetics within Density-Functional Theory Using
(16) Du, H.-L.; Chatti, M.; Hodgetts, R. Y.; Cherepanov, P. V.; Revised Perdew-Burke-Ernzerhof Functionals. Phys. Rev. B 1999, 59,
Nguyen, C. K.; Matuszek, K.; MacFarlane, D. R.; Simonov, A. N. 7413−7421.
Electroreduction of Nitrogen with Almost 100% Current-to-Ammonia (33) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone
Efficiency. Nature 2022, 609, 722−727. Integrations. Phys. Rev. B 1976, 13, 5188−5192.
(17) Li, K.; Shapel, S. G.; Hochfilzer, D.; Pedersen, J. B.; Krempl, K.; (34) Henkelman, G.; Jónsson, H. A Dimer Method for Finding
Andersen, S. Z.; Sažinas, R.; Saccoccio, M.; Li, S.; Chakraborty, D.; Saddle Points on High Dimensional Potential Surfaces Using Only
et al. Increasing Current Density of Li-Mediated Ammonia Synthesis First Derivatives. J. Chem. Phys. 1999, 111, 7010−7022.
with High Surface Area Copper Electrodes. ACS Energy Lett. 2022, 7, (35) Heyden, A.; Bell, A. T.; Keil, F. J. Efficient Methods for Finding
36−41. Transition States in Chemical Reactions: Comparison of Improved
(18) Westhead, O.; Spry, M.; Bagger, A.; Shen, Z.; Yadegari, H.; Dimer Method and Partitioned Rational Function Optimization
Favero, S.; Tort, R.; Titirici, M.; Ryan, M. P.; Jervis, R.; et al. The Role Method. J. Chem. Phys. 2005, 123, 224101.
of Ion Solvation in Lithium Mediated Nitrogen Reduction. J. Mater. (36) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and
Chem. A 2023, 11, 12746−12758. Accurate Ab Initio Parametrization of Density Functional Dispersion
(19) Du, H.-L.; Matuszek, K.; Hodgetts, R. Y.; Ngoc Dinh, K.; Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,
Cherepanov, P. V.; Bakker, J. M.; MacFarlane, D. R.; Simonov, A. N. 132, 154104.
The Chemistry of Proton Carriers in High-Performance Lithium- (37) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping
Mediated Ammonia Electrosynthesis. Energy Environ. Sci. 2023, 16, Function in Dispersion Corrected Density Functional Theory. J.
1082−1090. Comput. Chem. 2011, 32, 1456−1465.
(20) Chorkendorff, I.; Fu, X.; Xu, A.; Pedersen, J. B.; Li, S.; Sažinas, (38) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Liquid
R.; Zhou, Y.; Andersen, S.; Saccoccio, S.; Deissler, N. H.; Mygind, J.; Metals. Phys. Rev. B 1993, 47, 558−561.
et al. Phenol as Proton Shuttle and Buffer for Lithium-Mediated (39) Kresse, G.; Hafner, J. Ab Initio Molecular-Dynamics Simulation
Ammonia Electrosynthesis. Research Square 2023, DOI: 10.21203/ of the Liquid-Metal–Amorphous-Semiconductor Transition in
rs.3.rs-3166795/v1. Germanium. Phys. Rev. B 1994, 49, 14251−14269.

1204 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

(40) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab

Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys.
Rev. B 1996, 54, 11169−11186.
(41) Kresse, G.; Furthmüller, J. Efficiency of Ab-Initio Total Energy
Calculations for Metals and Semiconductors Using a Plane-Wave
Basis Set. Comput. Mater. Sci. 1996, 6, 15−50.
(42) Bettens, T.; Pan, S.; De Proft, F.; Frenking, G.; Geerlings, P.
Alkaline Earth Metals Activate N2 and CO in Cubic Complexes Just
Like Transition Metals: A Conceptual Density Functional Theory and
Energy Decomposition Analysis Study. Chem.�Eur. J. 2020, 26,
12785−12793.
(43) Carro, P.; Choi, J.; MacFarlane, D.; Simonov, A. N.; Doña-
Rodriguez, J. M.; Azofra, L. M. Competition between Metal-Catalysed
Electroreduction of Dinitrogen, Protons, and Nitrogen Oxides: A
DFT Perspective. Catal. Sci. Technol. 2022, 12, 2856−2864.
(44) Bagger, A.; Tort, R.; Titirici, M.-M.; Walsh, A.; Stephens, I. E.
L. Electrochemical Nitrogen Reduction: The Energetic Distance to
Lithium. ACS Energy Lett. 2024, 9, 4947−4952.
(45) Goyal, I.; Kani, N. C.; Olusegun, S. A.; Chinnabattigalla, S.;
Bhawnani, R. R.; Glusac, K. D.; Singh, A. R.; Gauthier, J. A.; Singh, M.
R. Metal Nitride as a Mediator for the Electrochemical Synthesis of
NH3. ACS Energy Lett. 2024, 9, 4188−4195.
(46) Li, S.; Fu, X.; Nørskov, J. K.; Chorkendorff, I. Towards
Sustainable Metal-Mediated Ammonia Electrosynthesis. Nat. Energy
2024, 9, 1344−1349.

1205 https://doi.org/10.1021/acs.jpcc.4c07438
J. Phys. Chem. C 2025, 129, 1198−1205