Redox-Mediated Ambient Electrolytic Nitrogen Reduction for Hydrazine and Ammonia Generation

Article abstract of DOI:10.1002/anie.202105536

This work presents a redox-mediated electrolytic nitrogen reduction reaction (RM-eNRR) using polyoxometalate (POM) as the electron and proton carrier which frees the TiO₂-based catalyst from the electrode and shifts the reduction of nitrogen to a reactor tank. The RM-eNRR process has achieved an ammonium production yield of 25.1 μg h^?1 or 5.0 μg h^?1 cm^?2 at an ammonium concentration of 6.7 ppm. With high catalyst loading, 61.0 ppm ammonium was accumulated in the electrolyte upon continuous operation, which is the highest concentration detected for ambient eNRR so far. The mechanism underlying the RM-eNRR was scrutinized both experimentally and computationally to delineate the POM-mediated charge transfer and hydrogenation process of nitrogen molecule on the catalyst. RM-eNRR is expected to provide an implementable solution to overcome the limitations in the conventional eNRR process.

Full text of DOI:10.1002/anie.202105536

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
International Edition Chemie
www.angewandte.org
Accepted Article
Title: Redox-mediated Ambient electrolytic Nitrogen Reduction for
Hydrazine and Ammonia Generation
Authors: Xun Wang, Jing Yang, Manohar Salla, Shibo Xi, Yi Yang,
Mengsha Li, Feifei Zhang, Mingke Zhu, Songpeng Huang,
Shiqiang Huang, Yongwei Zhang, and Qing Wang
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202105536
Link to VoR: https://doi.org/10.1002/anie.202105536

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
Redox-mediated Ambient electrolytic Nitrogen Reduction for
Hydrazine and Ammonia Generation
[
a]
[b]
[a]
[c]
[a]
[a]
[a]
Xun Wang , Jing Yang , Manohar Salla , Shi-Bo Xi , Yi Yang , Meng-Sha Li , Fei-Fei Zhang ,
[
a]
[a]
[a]
*[b]
*[a]
Ming-Ke Zhu , Song-Peng Huang , Shi-Qiang Huang , Yong-Wei Zhang , Qing Wang
[
[
[
a]
b]
c]
X. Wang, Dr. M. Salla, Y. Yang, Dr. M-S. Li, Dr. F-F. Zhang, M-K. Zhu, S-P. Huang, Prof. Q. Wang
Department of Materials Science and Engineering, Faculty of Engineering
National University of Singapore, 117576, Singapore
E-mail: msewq@nus.edu.sg
Dr. J. Yang, Prof. Y-W, Zhang
Institute of High Performance Computing
A*STAR
Singapore 138632, Singapore
Dr. S-B. Xi
Singapore Synchrotron Light Source
National University of Singapore
5
Research Link, Singapore 117603, Singapore
Supporting information for this article is given via a link at the end of the document.
Abstract: Ambient electrolytic nitrogen reduction reaction (eNRR) is
extensively studied as a potential alternative to the capital and energy
intensive Habor-Bosch process for ammonia production. Most lab-
based eNRRs are performed with catalyst directly loaded on the
electrode surface, with which the catalyst loading, gas flow and
current density are constrained leading to a low rate of production.
Herein, we present a redox-mediated electrolytic nitrogen reduction
reaction (RM-eNRR) using polyoxometalate (POM) as the electron
possible. The eNRRs, especially those in water-based systems,
could potentially produce ammonia in an environmentally
sustainable manner, resulting in less carbon footprint and capital
cost.[
3,4]
Currently, most lab-based eNRRs are performed with
static cell setup[5–7], with catalyst directly loaded onto the working
electrode. As a result, catalyst loading is usually limited to 0.1-1
mg cm-2 .[8–15] On one hand, a lower loading value will inevitably
result in inaccuracy and poor reproducibility, while on the other
and proton carrier which frees the Fe-TiO
2
catalyst from the electrode
hand, a high loading value may compromise electrode
and shifts the reduction of nitrogen to a reactor tank. The RM-eNRR
conductivity, leading to low catalyst utilization and high cost.
-
1
process has achieved an ammonium production yield of 25.1 μg h or
Moreover, in a static cell setup, the gas flow is limited given the
-
1
-2
[16,17]
5
.0 μg h cm at an ammonium concentration of 6.7 ppm. With excess
space constraint in the electrode compartment.
Hence, the
catalyst, 61.0 ppm ammonium was accumulated in the electrolyte
upon continuous operation, which is the highest concentration
detected for ambient eNRR so far. The mechanism underlying the
RM-eNRR was scrutinized both experimentally and computationally
to delineate the POM-mediated charge transfer and hydrogenation
process of nitrogen molecule on the catalyst. RM-eNRR is expected
to provide an implementable solution to overcome the limitations in
the conventional eNRR process.
majority of aqueous eNRR studies based on static cells operate
at low current density with scant detected ammonia concentration.
Herein, we present
a redox-mediated electrolytic nitrogen
reduction reaction (RM-eNRR) using polyoxometalate (POM) as
the electron and proton carrier at ambient conditions. Resorting to
a redox-mediated process, NRR is liberated from the electrode
surface and occurs inside a separate reactor tank (Figure 1a, b).
Upon operation, the redox mediator POM is firstly reduced on the
electrode and circulated into the reactor tank fed with N
2
, in which
N
2
is reduced to form ammonia or hydrazine on the surface of the
Introduction
catalyst. As such, the catalytic reaction is decoupled from the
electrochemical reaction in the electrode compartment. The
current is associated with the electrochemical reduction rate of
POM regardless of the loading of catalyst in the reactor tank.
Hence, in RM-eNRR, the current density can be as high as 100
mA cm , which is considerably larger than the state-of-the-art
2
eNRR process. Moreover, as N is fed into the reactor tank rather
than to the electrode, the gas flow rate can be sufficiently strong
to ensure a fast NRR reaction on the catalyst bed.
Nitrogen reduction products play key roles in various human
activities, of which ammonia is the second most-produced
chemical globally. In the chemical industry, ammonia is an
important precursor for fertilizers, dyes, explosives, etc. In the
energy sector, ammonia is also an excellent energy carrier. Liquid
-2
-
1
ammonia has a high energy density of 4.33 kWh L (note that of
liquid hydrogen is only 2.36 kWh L ).
-1 [1,2]
Besides, hydrazine, the
intermediate nitrogen reduction product, has also been widely
used as rocket fuels, blowing agent and precursor to pesticides
and pharmaceuticals. However, the nitrogen molecule is very
stable due to its strong triple bond. Despite it has been optimized
for a century, the energy and capital-intensive Haber-Bosch
process3,4 is still the most prominent method for industrial nitrogen
reduction. Attributed to electrochemistry, ambient condition
electrolytic nitrogen reduction reaction (eNRR) has become
POM has a long history of being used as a catalyst.[18] The
electron-rich oxygen atom serves as the Lewis base active site to
[19]
donate electrons , while the transition metal center can accept
electrons due to the unoccupied d-orbitals[20]. Also, due to the
distinct highest occupied molecular orbital – lowest unoccupied
molecular orbital level of POM, it can also work as a
1
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
Figure 1. (a) Schematic of the redox-mediated electrolytic nitrogen reduction reaction (RM-eNRR) setup based on a flow cell configuration. Fe-TiO
on different substrates is loaded in the reactor tank. Negolyte: 0.05 M LiPOM solution in 2 M sulfuric acid; Posolyte: 2 M sulfuric acid. Acid trap: 2 M sulfuric acid.
b) Energy diagram showing the working principle of the RM-eNRR reaction based on a close-loop electrochemical-chemical cycle in the cathodic compartment,
2
catalyst coated
(
paired with oxygen evolution reaction in the anodic compartment.
photocatalyst[21]. POM has been employed in various organic
is the highest concentration detected for eNRR so far. Such a
close-loop electrochemical-chemical reaction mediated by redox
species provides ample room to overcome the limitations of
conventional eNRR.
syntheses, oxygen/hydrogen evolution, and CO
2
reduction
processes.[
Li
capability to reversibly store 18 electrons at a potential of around
0.40 V vs. RHE[23] and has a reasonably good water solubility of
0.37 M. Due to the low redox potential and high charge
18,22]
Here
we
have
specifically chosen
6
2 18 2
P W O62·28H O (POM) as the redox mediator considering its
-
~
Results and Discussion
concentration, the reduced POM electrolyte provides a favorable
environment for NRR in the presence of a heterogeneous catalyst.
Catalyst and POM Characterizations. The 1%Fe-TiO
was prepared by sonicating 3% (Fe/Ti weight ratio) FeCl
powder in water. Due to the loss of Fe3+ ion in the
P25 TiO
supernatant, the final powder catalyst has around 1% Fe to Ti
ratio (denoted as 1%Fe-TiO ), which is confirmed by the EDX and
2
catalyst
3
with
Transition metal oxides have been widely studied as
2
electrocatalysts for eNRR, such as TiO
etc.[8,10,12,24–26] To couple with POM in acidic conditions, TiO
selected as the catalyst. Considering pristine anatase TiO lacks
active sites for N adsorption and simple thermal treatment of TiO
2 2 2 3 2 3
, SnO , Bi O , Fe O ,
2
is
2
2
ICP results in Figure S2, S3, and Table S1. The P25 powder is
composed of 15% rutile and 85% anatase, as revealed by the
XRD pattern in Figure S4. The doping of 1% Fe does not
2
2
cannot maintain the stability of oxygen vacancy at ambient
conditions[24], here Fe3+ is introduced to create and stabilize
significantly affect the lattice structure of P25 TiO
high-resolution transmission electron microscopy (HR-TEM)
image of 1%Fe-TiO in Figure S5a shows the lattice fringes with
a d-spacing of 0.352 nm, which corresponds to the (101) planes
of anatase TiO (Figure S5b). Besides, since the precursor is P25
2
powder. The
oxygen vacancies in the TiO
TiO exhibits a reduction onset potential matching that of POM
Figure S1). When POM is at its highly reduced state, it can
spontaneously inject electrons to the TiO matrixand facilitate the
reduction process of the adsorbed nitrogen molecules as shown
2
matrix. More importantly, anatase
2
2
(
2
2
powder, the rutile phase is also captured by scanning
transmission electron microscopy (STEM) in Figure S5c-d.
2
in Figure 1b. Moreover, the Fe-TiO catalyst also possesses
adequate chemical stability in acidic electrolyte.[27] In 2 M sulfuric
acid, the system has achieved an ammonium production yield of
X-ray absorption near edge structure (XANES) profile together
-1
-2
2
5.1 μg h-1 or5.0 μg h cm at an ammonium concentration of 6.7
with the electron spin resonance (ESR) spectrum of 1%Fe-TiO
2
ppm. With excess Fe-TiO
accumulated in the electrolyte upon continuous operation, which
2
catalyst, 61.0 ppm ammonium was
prove the existence of oxygen vacancy inside the catalyst. As
illustrated in Figure 2a, the negative shift of Ti K-edge position of
2
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
cm-2 in Ar. The charged POM electrolyte was then bubbled with
N for 24 hours at room temperature. Surprisingly, nearly no
2
ammonia was detected in the electrolyte. Instead, a substantial
amount of hydrazine was predominantly present in the solution.
As the hydrazine detection could not be done with Watt and
Chrisp method[31] due to the strong heteropoly blue effect of the
solution interfering with standard UV-Vis reading, the amount of
hydrazine was determined with high-performance liquid
chromatography (HPLC) using 2-hydroxy-1-naphthaldehyde
HNA) as coloring agent[ (See Determination of Hydrazine
32]
(
section and Figure S8a-b, S9 in SI for details). The area of the
peak intensity is directly proportional to the hydrazine
concentration, which is 8.2 ppm under this condition (Figure 3a).
Although ammonia is not obtained under these conditions, these
results reveal an intriguing reaction pathway for the POM-
2
mediated reduction of N at the early stage of hydrogenation
reactions, which would then provide useful information delineating
the reaction mechanism of RM-eNRR for the further reduction of
nitrogen to ammonia.
Figure 2. (a) X-ray absorption near edge structure (XANES) profile of Ti K-
edge in pristine P25 TiO
the reduction of Ti. (b) ESR spectra of the prepared 1%Fe-TiO
TiO . (c) W L-edge XANES spectra of POM showing the reduction of W after it
2
, 1%Fe-TiO
2
and in-situ reduced Fe-TiO
2
illustrating
2
vs. pristine P25
2
was reduced with 18 equivalent electrons. (d) Fourier-transform infrared
spectroscopy (FTIR) measurement of POM charged to different numbers of
equivalent electrons and POM after one-time NRR reaction.
2 2
Fe-TiO as compared to pristine TiO suggests the possible
electron transfer from oxygen vacancy to Ti(IV). This is further
corroborated by the increase in Ti(III) 2p1/2 peak intensity in the
XPS spectrum of the 1%Fe-TiO
ESR spectrum in Figure 2b, the 1%Fe-TiO
peak at g=2.004, corresponding to the characteristic peak of
oxygen vacancy.[29] In contrast, the pristine P25 TiO
powder does
2
sample in Figure S6.[28] From the
catalyst displays a
2
2
not possess an observable peak in this region. These oxygen
vacancies serve as important active sites for nitrogen adsorption
and electron transfer, which will be discussed in the subsequent
section.
Figure 3. (a) High-performance liquid chromatography (HPLC) data of reduced
POM electrolyte after N
b) Galvanostatic intermittent titration technique (GITT) charging profile of RM-
eNRR electrolyzer. Negolyte: 20 mL 0.05 M LiPOM in 2 M H SO ; Posolyte: 50
mL 2 M H SO . (c) UV-Vis absorption spectra of the posolyte acid stained with
2 2
pumping at room temperature with pure TiO catalyst.
The electrochemical properties of POM were firstly evaluated with
a POM-CeSO flow cell in sulfuric acid, which shows that 18
4
(
2
4
electrons could be reversibly stored in POM with a coulombic
efficiency of 98.2% (Figure S1). The XANES profile together with
XPS and FTIR spectra reveals the reduction of tungsten as POM
is charged electrochemically. As shown in Figure 2c, the negative
shift of W L-edge position suggests an overall decrease in the
tungsten oxidation state. In addition, the XPS data in Figure S7
depicts an increase in W(V) peaks suggesting the reduction of
tungsten. Besides, the FTIR result in Figure 2d illustrates a drop
in peak intensity around 950 cm-1 which indicates the decrease in
W-O bond[30] and tungsten as charge storage center in POM.
2
4
indophenol blue indicator after exchange with the one-time charged negolyte at
different conditions. (d) The concentration of ammonia detected at different
2
reaction conditions in (c). Here Fe-TiO catalyst was loaded in a flow-bed
-
2
manner at 0.1 mg cm . Detailed conditions regarding state of charging,
electrolyte concentration and feeding gases are tabulated in Table S2.
Production of Ammonia. Similar tests were conducted with the
above flow cell setup while 1%Fe-TiO
2
catalyst was loaded on 5
2
pieces of 1 cm carbon paper in the reactor tank. Each had a
Production of Hydrazine. The RM-eNRR tests were conducted
with a flow cell setup (Figure 1a). The negolyte consists of 0.05 M
-2
loading of 0.1 mg cm . Due to the low density of carbon paper,
they floated on the POM electrolyte solution while gas was fed
from the bottom of the reactor tank, imitating the industrial three-
phase flow-bed reactor. Here, the negolyte consisting of 0.05 M
LiPOM in 10 mL 2 M H
2 4
SO while the posolyte consists of 50 mL
2
M H SO . A Nafion 115 membrane was used to separate the
2
4
two electrode compartments. Prior to the test, the catholyte was
bubbled with Ar for 30 minutes and sealed. The test was first
carried out with P25 TiO as the catalyst loaded on polypropylene
2
substrate, and the POM molecule in the negolyte was charged to
the most reduced state with 18 equivalent electrons[23] at 100 mA
LiPOM was dissolved in 20 mL of 2 M H
2 4
SO while the posolyte
was 50 mL of 2 M H SO . The electrolytes were bubbled with Ar
2
4
for 30 minutes before starting the experiment and the negolyte
was then charged with 18 equivalent electrons under Ar flow and
-
slightly overcharged (to 18.5 e ) to ensure that POM is at the most
3
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
Figure 4. (a) UV-Vis absorption spectra of the negolyte acid diluted 5 times and colored with indophenol blue indicator after different
charging-reaction cycles, and the exchanged ammonium concentration in the posolyte. The Ar reference was tested without dilution.
(
b) 14NNMR spectrum of non-diluted negolyte after 6 charging-reaction cycles and its comparison with standard solutions. The change
in chemical shift is due to the pH difference of the sample solution after long cycling. (c) XPS spectrum of N1s detected on Fe-TiO
catalyst soaked in 0.05 M 18-electron reduced POM bubbled with N
2
2
.
reduced state. N
2
was fed after the charging was stopped to avoid
trap containing 2 M sulfuric acid to remove ammonium. In order
to eliminate NO interference, we also attempted to purify the fed
gases with three sequential traps which comprise 2 M sulfuric acid
with 1% H , 1 M NaOH and 2 M sulfuric acid, respectively as
shown in Scheme S1.
difference in ammonium yield as compared to the setup with only
a single acid trap (Figure S13). Also, a direct test was performed
by bubbling nitrogen gas, passed through one or three traps, into
2 M sulfuric acid. As depicted in Figure S14, both methods have
NRR on the electrode. It is also worth mentioning that the
charging was carried out with the galvanostatic intermittent
titration technique (GITT) method to avoid excessive local
overpotentials that may cause precipitation of LiPOM (Figure 3b).
Due to the strong interference caused by POM, ammonium
x
2
O
2
[
34,35]
However, there was no significant
cannot be directly determined with the indophenol blue method
33]
(
Figure S11).[ Instead, after 4 hours of N
2
bubbling, the gas
supply was stopped and then the POM electrolyte underwent
membrane exchange in the same setup, which allows ammonium
to diffuse across the Nafion membrane to the posolyte. After 96
hours of exchange, the posolyte ammonium concentration was
tested, and the obtained concentration was converted back to the
initial concentration in the negolyte (Note: the 96 hours exchange
is only for the purpose of precise determination of ammonium, not
necessarily mean the whole RM-eNRR takes such long time). The
optimal exchange time was determined by a simple exchange test
with ammonium sulfate solution of known concentration in 2 M
x
shown no sign of NO contamination. Hence, to simplify the
experimental setup, most of the tests were performed with only
single acid trap. These controlled experiments unequivocally
eliminate the effect from potential contaminations and prove that
ammonium is primarily produced by the proposed RM-eNRR
2
process on Fe-TiO catalyst.
In a separate test, catalyst loading was increased by 10 times and,
instead of using carbon paper as substrate, porous titanium foam
H
2
SO
4
(Figure S12). The concentration achieved after one-time
was selected. After 6 consecutive electrolyte charging and N
2
charging for the flow-bed test is 6.7 ppm and the corresponding
yield is 25.1 μg h-1 or 5.0 μg h-1 cm . Besides, there is only
minimal amount of hydrazine detected (Figure S10). We also
compared various factors that affect ammonium production. The
amount of ammonium produced was found to depend on POM
concentration and the number of equivalent charges stored in
POM (Figure 3d).
reduction cycles, an ammonium concentration of 40.1 ppm was
accumulated as detected by the indophenol blue method (Figure
4a) and 52.5 ppm by 14N nuclear magnetic resonance (NMR)
method following reported protocol (Figure 4b, S15).[36] After 9
electrochemical-chemical reaction cycles, it further reached 61.0
ppm. Correspondingly, the ammonium yield is determined to be
25.4 μg h-1 and 5.1 μg h-1 cm . The presence of ammonium
-2
-2
species in the system was also confirmed with XPS measurement
+
The results obtained here were further validated with two
controlled experiments. The first experiment follows the same
conditions as the flow-bed tests but without catalyst loaded on the
carbon paper, for which only a trace amount of ammonium was
collected after the exchange process, indicating that the reduced
where distinct signals of N1s from NH
on the surface of Fe-TiO
4
and NH
3
were detected
2
after reaction (Figure 4c). Due to the
high concentration of ammonium accumulated, the electrolyte
released a discernable ammonia scent when neutralized with
NaOH. Unfavorably, the overall faradaic efficiency (FE) is only
1.2% because hydrogen was also spontaneously produced as
proved by the gas chromatograms in Figure S16, where a
2
POM itself could hardly reduce N despite its negative redox
potential. The second experiment had Ar purged in the system in
the presence of catalyst during the entire experiment (denoted as
Argon, Figure 3c-d), for which only around 0.3 ppm ammonium
was detected from the exchanged solution, which could be taken
as the background ammonium concentration from impurity or
contaminations for various reasons. In addition, despite the fed
gases has 99.995% purity, they were further purified with an acid
2
significant amount of H was detected in a sealed glass vial
containing the reduced POM after 15 h. Hence, the self-discharge
of charged POM via HER inevitably undermines the FE and the
yield of the RM-eNRR system. Leveraging the operational
flexibility of the redox-flow electrolytic cell, it is believed that there
is an ample room to improve the reaction yield and mitigate the
4
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
Figure 5. DFT calculations of proton-coupled charge transfer between reduced POM/Fe-TiO
2
and N
2
, and the corresponding N
2
bond length changes at different
stages of reduction and hydrogenation of N
process and further activation of N molecule in the presence of reduced POM. (III) Formation of *NH
or (III') release of hydrazine to surrounding electrolyte when pure anatase TiO is used as the catalyst. (IV) Charge transfer from reduced POM to *NH
2
molecule. (I) N
2
molecule adsorption and activation on the protonated Fe-TiO
2
catalyst surface. (II) First hydrogenation
after multiple hydrogenation steps
NH and
2
2
NH on protonated Fe-TiO
2
2
2
2
2
breakage of N-N bond eventually forming ammonia. Red, green, white, blue, grey, orange, purple and yellow spheres represent O, Ti, H, N, W, P, Fe atoms and
oxygen vacancy, respectively.
side reaction. For instance, the volume and concentration of POM
negolyte can be increased to promote the ammonium yield (see
detailed calculations in SI). Besides, the catalyst-to-electrolyte
ratio could be further optimized to an industrial fluidized bed, so
that the NRR reaction rate would be maximized at the three-phase
interface[26], which could subsequently enhance the FE as a result
of the competition between NRR and HER.
immersed in POM electrolyte charged to 18 equivalent electrons.
This gives direct proof of the reduction of surface Ti(IV) to Ti(III)
by POM. Hence, to better represent the actual conditions, a
2
protonated Fe-TiO model is applied in the DFT calculations in
which the surface Ti(III) atoms are saturated with -OH. This is also
2
in line with the literature that TiO surface has a strong tendency
to form hydrogen bond with proton in acidic environment[40]. DFT
calculations of the charge distribution was first conducted on the
Recently, false-positive results in eNRR have drawn attention to
researchers.[36–38] The contamination from the ambient
environment brings huge errors in results that are less than 1
ppm.[6,36,39] In contrast, the detected ammonium concentration in
RM-eNRR here is nearly 2 to 3 orders of magnitude higher than
most of the reported results (Table S4). Also, our study has taken
all the precautions to minimize the contaminations that could lead
to uncertainty (Scheme S2). Moreover, controlled experiments
were performed to validate the obtained results. We believe that
the RM-eNRR provides a credible method for the electrolytic
production of ammonium at ambient conditions.
traditional TiO
detailed information of the TiO
2
and protonated Fe-TiO
electrode electronic structure
model, the surface Ti atoms host
2
models to provide
2
changes. For traditional TiO
2
-
(101) model), which is increased to 64.44 e-
61.51 e (a 4x4 TiO
2
after Fe doping and protonation. The 5% increase in the total
electron of the surface Ti atoms is broadly consistent with XANES
and XPS results (Figure 2a, S17). Then, a detailed charge
transfer analysis was carried out to study the role of POM at
different charge transfer steps. The RM-eNRR process involved
with reduced POM is proposed and summarized in Figure S18.
Here, we focus on the reduction of N
that the transformation of N to *NNH is the very first and one of
the most difficult steps, and that NH NH is a critical intermediate
compound for the formation of ammonia entailing concerted
surface chemistry of Fe-TiO catalyst. It is noted that NH NH has
been reported as the final product catalyzed by a single Fe-
oxygen vacancy (V ) pair and only with multiple oxygen vacancies
(i.e. 2V next to Fe atom) could ammonium be obtained . Here,
our calculation shows that even with the ubiquitous single Fe-V
pair, NH NH is able to be reduced to ammonia when it is
mediated by POM.
2 2 2
and NH NH considering
2
Reaction Mechanism. Density functional theory (DFT)
calculations were performed to obtain an in-depth understanding
of the mechanism underlying the redox-mediated eNRR process
involving proton-coupled electron transfer (Computational details
2
2
2
2
2
are given in SI). In highly reduced POM electrolyte, Ti(IV) on TiO
2
O
[
8]
surface tends to be reduced to Ti(III), as revealed by the CV
measurement in Figure S1. The Ti K-edge XANES spectra
O
O
(
Figure 2a) of 1%Fe-TiO
2
catalyst also show a shift of the edge
2
2
position towards lower binding energy when the catalyst is
5
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
With the assistance of Fe and V
O
, N
2
is first adsorbed onto the
In summary, the RM-eNRR based on
a
close-loop
protonated Fe-TiO
are 0.71 e transferred from the reduced catalyst substrate to the
2
surface (Figure 5(I)). Upon adsorption, there
electrochemical-chemical cycle presents an implementable way
for nitrogen reduction, which breaks the limit of traditional eNRR
when it operates upon an electrolytic flow cell. With reduced POM
-
N
1
2
molecule and the nitrogen triple bond length is increased to
.19 Å (1.11 Å for isolated N ) indicating an activation of the N
molecule. The *N (* denotes adsorption) is further activated by
the reduced POM in the vicinity of the surrounding electrolyte
2
2
as
a
charge carrier that functions synergistically with
a
2
heterogeneous catalyst 1%Fe-TiO
2
, the reduction of nitrogen is
shifted from the electrode surface to a separate catalyst bed with
enhanced accessibility of the reactants. A high ammonium yield
of 25.4 µg h-1 at an unprecedented ammonium concentration of
61.0 ppm has been achieved with RM-eNRR under ambient
conditions. Hydrazine can also be produced as a different NRR
solution. The interaction of reduced POM with *N
bond leads to another 0.29 e transferred into *N
bond length is further elongated to 1.25 Å. An H atom is
concomitantly attached to *N forming *NNH (Figure 5(II)) in the
2
via hydrogen
and the N≡N
-
2
2
first hydrogenation step. Such a coupled reduction process takes
place in the subsequent hydrogenation reactions. That is, more
electrons are transferred from the instantaneously regenerated
Ti(III) (by the highly reduced POM) to *NNH along with additional
electrons donated from the reduced POM in solution. Given the
product when switching the heterogeneous catalyst to pure TiO
The compelling evidence from the extensive experimental and
DFT studies reveal that both the Fe-TiO catalyst and reduced
2
.
2
POM collectively play critical roles in the activation of the nitrogen
molecule, subsequent charge transfer and stepwise
hydrogenation process. We believe this work opens up a new way
to the electrochemical nitrogen reduction with enhanced
ammonium yield. It is anticipated that with an optimized three-
phase reactor and redox molecules in neutral conditions that
could suppress HER, the reaction yield would be further promoted
at an enhanced Faradaic efficiency.
presence of NH
the subsequent steps follow an associative alternating
pathway.[7,41] When the reaction proceeds towards the *NH
NH
2 2
NH as an intermediate product, we propose that
2
2
step, the hydrogenation of the intermediate compound and
breakage of the N-N bond are necessary to obtain ammonia. To
reveal the crucial role of Fe, we first studied the case on pure TiO
In the absence of Fe-doping in TiO , NH NH adsorbs onto the
protonated TiO surface via a hydrogen bond with an adsorption
energy of -0.74 eV, which however does not involve any charge
transfer into *NH NH . Considering NH NH stays intact in the
reduced POM electrolyte, it interprets why hydrazine is obtained
as the final product when pure TiO was used (Figure 5(III')). We
then studied the case on Fe-TiO . In contrast, with Fe-doping the
Fe-TiO substrate exerts much stronger interactions on *NH NH
2
.
2
2
2
2
Acknowledgements
2
2
2
2
This research is supported by the National Research
Foundation, Prime Minister’s Office, Singapore under its
Investigatorship Programme (Award No. NRF-NRFI2018-06).
2
2
2
2
2
with an adsorption energy of -1.36 eV, which is deemed to hold
the molecule onto the surface. However, DFT calculations show
2
Keywords: Redox mediated N reduction • Ammonium
no electron transfer from the substrate to *NH
suggesting that the Fe-TiO substrate alone is insufficient to
catalyze the formation of NH from *NH NH . The POM in solution
2
NH
2
as well,
production • Hydrazine production • Flow battery based NRR
•Decoupled catalyst and electrode
2
3
2
2
plays the crucial role in this step. As shown in Figure 5(IV), the
[1]
A. Valera-Medina, H. Xiao, M. Owen-Jones, W. I. F. David, P. J.
Bowen, Prog. Energy Combust. Sci. 2018, 69, 63–102.
R. Lan, S. Tao, Front. Energy Res. 2014, 2, 35.
-
reduced POM donates 0.83 e to *NH
2
NH
2
, which triggers the N-
[
2]
N bond breakage and eventually leads to the formation of *NH
It should be noted that the charge transfer was calculated
between a fully reduced POM and *NH NH
3
.
[3]
J. G. Chen, R. M. Crooks, L. C. Seefeldt, K. L. Bren, R. Morris
Bullock, M. Y. Darensbourg, P. L. Holland, B. Hoffman, M. J. Janik,
A. K. Jones, M. G. Kanatzidis, P. King, K. M. Lancaster, S. V.
Lymar, P. Pfromm, W. F. Schneider, R. R. Schrock, Science (80-. ).
2
2
.
2
018, 360, DOI 10.1126/science.aar6611.
M. Li, H. Huang, J. Low, C. Gao, R. Long, Y. Xiong, Small Methods
019, 3, DOI 10.1002/smtd.201800388.
[
4]
Hence, based on the above DFT calculations together with the
experimental results, the following could be concluded for the
2
[5]
G. Soloveichik, Nat. Catal. 2019, 2, 377–380.
C. Tang, S.-Z. Qiao, Chem. Soc. Rev. 2019, DOI
[
[
6]
7]
redox-mediated NRR reaction: (1) doping of Fe into TiO
introduces and stabilizes oxygen vacancies that can effectively
activate N molecule. It also enhances the adsorption and
activation of *NH NH for the eventual formation of ammonia; (2)
the surface Ti(III) atoms serve as donors transferring electrons in
the early hydrogenation process of N , while POM is the main
contributor donating electrons in the later hydrogenation process
that forms ammonia; (3) the POM molecules not only function as
charge carriers but also donate electrons directly and indirectly
2
matrix
10.1039/c9cs00280d.
X. Cui, C. Tang, Q. Zhang, Adv. Energy Mater. 2018, 8, 1–25.
T. Wu, X. Zhu, Z. Xing, S. Mou, C. Li, Y. Qiao, Q. Liu, Y. Luo, X.
Shi, Y. Zhang, X. Sun, Angew. Chemie - Int. Ed. 2019, 58, 18449–
[8]
2
2
2
18453.
[9]
M. Wang, S. Liu, T. Qian, J. Liu, J. Zhou, H. Ji, J. Xiong, J. Zhong,
C. Yan, Nat. Commun. 2019, 10, 1–8.
L. Zhang, M. Cong, X. Ding, Y. Jin, F. Xu, Y. Wang, L. Chen, L.
Zhang, Angew. Chemie Int. Ed. 2020, DOI
10.1002/anie.202003518.
L. Hui, Y. Xue, H. Yu, Y. Liu, Y. Fang, C. Xing, B. Huang, Y. Li, J.
Am. Chem. Soc. 2019, 141, 10677–10683.
2
[
[
10]
11]
(
via Ti(III)) to N
2
throughout the NRR process. The above DFT
[12]
Y. C. Hao, Y. Guo, L. W. Chen, M. Shu, X. Y. Wang, T. A. Bu, W. Y.
Gao, N. Zhang, X. Su, X. Feng, J. W. Zhou, B. Wang, C. W. Hu, A.
X. Yin, R. Si, Y. W. Zhang, C. H. Yan, Nat. Catal. 2019, 2, 448–456.
T. Xu, D. Ma, C. Li, Q. Liu, S. Lu, A. M. Asiri, C. Yang, X. Sun,
Chem. Commun. 2020, 56, 3673–3676.
X. Yu, P. Han, Z. Wei, L. Huang, Z. Gu, S. Peng, J. Ma, G. Zheng,
Joule 2018, 2, 1610–1622.
Y. Liu, Q. Li, X. Guo, X. Kong, J. Ke, M. Chi, Q. Li, Z. Geng, J.
Zeng, Adv. Mater. 2020, DOI 10.1002/adma.201907690.
L. F. Greenlee, J. N. Renner, S. L. Foster, ACS Catal. 2018, 8,
calculation gives a snapshot of the most critical reaction steps. An
extensive computational study would be required in the future to
describe the intricate stepwise proton-coupled charge transfer
process of NRR.
[
[
[
[
[
13]
14]
15]
16]
17]
7820–7827.
E. L. Clark, J. Resasco, A. Landers, J. Lin, L. T. Chung, A. Walton,
C. Hahn, T. F. Jaramillo, A. T. Bell, ACS Catal. 2018, 8, 6560–6570.
Conclusion
6
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
[
[
18]
19]
S.-S. Wang, G.-Y. Yang, Chem. Rev. 2015, 115, 4893–4962.
W. Zhang, C. W. Haskins, Y. Yang, M. Dai, Org. Biomol. Chem.
2014, 12, 9109–9112.
[
20]
M. S. S. Balula, I. C. M. S. Santos, M. M. Q. Simões, M. G. P. M. S.
Neves, J. A. S. Cavaleiro, A. M. V. Cavaleiro, J. Mol. Catal. A
Chem. 2004, 222, 159–165.
[
[
21]
22]
Y. Guo, C. Hu, J. Mol. Catal. A Chem. 2007, 262, 136–148.
Z.-H. Wang, X.-F. Wang, Z. Tan, X.-Z. Song, Mater. Today Energy
2
020, 100618.
J. J. Chen, M. D. Symes, L. Cronin, Nat. Chem. 2018, 10, 1042–
047.
[
[
[
[
23]
24]
25]
26]
1
L. Yang, T. Wu, R. Zhang, H. Zhou, L. Xia, X. Shi, H. Zheng, Y.
Zhang, X. Sun, Nanoscale 2019, 11, 1555–1562.
A. Huang, Y. He, Y. Zhou, Y. Zhou, Y. Yang, J. Zhang, L. Luo, Q.
Mao, D. Hou, J. Yang, J. Mater. Sci. 2019, 54, 949–973.
S. Chen, S. Perathoner, C. Ampelli, C. Mebrahtu, D. Su, G. Centi,
ACS Sustain. Chem. Eng. 2017, 5, 7393–7400.
[
[
[
27]
28]
29]
R. Schenk, Springer, Berlin, Heidelberg, 2001, pp. 145–170.
H. Khan, I. K. Swati, Ind. Eng. Chem. Res. 2016, 55, 6619–6633.
J. Zhang, Z. Jin, C. Feng, L. Yu, J. Zhang, Z. Zhang, J. Solid State
Chem. 2011, 184, 3066–3073.
[
30]
N. Sharma, M. Deepa, P. Varshney, S. A. Agnihotry, J. Non. Cryst.
Solids 2002, 306, 129–137.
[
[
31]
32]
G. W. Watt, J. D. Chrisp, Anal. Chem. 1952, 24, 2006–2008.
J. Wang, S. Yang, K. Zhang, J. Pharm. Biomed. Anal. 2016, 126,
1
41–147.
W. T. Bolleter, C. J. Bushman, P. W. Tidwell, Anal. Chem. 1961, 33,
92–594.
[
[
33]
34]
5
S. Liu, T. Qian, M. Wang, H. Ji, X. Shen, C. Wang, C. Yan, Nat.
Catal. 2021, 1–10.
[
[
35]
36]
E. B. Myers, T. J. Overcamp, Environ. Eng. Sci. 2002, 19, 321–327.
S. Z. Andersen, V. Čolić, S. Yang, J. A. Schwalbe, A. C. Nielander,
J. M. McEnaney, K. Enemark-Rasmussen, J. G. Baker, A. R. Singh,
B. A. Rohr, M. J. Statt, S. J. Blair, S. Mezzavilla, J. Kibsgaard, P. C.
K. Vesborg, M. Cargnello, S. F. Bent, T. F. Jaramillo, I. E. L.
Stephens, J. K. Nørskov, I. Chorkendorff, Nature 2019, 570, 504–
508.
[
37]
G. Qing, R. Ghazfar, S. T. Jackowski, F. Habibzadeh, M. M.
Ashtiani, C. Chen, M. R. Smith, T. W. Hamann, n.d., DOI
10.1021/acs.chemrev.9b00659.
[
[
38]
39]
Y. Chen, H. Liu, N. Ha, S. Licht, S. Gu, W. Li, Nat. Catal. 2020, 1–7.
D. R. MacFarlane, P. V. Cherepanov, J. Choi, B. H. R. Suryanto, R.
Y. Hodgetts, J. M. Bakker, F. M. Ferrero Vallana, A. N. Simonov,
Joule 2020, DOI 10.1016/j.joule.2020.04.004.
[
[
40]
41]
G. Tabacchi, M. Fabbiani, L. Mino, G. Martra, E. Fois, Angew.
Chemie Int. Ed. 2019, 58, 12431–12434.
C. V. S. Kumar, V. Subramanian, Phys. Chem. Chem. Phys. 2017,
19, 15377–15387.
7
This article is protected by copyright. All rights reserved.

Angewandte Chemie International Edition
10.1002/anie.202105536
RESEARCH ARTICLE
Entry for the Table of Contents
This work presents a redox mediated electrolytic nitrogen
reduction reaction (RM-eNRR) system that is able to over come
the limitation of traditional electrolytic nitrogen reduction. The
RM-eNRR system decouples the catalyst from the electrode
which gives the system huge versatility in catalyst design,
loading and gas flow. Besides, redox mediator is utilized to
transfer the charge from power source to the catalyst and
simultaneously facilitate the nitrogen reduction process. By
modifying the catalyst, either hydrazine or ammonium can be
obtained selectively. Overall, the RM-eNNR process has
achieved ammonium production yield of 25.1 μg h-1 and 61 ppm
ammonium can be accumulated via continuous operation.
8
This article is protected by copyright. All rights reserved.