Molybdenum-catalysed ammonia production with samarium diiodide and alcohols or water

Article abstract of DOI:10.1038/s41586-019-1134-2

The production of ammonia from nitrogen gas is one of the most important industrial processes, owing to the use of ammonia as a raw material for nitrogen fertilizers. Currently, the main method of ammonia production is the Haber–Bosch process, which operates under very high temperatures and pressures and is therefore very energy-intensive¹. The transition-metal-catalysed reduction of nitrogen gas^2–6 is an alternative method for the formation of ammonia. In these reaction systems, metallocenes or potassium graphite are typically used as the reducing reagent, and conjugate acids of?pyridines or related compounds are used as a proton source. To develop a next-generation nitrogen-fixation system, these reagents should be low cost, readily available and environmentally friendly. Here we show that the combination of samarium(ii) diiodide (SmI₂) with alcohols or water enables the fixation of nitrogen to be catalysed by molybdenum complexes under ambient conditions. Up to 4,350 equivalents of ammonia can be produced (based on the molybdenum catalyst), with a turnover frequency of around 117 per minute. The amount of ammonia produced and its rate of formation are one and two orders of magnitude larger, respectively, than those achieved in artificial reaction systems reported so far, and the formation rate approaches that observed with nitrogenase enzymes. The high reactivity is achieved by a proton-coupled electron-transfer process that is enabled by weakening of the O–H bonds of alcohols and water coordinated to SmI₂. Although the current reaction is not suitable for use on an industrial scale, this work demonstrates an opportunity for further research into catalytic nitrogen fixation.

Full text of DOI:10.1038/s41586-019-1134-2

Letter
https://doi.org/10.1038/s41586-019-1134-2
Molybdenum-catalysed ammonia production with
samarium diiodide and alcohols or water
Yuya Ashida1, Kazuya Arashiba1, Kazunari Nakajima2 & Yoshiaki Nishibayashi1*
The production of ammonia from nitrogen gas is one of the most
To overcome this difficulty, we considered the combination of samar-
important industrial processes, owing to the use of ammonia as a ium(ii) diiodide (SmI2) and simple alcohols or water as the reducing
raw material for nitrogen fertilizers. Currently, the main method of reagent and proton source, respectively. SmI2 is well established as an
ammonia production is the Haber–Bosch process, which operates effective reducing reagent of various functional groups in organic reac
-
under very high temperatures and pressures and is therefore very tions13,14. As a result, we have found a highly efficient method that can
energy-intensive1. The transition-metal-catalysed reduction catalytically form ammonia from nitrogen gas using the combination
of nitrogen gas2–6 is an alternative method for the formation of of SmI2 and ethylene glycol or water in the presence of molybdenum
ammonia. In these reaction systems, metallocenes or potassium complexes as catalysts. In this reaction system, a stoichiometric amount
graphite are typically used as the reducing reagent, and conjugate of SmI2 as a reducing reagent is required for the reduction of nitrogen
acids of pyridines or related compounds are used as a proton source. gas to ammonia. Hereafter, we have defined the combination of SmI2
To develop a next-generation nitrogen-fixation system, these and alcohols or water as reaction system B (Fig. 1a).
reagents should be low cost, readily available and environmentally
We reacted nitrogen gas at atmospheric pressure with 180 equiv-
friendly. Here we show that the combination of samarium(ii) alents (equiv.) (to molybdenum catalyst 1a) of SmI2 as a reducing
diiodide (SmI2) with alcohols or water enables the fixation of reagent and 180 equiv. (to 1a) of ethylene glycol as a proton source, in the
nitrogen to be catalysed by molybdenum complexes under ambient presence of 1a, in tetrahydrofuran (THF) at room temperature for 0.5 h.
conditions. Up to 4,350 equivalents of ammonia can be produced This gave 39.7 equiv. of ammonia based on the molybdenum atom
(based on the molybdenum catalyst), with a turnover frequency of (65% yield based on SmI2) together with 15.0 equiv. of hydrogen gas
around 117 per minute. The amount of ammonia produced and based on the molybdenum atom (16% yield based on SmI2) (Fig. 1b).
its rate of formation are one and two orders of magnitude larger, Separately, we confirmed that there was no formation of ammonia
respectively, than those achieved in artificial reaction systems in the absence of either molybdenum complexes or nitrogen gas.
reported so far, and the formation rate approaches that observed Although the use of a longer reaction time (18 h) did slightly increase
with nitrogenase enzymes. The high reactivity is achieved by the yield of ammonia (up to 42.8 equiv. based on the Mo atom), 21.7
a proton-coupled electron-transfer process that is enabled by equiv. of ammonia were produced in just the first minute of the reac-
weakening of the O–H bonds of alcohols and water coordinated to tion. The time profile of the catalytic reaction is shown in Fig. 1c, in
SmI2. Although the current reaction is not suitable for use on an which it can be seen that ammonia formation is almost complete in
industrial scale, this work demonstrates an opportunity for further 0.5 h. The turnover frequency (TOF) of ammonia formation—which
research into catalytic nitrogen fixation.
is determined as the number of moles ammonia produced per molyb-
In recent years, transition-metal dinitrogen complexes have been denum atom in the initial 1 min—is 21.7 min−1 (1,300 h−1). This is
found to be effective catalysts for the formation of ammonia under mild nearly two orders of magnitude larger than that observed7 in reaction
reaction conditions2–6. Our group has reported molybdenum iodide system A (TOF = 28 h−1).
complexes that bear a pyridine-based PNP-pincer ligand [MoI3(PNP)]
Next, because the reducing ability of SmI2 depends on the nature
(1a; PNP = 2,6-bis(di-tert-butylphosphinomethyl)pyridine) and cat- of the alcohol15, we investigated these catalytic reactions using other
alyse the production of ammonia under ambient conditions, in which alcohols as proton sources. For comparison, all catalytic reactions
decamethylcobaltocene (CoCp*2) (Cp* = η5-C5Me5) and 2,4,6- were carried out for 18 h. Typical results are shown in Table 1. When
trimethylpyridinium trifluoromethanesulfonate ([ColH]OTf) are used methanol (MeOH), ethanol (EtOH), 2-propanol (iPrOH), tert-butyl
as a reducing reagent and a proton source, respectively7. Hereafter, alcohol (tBuOH), 2,2,2-trifluoroethanol (CF3CH2OH) and phenol
we define this system as reaction system A (Fig. 1a). However, so far, (PhOH) were used as proton sources instead of ethylene glycol, the
the catalytic reductions of nitrogen gas to ammonia using transition- amount of ammonia produced—based on the molybdenum atom
metal dinitrogen complexes have required reducing reagents and of the catalyst—was smaller (Table 1, entries 1–7). [ColH]OTf was
proton sources that are expensive and not readily available2–7. Simple not an effective proton source in this reaction (Table 1, entry 8),
alcohols and water are considered to be good candidates for proton although we had previously found that this triflate was the most
sources. However, the use of these compounds as proton sources is effective proton source for ammonia formation in combination
difficult because they can react with low-valent transition metal com- with CoCp*2 as a reducing reagent7. Conversely, no ammonia for-
plexes to give oxygen-coordinated complexes—for example metal–oxo mation was observed at all when CoCp*2 was used as a reducing
and related complexes8—which generally inhibit the regeneration of reagent in place of SmI2 (Table 1, entry 9), although CoCp*2 (Ered in
the corresponding catalytically active metal dinitrogen complexes9. acetonitrile = –1.85 V versus the ferrocene/ferrocenium ion redox
Although water and some simple alcohols have previously been used couple (Fc/Fc+)16) has a higher reducing ability than SmI2 (Ered in
as proton sources in the formation of ammonia under mild conditions, MeCN = –1.22 V versus Fc/Fc+17). These results clearly indicate
these reactions were stoichiometric9,10; to our knowledge, their success- that the combination of SmI2 and ethylene glycol—that is, reac-
ful use in the catalytic formation of ammonia under mild conditions tion system B—is an effective reaction system for the formation of
has not yet been reported11,12
.
ammonia.
1Department of Systems Innovation, School of Engineering, The University of Tokyo, Tokyo, Japan. 2Frontier Research Center for Energy and Resources, School of Engineering, The University of
Tokyo, Tokyo, Japan. *e-mail: ynishiba@sys.t.u-tokyo.ac.jp
5 3 6
| N A t U r e | V O L 5 6 8 | 2 5 A P r I L 2 0 1 9

Letter reSeArCH
Reaction system A (previous work)
a
c
80.0
60.0
NH3
H2
cat. [Mo]
RT
N2
+
6CoCp*2 + 6[ColH]OTf
2NH3 + 6[CoCp*2][OTf] + 6Col
42.8
39.7
Reducing
reagent
Proton
source
35.7
(1 atm)
40.0
20.0
26.8
3.8
21.7
0.7
Reaction system B (this work)
6ROH
22.2
18
15.0
0.5
17.7
6Sml2(OR)
cat. [Mo]
RT
0.0
N2
+
6Sml2
+
or
2NH3
+
or
0
0.25
6H2O
6Sml2(OH)
Reaction time (h)
Reducing
reagent
Proton
source
up to 4,350 equiv./Mo
up to 7,000 h–1
(1 atm)
d
9
(0.025 μmol)
N2
+
6Sml2
+
6ROH
2NH3 + 6Sml2(OR)
H2
b
THF, RT, 4 h
1a
(1 atm) 14,400 equiv. 14,400 equiv.
(0.002 mmol)
N2
+
6Sml2
+
6ROH
2NH3 + 6Sml2(OR)
ROH
NH3
THF, RT, 0.5 h
HOCH2CH2OH
3,650 250 equiv./Mo 1,600 150 equiv./Mo
(1 atm) 180 equiv. 180 equiv.
R = CH2CH2OH
39.7 4.9
equiv./Mo
65 8%a
76 6%a
22 2%a
H2O
4,350 150 equiv./Mo
91 4%a
150 100 equiv./Mo
2
1%a
Fig. 1 | Catalytic reduction of nitrogen gas to ammonia in the presence
of molybdenum complexes. a, Previous and current reaction systems for
catalytic ammonia formation. b, Catalytic reduction of nitrogen gas (1 atm)
in the presence of 1a using SmI2 and ethylene glycol in THF at room
temperature. c, Time profile of the catalytic formation of ammonia and
dihydrogen using 1a. Data are mean of multiple individual experiments
(at least 2), with error bars representing the s.d. d, Results of the catalytic
reaction using larger amounts of SmI2 and ethylene glycol or water in the
presence of catalyst 9. aYield based on SmI2. SmI2(THF)2 was used as the
source of SmI2.
This combination was then applied to reactions using other molyb- 5 equiv. of SmI2 in THF at room temperature for 5 min under 1 atm of
denum complexes7 as catalysts. Tribromide and trichloride complexes N2 gave the corresponding molybdenum nitride complex [MoI(≡N)
of molybdenum that bear a pyridine-based PNP-type pincer ligand (PNP)] (4) in 76% yield (based on nuclear magnetic resonance (NMR)
[MoX3(PNP)] (1b, X = Br; 1c, X = Cl) were found to be more effective spectroscopy) (Fig. 2a). Complex 4 was previously isolated and charac-
catalysts than 1a (Extended Data Table 1, entries 1 and 2). Conversely, terized using reaction system A7. Notably, reactions of 1b and 1c gave
a dinitrogen-bridged dimolybdenum complex bearing pyridine-based the same complex, 4, in yields of 74% and 85% respectively (based on
PNP-pincer ligands [Mo(N2)2(PNP)]2(μ-N2) (2) and a molybdenum NMR). This shows that molybdenum halide complexes readily under-
trichloride complex bearing a triphosphine (PPP) ligand [MoCl3(PPP)] went ligand exchange with the iodide derived from SmI2. When 4 was
(3; PPP = bis(di-tert-butylphosphinoethyl)phenylphosphine) were less allowed to react with 3 equiv. of SmI2 and 3 equiv. of ethylene glycol in
effective than 1a (Extended Data Table 1, entries 3 and 4).
THF at room temperature for 1 h under 1 atm of argon, ammonia was
To obtain more information on the reaction pathway, we carried out produced quantitatively (Fig. 2b). Separately, we confirmed that 4 has
various stoichiometric and catalytic reactions. The reaction of 1a with a similar catalytic activity to 1 in the formation of ammonia (Table 1,
Table 1 | Catalytic nitrogen fixation using typical alcohols as proton sources in the presence of molybdenum complexes as catalysts
Mo catalyst
N2
+
6SmI2
+
6ROH
2NH3
+
6SmI2(OR) (+ H2)
THF, RT, 18 h
(1 atm)
Mo catalyst
N
N
N
I
I
tBu2P
N
Cl
PtBu2
O
PtBu2
PtBu2
N
PtBu2
PtBu2
N
N
N
N
N
Mo
I
I
N
Mo
Mo
N
N
Mo
Mo Cl
Cl
tBu2
N
Mo
I
I
N
N
P
P
P
N
P
P
P
tBu2
N
N
tBu2
tBu2
OH2
tBu2
tBu2
N
1a
4
5
8
9
Entry
1
Mo catalyst
Alcohol
NH3 production (equiv. based on Mo)
NH3 yield (%)a
70 2b
28 1b
24
H2 production (equiv. based on Mo)
H2 yield (%)a
1a
1a
1a
1a
1a
1a
1a
1a
1a
4
HOCH2CH2OH
MeOHc
EtOHc
iPrOHc
tBuOHc
CF3CH2OHc
PhOHc
[ColH][OTf]c
HOCH2CH2OH
HOCH2CH2OH
HOCH2CH2OH
HOCH2CH2OH
HOCH2CH2OH
42.8 1.5b
17.2 1.1b
14.5
22.2 4.5b
40.7 1.9b
41.6
24 4b
44 12b
2
3
45
57
45
60
50
68
0
4
11.8
19
52.2
5
7.7
13
41.2
6
13.8
22
55.4
7
16.5
27
46.1
8
14.4
23
63.0
9d
10
11
12
13
0
0
0
50.0 0.1b
85 1b
7.7 1.0b
9
1b
5
44.1
76
17.9
20
10 1b
8
53.3 2.1b
55.0 0.9b
88 4b
92 2b
9.1 0.7b
4.0 0.8b
9
5
1b
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.
bData are mean of multiple individual experiments (at least 2) with error bars representing the s.d.
cA proton source (0.72 mmol; 360 equiv. based on the molybdenum atom of 1a) was used.
dCoCp*2 was used as the reducing reagent instead of SmI2.
2 5 A P r I L 2 0 1 9
|
V O L 5 6 8
|
N A t U r e | 5 3 7

reSeArCH Letter
a
a
b
X
N
I
N2
(1 atm)
PtBu2
PtBu2
I
HO
O
Sm
OH
OH
H
O
H
O
N
Mo
X
X
N
Mo
+
SmI2
SmI2 + 3HOCH2CH2OH
THF
RT, 5 min
5 equiv.
THF
RT, 5 min
P
PtBu2
HO
tBu2
4
n
I
1a: X = I
1b: X = Br
1c: X = Cl
76%a (from 1a)
74%a (from 1b)
85%a (from 1c)
6
63%
tBu2
P
I4
1a
(0.002 mmol)
Recryst.
(EtOH
hexane)
OH
HO
I
HO
OH
HO
Mo
N
N
2 (1atm)
O
H
Sm
O
Et
N
SmI2
HOCH2CH2OH
900 equiv.
Sm
OH
+
Et
O
N
THF
H
O
N
P
180 equiv.
RT, 18 h
tBu2
tBu2P
OH
HO
Mo
I
P
OH
HO
tBu2
7a
Ι
74%
b
c
N
Ar
(1 atm)
PtBu2
I
1a
(0.002 mmol)
2 (1atm)
I
O
O
HOCH2CH2OH
3 equiv.
+
+
SmI2
3 equiv.
NH3
N
Mo
I
THF
RT, 1 h
O
N
O
>99%
P
Sm
I
O
O
O
Sm
SmI2
HOCH2CH2OH
180 equiv.
+
tBu2
THF
RT, 18 h
180 equiv.
HO
O
Sm
O
I
OH
4
O
Fig. 2 | Stoichiometric reactions of ammonia formation in reaction
system B. a, Stoichiometric reduction of molybdenum trihalide complexes
(1a–c) to a molybdenum nitride complex (4) under 1 atm of N2.
b, Stoichiometric formation of ammonia from 4. aYield based on NMR.
SmI2(THF)2 was used as the source of SmI2.
I
7b
Fig. 3 | Samarium species in the catalytic reaction. a, Ethylene glycol-
bridged polysamarium(ii) complex (6) obtained from the reaction
mixture using SmI2 and 3 equiv. of ethylene glycol. b, Oxygen-bridged
disamarium(iii) complex (7a) obtained from the reaction mixture of the
catalytic reaction using 180 equiv. (to 1a) of SmI2 and 900 equiv. (to 1a) of
ethylene glycol (after recrystallization from ethanol/hexane). c, Oxygen-
bridged trisamarium (iii) complex (7b) obtained from the reaction mixture
of the catalytic reaction using 180 equiv. (to 1a) of SmI2 and 180 equiv.
(to 1a) of ethylene glycol. SmI2(THF)2 was used as the source of SmI2.
entry 10), which indicates that 4 is involved as one of the key reac-
tive intermediates in this catalytic reaction. We also confirmed that 4
was recovered unreacted from reactions using either SmI2 or ethylene
glycol.
We have observed the same result in both stoichiometric and cata-
lytic reactions using 1 and 4 in the reaction system A7. On the basis of
the experimental results and density functional theory calculations,
we have previously proposed a reaction pathway via direct cleavage of
the nitrogen–nitrogen triple bond of the bridging dinitrogen ligand
in [MoI(PNP)–N≡N–MoI(PNP)] (I)7 (Extended Data Fig. 1a). The
results presented in this paper may suggest that ammonia formation
using reaction system B proceeds via a similar pathway as reaction
system A.
(to 1a) of ethylene glycol, although we isolated only a small amount of
this complex in a pure form (Fig. 3c). Detailed molecular structures of
7a and 7b were confirmed by X-ray analysis. In both complexes, the
samarium centre had been oxidized from Sm(ii) to Sm(iii) together
with the loss of 1 equiv. of protons (based on the samarium atom),
showing that SmI2 and ethylene glycol may formally act as a one-
electron reducing reagent and also as a proton source in the process of
protonation and reduction in this reaction system.
Next, the reactivity of a molybdenum–oxo complex was investi-
gated in detail. The reaction of a cationic molybdenum oxo complex
bearing an aqua ligand [MoI(=O)(OH2)(PNP)]I (5) (Extended Data
Fig. 1b) with 5 equiv. of SmI2 in THF at room temperature for 1 h
under 1 atm of N2 gave the molybdenum nitride complex 4 in 36%
yield (based on NMR). This suggests that, in the presence of SmI2, an
oxygen atom can be abstracted from 5 to give 4 via direct cleavage of
the nitrogen–nitrogen triple bond of the bridging dinitrogen ligand in
complex I. Separately, we confirmed that 5 has a similar catalytic activ-
ity to 1 and 4 under the same reaction conditions (Table 1, entry 11).
This shows that the formation of molybdenum oxo complexes—including
5—does not interfere with the formation of ammonia, even if the com-
plexes are formed during the catalytic reaction in reaction system B.
Next, we turned our attention to the samarium species in this cata-
lytic reaction. We did not isolate any samarium(ii) complexes from the
reaction of SmI2 with the same number of equivalents of ethylene glycol
in THF at room temperature. However, from the reaction of SmI2 with
3 equiv. (to SmI2) of ethylene glycol, we isolated (in 63% yield) an eth-
ylene glycol-bridged polysamarium(ii) complex bearing two ethylene
glycol molecules and one THF molecule (6) (Fig. 3a). The molecular
structure of 6 was confirmed by X-ray analysis. Some samarium(ii)
complexes with alcohols as ligands have already been prepared via the
reaction of SmI2 with various alcohols15.
To obtain more information on the reaction pathway, we estimated
the kinetic isotope effect of the catalytic reaction using deuterated eth-
ylene glycol (DOCH2CH2OD) as a proton source. The ratio of the rate
constants of the reactions involving hydrogen and deuterium (kH/kD)
is 2.0, which is consistent with that of proton-coupled electron transfer
reactions with SmI218–21. On the basis of the kinetic isotope effect and
the unique reactivity of the combination of SmI2 and ethylene glycol, we
consider that ammonia formation in reaction system B proceeds via a
proton-coupled electron transfer pathway18–21 from samarium ethylene
glycol complexes to nitrogenous ligands on the molybdenum atom of
nitride, imide and amide complexes3,22,23. At present, we consider that
SmI2 and ethylene glycol are not separate electron and proton sources,
but rather they work together24–26. Thus, we propose that ethylene glycol
first coordinates to SmI2, resulting in the weakening of the O–H bond;
this enables the delivery of the hydrogen atoms to nitrogenous lig-
ands on the molybdenum catalyst. It is known that the O–H bonds in
the SmI2–ethylene glycol complex are sufficiently reactive to enable
N–H bond formation in reduced molybdenum ammine and -amide
complexes3,22,23. For example, when water complexes with SmI2,
the effective O–H bond strength in [Sm(H2O)n]2+ is 26 kcal mol−1
(ref. 21), compared that of free water at 111 kcal mol−1 (see below).
Next we investigated catalytic reactions using other molybde-
num complexes as catalysts. We recently found that a molybdenum
complex bearing N-heterocyclic carbene- and phosphine-based
PCP-pincer ligands [Mo(N2)2(PCP)]2(μ-N2) (8; PCP = 1,3-bis((di-tert-
butylphosphino)methyl)benzimidazol-2-ylidene) was a more effective
catalyst than molybdenum complexes bearing pyridine-based PNP-
pincer ligands27. Accordingly, complex 8 and a molybdenum trichloride
Conversely, an oxygen-bridged disamarium(iii) complex (7a) was
obtained in 74% isolated yield (after recrystallization in ethanol/
hexane) from the reaction mixture of the catalytic reaction using 180
equiv. (to 1a) of SmI2 and 900 equiv. (to 1a) of ethylene glycol (Fig. 3b).
Additionally, an oxygen-bridged trisamarium(iii) complex (7b) was
obtained after reaction with 180 equiv. (to 1a) of SmI2 and 180 equiv.
5 3 8
| N A t U r e | V O L 5 6 8 | 2 5 A P r I L 2 0 1 9

Letter reSeArCH
a
+
6e–
+
6H+
2NH3
PtBu2
N2
1 atm
This work
N
N
K(THF)2
I
I
Cl
Cl
PtBu2
PtBu2
PtBu2
PiPr2
PiPr2
N
N
iPr2P
Os
Si
N
Mo
I
I
N
Mo
I
I
Nitrogenase
40–120 min–1
Mo Cl
Cl
Mo Cl
P
Cl
N
N
P
P
P
tBu2
tBu2
tBu2
tBu2
1a
1a
9
9
117.0 min–1
112.9 min–1
21.7 min–1
0.5 min–1
(28 h–1
2 min–1
(120 h–1
)
Turnover
frequency
(7,000 h–1
)
(6,800 h–1
)
(1,300 h–1
)
)
Reducing reagent
Proton source
SmI2
SmI2
H2O
RT
SmI2
CoCp*2
CoCp*2
HOCH2CH2OH
RT
HOCH2CH2OH
RT
[ColH]OTf
[H2NPh2]OTf
RT
ref. 7
–78 ºC
ref. 6
RT
ref. 28
b
400.0
300.0
291.6
200.0
100.0
168.8
135.6
94.9
44.4
26.4
0.0
a
[H][BArF
]
[Ph2NH2]OTfc
Cp*2Coc
a
Proton source
[LutH][OTf]a [LutH][BArF
]
[ColH][OTf]b
4
4
2+ d
Sm(H2O)n
a
Reducing reagent
Cp*2Cr a
Cp*2Cob
Cp2Co a
KC8
System A
System B
Fig. 4 | Typical turnover frequencies and chemical overpotentials.
are corrected from the reported value according to a literature method31.
a, Comparison of typical TOFs obtained in the current reaction and known aThe value reported in ref. 29. bThe system was reported by our group7.
catalytic reactions. b, Chemical overpotential is calculated as 6(48.6 −
BDFE), where effective bond dissociation free energy (BDFE) is calculated
as 1.37(pKa) + 23.06(Ered) + C (where C is the H+/H• standard reduction
potential). C(MeCN) = 54.9 kcal mol−1 (ref. 30). Reduction potentials
pKa(MeCN) = 15.0 (ref. 32), Ered = −1.85 V versus Fc/Fc+ (ref. 16), cThe
system was reported in ref. 6. pKa(MeCN) = 6.0 (ref. 32), Ered = −1.85 V
versus Fc/Fc+ (ref. 16). dThe BDFE value reported in ref. 21 (26 kcal mol−1
was used.
)
complex bearing a PCP-type pincer ligand [MoCl3(PCP)] (9) were for the catalytic reduction of nitrogen gas to ammonia under ambient
found to have a similar reactivity to 1a (Table 1, entries 12 and 13). This reaction conditions. We also observed the formation of 112.9 equiv.
prompted us to carry out the catalytic reaction using larger amounts of ammonia (based on the molybdenum atom) in 1 min, where the
of SmI2 and ethylene glycol in the presence of 9 as a catalyst, because TOF is 112.9 min−1 (6,800 h−1) (Extended Data Table 5, entry 4). It
the formation of only a small amount of hydrogen gas was observed is notable that the TOF observed in water (112.9 min−1) is almost the
when 9 was used as a catalyst. The reaction of nitrogen gas at 1 atm same as that in ethylene glycol (117.0 min−1), although the amount of
with 14,400 equiv. (to 9) of SmI2 and 14,400 equiv. (to 9) of ethylene ammonia produced in water (4,350 equiv.) is substantially higher than
glycol in the presence of 9 in THF at room temperature for 4 h gave that produced in ethylene glycol (3,650 equiv).
3,650 equiv. of ammonia (based on the molybdenum atom; 76% yield
For comparison, typical results for this reaction together with those
based on SmI2) (Fig. 1d). The amount of ammonia produced was ten from previous reactions6,7 and values for nitrogenase enzymes28 are
times larger than that previously obtained7 in reaction system A. The summarized in Fig. 4a. The highest TOF (117.0 min−1) was achieved
high performance of 9 as a catalyst prompted us to measure the TOF using 9 as a catalyst in reaction system B, which is substantially higher
of 9 in reaction system B. We observed the formation of 117.0 equiv. of than that (0.5 min−1) using 1a as a catalyst in reaction system A. A
ammonia (based on the molybdenum atom) in 1 min, giving a TOF of previous study has also reported the osmium-catalysed formation of
117.0 min−1 (7,000 h−1) (Extended Data Table 5, entry 2). The result ammonia under mild reaction conditions, in which the TOF is 2 min−1
indicates that 9 has a higher catalytic performance than 1c, in terms (ref. 6). It is noteworthy that the TOF of the current reaction system
of both the amount of ammonia produced and its formation rate in (117.0 min−1) is close to that observed in nitrogenase enzymes (TOF
reaction system B (see above).
Notably, the reaction of nitrogen gas at 1 atm with 14,400 equiv. (to 9)
40–120 min−1)28.
Recently the energetic efficiency of ammonia formation was
of SmI2 and 14,400 equiv. (to 9) of water in the presence of 9 in THF at reported, in which the chemical overpotentials were calculated on
room temperature for 4 h gave 4,350 equiv. of ammonia (based on the the basis of effective bond dissociation free energy29. The chemical
molybdenum atom; 91% yield based on SmI2) together with 150 equiv. overpotential refers to excess thermodynamic driving force using the
of hydrogen gas (based on the molybdenum atom; 2% yield based on combination of a reducing reagent and a proton source in comparison
SmI2). This indicates that water acts as a more effective proton source to ammonia formation from hydrogen gas. Typical chemical overpoten-
than alcohols in this reaction system. To our knowledge, this is the tial values for ammonia formation are shown in Fig. 4b. In the reaction
first successful example of the direct use of water as a proton source systems reported in refs 2,4,5, the respective values of 44.4 kcal mol−1
,
2 5 A P r I L 2 0 1 9
| V O L 5 6 8 | N A t U r e | 5 3 9

reSeArCH Letter
26.4 kcal mol−1 and 291.6 kcal mol−1 were calculated on the basis of the
nature of the reducing reagents and proton sources29. We have also cal-
culated the chemical overpotentials for the current reaction system B.
The pKa of water reduces upon coordination and, in combination with
a reducing Sm(ii)/Sm(iii) couple, results in an effective O–H bond
strength of 26 kcal mol−1 (ref. 21) in [Sm(H2O)n]2+, compared to an
O–H bond dissociation free energy of 111 kcal mol−1 in free water.
When the effective value for [Sm(H2O)n]2+ is used, the calculated
chemical overpotential value is 135.6 kcal mol−1. Although this value
is very far from the ideal reaction, this presents an opportunity for
further research in the field of catalytic nitrogen fixation.
18. Chciuk, T. V., Anderson, W. R. Jr & Flowers, R. A. II. High-afnity proton donors
promote proton-coupled electron transfer by samarium diiodide. Angew. Chem.
Int. Ed. 55, 6033–6036 (2016).
19. Chciuk, T. V., Anderson, W. R. Jr & Flowers, R. A. II. Proton-coupled electron
transfer in the reduction of carbonyls by samarium diiodide–water complexes.
J. Am. Chem. Soc. 138, 8738–8741 (2016).
20. Chciuk, T. V., Anderson, W. R. Jr & Flowers, R. A. II. Interplay between substrate
and proton donor coordination in reductions of carbonyls by SmI2–water
through proton-coupled electron-transfer. J. Am. Chem. Soc. 140,
15342–15352 (2018).
21. Kolmar, S. S. & Mayer, J. M. SmI2(H2O)n reduction of electron rich enamines by
proton-coupled electron transfer. J. Am. Chem. Soc. 139, 10687–10692 (2017).
22. Bezdek, M. J., Guo, S. & Chirik, P. J. Coordination-induced weakening of
ammonia, water, and hydrazine X–H bonds in a molybdenum complex. Science
354, 730–733 (2016).
23. Bezdek, M. J. & Chirik, P. J. Interconversion of molybdenum imido and amido
complexes by proton-coupled electron transfer. Angew. Chem. Int. Ed. 57,
2224–2228 (2018).
24. Jonas, R. T. & Stack, T. D. P. C–H bond activation by a ferric methoxide complex:
a model for the rate-determining step in the mechanism of lipoxygenase. J. Am.
Chem. Soc. 119, 8566–8567 (1997).
25. Spiegel, D. A., Wiberg, K. B., Schacherer, L. N., Medeiros, M. R. & Wood, J. L.
Deoxygenation of alcohols employing water as the hydrogen atom source.
J. Am. Chem. Soc. 127, 12513–12515 (2005).
26. Pozzi, D., Scanlan, E. M. & Renaud, P. A mild radical procedure for the reduction
of B-alkylcatecholboranes to alkanes. J. Am. Chem. Soc. 127, 14204–14205
(2005).
27. Eizawa, A. et al. Remarkable catalytic activity of dinitrogen-bridged
dimolybdenum complexes bearing NHC-based PCP-pincer ligands toward
nitrogen fxation. Nat. Commun. 8, 14874 (2017).
28. Brown, K. A. et al. Light-driven dinitrogen reduction catalyzed by a
CdS:nitrogenase MoFe protein biohybrid. Science 352, 448–450 (2016).
29. Pappas, I. & Chirik, P. J. Catalytic proton coupled electron transfer from metal
hydrides to titanocene amides, hydrazides and imides: determination of
thermodynamic parameters relevant to nitrogen fxation. J. Am. Chem. Soc. 138,
13379–13389 (2016).
30. Warren, J. J., Tronic, T. A. & Mayer, J. M. Thermochemistry of proton-coupled
electron transfer reagents and its implications. Chem. Rev. 110, 6961–7001
(2010).
31. Pavlishchuk, V. V. & Addison, A. W. Conversion constants for redox potentials
measured versus diferent reference electrodes in acetonitrile solutions at
25°C. Inorg. Chim. Acta 298, 97–102 (2000).
Online content
Any methods, additional references, Nature Research reporting summaries, source
data, statements of data availability and associated accession codes are available at
https://doi.org/10.1038/s41586-019-1134-2.
Received: 14August 2018;Accepted: 28 February 2019;
Published online 24April 2019.
1. Liu, H. Ammonia Synthesis Catalysts: Innovation and Practice (World Scientifc,
Beijing, 2013).
2. Yandulov, D. V. & Schrock, R. R. Catalytic reduction of dinitrogen to ammonia at
a single molybdenum center. Science 301, 76–78 (2003).
3. Nishibayashi, Y. (ed.) Nitrogen Fixation (Topics in Organometallic Chemistry)
1st edn, Vol. 60 (Springer, Heidelberg, 2017).
4. Arashiba, K., Miyake, Y. & Nishibayashi, Y. A molybdenum complex bearing
PNP-type pincer ligands leads to the catalytic reduction of dinitrogen into
ammonia. Nat. Chem. 3, 120–125 (2011).
5. Anderson, J. S., Rittle, J. & Peters, J. C. Catalytic conversion of nitrogen to
ammonia by an iron model complex. Nature 501, 84–87 (2013).
6. Fajardo, J. Jr & Peters, J. C. Catalytic nitrogen-to-ammonia conversion by
osmium and ruthenium complexes. J. Am. Chem. Soc. 139, 16105–16108
(2017).
7. Arashiba, K. et al. Catalytic nitrogen fxation via direct cleavage of nitrogen–
nitrogen triple bond of molecular dinitrogen under ambient reaction conditions.
Bull. Chem. Soc. Jpn 90, 1111–1118 (2017).
8. Karunadasa, H. I., Chang, C. J. & Long, J. R. A molecular molybdenum-oxo
catalyst for generating hydrogen from water. Nature 464, 1329–1333
(2010).
32. Kaljurand, I. et al. Extension of the self-consistent spectrophotometric basicity
scale in acetonitrile to a full span of 28 pKa units: unifcation of diferent basicity
scales. J. Org. Chem. 70, 1019–1028 (2005).
9. Chatt, J., Pearman, A. J. & Richards, R. L. The reduction of mono-coordinated
molecular nitrogen to ammonia in a protic environment. Nature 253, 39–40
(1975).
10. Ohki, Y., Aoyagi, K. & Seino, H. Synthesis and protonation of N-heterocyclic-
carbene-supported dinitrogen complexes of molybdenum(0). Organometallics
34, 3414–3420 (2015).
Acknowledgements This project is supported by CREST, JST (JPMJCR1541).
We thank Grants-in-Aid for Scientific Research (numbers JP17H01201,
JP15H05798 and JP18K19093) from JSPS and MEXT. Y.A. is a recipient of the
JSPS Predoctoral Fellowships for Young Scientists. We also thank J. C. Peters
and S. Schneider for helpful discussions.
11. Li, H., Shang, J., Ai, Z. & Zhang, L. Efcient visible light nitrogen fxation with
BiOBr nanosheets of oxygen vacancies on the exposed {001} facets. J. Am.
Chem. Soc. 137, 6393–6399 (2015).
Reviewer information Nature thanks Robert Flowers and the other anonymous
reviewer(s) for their contribution to the peer review of this work.
12. Hirakawa, H., Hashimoto, M., Shiraishi, Y. & Hirai, T. Photocatalytic conversion of
nitrogen to ammonia with water on surface oxygen vacancies of titanium
dioxide. J. Am. Chem. Soc. 139, 10929–10936 (2017).
Author contributions Y.N. directed and conceived this project. Y.A., K.A. and
K.N. conducted the experimental work including X-ray analysis. All authors
discussed the results and wrote the manuscript.
13. Szostak, M., Fazakerley, N. J., Parmar, D. & Procter, D. J. Cross-coupling
reactions using samarium(ii) iodide. Chem. Rev. 114, 5959–6039 (2014).
14. Choquette, K. A. & Flowers, R. A. in Comprehensive Organic Synthesis II Vol. 1
(eds Knochel, P. & Molander, G. A.) Ch. 9, 278–343 (Elsevier, Amsterdam,
2014).
Competing interests Y.A., K.N. and Y.N. have filed a patent based on the work
described here (Japanese patent application number 2018-036967).
Additional information
15. Teprovich, J. A. Jr, Balili, M. N., Pintauer, T. & Flowers, R. A. II. Mechanistic studies
of proton-donor coordination to samarium diiodide. Angew. Chem. Int. Ed. 46,
8160–8163 (2007).
16. Robbins, J. L., Edelstein, N., Spencer, B. & Smart, J. C. Syntheses and electronic
structures of decamethylmetallocenes. J. Am. Chem. Soc. 104, 1882–1893
(1982).
17. Kuhlman, M. L. & Flowers R. A. II. Aggregation state and reducing power of the
samarium diiodide–DMPU complex in acetonitrile. Tetrahedron Lett. 41,
8049–8052 (2000).
Extended data is available for this paper at https://doi.org/10.1038/s41586-
019-1134-2.
Reprints and permissions information is available at http://www.nature.com/
reprints.
Correspondence and requests for materials should be addressed to Y.N.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2019
5 4 0
| N A t U r e | V O L 5 6 8 | 2 5 A P r I L 2 0 1 9

Letter reSeArCH
METhods
(0.12 mmol) was added to the stirred solution in the Schlenk flask in one portion,
General methods. All manipulations were carried out under an atmosphere and the mixture was stirred at room temperature for 2 h. After the reaction, the
of nitrogen or argon by using standard Schlenk techniques or glove-box tech- amount of generated dihydrogen (0.0284 mmol, 47% yield based on SmI2(THF)2)
niques unless otherwise stated. All solvents were dried by general methods and was quantified by GC. The amount of ammonia (0.0416 mmol, >99% yield based
degassed before use. [MoI3(PNP)] (1a)7, [MoBr3(PNP)] (1b)7, [MoCl3(PNP)] on SmI2(THF)2) was determined by the indophenol method36.
(1c)4, [Mo(N2)2(PNP)]2(μ-N2) (2)4, [MoCl3(PPP)] (3)33, [MoI(≡N)(PNP)] (4)7, Reaction of 4 with 1 equivalent of SmI2(THF)2 or 1 equivalent of ethylene glycol
[Mo(N2)2(PCP)]2(μ-N2) (8)27, [MoCl3(PCP)] (9)27, [SmI2(THF)2]34, [ColH]OTf under argon. In an argon-filled glove box, to a mixture of 4 (1.6 mg, 0.0025 mmol)
(ref. 33) and CoCp*2 (ref. 35) were prepared according to the literature methods. and triphenylphosphine as an internal standard in an NMR sample tube was added
Other reagents were purchased commercially and used as received.
THF (0.5 ml). Then THF solution (0.1 ml) containing ethylene glycol (0.0025
1H NMR (400 MHz), 31P{1H} NMR (162 MHz), and 15N{1H} NMR (41 MHz) mmol) or SmI2(THF)2 (0.0025 mmol) was added to the solution. No consumption
spectra were recorded on a JEOL ECS400 spectrometer in suitable solvents, and of 4 was detected by 31P{1H} NMR in THF.
spectra were referenced to residual solvent (1H) or external standard (31P{1H}: Preparation of [Mo(O)I(PNP)(H2O)]I (5). To a solution of [MoI3(PNP)] (1a)
85% H3PO4, 15N{1H}: CH3NO2). Ultraviolet–visible (UV–vis) absorption spectra (174 mg, 0.199 mmol) in THF (16 ml) were added pyridine (38.0 μl, 0.473 mmol)
were recorded on a Shimadzu MultiSpec-1500 and a Shimadzu UV-1850. Mass and water (7.0 μl, 0.39 mmol) under a nitrogen atmosphere, and the mixture was
spectra were recorded on a JEOL Accu TOF JMS-T100LP. Evolved dihydrogen stirred at 50°C for 14 h (Extended Data Fig. 1b). The resultant yellow-green sus-
was quantified by gas chromatography (GC) using a Shimadzu GC-8A equipped pension was concentrated under reduced pressure, then the residue was washed
with a thermal conductivity detector and a ShinCarbon ST column (6 m × 3 mm). with benzene (5 ml × 3). Volatiles were removed in vacuo, and THF (5 ml) was
Elemental analyses were performed at Microanalytical Center of The University added to the residue. The THF solution was filtered through Celite, the filter cake
of Tokyo.
was washed with THF (5 ml × 1), and the combined filtrate was concentrated
Catalytic reduction of dinitrogen to ammonia with alcohols. A typical experi- under reduced pressure. To the residue was added CH2Cl2 (4 ml), then slow addi-
mental procedure for the catalytic reactions is described below. In a nitrogen-filled tion of hexane (20 ml) afforded 5 as dark yellow crystals, which were collected by
glove box, to a mixture of 1a (1.7 mg, 0.0020 mmol) and SmI2(THF)2 (198 mg, filtration, washed with hexane (5 ml × 3) and Et2O (5 ml × 3), and dried in vacuo
0.361 mmol) placed in a 50 ml Schlenk flask was added THF (6 ml). Next, ethylene (46.1 mg, 0.0591 mmol, 30% yield). 1H NMR (THF-d8): δ 7.96 (t, J = 7.9 Hz, ArH,
glycol (20 μl, 0.36 mmol) was added in one portion, and the mixture was stirred at 1H), 7.78 (d, J = 7.9 Hz, ArH, 2H), 6.10 (br, OH2, 2H), 4.81–4.72 (m, CH2P, 2H),
room temperature for 30 min. After the reaction, the amount of generated dihydro- 3.94–3.88 (m, CH2P, 2H), 1.47 (pseudo t, J = 6.6 Hz, PtBu2, 18H), 1.40 (pseudo t,
gen (0.0291 mmol, 14.9 equiv. based on the molybdenum atom, 16% yield based J = 6.6 Hz, PtBu2, 18H). 31P{1H} NMR (THF-d8): δ 56.9 (s). ESI-TOF-MS (THF):
on SmI2(THF)2) was quantified by GC. Aqueous potassium hydroxide solution 636 (m/z). Anal calcd for C23H45I2MoNO2P2: C, 35.45; H, 5.82; N, 1.80. Found: C,
(30 wt%, 5 ml) was added to the reaction mixture. The mixture was evaporated 35.24; H, 5.56; N, 1.84. Pyridine acted as a base to deprotonate water to produce
under reduced pressure, and the distillate was trapped in a dilute H2SO4 solu- the corresponding molybdenum oxo-aquo complex 5.
tion (0.5 M, 10 ml). The amount of ammonia (0.0839 mmol, 43.1 equiv. based on Reaction of 5 with 5 equiv. of SmI2(THF)2 under N2. A mixture of 5 (3.9 mg,
the molybdenum atom, 70% yield based on SmI2(THF)2) present in the H2SO4 0.0050 mmol) and SmI2(THF)2 (13.7 mg, 0.0250 mmol) in THF (1 ml) placed
solution was determined by the indophenol method36. No hydrazine was detected in a 20 ml Schlenk flask was stirred at room temperature for 1 h under N2
by the p-(dimethylamino)benzaldehyde method37. The consumption of SmI2 was (1 atm) (Extended Data Fig. 1b). After volatiles were removed in vacuo, the yield
confirmed by the colour change of the reaction mixture from deep blue to yellow. of [Mo(≡N)I(PNP)] (4)7 (0.0018 mmol, 36%) in the residual solid was deter-
Catalytic reactions in various solvents were carried out using the typical exper- mined by 1H NMR in THF-d8 with hexamethylbenzene as an internal stand-
imental procedure (Extended Data Table 2).
ard. Separately, we observed the formation of 15N-labelled molybdenum nitride
After the catalytic reaction, we tried to identify the resting state of molybdenum complex [MoI(≡15N)(PNP)] (4ʹ) by ESI-TOF-MS analysis from the reaction of
complexes. Only the formation of molybdenum nitride complex [Mo(≡N)I(PNP)] [MoI(=O)(OH2)(PNP)]I (5) with 10 equiv. of SmI2 under 1 atm of 15N2 gas.
(4) was observed from the reaction mixture by electrospray ionization time-of- Preparation of ethylene glycol-bridged polysamarium(ii) complex (6). In an
flight mass spectrometry (ESI-TOF-MS) (m/z = 634). No identified molybdenum argon-filled glove box, to a solution of SmI2(THF)2 (197 mg, 0.360 mmol) in
complexes except for 4 were observed by NMR and electron paramagnetic THF (3.5 ml) placed in a 20 ml Schlenk flask was added ethylene glycol (60.0 μl,
resonance spectroscopies.
1.08 mmol) in one portion. The mixture was stirred at room temperature for
Catalytic reduction of dinitrogen to ammonia under 15N2. In a 50 ml 5 min. After the solution was cooled and kept at −30°C for 4 days, green crystals
Schlenk flask were placed 1c (1.2 mg, 0.0020 mmol) and SmI2(THF)2 (197 mg, were formed. The crystals were washed with Et2O (2 ml × 3) and dried under
0.360 mmol). The flask was evacuated and backfilled with 15N2. Then ethylene vacuum to afford 6·C4H8O (165 mg, 0.225 mmol, 63% yield). Anal calcd for
glycol (20 μl, 0.36 mmol) in THF (6 ml) was added to the stirred solution in the C10H26I2O7Sm·C4H8O: C, 22.89; H, 4.67. Found: C, 23.24; H, 4.47. A single crystal
Schlenk flask in one portion, and the mixture was stirred at room temperature for suitable for X-ray analysis was obtained by heat recrystallization from a THF solu-
18 h. KOtBu (4 mmol) in MeOH (5 ml) was added to the resultant yellow solution, tion of a reaction mixture of SmI2(THF)2 and ethylene glycol at −30°C for 4 days.
and the mixture was stirred at room temperature for 20 min. The volatile compo- Preparation of oxygen-bridged disamarium(iii) complex (7a). In a nitrogen-
nents in the mixture were collected by trap-to-trap distillation to the Schlenk flask filled glove box, to a mixture of 1a (1.7 mg, 0.0020 mmol) and SmI2(THF)2 (197
to which was added a solution of 2 M HCl in Et2O (5 ml, 10 mmol). The obtained mg, 0.360 mmol) placed in a 50 ml Schlenk flask was added THF (6 ml). Then
colourless solution was dried in vacuo to afford a colourless solid which contained ethylene glycol (100 μl, 1.80 mmol) was added to the stirred solution in the Schlenk
15NH4Cl. The amount of 15NH4Cl (0.0797 mmol, 39.7 equiv. based on the molyb- flask in one portion, and the mixture was stirred at room temperature for 18 h.
denum atom, 67% NMR yield based on SmI2(THF)2) was determined by 1H NMR After volatiles were removed in vacuo, the solid was extracted with EtOH (5 ml).
in DMSO-d6 with 1,1,2,2-tetrachloroethane as an internal standard and [ColH] Slow addition of hexane (30 ml) to the EtOH solution afforded 7a as colourless
[OTf] as an acid. The 1H NMR spectrum of this solid is shown in Extended Data crystals, which were collected by filtration, washed with hexane (5 ml × 3) and
Fig. 2. 1H NMR (DMSO-d6): δ 7.32 (d, JNH = 71.6 Hz, 15NH4Cl). 15N{1H} NMR dried in vacuo (169 mg, 0.133 mmol, 74% yield). Anal calcd for C16H46I4O14Sm2:
(DMSO-d6): δ −355.7 (s, 15NH4Cl).
C, 15.12; H, 3.65. Found: C, 14.84; H, 3.72. A single crystal suitable for X-ray
Reaction of 1 with 5 equivalents of SmI2(THF)2 under N2. A mixture of 1a analysis was obtained by two-layer recrystallization by the slow addition of hexane
(8.7 mg, 0.010 mmol) and SmI2(THF)2 (27.6 mg, 0.050 mmol) in THF (2 ml) (30 ml) to a EtOH (5 ml) solution of 7a.
placed in a 20 ml Schlenk flask was stirred at room temperature for 5 min under Preparation of oxygen-bridged trisamarium(iii) complex (7b). In a nitrogen-
N2 (1 atm). After volatiles were removed in vacuo, formation of 4 was confirmed filled glove box, to a mixture of 1a (1.7 mg, 0.0020 mmol) and SmI2(THF)2
by 1H NMR and ESI-TOF-MS7. The NMR yield of 4 (0.0076 mmol, 76%) in the (197 mg, 0.360 mmol) placed in a 50 ml Schlenk flask was added THF (6 ml).
residual solid was determined with hexamethylbenzene as an internal standard. Then ethylene glycol (20 μl, 0.36 mmol) was added to the stirred solution in the
Separately, we observed the formation of the 15N-labelled molybdenum nitride Schlenk flask in one portion, and the mixture was stirred at room temperature for
complex [MoI(≡15N)(PNP)] (4ʹ) by ESI-TOF-MS analysis from the reaction of 18 h. After the reaction, slow addition of hexane (40 ml) to the reaction mixture
1a with 5 equiv. of SmI2 under 1 atm of 15N2 gas.
afforded 7a as colourless crystals for X-ray analysis. These crystals contain a small
When 1b was used as a starting material, the yield of 4 in the residual solid was amount of 7b·2C4H8O.
0.0074 mmol (74% NMR yield). When 1c was used as a starting material, the yield Kinetic study of catalytic reactions and stoichiometric. Kinetic study of the
of 4 in the residual solid was 0.0086 mmol (85% NMR yield).
catalytic reaction was carried out by tracing the consumption of SmI2 by UV–vis
Reaction of 4 with 3 equivalents of SmI2(THF)2 and 3 equivalents of ethylene absorption spectroscopy at room temperature. A typical procedure is described
glycol under argon. In an argon-filled glove box, to a mixture of 4 (25.4 mg, below. A THF solution containing SmI2 (6.0 μmol) and ethylene glycol (6.0 μmol)
0.0402 mmol) and SmI2(THF)2 (65.8 mg, 0.120 mmol) placed in a 50 ml Schlenk was added into a quartz glass cell (1 cm × 1 cm × 4 cm) with a septum in a nitrogen-
flask was added THF (3 ml). Then THF solution (3 ml) containing ethylene glycol filled glove box. The THF solution of 1a (0.3 μmol) was added to the quartz glass

reSeArCH Letter
cell using a syringe with stirring. The total amount of solution was adjusted to be Catalytic reduction of hydrazine to ammonia under argon. In an argon-filled
3.0 ml. The absorbance of SmI2 at 560 nm was measured for 5 min. The kH/kD value glove box, to a mixture of 1a (1.7 mg, 0.0020 mmol) and SmI2(THF)2 (197 mg,
of 2.0 was calculated from the consumption of SmI2 during the period from 10 to 0.360 mmol) placed in a 50 ml Schlenk flask was added THF (4 ml). Then THF
20 s after the start of the reaction by using HOCH2CH2OH and DOCH2CH2OD as solution (2 ml) containing hydrazine (0.06 mmol) and ethylene glycol (0.06 mmol)
proton sources under typical analytical conditions described above.
was added to the mixture, and the reaction mixture was stirred at room temperature
To obtain more information on the proposed reaction pathway, we carried out for 18 h. After the reaction, the amount of generated dihydrogen (0.0097 mmol,
kinetic studies at different initial concentrations of 1a for the stoichiometric reaction 5.0 equiv. based on the molybdenum atom, 16% yield based on SmI2(THF)2) was
of 1a with 2.5 equiv. of SmI2 under 1 atm of N2 gas to 4. We observed a second-order quantified with GC. The amount of ammonia (0.0995 mmol, 50.0 equiv. based
rate dependence on 1a in the stoichiometric reaction (Extended Data Fig. 3). This on the molybdenum atom, 83% yield based on hydrazine) was determined by the
indicates that the formation of the dinitrogen-bridged dimolybdenum complex may indophenol method36. The amount of hydrazine (0.00585 mmol, 10% recovery)
be involved as the rate-determining step of the stoichiometric reaction.
was determined by the p-(dimethylamino)benzaldehyde method37. When the cat-
Catalytic reduction of dinitrogen to ammonia with water. A typical experimental alytic reaction was performed in the absence of 1a, the amount of ammonia was
procedure for catalytic reduction with water is described below. To 1a (1.7 mg, 0.0434 mmol (36% yield based on hydrazine). The results are shown in Extended
0.0020 mmol) and SmI2(THF)2 (199 mg, 0.363 mmol) placed in a 50 ml Schlenk Data Table 6.
flask was added THF (4 ml) under N2. Then THF solution (2 ml) containing X-ray crystallography. Diffraction data for 5, 6, 7a and 7b·2C4H8O were col-
H2O (0.36 mmol) was slowly added to the stirred solution in the Schlenk flask lected for the 2θ range of 4° to 55° on a Rigaku RAXIS RAPID imaging plate area
with a syringe pump over a period of 0.5 h. After the addition of the water, the detector with graphite-monochromated Mo Kα radiation (λ = 0.71075 Å), with
mixture was further stirred at room temperature for 17.5 h. After the reaction, the VariMax optics. Intensity data were corrected for Lorenz-polarization effects and
amount of generated dihydrogen (0.0702 mmol, 36.0 equiv. based on the molybde- for empirical absorption (ABSCOR). The structure solution and refinements were
num atom, 39% yield based on SmI2(THF)2) of the catalytic reaction was quantified carried out by using the CrystalStructure crystallographic software package38. The
by GC. Aqueous potassium hydroxide solution (30 wt%, 5 ml) was added to the positions of non-hydrogen atoms were determined by direct methods (SIR 9739 for
reaction mixture. The mixture was evaporated under reduced pressure, and the 5, 7a, 7b·2C4H8O and SHELXS40 for 6) and subsequent Fourier syntheses (SHELXL
distillate was trapped in a dilute H2SO4 solution (0.5 M, 10 ml). The amount of version 2016/641) and were refined on Fo2 using all unique reflections by full-matrix
ammonia (0.0522 mmol, 26.8 equiv. based on the molybdenum atom, 43% yield least-squares with anisotropic thermal parameters. All the hydrogen atoms were
based on SmI2(THF)2) present in the H2SO4 solution was determined by the indo- placed at the calculated positions with fixed isotropic parameters, except for the
phenol method36. The results of the catalytic reactions using various molybdenum H(1) and H(5) atoms in 5, the H(9), H(10) and H(11) atoms in 7a and the H(39)
complexes as catalysts are described in Extended Data Table 3.
and H(40) atoms in 7b·2C4H8O. The positions of the H(1) and H(5) atoms in 5, the
Catalytic reactions using larger amounts of reducing reagent and proton source. H(9), H(10) and H(11) atoms in 7a and the H(39) and H(40) atoms in 7b·2C4H8O
When ethylene glycol was used as a proton source, catalytic reactions were per- were determined on appropriate peaks in the difference Fourier maps and further
formed using smaller amount of catalysts for 4 h using the typical experimental refined isotropically. The crystal of 6 is a merohedral twin with a minor component
procedure. When water was used as a proton source, the experimental procedure comprising 3.15% of the crystal.
for the catalytic reactions is described below. To 9 (0.016 μg, 0.025 μmol) and
SmI2(THF)2 (197 mg, 0.359 mmol) placed in a 50 ml Schlenk flask was added
Data availability
THF (1 ml) under N2. Then THF solution (2 ml) containing H2O (0.36 mmol) was
slowly added to the stirred solution in the Schlenk flask with a syringe pump over
a period of 1 h. After the addition of the H2O, the mixture was further stirred at
room temperature for 3 h. After the reaction, the amount of generated dihydrogen
(0.0017 mmol, 69 equiv. based on the molybdenum atom, 1% yield based on
SmI2(THF)2) of the catalytic reaction was quantified by GC. Aqueous potassium
hydroxide solution (30 wt%, 5 ml) was added to the reaction mixture. The mixture
was evaporated under reduced pressure, and the distillate was trapped in a dilute
H2SO4 solution (0.5 M, 10 ml). The amount of ammonia (0.111 mmol, 4,440 equiv.
based on the molybdenum atom, 88% yield based on SmI2(THF)2) present in
the H2SO4 solution was determined by the indophenol method36. The results are
shown in Extended Data Table 4.
Crystallographic data for the reported structures have been deposited at the
Cambridge Crystallographic Data Centre, under deposition numbers: CCDC
185998 (5), 1857999 (6), 1858000 (7a) and 1858001 (7b·2C4H8O). All other data
supporting the findings of this study are available within the paper or from the
corresponding author upon reasonable request.
33. Arashiba, K. et al. Catalytic reduction of dinitrogen to ammonia by use of
molybdenum–nitride complexes bearing a tridentate triphosphine as
catalysts. J. Am. Chem. Soc. 137, 5666–5669 (2015).
34. Watson, P. L., Tulip, T. H. & Williams, I. Defuorination of perfuoroolefns by
divalent lanthanoid reagents: activating C–F bonds. Organometallics 9,
1999–2009 (1990).
35. Yandulov, D. V. & Schrock, R. R. Synthesis of tungsten complexes that contain
hexaisopropylterphenyl-substituted triamidoamine ligands, and reactions
relevant to the reduction of dinitrogen to ammonia. Can. J. Chem. 83, 341–357
(2005).
36. Weatherburn, M. W. Phenol-hypochlorite reaction for determination of
ammonia. Anal. Chem. 39, 971–974 (1967).
37. Watt, G. W. & Chrisp, J. D. Spectrophotometric method for determination of
hydrazine. Anal. Chem. 24, 2006–2008 (1952).
38. CrystalStructure v.4.3.: Single Crystal Structure Analysis Software (Rigaku
Corp., Tokyo and MSC, Texas, 2018).
39. Altomare, A. et al. SIR97: a new tool for crystal structure determination and
refnement. J. Appl. Crystallogr. 32, 115–119 (1999).
40. Sheldrick, G. M. A short history of SHELX. Acta Crystallogr. A 64, 112–122
(2008).
41. Sheldrick, G. M. Crystal structure refnement with SHELXL. Acta Crystallogr. C
71, 3–8 (2015).
Catalytic reactions to determine turnover frequencies. When ethylene glycol was
used as a proton source, catalytic reactions were performed for 1 min using the typical
experimental procedure. In the case that water was used as a proton source, the
experimental procedure for the catalytic reactions is descried below. To 9 (0.6 mg,
0.0009 mmol) and SmI2(THF)2 (198 mg, 0.361 mmol) placed in a 50 ml Schlenk
flask was added THF (1 ml) under N2. Then THF solution (2 ml) containing H2O
(0.36 mmol) was added in one portion, and the mixture was stirred at room tempera-
ture for 1 min. After the reaction, aqueous potassium hydroxide solution (30 wt%, 5 ml)
was added to the reaction mixture. The mixture was evaporated under reduced
pressure, and the distillate was trapped in a dilute H2SO4 solution (0.5 M, 10 ml).
The amount of ammonia (0.106 mmol, 112.9 equiv. based on the molybdenum atom,
88% yield based on SmI2(THF)2) present in the H2SO4 solution was determined by
the indophenol method36. The results are shown in Extended Data Table 5.

Letter reSeArCH
Extended Data Fig. 1 | Reactions via direct nitrogen cleavage pathway.
a, A reaction pathway via direct cleavage of the nitrogen–nitrogen triple
bond. b, Synthesis of the molybdenum oxo complex (5) from 1a and
water, and reduction of 5 to 4 via direct nitrogen cleavage of the nitrogen–
nitrogen triple bond. SmI2(THF)2 was used as the source of SmI2.aYield
based on NMR.

reSeArCH Letter
Extended Data Fig. 2 | 1H NMR spectra of catalytic reduction of
dinitrogen to ammonia under 15N2. a–c, 1H NMR (DMSO-d6) spectra of
product from catalytic reaction with 1c under 15N2 and [ColH]OTf (a),
authentic sample of the mixture of 14NH4Cl and [ColH]OTf (b) and
authentic sample of the mixture of 15NH4Cl and [ColH]OTf (c).

Letter reSeArCH
reaction rate (kobs, in abs s−1) was determined from the formation rate of
the absorbance of 4 at 828 nm. a, Typical time profile of the formation of 4
observed at 828 nm with SmI2 (1.0 mM) and 1a (0.4 mM) in THF at room
temperature. b, Rate of formation of 4 at various concentrations of 1a.
c, Rate dependence of the formation of 4 on concentration of 1a in THF.
d, UV–vis absorption spectra between 380 and 1,100 nm of 1a (0.48 mM
in THF; blue line), 4 (0.60 mM in THF; red line), and SmI2 (1.2 mM in
THF; green line).
Extended Data Fig. 3 | Kinetic study of stoichiometric reactions. A
kinetic study of the stoichiometric reaction was carried out by monitoring
the formation of 4 by UV–vis spectroscopy at room temperature. A typical
procedure is described below. A THF solution containing 1a (1.2 μmol)
was added into a quartz glass cell (1 cm × 1 cm × 4 cm) with a septum in a
nitrogen-filled glove box. The THF solution of SmI2 (3.0 μmol) was added
to the quartz glass cell using a syringe with stirring. The total amount of
solution was adjusted to be 3.0 ml. The spectra were measured every 0.4 s,
and the rate was determined from the time profile of the initial 100 s. The

reSeArCH Letter
Extended data Table 1 | Catalytic reactions with various
molybdenum complexes as catalysts
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.
bData are mean of multiple individual experiments (at least 2) with error bars representing the s.d.

Letter reSeArCH
Extended data Table 2 | Catalytic reactions in various solvents using
typical experimental procedure
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.
bData are mean of multiple individual experiments (at least 2) with error bars representing the s.d.
cThe reaction was carried out under 1 atm of argon.
dThe reaction was carried out in the absence of 1a.

reSeArCH Letter
Extended data Table 3 | Catalytic reactions using water as a proton source
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.

Letter reSeArCH
Extended data Table 4 | Catalytic reactions using larger amounts of
reducing reagent and proton source using 1a or 9 as catalysts
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.
b Data are mean of multiple individual experiments (at least 2) with error bars representing the s.d.
cTHF (2 ml) was used as a solvent.
dH2O was used as a proton source.
n.d., not determined.

reSeArCH Letter
Extended data Table 5 | determination of the turnover frequencies
of catalytic reactions
SmI2(THF)2 was used as the source of SmI2.
aYield based on SmI2.

Letter reSeArCH
Extended data Table 6 | Catalytic reduction by using hydrazine as a
substrate instead of dinitrogen gas