One-pot, room-temperature conversion of dinitrogen to ammonium chloride at a main-group element

Article abstract of DOI:10.1038/s41557-020-0520-6

The industrial reduction of dinitrogen (N₂) to ammonia is an energy-intensive process that consumes a considerable proportion of the global energy supply. As a consequence, species that can bind N₂ and cleave its strong N–N bond under mild conditions have been sought for decades. Until recently, the only species known to support N₂ fixation and functionalization were based on a handful of metals of the s and d blocks of the periodic table. Here we present one-pot binding, cleavage and reduction of N₂ to ammonium by a main-group species. The reaction—a complex multiple reduction–protonation sequence—proceeds at room temperature in a single synthetic step through the use of solid-phase reductant and acid reagents. A simple acid quench of the mixture then provides ammonium, the protonated form of ammonia present in fertilizer. The elementary reaction steps in the process are elucidated, including the crucial N–N bond cleavage process, and all of the intermediates of the reaction are isolated. [Figure not available: see fulltext.].

Full text of DOI:10.1038/s41557-020-0520-6

Articles
https://doi.org/10.1038/s41557-020-0520-6
One-pot, room-temperature conversion of
dinitrogen to ammonium chloride at a
main-group element
Marc-André Légaréꢀ ꢀ1,2,3, Guillaume Bélanger-Chabot1,2,4, Maximilian Rangꢀ ꢀ1,2, Rian D. Dewhurstꢀ ꢀ1,2,
Ivo Krummenacher1,2, Rüdiger Bertermann1,2 and Holger Braunschweigꢀ ꢀ1,2
ꢀ✉
The industrial reduction of dinitrogen (N2) to ammonia is an energy-intensive process that consumes a considerable proportion
of the global energy supply. As a consequence, species that can bind N2 and cleave its strong N–N bond under mild conditions
have been sought for decades. Until recently, the only species known to support N2 fixation and functionalization were based
on a handful of metals of the s and d blocks of the periodic table. Here we present one-pot binding, cleavage and reduction of
N2 to ammonium by a main-group species. The reaction—a complex multiple reduction–protonation sequence—proceeds at
room temperature in a single synthetic step through the use of solid-phase reductant and acid reagents. A simple acid quench
of the mixture then provides ammonium, the protonated form of ammonia present in fertilizer. The elementary reaction
steps in the process are elucidated, including the crucial N–N bond cleavage process, and all of the intermediates of the reaction
are isolated.
ue to its presence in amino acids, the element nitrogen is of binding, partially reducing and even coupling N2 (refs. 18–20).
essential for life on Earth. However, its extremely stable Following our work, a number of porous carbon-based nanoma-
D
elemental form (dinitrogen, N2) makes it difficult to uti- terials were reported to effect the electrocatalytic reduction of N2
lize nitrogen either biologically or synthetically. The century-old to various nitrogen-containing products, including ammonia21,22
.
Haber–Bosch process1, which converts N2 and H2 into ammonia Our transition-metal-free boron fixation processes and subsequent
over an iron catalyst at high temperatures (400–600°C) and pres- chemistry led to a number of species containing central N2 (and N4)
sures (20–40MPa), is still the major source of ammonium fertilizer units bound to boron-containing fragments; however, the complete
and other synthetic nitrogen compounds worldwide. Consequently, cleavage of the strong N–N bond of N2, a critical step in the poten-
over 1% of the world’s generated energy is dedicated to the synthe- tial synthesis of ammonia (and amines) from N2, remains elusive in
sis of ammonia. The importance of this energy-intensive process, a main-group complex.
and the 1965 discovery of transition metal complexes containing
Although a number of different N2 binding modes with tran-
bound N2 units2, has motivated over half a century of research into sition metals exist, recent advances in the catalytic synthesis of
transition-metal-catalysed fixation of N2 and its conversion into ammonia from N2 using molecular transition metal species have
ammonia and other useful nitrogen compounds3–10. Although this focused on complexes featuring N2 as a terminally bound ligand23.
research has yet to uncover a practical method for N2 reduction In such complexes, a reduction–protonation cascade involving six
under mild conditions, progress towards nitrogen fixation and con- electrons and six protons can lead to the total reduction of N2 to
version by molecular complexes remains the exclusive preserve of two equivalents of ammonia, according to a mechanism known as
transition metals because other elements of the periodic table gen- the Chatt cycle, which is relevant to biological nitrogenase chem-
erally lack the advantageous oxidation-state and orbital diversity of istry24. Some side-on bridging N2 transition metal complexes have
transition metals.
also been hydrogenated in a stepwise manner, albeit through a com-
Outside the d- and f-block metals, until very recently only the pletely different mechanism25. In contrast, the reactivity of end-on
alkali metal lithium was known to react with N2 at room tem- bridging N2 complexes is dominated by N2 release and photolytic or
perature to give an isolable product11 and, in both biological and thermal cleavage of the bridging N2 ligand, producing two transition
synthetic settings, transition metals were the only elements with a metal nitride complexes, the latter being of relevance to ammonia
demonstrated ability to enable the fixation and functionalization of synthesis23,26–28. Indeed, the rate-limiting step of the Haber–Bosch
N2. In 2018 we reported the reaction of a p-block molecule with N2 process is the cleavage of the N–N triple bond on the catalyst to
under transition-metal-free conditions by taking advantage of the form surface nitrides29. However, in transition metal complexes,
emerging concept of main-group metallomimetics: p-block com- the direct cleavage of a bridging N2 ligand—a six-electron reduc-
pounds able to react like transition metal complexes12–14. In this way, tion process—typically requires strongly reducing metal centres or
transient dicoordinate borylenes15–17 of the form [L→B–R] (where unusual multimetallic mechanisms30. We hypothesized that, under
L is a neutral Lewis donor unit, for example, a carbene, and R is the right circumstances, binding N2 in an end-on bridging fashion
an anionic substituent, for example, aryl) were shown to be capable could allow the ammonia synthesis process to be separated into
1Institute for Inorganic Chemistry, Julius-Maximilians-Universität Würzburg, Würzburg, Germany. 2Institute for Sustainable Chemistry & Catalysis with
Boron, Julius-Maximilians-Universität Würzburg, Würzburg, Germany. 3Present address: Department of Chemistry, McGill University, Montréal, Québec,
4
✉
Canada. Present address: Département de Chimie, Université Laval, Québec, Québec, Canada. e-mail: h.braunschweig@uni-wuerzburg.de
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
Dur
B
Dur
B
K
N
Dur
B
H
N
Dur
B
N2, KC8
KC8
H2O
N
CAAC
CAAC
CAAC
CAAC
CAAC
2
N
B
N
K
B
N
H
B
CAAC
CAAC
Br
Dur
Dur
Dur
N2, KC8
B(OH)3
1
2
3
4
(previously reported)
KC8, B(OH)3
(previously reported)
(previously reported)
(1)
N2, KC8, B(OH)3
One
pot
(2) Aqueous HCl
KC8, trace water
2 NH4Cl
KC8
Aqueous HCl
Dur
B
Dur
B
Dur
Dur
B
B(OH)3
KC8
B(OH)3
H
H
B
K
K
2
2
CAAC
CAAC
CAAC
CAAC
N
H
N
H
N
H
N
H
2
2
H
H
8
7
6
5
N2H4, UV irradiation, –CO
Dur
N–N activation
CAAC
N
Dur =
CAAC =
C
B
CO
Fig. 1 | A one-pot, borylene-mediated synthesis of ammonium chloride from N2, and elucidation of the individual reduction–protonation steps involved.
The solid-state structures of 5–8 are presented with 50% probability displacement ellipsoids. Peripheral hydrogen atoms and carbon ellipsoids have been
removed for clarity. All depicted hydrogen atoms, except the carbon-bound hydrogen atom in 8, were crystallographically located. Atom codes: grey,
carbon; white, hydrogen; green, boron; purple, potassium; blue, nitrogen. N2 was applied at pressures of ~5ꢀatm. KC8 and B(OH)3 were applied as 10–15
molar equivalents relative to boron. All reactions involving N2 were kept below –50ꢀ°C until application of gaseous N2, after which they were warmed to
room temperature. All other reactions were conducted at room temperature. Selected bond distances (Å) for 5: B–N 1.399(2) (average), B–CCAAC 1.561(3)
(average); for 6: B–N 1.415(1), B–CCAAC 1.535(2); for 7: B–N 1.505(2) (average), B–CCAAC 1.465(2) (average); for 8: B–N 1.376(2), B–CCAAC 1.616(3). Purple
dots denote unpaired electrons.
three sequential two-electron/two-proton steps, which may allow adducts, in which reduction predominantly occurs at the transition
it to proceed with lower energy barriers than a sequence involv- metal centre or auxiliary ligands33. Our recent discovery of direct
ing a six-electron reduction followed by addition of six protons as binding of dinitrogen to boron in the complex [{(CAAC)(Dur)
described above23,31. A similar approach was recently employed by B}2(N2)] (2, where CAAC is 1-(2,6-diisopropylphenyl)-3,3,5,5-tetr
Liddle and co-workers32 in the reduction of N2 to ammonia on tita- amethylpyrrolidin-2-ylidene34 and Dur is 2,3,5,6-tetramethylphenyl;
nium through an initial four-electron reduction step. However, to Fig. 1)18–20 provided a platform to test the possibility of repeated
our knowledge a cascade mechanism of two-electron/two-proton reduction–protonation steps from an end-on bridging N2 com-
steps for complete N2 reduction from an end-on bridging complex plex and ultimately the cleavage of the N–N triple bond. Indeed,
has never been fully characterized.
the first two-electron/two-proton reduction–protonation cycle was
In this context, we envisaged that using p-block elements demonstrated to be facile in our previous report of the syntheses of
for nitrogen fixation could allow the discovery of new N2 reduc- the dipotassium compound K2[{(CAAC)(Dur)B}2(N2)] (3) and its
tion chemistry distinct from that of transition metals. In contrast doubly protonated diradical congener [{(CAAC)(Dur)B}2(N2H2)]
to transition metals, compounds of the light p-block elements do (4, Fig. 1). We hypothesized that repeated reduction–protonation
not undergo reduction events in a partially filled d-electron sub- cycles starting from 4 may lead to complete N–N cleavage and ulti-
shell. As such, the binding of N2 by an element with limited redox mately a p-block-element-based process for the synthesis of ammo-
flexibility, such as boron, could lead to an N-centred redox chem- nia, which would also constitute a stepwise dinitrogen-to-ammonia
istry distinct from those common in classical transition metal N2 mechanism from a bridging N2 complex.
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
Dur
B
2
CAAC
Br
2e–
2NH4
Not observed
NH3
2Br–
Acid
quench
N2
2
CAAC
B
Dur
A
6e–
6H
Dur
B
Dur
B
H
N
CAAC
2
CAAC
CAAC
H
N
H
N
B
Isolated
this work
Dur
Isolated
2e–
2H+
6e–
4H
2e–
Dur
Dur
B
–
B
N
CAAC
2
–
CAAC
CAAC
N
N
B
–
Isolated
this work
H
H
Dur
Isolated
2H
2e–
4e–
4H
2e–
2H
Dur
B
H
Dur
B
N
CAAC
H
2
CAAC
N
B
N
CAAC
4e–
2H
Isolated
this work
Dur
B
H
Dur
H
Isolated
–
2
CAAC
N
H
2e–
2H+
Isolated
this work
Fig. 2 | schematic of the overall mechanism of the borylene-mediated ammonium synthesis. Initial reduction of the radical [(CAAC)BBrDur]
provides hypothetical borylene species A, from which sequential two-electron and two-proton steps and a final acid quench step provide ammonium.
All seven reduction–protonation processes can be performed in a single step and reaction vessel using the solid-state reductant KC8 and solid-state
protonating agent B(OH)3; a subsequent acid quench can also be performed in the same vessel without an isolation step. The overall process therefore
constitutes a one-pot reaction. Compounds in blue boxes have been reported previously18. Text in orange circles denotes the formal reduction–
protonation extent of the N2 unit.
Results and discussion
is effectively planar (torsion angles of 13.5(3)° and 17.6(3)°). These
Outcomes of reduction of the hydrazino diradical species. Upon geometric features, as well as the average B–N and B–CCAAC bond
exposure of hydrazino diradical 4 (Fig. 1) to an excess of the reduc- distances (1.399(2) and 1.561(3)Å, respectively) and the EPR signa-
tant potassium graphite (KC8) in benzene, the deep blue reaction ture of 5, are consistent with a delocalized (captodatively stabilized)
mixture instantaneously turned black, and the B NMR spectrum C-centred CAAC–B radical35.
of this mixture showed no signal. Filtration of the mixture and
11
The outcome of the reduction of 4 with KC8 is highly depen-
recrystallization of the resulting black solution in hexanes afforded dent on the moisture accompanying samples of the former, which
a dark crystalline solid (78% yield), which was found to be para- itself is prepared using water. Samples of 4 dried extensively by the
magnetic, as evidenced by the strong signal observed in its electron use of azeotropic distillation and 4Å molecular sieves produce 5 as
paramagnetic resonance (EPR) spectrum (Supplementary Fig. 12). described above; however, reduction of incompletely dried samples
Single-crystal X-ray diffraction (SCXRD) confirmed this species produces colourless solutions, indicative of a different product.
to be the radical amide K[(CAAC)(Dur)B(NH)] (5, Fig. 1), arising Removal of solvent and recrystallization of the latter resulted in iso-
11
from the direct reductive cleavage of the N–N bond of 4. Potassium lation of a colourless diamagnetic solid with an B NMR spectro-
amide radical 5 exists in the solid state as a dimer, held together scopic resonance at δ=46ppm and a crystallographically-derived
by the two potassium atoms, each of which binds to both amide structure corresponding to the aminoborane [B(CAACH)(Dur)
nitrogen atoms and two further arene rings. The core N–B–C–N (NH2)] (8, 84% yield; Fig. 1), the product of a formal addition of
unit of 5, the π system of which is expected to host five electrons, one N–H bond of ammonia across the C=B double bond of the
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
hypothetical borylene [(CAAC)B(Dur)] (A, Fig. 2). Although the both species having chemically identical [B(CAAC)(Dur)(NH)]
two enantiomers of the chiral compound 8 are superimposed in its units (B–N bond lengths, 1.415(1)Å (6), 1.435Å (4); B–CCAAC bond
solid-state structure and cause disorder, 8 is clearly identifiable as a lengths, 1.535(2)Å (6), 1.512(3)Å (4))18. The EPR spectrum of 6
conventional aminoborane, with a short bond between the boron (Supplementary Fig. 16) is virtually indistinguishable from that
atom and the π-donating nitrogen atom (1.376(2)Å). Interestingly, of amide radical 5 (Supplementary Fig. 12), which is understand-
with regards to the stoichiometry of N2, the formation of 8 corre- able given that the majority of the spin density of these two radicals
sponds to a further two-electron/four-proton transfer to 5, effec- likely resides on the CAAC carbene C atom and a similarly delo-
tively completing the full six-electron/six-proton reduction of N2 calized N–B–C–N core (see above; average B–N and B–CCAAC
to NH3, with the latter still being bound to the boron-containing bond lengths, 1.399(2) and 1.561(3)Å, respectively). Compound
species (Fig. 2). This identification of 8 as a proammonia complex 6 corresponds to the product of two reduction–protonation
was subsequently underlined by its hydrolysis with concentrated sequences from the borylene–dinitrogen adduct 2, and thereby
hydrochloric acid (a non-redox process), leading to the detection of occupies the same conceptual space as hydrazine (N2H4) on the
14
ammonium by N{1H} NMR spectroscopy and to the decomposi- nitrogen-to-ammonia reduction pathway. Intrigued by this similar-
tion of the boron-containing moieties to a mixture in which we were ity, we irradiated a benzene solution of [(CAAC)(Dur)B(CO)]36—a
only able to identify DurB(OH)2 and [CAAC–H]+Cl– by SCXRD.
process assumed to generate the putative borylene A by photode-
carbonylation—in the presence of hydrazine, providing 6 in 27%
One-pot synthesis of ammonium from dinitrogen. Encouraged yield after purification (Fig. 1, bottom). This reaction highlights
by the cascade nature of the reduction–protonation steps implied the relationship of 6 with hydrazine on the dinitrogen-to-ammonia
by the synthesis of 8 from 4 in the presence of both reductant (KC8) reduction pathway and reveals an intriguing and unprecedented
and protonating agent (water), probably through 5 as an interme- generation of boron-containing radicals by homolytic N–N
diate, we envisaged the possibility of preparing the proammonia bond cleavage.
borane 8 directly from boryl radical 1, the precursor required for
Exposure of a bright orange, 11B-NMR-silent solution of 6 to
the initial N2 fixation, and N2. A challenge that we identified early excess KC8 was accompanied by a colour change to deep purple,
on, however, was the unwanted potential reactions of the puta- the mixture exhibiting a new, broad 11B NMR spectroscopic sig-
tive borylene intermediate [(CAAC)B(Dur)] (A, Fig. 2) and/or nal (δ=6ppm). Filtration of the mixture and thorough extrac-
the reductant (KC8) with the proton source (water). For this rea- tion led to essentially quantitative isolation of dark red crystals,
son, we chose the weakly protic reagent boric acid (B(OH)3) as the the molecular structure of which was confirmed by SCXRD to be
proton source, reasoning that its poor solubility in organic solvents that of the CAAC-stabilized boryl anion K[(CAAC)(Dur)B(NH2)]
would limit its reactivity with intermediate A or KC8. Preliminary (7, 92% yield; Fig. 1). The solid-state structure of boryl anion dimer
tests hinted at the plausibility of this strategy: stirring a toluene 7 shows the presence of very short B–C bonds (1.464(2), 1.465(2)Å)
solution of 2 with excess KC8 and B(OH)3 for 16h led to isolation and relatively long C–N (1.471(2), 1.470(2)Å) and B–N bonds
of 8 in 67% yield, while reduction of 1 in the presence of B(OH)3 (1.506(2), 1.503(2)Å), in line with those of related CAAC-stabilized
provided the C–H activation product18 observed when the reac- boryl anions37,38 and consistent with their alternative description as
tion is performed in the absence of B(OH)3, suggesting that A is zwitterionic alkylideneborates with B=C double bonds. Although
inert towards B(OH)3. Thereby, room-temperature reduction of 1 isolable, 7 is highly reactive, undergoing facile protonation to pro-
with KC8 (10 equiv.) in the presence of N2 (or 15N2) gas (5 or 3atm, ammonia compound 8 with extended handling and/or storage.
respectively) and B(OH)3 (15 equiv.) provided ammonia adduct Selective protonation of 7 can be achieved using B(OH)3, affording
8 (or 8-15N) after 4h of reaction (near-full conversion by NMR 8 as the sole product according to NMR spectroscopy, and thus clos-
spectroscopy; 65% or 30% isolated yield of 8 and 8-15N, respec- ing the reaction sequence from 1 to 8 and ultimately [NH4]Cl.
tively). Repeating the N2 reaction with a direct acid-quench
step (aqueous HCl, 11M) provided a 27% yield of ammonium Conclusions
chloride, the conjugate acid of ammonia. This relatively low yield This work, as summarized in Fig. 2, comprises the one-pot,
is probably due to an inability to completely extract the product two-step chemical conversion of N2 to ammonium without tran-
from the complex reaction mixture. This reaction constitutes a sition metals, and the isolation of the intermediate products
simple one-pot, two-step protocol for the transition-metal-free of every two-electron/two-proton step in the process. Notably,
reduction of N2 to ammonium. Production of [15NH4]Cl was also this dinitrogen-to-ammonium process proceeds by a stepwise
confirmed by 15N{1H} insensitive-nuclei-enhanced polarization reduction–protonation mechanism via an end-on bridging N2
transfer NMR spectroscopy upon acidolysis of the labelled ammo- species. It should be noted that the B(OH)3 used as protonat-
nia adduct 8-15N.
ing agent in this work is very mild compared with those used in
transition-metal-based catalytic processes for N2 conversion, which
Isolation of remaining intermediates in the dinitrogen-to- are often relatively strong organic acids (for example, 2,6-lutidinium
ammonium process. A further advantage of the use of B(OH)3 is salts), triflic acid or tetra(fluoroaryl)borate superacids5,8,28, indicat-
that, in contrast to water, it provides excellent control over proton- ing the existence of highly nucleophilic nitrogen atoms throughout
ation reactions and, because of its poor solubility, can be easily and our process. The success of these reactions can thus be attributed to
completely removed from the products. This, in turn, allowed us the unwillingness of the electron-rich boron atoms in our system
to perform each individual reduction and protonation step of the to accept further electrons (in contrast with many transition metal
dinitrogen-to-ammonium process and isolate all of the remaining complexes), forcing the reduction steps to occur predominantly
intermediates, confirming the reductive cascade nature of the reac- on the N2 moiety and thereby facilitating N–N cleavage and pro-
tion. We found that exposing a near-black benzene solution of 5 tonation. Thus, in this regard, the limited metallomimicry of the
to excess B(OH)3 led to a complete colour change of the mixture borylene system is a distinct advantage, promoting a facile cascade
to bright orange within 20min. Filtration of the reaction mixture mechanism with reduction events largely centred on nitrogen. The
and slow evaporation of the solvent resulted in isolation of boryl one-pot chemical conversion of N2, while not a catalytic process, is a
radical [(CAAC)(Dur)B(NH2)] (6) as bright orange crystals (90% convincing demonstration of the ability of main-group elements in
yield), the structure of which was confirmed by SCXRD (Fig. 1). low oxidation states not only to activate very inert small molecules
The structure of 6 is similar to that of the related amide radical 5, but also to mediate their conversion through cascades of redox and
but more closely resembles one half of the structure of diradical 4, protonation processes.
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
19. Légaré, M.-A. et al. Te reductive coupling of dinitrogen. Science 363,
Online content
1329–1332 (2019).
Any methods, additional references, Nature Research report-
ing summaries, source data, extended data, supplementary infor-
mation, acknowledgements, peer review information; details of
author contributions and competing interests; and statements of
data and code availability are available at https://doi.org/10.1038/
s41557-020-0520-6.
20. Broere, D. L. J. & Holland, P. L. Boron compounds tackle dinitrogen. Science
359, 871 (2018).
21. Liu, Y. et al. Facile ammonia synthesis from electrocatalytic N2 reduction
under ambient conditions on N-doped porous carbon. ACS Catal. 8,
1186–1191 (2018).
22. Qiu, W. et al. High-performance artifcial nitrogen fxation at ambient
conditions using a metal-free electrocatalyst. Nat. Commun. 9, 3485 (2018).
23. Burford, R. J. & Fryzuk, M. D. Examining the relationship between
coordination mode and reactivity of dinitrogen. Nat. Rev. Chem. 1, 0026
(2017).
24. Chatt, J., Pearman, A. J. & Richards, R. L. Te reduction of mono-coordinated
molecular nitrogen to ammonia in a protic environment. Nature 253,
39–40 (1975).
25. Pool., J. A., Lobkovsky, E. & Chirik, P. J. Hydrogenation and cleavage of
dinitrogen to ammonia with a zirconium complex. Nature 427, 527–530 (2004).
26. Laplaza, C. A. & Cummins, C. C. Dinitrogen cleavage by a three-coordinate
molybdenum(iii) complex. Science 268, 861–863 (1995).
27. Curley, J. J., Cook, T. R., Reece, S. Y., Müller, P. & Cummins, C. C. Shining
light on dinitrogen cleavage: structural features, redox chemistry, and
photochemistry of the key intermediate bridging dinitrogen complex. J. Am.
Chem. Soc. 130, 9394–9405 (2008).
28. Tompson, N. B., Green, M. T. & Peters, J. C. Nitrogen fxation via a terminal
Fe(iv) nitride. J. Am. Chem. Soc. 139, 15312–15315 (2017).
29. Hellman, A. et al. Predicting catalysis: understanding ammonia synthesis
from frst-principles calculations. J. Phys. Chem. B 110, 17719–17735 (2006).
30. Rodriguez, M. M., Bill, E., Brennessel, W. W. & Holland, P. L. N2 reduction
and hydrogenation to ammonia by a molecular iron–potassium complex.
Science 334, 780–783 (2011).
Received: 14 May 2020; Accepted: 1 July 2020;
Published: xx xx xxxx
References
1. Notman, N. Haber–Bosch power consumption slashed. Chemistry World
(21 October 2012).
2. Allen, A. D., Senof, C. V. Nitrogenopentammineruthenium(ii) complexes.
J. Chem. Soc. D 621–622 (1965).
3. Burford, R. J., Yeo, A. & Fryzuk, M. D. Dinitrogen activation by group 4 and
group 5 metal complexes supported by phosphine-amido containing ligand
manifolds. Coord. Chem. Rev. 334, 84–99 (2017).
4. Tanaka, H., Nishibayashi, Y. & Yoshizawa, K. Interplay between theory and
experiment for ammonia synthesis catalyzed by transition metal complexes.
Acc. Chem. Res. 49, 987–995 (2016).
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. Fryzuk, M. D. N2 coordination. Chem. Commun. 49, 4866–4868 (2013).
7. Hazari, N. Homogeneous iron complexes for the conversion of dinitrogen
into ammonia and hydrazine. Chem. Soc. Rev. 39, 4044–4056 (2010).
8. Schrock, R. R. Catalytic reduction of dinitrogen to ammonia at a single
molybdenum center. Acc. Chem. Res. 38, 955–962 (2005).
9. MacKay, B. A. & Fryzuk, M. D. Dinitrogen coordination chemistry: on the
biomimetic borderlands. Chem. Rev. 104, 385–401 (2004).
10. Hidai, M. & Mizobe, Y. Recent advances in the chemistry of dinitrogen
complexes. Chem. Rev. 95, 1115–1133 (1995).
31. Bazhenova, T. A. & Shilov, A. E. Nitrogen fxation in solution. Coord. Chem.
Rev. 144, 69–145 (1995).
32. Doyle, L. R. et al. Catalytic dinitrogen reduction to ammonia at a triamidoamine–
titanium complex. Angew. Chem. Int. Ed. 57, 6314–6318 (2018).
33. Bezdek, M. J., Guo, S. & Chirik, P. J. Terpyridine molybdenum dinitrogen
chemistry: synthesis of dinitrogen complexes that vary by fve oxidation
states. Inorg. Chem. 55, 3117–3127 (2016).
11. Greenwood, N. N. & Earnshaw, A. Chemistry of the Elements 2nd edn
(Elsevier Butterworth-Heinemann, 2005).
12. Power, P. P. Main-group elements as transition metals. Nature 463,
171–177 (2010).
34. Soleilhavoup, M. & Bertrand, G. Cyclic (alkyl)(amino)carbenes (CAACs):
stable carbenes on the rise. Acc. Chem. Res. 48, 256–266 (2015).
35. Bissinger, P. et al. Isolation of a neutral boron-containing radical stabilized by
a cyclic (alkyl)(amino)carbene. Angew. Chem. Int. Ed. 53, 7360–7363 (2014).
36. Braunschweig, H. et al. Main-group metallomimetics: transition metal-like
photolytic CO substitution at boron. J. Am. Chem. Soc. 139, 1802–1805 (2017).
37. Arrowsmith, M. et al. Direct access to a CAAC-supported dihydrodiborene
and its dianion. Chem. Commun. 54, 4669–4672 (2018).
13. Martin, D., Soleilhavoup, M. & Bertrand, G. Stable singlet carbenes as mimics
for transition metal centers. Chem. Sci. 2, 389–399 (2011).
14. Légaré, M.-A., Pranckevicius, C. & Braunschweig, H. Metallomimetic
chemistry of boron. Chem. Rev. 119, 8231–8261 (2019).
15. Kinjo, R., Donnadieu, B., Celik, M. A., Frenking, G. & Bertrand, G. Synthesis
and characterization of a neutral tricoordinate organoboron isoelectronic
with amines. Science 333, 610–613 (2011).
38. Arrowsmith, M. et al. Facile synthesis of a stable dihydroboryl {BH2}− anion.
Angew. Chem. Int. Ed. 57, 15272–15275 (2018).
16. Braunschweig, H. et al. Multiple complexation of CO and related ligands to a
main-group element. Nature 522, 327–330 (2015).
17. Soleilhavoup, M. & Bertrand, G. Borylenes: an emerging class of compounds.
Angew. Chem. Int. Ed. 56, 10282–10292 (2017).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
18. Légaré, M.-A. et al. Nitrogen fxation and reduction at boron. Science 359,
896–900 (2018).
© The Author(s), under exclusive licence to Springer Nature Limited 2020
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
The fourth method was reduction–protonation of 15N2 using 1 and a KC8/B(OH)3
mixture to yield 8-15N. [(CAAC)BBr(Dur)] (1, 75.0mg, 147 μmol), B(OH)3
(136mg, 2.20mmol) and KC8 (299mg, 2.20mmol) were placed in a glass reactor
equipped with a Teflon valve. Toluene was degassed under vacuum and condensed
over the solid mixture at –196°C. The solvent was warmed to approximately
−80°C, at which point the reactor was flushed with 15N2. Further 15N2 (in a
calibrated volume) was then condensed into the reactor at –196°C and the reactor
was sealed, ultimately providing a pressure of ~3.3atm at the end of the reaction
at room temperature. The mixture was allowed to warm to room temperature,
stirred for 4h and subsequently filtered. The solid residue was extracted with four
portions (~3ml each) of benzene. The combined liquid fractions were evaporated
to dryness to give a residue in which 8-15N was identified as the main product.
This solid was rinsed with pentane to yield pure 8-15N as colourless crystals (20mg,
30%). The pentane fraction was evaporated to dryness to yield a further white solid
in which 8-15N is also the main component (39mg), indicating a high conversion
to 8-15N. The NMR spectra of 8-15N are identical to those of 8 apart from the
following data. 1H NMR (500MHz, 298K, C6D6,): δ 3.51 (dd, 1J1H–15N =80Hz,
2J1H–1H =1.6Hz, 1H, NH2), 3.09 (dd, 1J1H–15N =80Hz, 2J1H–1H =1.6Hz, 1H, NH2) ppm.
13C{1H} NMR (126MHz, 298K, C6D6): 134.7 (d, 3J13C–15N =0.8Hz, Dur(C-Me)),
40.7 (d, 3J13C–15N =0.8Hz, CAAC(BCMe2)) ppm. 15N NMR (51MHz, 298K, C6D6):
δ –321.0 (CAAC(NDip))–282.1 (NH2) ppm. Infrared (ATR, solid, 15N–H, cm–1):
3465, 3374. HRMS (toluene, LIFDI): [M+] 447.3790 (calculated 447.3797). See
Supplementary Figs. 4, 5, 7 and 9 for spectra of 8-15N.
Methods
General synthetic considerations. All syntheses were carried out under an
atmosphere of argon or nitrogen in argon-flled gloveboxes or using standard
Schlenk techniques. Compounds 1–4, as well as [(CAAC)B(CO)Dur], were
prepared according to literature procedures18,35. Solvents were dried by distillation
over appropriate drying agents39 and stored over molecular sieves. C6D6 was dried,
degassed and stored over molecular sieves. Anhydrous hydrazine was obtained by
distilling concentrated aqueous hydrazine over sodium hydroxide under an inert
atmosphere. Hydrazine was introduced to reaction mixtures by condensation
under vacuum and the amounts used were estimated by pressure measurements in
a vacuum line with calibrated volumes.
Synthesis of new compounds. Synthesis of amide radical 5. A blue solution of 4
(20mg, 22μmol) in C6D6 (~1.5ml) was stirred overnight over molecular sieves
(4Å) to remove trace moisture arising from the previous hydrolysis step. Once
sufciently dry, the solution was added to solid KC8 (30mg, 222μmol) with
stirring. Te mixture immediately turned very dark green (almost black). Afer
~5min, the solution was fltered and analysed by 11B NMR spectroscopy, which
did not show any signal. Te residue was extracted with more benzene (3×1ml)
and the combined organic fractions were subsequently evaporated to dryness and
the residue was recrystallized from a cooled hexanes solution. Te product was
obtained as black crystals suitable for X-ray difraction analysis (17mg, 78%).
EPR (continuous wave, X band, toluene, room temperature): g (the magnetic
feld at which an EPR line is observed)=2.0030, a(11B)=5.6MHz, a(14N)=18.0
and 15.5MHz (a = hyperfne splitting constant). Te spectrum shows additional
partially resolved hyperfne couplings (a=0.5G), probably due to the NH
hydrogen, which could not, however, be successfully included in the simulation.
High-resolution mass spectrometry (HRMS) (toluene, liquid injection feld
desorption ionization (LIFDI)): [M–K+H]+ 445.3747 (calculated 445.3749).
Infrared spectroscopy of 5 revealed only very weak bands in the N–H stretch
region. Tis situation is attributed to the dimeric nature of 5. See Supplementary
Figs. 10–13 for spectra of 5.
General method for the qualitative hydrolysis of 8 and 8-15N to form 14/15NH4Cl.
A drop of concentrated aqueous HCl (~37%) was added to NMR samples of 8
and 8-15N in C6D6. The tubes were vigorously shaken and demineralized water
with ~10% D2O (~0.2ml) was added. The organic layer was separated and
removed and NH4Cl was identified by 14N{1H} (from 8) or 15N{1H} (from 8-15N)
insensitive-nuclei-enhanced polarization transfer NMR spectroscopy, which
showed a single resonance in both cases at δ=–359 and –360 ppm, respectively.
The 14/15N NMR shifts and the 15N–H J-coupling constant of 73Hz are identical to
the literature data of NH4Cl.
Synthesis of aminoborane 8. The first method was one-pot reduction–protonation
of 2 using a KC8/B(OH)3 mixture. Toluene (1.5ml) was rapidly added to a solid
mixture of 2 (20mg, 23μmol), B(OH)3 (14mg, 226 μmol) and KC8 (31mg,
226 μmol). The suspension was stirred overnight, filtered and the residue was
extracted once with toluene (0.5ml). All volatiles were evaporated from the
filtrate in vacuo to afford 8 as a colourless solid, which was further purified by
recrystallization from a saturated pentane solution at −30°C (13.0mg, 67% yield).
1H NMR (500MHz, 298K, C6D6): δ 7.2 (m, 2H, 2,6-diisopropylphenyl (Dip)(m-Ar)
overlapping with Dip(p-Ar)), 7.14 (m, 1H, Dip(m-Ar)), 6.87 (s, 1H, Dur(Ar))
4.29 (septet, 3J=6.8Hz, 1H, Dip(CHMe2)), 4.08 (s, 1H, CAAC-H), 3.60 (septet,
3J=6.7Hz, 1H, Dip(CHMe2)), 3.52 (d, 2J=1.6Hz, 1H, NH2), 3.09 (d, 2J=1.6Hz,
1H, NH2), 2.25 (s, 3H, Dur(Me)), 2.21 (s, 3H, Dur(Me)), 2.13 (s, 6H, Dur(Me)),
1.84 (d, 2J=12.4Hz, 1H, CAAC(CH2)), 1.71 (d, 2J=12.4Hz, 1H, CAAC(CH2)),
1.44 (d, 3J=6.7Hz, 3H, Dip(CHMe2)), 1.38 (d, 3J=6.8Hz, 3H, Dip(CHMe2)), 1.37
(s, 3H, CAAC(BCMe2), 1.34 (d, 3J=6.8Hz, 3H, Dip(CHMe2)), 1.32 (d, 3J=6.7Hz,
3H, Dip(CHMe2)), 1.26 (s, 3H, CAAC(NCMe2), 1.10 (s, 3H, CAAC(NCMe2),
0.92 (s, 3H, CAAC(BCMe2) ppm. 13C{1H} NMR (126MHz, 298K, C6D6): δ 153.1
(Dip(o-Ar)), 152.2 (Dip(o-Ar)), 144.4 (broad, Dur(B–C)), 141.8 (Dip(Ar–N)),
134.7 (Dur(C–Me)), 134.2 (Dur(C–Me)), 133.7 (Dur(C–Me)), 132.7 (Dur(C–Me)),
130.9 (Dur(C–H)), 127.2 (Dip(p-Ar)), 125.1 (Dip(m-Ar)), 124.9 (Dip(m-Ar)),
68.4 (broad, CAAC(BC–H)), 61.8 (CAAC(NCMe2)), 60.1 (CAAC(CH2),
40.7 (CAAC(BCMe2)), 30.8 (CAAC(NCMe2)), 29.7 (Dip(CHMe2)), 28.9
One-pot synthesis of NH4Cl from N2, 1 and a KC8/B(OH)3 mixture. A solid
mixture of 1 (104mg, 196 μmol), B(OH)3 (200mg, 3.23mmol) and KC8 (420mg,
3.11mmol) was cooled to −78°C in a closed pressure flask. Toluene was then
added very slowly to the mixture via a cannula, after which the flask was
immediately pressurized with N2 (~5atm) with vigorous stirring. The suspension
was allowed to warm slowly to room temperature and then stirred for 4h, after
which concentrated aqueous HCl (11M, 2ml) was added drop-wise at 0°C.
The resulting biphasic mixture was stirred overnight at room temperature, then
extracted with a phosphate buffer solution (0.5M, pH 7) until 1l of aqueous phase
was obtained in a volumetric flask. The ammonium chloride concentration of this
solution was quantified using the indophenol method (yield 53μmol, 27%)40.
Synthesis of aminoboryl radical 6. The first method was protonation of 5. A black
solution of 5 (12mg, 12μmol) was added to solid B(OH)3 in a vial (15mg, 250μmol,
10 equiv. relative to B) and stirred at room temperature for 20min. During this
time, the solution turned bright orange, with no hint of the dark green colour of 5
being apparent. The reaction mixture was filtered through a glass filter in a Pasteur
pipette in a glovebox and the colourless residue was extracted with additional C6D6.
NMR analysis of the filtrate did not reveal the formation of a new diamagnetic
substance. The filtrate was slowly evaporated to dryness, leading to the isolation of
6 as orange crystals (10mg, 90%). Radical 6 was characterized by SCXRD and EPR
spectroscopy. Their structural similarities result in the EPR spectrum of 6 being
indistinguishable from that of radical 5; however, analysis of the N–H stretching
bands of 5 and 6 in their infrared spectra enables them to be distinguished and
can be used to confirm the protonation of 5 to 6. EPR (continuous wave, X band,
toluene, room temperature): g=2.0030, a(11B)=5.8MHz, a(14N)=18.0 and
15.6MHz. The spectrum shows additional partially resolved hyperfine couplings
(a=0.5G), probably due to the NH2 hydrogen nuclei, although these could not,
however, be successfully included in the simulation. Infrared (ATR, solid, N–H,
cm–1): 3498, 3402. HRMS (toluene, LIFDI): [M]+ 445.3747 (calculated 445.3749).
See Supplementary Figs. 14–17 for spectra of 6.
(CAAC(BCMe2), 28.3 (CAAC(NCMe2), 28.2 (CAAC(BCMe2), 27.7 (Dip(CHMe2)),
26.2 (Dip(CHMe2)), 26.1 (Dip(CHMe2)), 24.9 (Dip(CHMe2)), 24.7 (Dip(CHMe2)),
21.5 (Dur(Me)), 20.6 (Dur(Me)), 20.1 (Dur(Me)), 20.0 (Dur(Me)) ppm. 11
B
NMR (160MHz, 298K, C6D6): 46 ppm. HRMS (toluene, LIFDI): [M+] 446.3821
(calculated 446.3838). Infrared (attenuated total reflection (ATR), solid, N–H
stretch, cm–1): 3474, 3379. See Supplementary Figs. 1–3, 6 and 8 for spectra of 8.
The second method involved reduction of 4 to 8. A blue solution of 4 (40mg,
45μmol) in C6D6 (~1ml), containing moisture from its synthesis by hydrolysis,
was added to solid KC8 (60mg, 444μmol) with stirring, and the solution rapidly
became colourless. After ~30min, the mixture was filtered to give a colourless
solution that was analysed by 11B NMR spectroscopy, showing a new signal at
δ=46ppm. The residue was extracted with more benzene (3×1ml) and the
combined organic fractions were subsequently evaporated to dryness under
vacuum and the residue was recrystallized from a cooled hexanes solution to give 8
as colourless crystals (34mg, 84%).
The second method was photolytic reaction of hydrazine with borylene
carbonyl [DurB(CO)(CAAC)]. A solution of [DurB(CO)(CAAC)] (41mg, 90μmol)
in C6D6 was placed in a J.-Young-style NMR tube. Anhydrous hydrazine (~50 μmol)
was condensed into the mixture at –196°C. After warming to room temperature,
the thawed biphasic mixture was placed under a mercury vapor lamp (current
19A, voltage 26V) for 18h, resulting in a colour change from orange to red-orange.
EPR and NMR spectroscopic analyses suggested that a single paramagnetic species
comprised the bulk of the mixture. The volatile components were removed under
vacuum and the red-orange solid was washed several times with hexane and
dissolved in toluene/hexane. A crop of red crystals was obtained upon standing
at –30°C for a few days, and these were washed twice with pentane (11mg, 27%).
Yields can be increased by processing the hexane washings and the recrystallization
supernatants. Compound 6 obtained in this way is spectroscopically identical to
samples obtained by protonation of 5 (see the first method).
The third method was reduction–protonation of N2 using 1 and a KC8/B(OH)3
mixture. A solid mixture of 1 (150mg, 295μmol), B(OH)3 (273mg, 4.42mmol) and
KC8 (598mg, 4.42mmol) in a pressure flask was cooled to −78°C. Toluene (~25ml)
was then added very slowly to the mixture via a cannula, after which the flask was
immediately pressurized with N2 (~5atm) under vigorous stirring. The suspension
was allowed to warm slowly to room temperature, stirred for 4h and subsequently
filtered. The residue was extracted once with toluene (15ml) and the organic
fraction was dried to afford 8 as a colourless solid, which was further purified by
recrystallization from a saturated pentane solution at −30°C (80.0mg, 65% yield).
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
and multilayer mirror monochromated Mo Kα radiation. Te crystal data of
Synthesis of aminoboryl anion dimer 7 by reduction of 6. An orange solution
of radical 6 (10mg, 22 μmol) in C6D6 (~1mL) was added to solid KC8 (30mg,
221μmol) in a vial and stirred at room temperature. The solution began to turn
deep purple immediately upon addition. After stirring for 15min, the mixture was
filtered through a glass filter in a Pasteur pipette and the residue was extracted
with additional C6D6. NMR spectroscopic analysis of the filtrate revealed the
formation of 7 as a single new diamagnetic substance. The filtrate was slowly
evaporated to dryness and rapidly rinsed with a small amount of hexane, leading
to the isolation of 7 as black crystals (10mg, 92%). Compound 7 is highly sensitive
to protonation by trace moisture and should be handled with care. 7 can be
quantitatively protonated to 8 by stirring it over B(OH)3 in benzene for 30min at
room temperature. 1H NMR (500MHz, 298K, C6D6): δ 7.16 (s, C6H6 solvate), 6.86
(d, 3J=7.4Hz, 4H, Dip(m-Ar)), 6.72 (t, 3J=7.4Hz, 2H, Dip(p-Ar)), 6.55 (s, 2H,
Dur(p-CH), 4.20 (septet, 3J=6.8Hz, 4H, Dip(CHMe2)), 2.51 (s, 12H, Dur(Me)),
2.05 (s, 12H, Dur(Me)), 1.99 (s, 4H, CAAC(CH2)), 1.44 (s, 12H, CAAC(CMe2),
1.42 (s, 12H, CAAC(CMe2), 1.28 (d, 3J=6.8Hz, 12H, Dip(CHMe2)), 1.02 (d,
3J=6.8Hz, 12H, Dip(CHMe2)), 0.52 (broad s, 1H, NH2), 0.49 (broad s, 1H, NH2)
ppm. 1H diffusion-ordered spectroscopy (DOSY) NMR experiments suggest
that the compound remains a dimer in benzene solution, which is consistent
with the observation of two different yet close 15N NMR signals for the NH2
groups, suggesting an unsymmetrical dimer. Although 15N–1H heteronuclear
single-quantum coherence spectroscopy and 15N–1H heteronuclear multiple-bond
correlation spectroscopy experiments allow us to locate unambiguously two N–H
(not NH2) protons (one on each inequivalent N) in the dimer, the remaining two
proton signals could not be located. This could tentatively be attributed to fluctional
H migration processes—possibly to the B and/or CCAAC atoms—that selectively
5–7 were collected on a Rigaku Synergy Dualfex Hypix difractometer with
a complementary metal–oxide–semiconductor area detector and multilayer
mirror-monochromated Cu Kα radiation. Te structures were solved using the
intrinsic phasing method42 refned with the SHELXL program43 and expanded
using Fourier techniques. All non-hydrogen atoms were refned anisotropically.
Hydrogen atoms were included in structure factor calculations. Unless otherwise
noted, all hydrogen atoms were assigned to idealized geometric positions.
Crystallographic data have been deposited with the Cambridge Crystallographic
Data Centre as supplementary publication numbers CCDC 1971191-1971194.
Tese data can be obtained free of charge from the Cambridge Crystallographic
Data Centre, www.ccdc.cam.ac.uk/data_request/cif.
Crystallographic details for 5. The coordinates of H1 (residues 1 and 2) were refined
but the distances between atoms N1 and H1 (residues 1 and 2) were kept at the
value of 0.91Å using the DFIX restraint. Crystal data: C60H90B2K2N4, λ=1.54184Å,
Mr =967.17, bronze-coloured blade, 0.193×0.050×0.039 mm3, triclinic space
group P1, a=10.7085(2)Å, b=16.6116(3)Å, c=18.0545(4)Å, α=69.131(2)°,
β=89.791(2)°, γ=74.733(2)°, V=2,880.52(11)Å3, Z=2, ρcalcd =1.115gcm–3,
μ=1.736mm–1, F(000)=1052, T=100(2)K, R1 =0.0479, wR2 =0.1249, 10,919
independent reflections [2θ≤140.15°] and 645 parameters (two restraints),
maximum/minimum residual electron density 0.85 eÅ–3 at 0.5729 0.7247 0.3598
[0.85Å from K1_2]/−0.62eÅ–3 at 0.5148 0.6492 0.3817 [0.69Å from K1_2]. See
Supplementary Fig. 27 for the solid-state structure of 5.
Crystallographic details for 6. Crystal data: C30H46BN2, λ=1.54184Å,
Mr =445.50, orange blade, 0.248×0.200×0.140mm3, triclinic space group
P1, a=9.49530(10)Å, b=11.55080(10)Å, c=12.9846(2)Å, α = 89.4810(10)°,
β=76.2920(10)°, γ=76.5540(10)°, V=1344.04(3)Å3, Z=2, ρcalcd =1.101gcm–3,
μ=0.464mm–1, F(000)=490, T=100(2)K, R1 =0.0390, wR2 =0.1060, 5,102
independent reflections [2θ≤140.14°] and 318 parameters (two restraints),
maximum/minimum residual electron density 0.24eÅ–3 at 0.2834 0.3467 0.8298
[0.75Å from C2]/−0.21eÅ–3 at 0.1947 0.3145 0.9227 [0.65Å from C2]. See
Supplementary Fig. 28 for the solid-state structure of 6.
affect only two N–H protons because of geometrical constraints in the dimer. 11
B
NMR (160MHz, 298K, C6D6): δ 25 ppm. 13C NMR (126MHz, 298K, C6D6): δ 157.5
(located by heteronuclear multiple-bond correlation spectroscopy, Dur(B-C)),
151.9 (Dip(o-Ar)), 150.0 (Dip(N-Ar)), 137.0 (Dur(m-Ar)), 133.0 (Dur(o-Ar)),
128.5 (C6H6 solvate), 128.1 (Dur(CH)), 124.6 (Dip(m-Ar)), 123.0 (Dip(p-CH)), 61.5
(CAAC(CH2)), 61.2 (CAAC(CMe2)), 42.5 (CAAC(CMe2)), 33.7 (CAAC(CMe2)),
28.8 (CAAC(CMe2)), 26.7 (Dip(CHMe2)), 25.9 (Dip(CHMe2)), 24.7 (Dip(CHMe2)),
22.0 (Dur(Me)), 20.4 (Dur(Me)) ppm. 15N NMR (shifts inferred from 15N–1H
heteronuclear single-quantum coherence spectroscopy and 15N–1H heteronuclear
multiple-bond correlation spectroscopy experiments, 51MHz, 298K, C6D6): δ
–344.3 (B–NH), –343.3 (B–NH), –296.6 (CAAC) ppm. HRMS (toluene, LIFDI):
[M–K+H]+ 446.3822 (calculated 446.3827). Infrared spectroscopy of 7 revealed
only very weak bands in the N–H stretch region. This situation is attributed to the
dimeric nature of 7. See Supplementary Figs. 18–26 for spectra of 7.
Crystallographic details for 7. The coordinates of H1 and H2 (residues 0 and 1)
were refined but the distances between atoms N1/H1 and N1/H2 (residues 0
and 1) were kept at the value of 0.91Å using the DFIX restraint. The molecule
co-crystallizes with a half benzene molecule that is disordered about an inversion
centre and was refined using an idealized geometry with the Afix 6 keyword.
The displacement parameters of atoms C1–C6 (benzene) were restrained to the
same value with similarity restraint SIMU. Uii displacement parameters of C1–
C6 (benzene) were restrained with the ISOR keyword to approximate isotropic
behaviour. Rigid bond restraints were applied to residue 2 (benzene) with the
keyword RIGU. Crystal data: C63H95B2K2N4, λ=1.54184Å, Mr =1008.24, orange
blade, 0.194×0.119×0.049mm3, triclinic space group P1, a=12.8434(2)Å,
b=13.1929(3)Å, c=18.2794(3)Å, α=81.594(2)°, β=89.1760(10)°, γ=80.114(2)°,
V=3,018.34(10)Å3, Z=2, ρcalcd =1.109gcm–3, μ=1.674mm–1, F(000)=1098,
T=100(2)K, R1 =0.0469, wR2 =0.1273, 11,468 independent reflections
[2θ≤140.144°] and 695 parameters (112 restraints), maximum/minimum residual
electron density 0.57eÅ–3 at −0.0076 0.0682 0.0524 [0.64Å from C4_2]/−0.42eÅ–3
at 0.7587 0.3262 0.3332 [0.68Å from K1]. See Supplementary Fig. 29 for the
solid-state structure of 7.
NMR spectroscopy. General NMR spectroscopic details. Solution NMR spectra
were acquired on a Bruker Avance I 400 NMR spectrometer (1H: 400.57MHz)
or a Bruker Avance I 500 spectrometer (1H: 500.13MHz). Chemical shifs (δ) are
given in ppm and internally referenced to the carbon nuclei (13C{1H}) or residual
protons (1H) of the solvent. 11B{1H} NMR spectra were referenced to [BF3·OEt2] as
an external standard. 14N and 15N spectra were referenced to nitromethane + 10%
CDCl3 as an external standard.
DOSY NMR spectroscopy experiments. 1H DOSY measurements of 7 and 8 in
C6D6 were performed at 298.15K in 5mm NMR tubes on a Bruker Avance Neo
400 spectrometer equipped with a 5mm BBFO iProbe (BBFO stands for broad
band fluorine observation) with a z-axis gradient coil capable of producing pulsed
magnetic field gradients of 51Gcm−1. Temperature calibration was performed with
a standard sample of 4% CH3OH in CD3OD. The gradient coil was calibrated to
4.995745Gmm−1 at 298.15K using a ‘doped water’ NMR sample (0.1mg GdCl3/mL
D2O + 1% H2O + 0.1% 13CH3OH, 40mm fill height). Data were acquired and
processed using Bruker Topspin 4.0.7 software. The DOSY data were recorded
with a stimulated echo BPP-LED sequence (longitudinal eddy current delay (LED)
sequence with bipolar gradient pulse pairs (BPP) for diffusion and additional
spoil gradients). The diffusion time Δ (50ms) was kept constant in each DOSY
experiment while the ‘smoothed square’ diffusion gradients were incremented
from 2% to 95% of maximum gradient strength in 24 linear steps. One-component
fittings of the gradient strength dependence of the signal areas were performed by
a Levenberg–Marquardt algorithm incorporated in the Topspin 4.0.7 software.
Crystallographic details for 8. The coordinates of H1 and H2 of residue 0 were
refined but the distances between atoms N1/H1 and N1/H2 were kept at the
value of 0.91Å using the DFIX restraint. One of the isopropyl moieties of the
2,6-diisopropylphenyl group could be modelled as a two-part disorder with a
refined ratio of 0.73:0.27, in which the isopropyl is found in two orientations
related by a rotation about the aryl–C axis. The protonated CAAC moiety is found
as two conformers, which was modelled as a two-part disorder with a refined ratio
of 0.74:0.26. The two disorders were refined independently (two free variables) but
may be related because H–H distances between the CAAC methyl groups and the
isopropyl groups are too short when the isopropyl disorder is not modelled. The
displacement parameters of atoms C8 and C8a of residue 0 were constrained to the
same value with the EADP keyword. The displacement parameters of atoms C1,
C1a, C3, C3a, C4, C4a, C7, C7a, C8, C8a, C2, C2a, C5, C5a, C6 and C6a of residue
0 and C8–C12a of residue 2 (2,6-diisopropylphenyl) were restrained to the same
value with similarity restraint SIMU. The distances between atoms C4/C7,
C4a/C7a, C4/C8, C4a/C8a, C2/C5, C2/C6, C2a/C6a, C2a/C5a of residue 0 were
restrained during refinement to the same value with the SADI restraint. The
Uii displacement parameters of atoms C4, C4a, C2 and C2a of residue 0 were
restrained with the ISOR keyword to approximate isotropic behaviour. The
1–2 and 1–3 distances in atoms C8a–C12a of residue 2 were restrained to the
same values with the SAME command. Crystal data: C30H47BN2, λ=0.71073Å,
Mr =446.50, colourless prisms, 0.537×0.377×0.172 mm3, triclinic space group
P1, a=9.372(4)Å, b=11.500(4)Å, c=13.271(9)Å, α=88.76(2)°, β=76.074(17)°,
γ=77.292(11)°, V=1353.6(12)Å3, Z=2, ρcalcd =1.096gcm–3, μ=0.062mm–1,
F(000)=492, T=100(2)K, R1 =0.0692, wR2 =0.1293, 5,549 independent reflections
Further spectroscopic and spectrometric details. EPR measurements at the
X band (9.38GHz) were carried out using a Bruker ELEXSYS E580 CW EPR
spectrometer equipped with an Oxford Instruments helium cryostat (ESR900)
and a MercuryiTC temperature controller. Spectral simulations were performed
using MATLAB 8.6.0.267246 (R2015b) and the EasySpin 5.1.9 toolbox41. The
EPR spectrometer was calibrated by using 2,2-diphenyl-1-picrylhydrazyl (dpph)
with a g value of 2.0037 as a standard. Solid-state infrared spectra were acquired
on a Bruker Alpha spectrometer using a setup with a Bruker diamond crystal
single-reflection ATR system. HRMS data were obtained from a Thermo Scientific
Exactive Plus spectrometer in LIFDI mode.
SCXRD. General considerations. Te crystal data of 8 were collected on a
Bruker X8-APEX II difractometer with a charge-coupled device area detector
NAtuRe CHeMIstRy | www.nature.com/naturechemistry

NATUre CHemisTry
Articles
[2θ≤52.744°] and 419 parameters (357 restraints), maximum/minimum residual
electron density 0.34eÅ–3 at 0.2788 0.5392 0.7730 [0.79Å from C3_1]/−0.31eÅ–3
at 0.0455 0.3944 0.9155 [0.11Å from C7A]. See Supplementary Fig. 29 for the
solid-state structure of 8.
Acknowledgements
We thank the Deutsche Forschungsgemeinschaft for financial support. M.-A.L. thanks
the Natural Sciences and Engineering Research Council of Canada for a postdoctoral
fellowship. G.B.-C. thanks the Alexander von Humboldt Foundation for a postdoctoral
fellowship.
Data availability
Crystallographic data for the structures reported in this article have been deposited
at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC
1971191 (8), 1971192 (7), 1971193 (5) and 1971194 (6). Copies of the data can be
obtained free of charge via https://www.ccdc.cam.ac.uk/structures/. Further data
supporting the findings of this study are available from the corresponding author
upon reasonable request.
Author contributions
Experiments were designed by M.-A.L., G.B.-C., R.D.D. and H.B. and performed by
M.-A.L., M.R. and G.B.-C. Data analysis was performed by M.-A.L., G.B.-C. and the
article was written by M.-A.L. and R.D.D. X-ray crystallography was performed by
G.B.-C. The EPR investigation was performed by I.K. and the NMR experiments were
performed by R.B. The project was overseen by H.B. All authors read and commented on
the manuscript.
References
39. Armarego, W. L. F. & Chai, C. L. L. Purifcation of Laboratory Chemicals
6th edn (Elsevier, 2009).
Competing interests
The authors declare no competing interests.
40. Chaney, A. L. & Marbach, E. P. Modifed reagents for determination of urea
and ammonia. Clin. Chem. 8, 130–132 (1962).
41. Stoll, S. & Schweiger, A. EasySpin, a comprehensive sofware package for
spectral simulation and analysis in EPR. J. Magn. Reson. 178, 42–55 (2006).
42. Sheldrick, G. SHELXT—integrated space-group and crystal-structure
determination. Acta Crystallogr. A 71, 3–8 (2015).
43. Sheldrick, G. A short history of SHELX. Acta Crystallogr. A 64,
112–122 (2008).
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/
s41557-020-0520-6.
Correspondence and requests for materials should be addressed to H.B.
Reprints and permissions information is available at www.nature.com/reprints.
NAtuRe CHeMIstRy | www.nature.com/naturechemistry