Characterization of an Fe
≡
N-NH
2
Intermediate Relevant to
Catalytic N
2
Reduction to NH
3
John S. Anderson
#
†
,
George E. Cutsail III
#
‡
,
Jonathan Rittle
#
†
,
Bridget A. Connor
†
,
William
A. Gunderson
‡,#
,
Limei Zhang
†,§,*
,
Brian M. Hoffman
‡,*
, and
Jonas C. Peters
†,*
†
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena,
California 91125, United States
‡
Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois
60208, United States
#
These authors contributed equally to this work.
Abstract
The ability of certain transition metals to mediate the reduction of N
2
to NH
3
has attracted broad
interest in the biological and inorganic chemistry communities. Early transition metals such as Mo
and W readily bind N
2
and mediate its protonation at one or more N atoms to furnish M(N
x
H
y
)
species that can be characterized and, in turn, extrude NH
3
. By contrast, the direct protonation of
Fe-N
2
species to Fe(N
x
H
y
) products that can be characterized has been elusive. Herein we show
that addition of acid at low temperature to [(TPB)Fe(N
2
)][Na(12-crown-4)] results in a new
S
=
1/2 Fe species. EPR, ENDOR, Mössbauer, and EXAFS analysis, coupled with a DFT study,
unequivocally assign this new species as [(TPB)Fe
≡
N-NH
2
]
+
, a doubly protonated hydrazido(2-)
complex featuring an Fe-to-N triple bond. This unstable species offers strong evidence that the
first steps in Fe-mediated nitrogen reduction by [(TPB)Fe(N
2
)][Na(12-crown-4)] can proceed
along a distal or `Chatt-type' pathway. A brief discussion of whether subsequent catalytic steps
may involve early or late stage cleavage of the N-N bond, as would be found in limiting distal or
alternating mechanisms, respectively, is also provided.
Since the discovery
1
and crystallographic characterization
2
of the FeMo-cofactor as the
active site of FeMo-nitrogenase, there has been substantial interest in elucidating the
mechanism of biological nitrogen reduction.
3
While synthetic model chemistry cannot
provide direct mechanistic information regarding enzymatic N
2
reduction, it can play a
crucial role in exploring the chemical viability of proposed pathways and in exposing new
reactivity patterns that help stimulate, frame, and constrain various hypotheses. Early
synthetic model work by Chatt, Hidai and others
4
as well as recent examples of molecular
Mo catalysts for N
2
reduction
5
lent credibility to the early proposal that the Mo center in the
FeMo-cofactor could in principle serve as the site of N
2
binding and reduction via a distal or
*
Corresponding Author
jpeters@caltech.edu, bmh@northwestern.edu, limei.zhang@unl.edu.
#
Current address: Department of Chemistry, Illinois College, 1101 West College Avenue, Jacksonville, IL, 62650, United States;
§
Department of Biochemistry and Redox Biology Center, University of Nebraska-Lincoln, Lincoln, NE 68588, United States.
Supporting Information
. Detailed experimental and spectroscopic data. This material is available free of charge via the Internet at
http://pubs.acs.org
.
HHS Public Access
Author manuscript
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Published in final edited form as:
J Am Chem Soc
. 2015 June 24; 137(24): 7803–7809. doi:10.1021/jacs.5b03432.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Chatt-type mechanism.
4a
,
6
More recent spectroscopic, biochemical, and structural studies of
FeMo-nitrogenase have suggested that iron is the more likely site for N
2
binding and
reduction in the FeMo-cofactor.
7
This hypothesis also reflects that, to date, iron is the only
transition metal known to be essential to nitrogenase activity, underscored by the
characterization of Fe-only nitrogenases.
3b
,
8
This state of affairs has motivated studies
towards the synthesis of Fe complexes that mimic steps in proposed N
2
reduction schemes
and/or stabilize candidate iron intermediates of biological nitrogen fixation.
9
Indeed, numerous synthetic studies underscore the idea that an Fe-N
2
binding site(s) may be
mechanistically relevant to biological N
2
fixation.
10
We have been interested in the
postulate that a single Fe center can mediate N
2
reduction via a flexible ligand environment
that enables N
2
coordination at Fe and facilitates the stabilization of intermediates exhibiting
Fe-N multiple bonding
en route
to NH
3
formation.
11
Isolated synthetic Fe complexes of a
tris(phosphine)borane (TPB) scaffold (Scheme 1) with bound N
2
or imido (NR
2−
) ligands
demonstrate important aspects of this single-site hypothesis.
11e,f
Moreover, through the use
of chlorosilanes, this scaffold permits the conversion of Fe-N
2
to Fe
≡
N-NR
2
, modeling the
first two steps in a Chatt-type mechanism. The generation and characterization of a parent
Fe
≡
N-NH
2
unit, either directly from Fe-N
2
or otherwise, proved more challenging.
Recently, three Fe-based catalysts for N
2
reduction to NH
3
were reported by our
laboratory.
12
These catalysts appear to be single-site
12a
,
13
and provide functional Fe-N
2
reduction systems for which systematic mechanistic studies can be conducted. In our
original report,
12a
a new
S
= 1/2 species was generated by the low temperature addition of
excess acid to [(TPB)Fe(N
2
)][Na(12-crown-4)
2
] (
1
) in the absence of exogenous reductant.
This doublet species was tentatively postulated to be the hydrazido(2-) complex
[(TPB)Fe
≡
N-NH ][BAr
F
4
] (BAr
F
4
= (3,5-(CF
3
)
2
C
6
H
3
)
4
B
−
) (
2
),
12a
but several alternative
structures such as those candidates depicted in Scheme 1 could not be excluded. Because
2
is a likely intermediate of catalytic nitrogen reduction by
1
(vide infra), its characterization
is of primary interest.
Herein we report combined EPR, ENDOR, and EXAFS spectroscopic studies that assign
2
as the parent hydrazido(2-) complex [(TPB)Fe
≡
N-NH
2
][BAr
F
4
].
57
Fe Mössbauer studies
indicate that this species constitutes a significant portion of the Fe-containing material upon
exposure of
1
to acid. Finally, the geometric constraints provided by the spectroscopic data
are corroborated by the optimized geometry of
2
obtained from Density Functional Theory
(DFT) methods.
Mixing of pre-catalyst
1
and 10 equivalents of HBAr
F
4
· 2 Et
2
O in thawing 2-MeTHF
solutions at −136 °C (Scheme 1) resulted in the disappearance of the dark red color
characteristic of
1
and the formation of a brown-yellow solution. Analysis by CW X-band
EPR indicates complete consumption of
1
and appearance of a new, more rhombic,
S
= 1/2
signal at 77 K (Figure 1B). Indeed, spin integration of the signal of
2
suggests a yield of
87(8)%. If stoichiometric acid is added to
1
at low temperature, rapid oxidation to neutral
S
= 1 (TPB)Fe(N
2
) with loss of 0.5 equiv H
2
occurs instead (eqn 1) and no EPR signal is
observed (SI).
12a
Anderson et al.
Page 2
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
(1)
(2)
The signal for
2
can be generated, albeit at lower intensity, with as little as 2 equivalents of
acid. EPR spectra collected at 10 K reveal weak, complicated signals at low field that are
consistent with one or more
S
= 3/2 species (SI), in addition to those stemming from
2
. The
EPR spectrum of
2
is distinct from the more axial signature of
1
(Figure 1A) and it is also
distinct from the previously reported EPR spectrum for (TPB)Fe(N
2
SiMe
3
) (
3
).
11g
The fact
that structurally characterized
3
, a model complex of the as yet uncharacterized complex
`(TPB)FeN
2
H', displays a more axial EPR spectrum than that of
2
suggests that
2
is
electronically distinct from
3
, and by extension is unlikely to be the parent (TPB)FeN
2
H
diazenido species (candidate
2'-E
in Scheme 1). Moreover, the requirement of ≥ 2 equiv of
acid strongly suggests an alternative structure to
2'-E
.
Metal hydrazido(2-) complexes typically display substantial M-N multiple bonding, and are
thus electronically similar to metal imido complexes.
5b
,
14
The cationic imido complex
[(TPB)Fe
≡
N-Ad][BAr
F
4
] (
4
) is isoelectronic to [(TPB)Fe
≡
N-NH
2
][BAr
F
4
], but was found
to be thermally stable. This imido complex has been crystallographically characterized (SI)
and also displays a rhombic EPR spectrum (Figure 1C). The rhombic EPR spectrum of
4
reflects a pseudotetrahedral iron center resulting from Fe-B elongation (Fe-B = 2.770 Å) as
the iron center slips above the P
3
plane. This geometry should place an unpaired spin in a
relatively nonbonding orbital arising from a
2
E
dx
2
−y
2
,xy
state. This electronic structure is
reminiscent of the ferrocenium cation and is also similar to various low spin L
3
Fe
III
≡
NR
imides that have been previously described.
15
,
9d
Fe-N triply bonded species supported by the TPB scaffold include [(TPB)Fe
≡
N-Ad][BAr
F
4
]
(this work) and (TPB)Fe
≡
N(4-OMe-Ph).
11e
These species have characteristically short Fe-
N distances of 1.660 Å and 1.668 Å respectively. Structural data was sought for
2
to probe
for an anticipated short
Fe
≡
N
-NH
2
triple bond. Iron K-edge X-ray extended X-ray
absorption fine structure (EXAFS) data were collected and allowed for the observation of
Fe-ligand distances in solution samples. These data were collected on frozen preparations
(SI) of
2
in 2-MeTHF and are shown in Figure 2. Three pronounced peaks are observed in
the Fourier transform spectrum, including one peak much shorter than expected for a singly-
bonded Fe-N pair. Indeed the EXAFS and Fourier transform spectra can be best fit with a
short Fe-N distance of 1.64 Å and two Fe-P distances of 2.28 Å and 2.42 Å in a 1:2 ratio.
Additionally, an Fe-B scattering path may be included in the fit at a distance of 2.67 Å,
although convolution from the Fe-P scatterers makes this assignment tentative.
The observation of two different Fe-P scatterers for a complex with three phosphorous
ligands can arise from various scenarios. If one assumes the EXAFS sample contains a
single species, the two distances observed could correspond to a distorted structure with two
long Fe-P distances and one short distance. A similar, though substantially lessened,
distortion is observed in [(TPB)Fe
≡
N-Ad][BAr
F
4
]. Alternatively, these disparate distances
Anderson et al.
Page 3
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
could arise from the presence of more than one species in the sample, each with a narrow
distribution of Fe-P distances, a scenario that seems more likely based on the observation
that oxidation of
1
by acid is a competing side reaction (eqn 1). The apparently larger than
expected Debye-Waller factor in the fitting for the Fe-N scatterer would thereby result from
the presence of multiple species; the short and long Fe-P distances correspond closely to
those observed in low-spin and high-spin iron complexes of the (TPB) ligand scaffold,
which display variable Fe-N distances.
11f–h
To further probe the possible presence of multiple iron species generated on addition of acid
to
1
at low temperature, Mössbauer analysis of in-situ generated solutions of
2
was
undertaken. In brief, an ethereal solution of
57
Fe-enriched [(TPB)
57
Fe(N
2
)][Na(Et
2
O)
x
] was
frozen and then layered with excess HBAr
F
4
· 2 Et
2
O in Et
2
O that was then also frozen. This
sample was then allowed to thaw to ~−110 °C and then mechanically stirred. The sample
was then transferred to a Mössbauer cup chilled at 77 K for analysis. The Mössbauer
spectrum of a representative sample is shown in Figure 3 and suggests the presence of three
primary iron-containing species. Similarly prepared samples showed the same features but
in variable ratios and hence caution must be exercised with respect to extrapolating the
population of
2
in this specific Mössbauer sample relative to the population of
2
in
independently prepared EXAFS and EPR samples. Nevertheless, the Mössbauer data
confirm the presence of three (TPB)Fe-species generated under relatively comparable
conditions. One of the three species present in the representative Mössbauer spectrum shown
(~20% of total Fe present; SI) can be definitively assigned as the neutral
S
= 1 Fe-N
2
adduct
(TPB)Fe(N
2
) by comparison with an authentic sample (
δ
= 0.56 mm/s; ΔE
Q
= 3.34 mm/s).
The two other species are present in approximately equal amounts (40% each of total Fe).
These two components can be simulated in multiple ways (see SI for a detailed discussion),
but comparison of the simulation parameters with those of independently characterized, pure
(TPB)Fe complexes suggests the simulation of Figure 3 to be the most reasonable. The
simulation parameters suggest one of them is the previously characterized
S
= 3/2 cation
[(TPB)Fe]
+
(
δ
= 0.75 mm/s; ΔE
Q
= 2.55 mm/s);
16
the other by default is therefore the iron
hydrazido(2-) cation
2
(
δ
= 0.35 mm/s; ΔE
Q
= 1.02 mm/s).
The presence of (TPB)Fe(N
2
) and [(TPB)Fe]
+
in the sample is consistent with our previous
finding that the addition of HBAr
F
4
· 2 Et
2
O to
1
leads to net oxidation of the complex to
generate (TPB)Fe(N
2
) via loss of H
2
.
12a
Likewise, (TPB)Fe(N
2
) can be further oxidized to
[(TPB)Fe]
+
via addition of HBAr
F
4
· 2 Et
2
O (eqn 2, SI). In sum, the available Mössbauer
data predicts that variously prepared samples of
2
may contain (TPB)Fe(N
2
) and [(TPB)Fe]
+
and this fact helps explain the additional P scatterer in the EXAFS data, as well as the larger
than expected Fe-N Debye-Waller factor. As (TPB)Fe(N
2
) is EPR-silent, and [(TPB)Fe]
+
does not display EPR signals at 77 K, this Fe speciation is fully consistent with the clean
spectrum observed by CW X-band EPR (Figure 1B).
While the aforementioned characterization data are most consistent with [(TPB)Fe
≡
N-NH
2
]
[BAr
F
4
], the techniques used are not sensitive to the presence or location of the protons. As
a direct probe of these protons, as well as other nuclei in
2
, we turned to electron nuclear
double resonance (ENDOR) spectroscopy. This technique is particularly attractive because it
Anderson et al.
Page 4
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
selectively monitors
S
= 1/2
2
regardless of the presence of the other species that complicate
the EXAFS and Mössbauer analysis.
Figure 4 presents a 2D field-frequency pattern of
1
H Davies pulsed Q-band ENDOR spectra
collected across the EPR envelope of
2
. In addition to ancillary ligand signals from weakly-
coupled protons with
A
< 4 MHz, the spectra show strongly-coupled signals that arise from
protonation of the N
2
ligand (Figure 4). The spectra collected at the `crystal-like' fields,
g
1
and
g
3
, show two distinct
1
H doublets,
A
g
1
(
1
H
α
) ~ 16 MHz, and
A
g
1
(
1
H
β
) ~ 10 MHz
(Figure 4, blue and red, respectively), indicating the presence of two well-defined,
inequivalent protons, consistent with assignment of
2
as exhibiting an unsymmetrically
bound di-protonated (-N-NH
2
) species, inconsistent with the proposed structures
2'-B
and
2'-E
(Scheme 1). Structure
2'-B
has two protons which are anticipated to be approximately
equivalent with couplings substantially larger than observed (see below), and structure
2'-E
should only have one proton signal.
As the magnetic field is increased, the
1
H
α
feature splits, with the outer peak reaching a
maximum
1
H coupling between
g
1
and
g
2
(11661 G),
A
max
= 18 MHz for
1
H
α
; with further
increase in field the pattern coalesces into the single doublet at
g
3
. The maximum coupling
for a metal-bound hydride (
A
max
~ 40 MHz) is much larger,
17
inconsistent with structures
2'-C
and
2'-D
, Scheme 1. The
1
H
α
pattern resembles that of a heme hydroperoxy Fe-O-OH
proton,
18
an appealing analogy to the Fe-N-NH
2
unit in
2
, and those analyses provide a good
starting point for analysis of the
1
H
α
hyperfine tensor; simulation of the
1
H
α
pattern (Figure
4, blue) yields a slightly rhombic tensor of
A
= +[18.0, 10.5, 8.0] MHz (see Fig 4 caption),
whose anisotropic contribution corresponds to an Fe-H
α
distance of d > 3.1 Å, with the Fe-
H
α
vector rotated away from g
1
, which coincides approximately with the Fe-N bond and the
molecular pseudo C
3
axis.
19
In short, the analysis requires a non-linear Fe-N-N geometry
(Fig S8).
The smaller
1
H
β
coupling (Fig 4, red) exhibits a 2D pattern that can be described by a
hyperfine tensor whose anisotropic contribution corresponds to an Fe-H
β
distance of d > 3.5
Å, with the Fe-H
β
vector along
g
1
(the Fe-N bond). The absolute signs of the
1
H
α
,
1
H
β
,
couplings are implied by the assumption that the dominant anisotropic component is
determined by through-space dipolar coupling, and were confirmed by the Variable Mixing
Time (VMT) Pulsed ENDOR Saturation and Recovery (PESTRE) protocol (SI). The
estimated Fe-
1
H
α
/
1
H
β
distances, along with the orientations of the hyperfine coupling
tensors relative to
g
, imply that both protons have been added to the
β
-N of a bound N
2
, to
generate a bent Fe
≡
N-NH
2
moiety, with a Fe-N-N bond angle
δ
~ 150° (SI). Notably, as
indicated above, the metrical parameters for the two protons are inconsistent with
every
other model visualized in Scheme 1.
Q-band
14/15
N ENDOR data was also acquired to further characterize the hydrazido ligand
of
2
(Figure 5). A
15
N ENDOR response is observed as a peak at
ν
+
= +9.8 MHz (Figure 5),
corresponding to
A
2
(
2
-
15
N) = +9.0 MHz,
A
2
(
2
-
14
N) = −6.4 MHz for the coordinated
nitrogen (upon scaling by the the gyromagnetic ratios of
14
N and
15
N (
γ ≡
g
n
(
14
N)/
g
n
(
15
N)
≡
A
(
14
N)/
A
(
15
N) = −0.71)). The
2
-
14
N ENDOR response has a quadrupole splitting of 3
P
=
2.7 MHz at
g
2
. Furthermore, the isotopically labeled
2
-(
15
N) sample has a resolved
Anderson et al.
Page 5
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
hyperfine coupling of
A
3
(
15
N) = +8.0 MHz near
g
3
. The corresponding
14
N ENDOR
response of
2
-(
14
N) is expected to have a coupling of
A
3
(
14
N) = −5.7 MHz, but is not
observed. The additional quadrupole splitting of the
14
N at
g
3
may orient
A
and
P
in such a
manner that the quadrupole ENDOR response is too broad at
g
3
and sharpest at
g
2
. No
distinguishable nitrogen ENDOR features were observed at fields below
g
2
for either
2
-
(
14
N) and
2
-(
15
N), but the single-field three-pulse ESEEM spectrum of
2
-(
14
N) (SI) is well
simulated with a single
14
N tensor of
A
(
2
-
14
N) = −[4.34, 7.18, 6.22] MHz and quadrupole
parameters of
e
2
qQ
/
h
= 1.76 (
P
= 0.88 MHz) and rhombicity factor,
η
= 0.64, in excellent
agreement with the observed
P
2
= 0.90 MHz from ENDOR spectroscopy. The presence of
quadrupole coupling with high rhombicity supports the idea of a non-linear Fe-N-N linkage.
11
B ENDOR spectra of
2
(SI, Fig. S10 and S11) show a coupling of
a
iso
~ −8.8 MHz,
completing the characterization of the axial ligands to Fe. As shown in Figure S11, the
11
B
nuclei exhibits a VMT-PESTRE response, typical of a nuclei with negative spin density and
a negative
A
. Using the
11
B as a `reference,' the observed
14
N nuclei possesses negative spin
density and
1
H
α
and
1
H
β
both exhibit positive spin density. The negative spin density
observed on the
11
B and
14
N nuclei is a result of the spin-polarization of the Fe-B bond by
the positive-spin density of the Fe center.
Finally, a nearly isotropic
31
P ENDOR signal from the phosphine ligand is simulated with
A
(
31
P
α
) = [43, 42, 40] MHz (Figure 4, green). Features of a second phosphine are observed
in Figure 5, around 10.0–10.5 MHz and assigned as the
ν
−
transitions of
31
P doublet with
A
~ 60 MHz; the
ν
+
partner of this
31
P response unfortunately falls underneath the intense
ENDOR response of weakly coupled solvent protons (
1
H) (~55 MHz) (Figure S12). This
observation of two
31
P couplings, one with
A
3
(
31
P
β
) ~ 64 MHz the other with
A
3
(
31
P
α
) = 40
MHz by ENDOR spectroscopy correlates well with the simulation of the X-band EPR
spectrum (Figure 1).
Computational studies were undertaken to compare the parameters of the theoretically
predicted structure with those obtained spectroscopically. An optimized geometry (SI) of
[(TPB)Fe
≡
N-NH
2
]
+
was found, utilizing the core of [(TPB)Fe
≡
NAd]
+
for an initial guess.
This geometry shows similar bonding metrics to those observed by EXAFS and ENDOR
analysis. Thus, ENDOR requires di-protonation of the remote N2, while EXAFS requires a
short Fe-N1 distance consistent with the calculated value 1.70 Å, and calculated Fe-P
distances of 2.29, 2.31, and 2.36 Å (Figure 6A). The characteristically short Fe-N distance
(i.e ≤ 1.7 Å) is consistently found when exploring a variety of functionals while the Fe-P
distances show slightly more variation (SI). In addition, the bending at N
α
required for the
interpretation of the ENDOR data is observed in the calculated gas-phase structure. Figure
6B depicts the predicted spin distribution in [(TPB)Fe
≡
N-NH
2
]
+
suggesting that the
majority of spin is carried by Fe with leakage to the NNH
2
ligand and B also evident. Other
candidate structures for
2
were similarly optimized and found to be higher energy isomers.
For example, a comparison of the energies of the hydrazido(2-) species versus a diazene
isomer (Scheme 1,
2
'-
A
) shows that the assigned [(TPB)Fe
≡
N-NH
2
]
+
isomer is lower in
energy by 17.0 (BP86) or 8.2 (M06L) kcal/mol (SI). Taken together the computational work
is also consistent with the assignment of
2
as [(TPB)Fe
≡
N-NH
2
]
+
.
Anderson et al.
Page 6
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
While
2
has been generated herein with excess acid at −136 °C in 2-MeTHF, N
2
reduction
catalysis by
1
was originally carried out at −78 °C in Et
2
O by addition of acid followed by
addition of reductant.
12a
As such, we sought to determine whether
2
could be detected in
mixtures more relevant to catalysis, prior to the addition of the reductant. X-band EPR
spectra of solutions prepared from the addition of acid to
1
at either −136 °C in 2-MeTHF or
at −78 °C in Et
2
O indicate that
2
is present in both preparations (SI, Fig. S15). Furthermore,
upon warming to −40 °C or to room temperature, both preparations show the growth of
S
=
3/2 signals concomitant with the decay of the signals of
2
. The identity of these
S
= 3/2
species is hard to determine due to convolution of the quartet signals. One such
S
= 3/2
species, [(TPB)Fe(NH
3
)]
+
, has been definitively assigned by inspection of the
1
H NMR
spectrum of a similarly-prepared mixture following warming to room temperature.
12a
Therefore,
2
decays at least in part to [(TPB)Fe(NH
3
)]
+
upon warming even without
exogenous reductant.
The combined spectroscopic data presented herein confirm the assignment of
2
as a
[(TPB)Fe
≡
N-NH
2
]
+
hydrazido(2-) complex. All of the likely alternative formulations for
2
can be conclusively excluded by presented data. The observation of a short Fe-N distance by
EXAFS spectroscopy conclusively excludes the two diazene formulations,
2'-A, B
, as well
as
2'-E
. In addition, an
η
1
diazene adduct (
2'-A
) should have two
1
H hyperfine couplings,
but with one much stronger than the two observed. An
η
2
diazene adduct (
2'-B
), is expected
to have two
1
H couplings larger than observed. Hydride formulations (
2'-C
,
2'-D
) are ruled
out by the
1
H couplings and metric parameters already discussed. Finally,
2'-E
would have
only one coupled proton.
To summarize, double protonation at the distal nitrogen, as invoked in a distal or Chatt type
mechanism, is viable for this iron catalyst and likely occurs in the Fe-mediated N
2
fixation
cycle catalyzed by
1
(Scheme 2). One or more downstream species arising from the
decomposition of
2
involve higher spin (
S
= 3/2) species based on X-band EPR data (see SI).
Plausible candidates for these
S
= 3/2 species include [(TPB)Fe(NH
3
)]
+
, (TPB)Fe(NH
2
), and
[(TPB)Fe]
+
; these species, each of which is shown in blue in Scheme 2, are compatible with
both
distal and alternating scenarios. Another candidate
S
= 3/2 species is the hydrazine
complex that would form on the alternating pathway, [(TPB)Fe(N
2
H
4
)]
+
, which was
previously characterized and shown to decompose to [(TPB)Fe(NH
3
)]
+
.
11h
A branching
scenario whereby the distal intermediate
2
isomerizes to an alternate diazene structure, such
as either
2'-A
or
2'-B
in Scheme 1, that then proceeds to (TPB)Fe-N
2
H
4
+
as a later-stage
intermediate of the alternating pathway, remains a mechanistic possibilty.
7
Likewise, formal
H-atom addition to
2
at the
α
N-atom (Scheme 1) provides an alternative possible branching
point from the distal to the alternating pathway. The key point we wish to underscore is that
the observation of
2
is not incompatible with late stage N-N cleavage via a hydrazine
intermediate.
Sorting out further mechanistic details for the present system warrants additional studies.
But one conclusion is clear: an imide-like [(TPB)Fe
≡
N-NH
2
]
+
species can be directly
generated at low temperature by protonation of the Fe-N
2
catalyst precursor, and solutions
of this species are known to decay to [(TPB)Fe(NH
3
)]
+
. This is a satisfying conclusion as it
Anderson et al.
Page 7
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
links the chemistry of iron-to-nitrogen multiple bonding to an N
2
fixation cycle. In this
context a conceptual parallel can be drawn to catalytic O
2
reduction wherein iron-to-oxygen
multiply bonded species (e.g., Fe=O) are generated as intermediates.
20
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
ACKNOWLEDGMENT
This work was supported by the NIH (GM 070757 to JCP, GM 111097 to BMH), the NSF (MCB-1118613 to
BMH, DGE-0824162 to GEC), an NSERC fellowship (to LZ), a Caltech Center for Environmental Microbial
Interactions fellowship (to LZ), and the Gordon and Betty Moore Foundation. We thank the staff at Beamline 9–3,
Stanford Synchrotron Radiation Lightsource (SSRL). SSRL is operated for the DOE and supported by OBER and
by the NIH, NIGMS (P41GM103393) and the NCRR (P31RR001209).
REFERENCES
1. Burgess BK. Chem. Rev. 1990; 90:1377–1406.
2. (a) Kim J, Rees DC. Science. 1992; 257:1677–1682. [PubMed: 1529354] (b) Einsle O, Tezcan FA,
Andrade SLA, Schmid B, Yoshida M, Howard JB, Rees DC. Science. 2002; 297:1696–1700.
[PubMed: 12215645] (c) Spatzal T, Aksoyoglu M, Zhang LM, Andrade SLA, Schleicher E, Weber
S, Rees DC, Einsle O. Science. 2011; 334:940–940. [PubMed: 22096190]
3. (a) Burgess BK, Lowe DJ. Chem. Rev. 1996; 96:2983–3011. [PubMed: 11848849] (b) Eady RR.
Chem. Rev. 1996; 96:3013–3030. [PubMed: 11848850]
4. (a) Chatt J, Dilworth JR, Richards RL. Chem. Rev. 1978; 78:589–625.(b) Hidai M, Mizobe Y.
Chem. Rev. 1995; 95:1115–1133.
5. (a) Yandulov DV, Schrock RR. Science. 2003; 301:76–78. [PubMed: 12843387] (b) Arashiba K,
Miyake Y, Nishibayashi Y. Nature Chem. 2011; 3:120–125. [PubMed: 21258384]
6. (a) Armstrong F. Nature. 1985; 317:576.(b) Pickett CJ, Talarmin J. Nature. 1985; 317:652.(c) Leigh
GJ. Science. 2003; 301:55. [PubMed: 12843380] (d) Schrock RR. Nature Chem. 2011; 3:95.
[PubMed: 21258376]
7. Hoffman BM, Dean DR, Seefeldt LC. Acc. Chem. Res. 2009; 42:609–619. [PubMed: 19267458]
8. (a) Bishop PE, Jarlenski DM, Hetherington DR. Proc. Natl. Acad. Sci. U. S. A. 1980; 77:7342.
[PubMed: 6938981] (b) Schneider K, Gollan U, Dröttboom M, Selsemeier-Voigt S, Müller A. Eur.
J. Biochem. 1997; 244:789. [PubMed: 9108249] (c) Hinnemann B, Norskov JK. Phys. Chem.
Chem. Phys. 2004; 6:843.
9. (a) Berry JF. Comments Inorg. Chem. 2009; 30:28.(b) Scepaniak JJ, Young JA, Bontchev RP,
Smith JM. Angew. Chem. Int. Ed. 2009; 48:3158.(c) Crossland JL, Tyler DR. Coord. Chem. Rev.
2010; 254:1883–1894.(d) Saouma CT, Peters JC. Coord. Chem. Rev. 2011; 255:920–937.
[PubMed: 21625302] (e) Bowman AC, Milsmann C, Bill E, Turner ZR, Lobkovsky E, DeBeer S,
Wieghardt K, Chirik PJ. J. Am. Chem. Soc. 2011; 133:17353. [PubMed: 21985461] (f) King ER,
Hennessy ET, Betley TA. J. Am. Chem. Soc. 2011; 133:4917. [PubMed: 21405138] (g)
Kuppuswamy S, Powers TM, Johnson BM, Bezpalko MW, Brozek CK, Foxman BM, Berben LA,
Thomas CM. Inorg. Chem. 2013; 52:4802. [PubMed: 23259457] (h) MacLeod KC, Holland PL. Nat
Chem. 2013; 5:559–565. [PubMed: 23787744]
10. For example see: Field LD, Guest RW, Turner P. Inorg. Chem. 2010; 49:9086. [PubMed:
20815362] Crossland JL, Balesdent CG, Tyler DR. Inorg. Chem. 2011; 51:439. [PubMed:
22017491] Rodriguez MM, Bill E, Brennessel WW, Holland PL. Science. 2011; 334:780.
[PubMed: 22076372] Saouma CT, Kinney RA, Hoffman BM, Peters JC. Angew. Chem. Int. Ed.
2011; 50:3446–3449. Balesdent CG, Crossland JL, Regan DT, López CT, Tyler DR. Inorg. Chem.
2013; 52:14178. [PubMed: 24256095] Chiang KP, Bellows SM, Brennessel WW, Holland PL.
Chemical Science. 2014; 5:267. Grubel K, Brennessel WW, Mercado BQ, Holland PL. J. Am.
Anderson et al.
Page 8
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
Chem. Soc. 2014; 136:16807. [PubMed: 25412468] Rittle J, McCrory CCL, Peters JC. J. Am.
Chem. Soc. 2014; 136:13853. [PubMed: 25184795]
11. (a) Betley TA, Peters JC. J. Am. Chem. Soc. 2004; 126:6252–6254. [PubMed: 15149221] (b)
MacBeth CE, Harkins SB, Peters JC. Can. J. Chem. 2005; 83:332.(c) Peters, JC.; Mehn, MP.
Activation of Small Molecules. Wiley; 2006. p. 81(d) Hendrich MP, Gunderson W, Behan RK,
Green MT, Mehn MP, Betley TA, Lu CC, Peters JC. Proc. Natl. Acad. Sci. U. S. A. 2006;
103:17107–17112. [PubMed: 17090681] (e) Lee Y, Mankad NP, Peters JC. Nat Chem. 2010(f)
Moret M-E, Peters JC. Angew. Chem. Int. Ed. 2011; 50:2063–2067.(g) Moret M-E, Peters JC. J.
Am. Chem. Soc. 2011; 133:18118–18121. [PubMed: 22008018] (h) Anderson JS, Moret M-E,
Peters JC. J. Am. Chem. Soc. 2013; 135:534–537. 2, 558–565. [PubMed: 23259776]
12. (a) Anderson JS, Rittle J, Peters JC. Nature. 2013; 501:84–87. [PubMed: 24005414] (b) Creutz SE,
Peters JC. J. Am. Chem. Soc. 2013; 136:1105–1115. [PubMed: 24350667] c) Ung G, Peters JC.
Angew. Chem. Int. Ed. 2015; 54:532.
13. While catalysis appears to be mediated by mono-metallic species, we cannot exclude a role for bi-
or multi-metallic intermediates.
14. (a) Hidai M, Kodama T, Sato M, Harakawa M, Uchida Y. Inorg. Chem. 1976; 15:2694.(b)
Gebreyes K, Zubieta J, George TA, Koczon LM, Tisdale RC. Inorg. Chem. 1986; 25:405.(c)
Yandulov DV, Schrock RR. Inorg. Chem. 2005; 44:1103. [PubMed: 15859292]
15. Brown SD, Betley TA, Peters JC. J. Am. Chem. Soc. 2002; 125:322. [PubMed: 12517130]
16. While we favor [(TPB)Fe]
+
, structurally and electronically similar
S
= 3/2 species that feature a
weakly bound axial ligand, including [(TPB)Fe(NH
3
)][BAr
F
4
] and [(TPB)Fe(N
2
H
4
)][BAr
F
4
],
cannot be strictly excluded as alternative candidates.
17. Kinney RA, Hetterscheid DGH, Hanna BS, Schrock RR, Hoffman BM. Inorg. Chem. 2010;
49:704. [PubMed: 20000748]
18. (a) Davydov R, Kofman V, Fujii H, Yoshida T, Ikeda-Saito M, Hoffman BM. J. Am. Chem. Soc.
2002; 124:1798. [PubMed: 11853459] (b) Davydov R, Makris TM, Kofman V, Werst DE, Sligar
SG, Hoffman BM. J. Am. Chem. Soc. 2001; 123:1403. [PubMed: 11456714]
19. (a) McNaughton RL, Chin JM, Weare WW, Schrock RR, Hoffman BM. J. Am. Chem. Soc. 2007;
129:3480. [PubMed: 17341079] (b) McNaughton RL, Roemelt M, Chin JM, Schrock RR, Neese
F, Hoffman BM. J. Am. Chem. Soc. 2010; 132:8645. [PubMed: 20429559] (c) Kinney RA,
McNaughton RL, Chin JM, Schrock RR, Hoffman BM. Inorg. Chem. 2011; 50:418. [PubMed:
21155580] (d) Lee Y, Kinney RA, Hoffman BM, Peters JC. J. Am. Chem. Soc. 2011; 133:16366.
[PubMed: 21954981] (e) Gunderson WA, Suess DLM, Fong H, Wang X, Hoffmann CM, Cut-sail
GE III, Peters JC, Hoffman BM. J. Am. Chem. Soc. 2014; 136:14998. [PubMed: 25244422] (f)
Cutsail GE III, Stein BW, Subedi D, Smith JM, Kirk ML, Hoffman BM. J. Am. Chem. Soc. 2014;
136:12323. [PubMed: 25137531]
20. Nam W. Acc. Chem. Res. 2007; 40:522. [PubMed: 17469792]
Anderson et al.
Page 9
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
SCHEME 1.
Anderson et al.
Page 10
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
SCHEME 2.
Plausible scenarios that would transform [Fe-N
2
]
−
catalyst
1
to [Fe-NH
3
]
+
where [Fe
≡
N-
NH
2
]
+
,
2
, is a key intermediate. Species shown in blue (along with spin states) have been
experimentally characterized.
Top path
illustrates a limiting distal mechanism,
bottom path
an alternating mechanism, and the dotted arrows illustrate hybrid paths that could shuttle
distal intermediate
2
to an alternating pathway that features late stage N-N cleavage. The
order of protons, electrons, and/or H-atoms are provided for bookkeeping purposes only.
Anderson et al.
Page 11
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 1.
(A) 77 K X-Band (9.388 GHz) EPR of [(TPB)Fe(N
2
)][Na(12-crown-4)
2
] (
1
). (B) 77 K X-
Band (9.409 GHz) EPR after addition of 10 equivalents of HBAr
F
4
· 2 Et
2
O to
1
to generate
[(TPB)Fe
≡
N-NH
2
][BAr
F
4
] (
2
). The parameters for the simulation of the spectrum are:
g
1
=
2.222,
g
2
= 2.091,
g
3
= 2.006, two
31
P nuclei coupling
A
3
(
31
P
α
) = 64 MHz,
A
3
(
31
P
β
) = 40
MHz determined from ENDOR spectroscopy and third phosphorus of
A
3
(
31
P
λ
) ≤ 20 MHz;
line broadening = 256, 113, and 41 MHz for
g
1
,
g
2
, and
g
3
respectively. (C) 77 K X-Band
(9.399 GHz) EPR of [(TPB)Fe
≡
N-Ad][BAr
F
4
] (
4
).
Anderson et al.
Page 12
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 2.
Fe K-edge Fourier transform of the EXAFS spectrum collected on a sample of
2
. Inset
shows the EXAFS oscillations. Scatterer distances (Å) are Fe-P = 2.28, Fe-P = 2.42, Fe-N =
1.64, and Fe-B = 2.67 with the phosphines present in a 1:2 ratio. Full fit parameters may be
found in the SI.
Anderson et al.
Page 13
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 3.
57
Fe Mössbauer spectrum obtained by the reaction of
57
Fe-enriched [(TPB)
57
Fe(N
2
)]
[Na(Et
2
O)
x
] with 5 equiv HBAr
F
4
· 2 Et
2
O in Et
2
O at −110 °C. Data is shown as black dots
and the combined simulation is shown in red. The individual sub-spectra representing
(TPB)Fe(N
2
), [(TPB)Fe][BAr
F
4
], and [(TPB)Fe
≡
N-NH
2
][BAr
F
4
] (
2
) are shown in blue,
orange, and green respectively and are offset for clarity. The spectrum was collected at 80 K
in the presence of a 50 mT applied magnetic field to minimize line broadening (fast
relaxation regime). Refer to the SI for simulation parameters and more detailed discussion.
Anderson et al.
Page 14
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 4.
2D field-frequency 35.049 GHz Davies ENDOR pattern of 2 (
black solid
) collected at 2
K.
1
H
α
,
blue
, A = [18.0, 10.5, 8.0] MHz,
β
= 30°;
1
H
β
,
red
, A = [11.5, 6.0, 4.5] MHz,
β
=
0°;
31
P,
green
, A = [43, 42, 40] MHz. Summed ENDOR simulation of
1
H
α
,
1
H
β
, and
31
P is
in black dashed line.
Conditions
:
π
= 60 ns;
τ
= 600 ns; T
RF
= 30 μs; repetition time = 20
ms; RF frequency randomly hopped.
Anderson et al.
Page 15
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 5.
Davies pulsed ENDOR spectra of
2
collected at 2 K for each
14
N and
15
N isotopologues in
red and blue, respectively. Observed 14/15N coupling are Larmor centered (triangles) split
by
A
and
P
for
14
N (goalposts). A stronger nitrogen coupling is observed in
2
following
protonation of
1
. At
g
2
:
A
2
(
2
-
15
N) = +9.0 MHz;
A
2
(
2
-
14
N) = −6.4 MHz; 3
P
(
2
-
14
N) = 2.7
MHz. At
g
3
:
A
3
(
2
-
15
N) = +8.0 MHz but the corresponding
A
3
(
2
-
14
N) = −5.7 MHz is
possibly further quadrupole split and not observed. The
ν
−
feature of
a
iso
(
2
-
11
B) = −8.7
MHz at ~11 to 12 MHz. The features observed at 10.0 and 10.5 MHz at
g
2
and
g
3
,
respectively, is the
ν
of
a
iso
(
2
-
31
P
β
) = 60 MHz.
11
B/3: peaks at
ν
B
/3, the third harmonic
of
11
B nuclear Larmor frequency. Parameters: microwave frequency = 34.88 GHz;
π
/2 =
100 ns;
τ
= 600 ns; T
RF
= 60 μs; repetition time = 20 ms.
Anderson et al.
Page 16
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript
FIGURE 6.
(A) Computed geometry for [(TPB)Fe
≡
N-NH
2
]
+
and (B) spin density plot for
[(TPB)Fe
≡
N-NH
2
]
+
. Both plots generated with the M06L functional with TZVP, SVP, and
6-31G(d) basis sets on Fe, P B N, C, and H respectively.
Anderson et al.
Page 17
J Am Chem Soc
. Author manuscript; available in PMC 2015 October 13.
Author Manuscript
Author Manuscript
Author Manuscript
Author Manuscript