Catalyst-Controlled Nitrene Transfer by Tuning Metal:Ligand Ratios: Insight into the Mechanisms of Chemoselectivity

Article abstract of DOI:10.1021/acs.organomet.7b00190

Catalyst-controlled, selective nitrene transfer is often challenging when both C-H and C - C bonds are present in a substrate. Interestingly, a simple change in the Ag(I):L ratio (L = bidentate N,N-donor ligand) enables tunable, chemoselective nitrene transfer that favors either C - C bond aziridination using an 1:1 Ag:L ratio (AgLOTf) or insertion into a C-H bond when the Ag:L ratio in the catalyst is 1:2 (AgL₂OTf). In this paper, mechanistic studies, coupled with kinetic profiling of the entire reaction course, are employed to examine the reasons for this unusual behavior. Steady-state kinetics were found to be similar for both AgLOTf and AgL₂OTf; both complexes yield electronically similar reactive intermediates that engage in nitrene transfer involving formation of a short-lived radical intermediate and barrierless radical recombination. Taken together, experimental and computational studies point to two effects that control tunable chemoselectivity: suppression of aziridination as the steric congestion around the silver center is increased in AgL₂OTf and a decrease in the rate of C-H insertion with AgLOTf in comparison to AgL₂OTf. The observation that the sterics of Ag catalysts can be varied, with minor effects on the electronic features of the putative nitrene, has important implications for the development of other silver catalysts that enable tunable, site-selective C-H bond aminations.

Full text of DOI:10.1021/acs.organomet.7b00190

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Article
pubs.acs.org/Organometallics
Catalyst-Controlled Nitrene Transfer by Tuning Metal:Ligand Ratios:
Insight into the Mechanisms of Chemoselectivity
Cale Weatherly,†,§ Juliet M. Alderson,† John F. Berry,*,† Jason E. Hein,*,‡
and Jennifer M. Schomaker*,†
†Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706, United States
‡Department of Chemistry, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z1
S
* Supporting Information
ABSTRACT: Catalyst-controlled, selective nitrene transfer is
often challenging when both C−H and CC bonds are
present in a substrate. Interestingly, a simple change in the
Ag(I):L ratio (L = bidentate N,N-donor ligand) enables
tunable, chemoselective nitrene transfer that favors either C
C bond aziridination using an ∼1:1 Ag:L ratio (AgLOTf) or
insertion into a C−H bond when the Ag:L ratio in the catalyst
is 1:2 (AgL2OTf). In this paper, mechanistic studies, coupled
with kinetic profiling of the entire reaction course, are employed to examine the reasons for this unusual behavior. Steady-state
kinetics were found to be similar for both AgLOTf and AgL2OTf; both complexes yield electronically similar reactive
intermediates that engage in nitrene transfer involving formation of a short-lived radical intermediate and barrierless radical
recombination. Taken together, experimental and computational studies point to two effects that control tunable
chemoselectivity: suppression of aziridination as the steric congestion around the silver center is increased in AgL2OTf and a
decrease in the rate of C−H insertion with AgLOTf in comparison to AgL2OTf. The observation that the sterics of Ag catalysts
can be varied, with minor effects on the electronic features of the putative nitrene, has important implications for the
development of other silver catalysts that enable tunable, site-selective C−H bond aminations.
nondirected and tunable intramolecular aminations.13 For
example, catalysts 1 and 2, based on AgOTf and 1,10-
phenanthroline (phen) (Scheme 1), enable selective aziridina-
INTRODUCTION
■
A central challenge in contemporary catalysis is the develop-
ment of methods for the oxidation of organic substrates that are
selective for single reactive sites, operate under catalyst-controlled
Scheme 1. Chemoselective Ag-Catalyzed Nitrene Transfer
Controlled by the Ag:Ligand Ratio
conditions (as opposed to typical substrate control), and are
tunable for different reactive sites in a molecule.1 Catalytic
nitrene transfer comprises a powerful method to install valuable
carbon−nitrogen bonds that are ubiquitous in natural products,
pharmaceuticals, bioactive molecules, and ligands for cataly-
sis.2,3 These reactions are promoted by an array of transition
metals in groups 7−11,4−10 including dinuclear Rh(II)
complexes that are typically supported by carboxylate or
carboxamidate bridging ligands. These popular catalysts display
high reactivity and excellent scope but tend to favor
aziridination over C−H insertion when a sulfamate nitrene
precursor is used in the presence of both CC and activated
C−H bonds.3,4,10 If nitrene insertion into an allylic C−H bond
is desired instead, researchers can employ dinuclear Ru catalysts
containing 2-hydroxypyridine supporting ligands12a or porphyr-
in- and phthalocyanine-supported metal catalysts.9,10,12b,c
However, few of these catalysts attain synthetically useful
chemoselectivities and site selectivities in unbiased substrates
with multiple, competing reactive sites,11,12 and fewer still can
be tuned to promote alternative chemoselective reactions
simply through ligand modifications.
tion or C−H amination of homoallylic and homoallenic
carbamates,13b,c where the reaction outcome is controlled by
simply varying the ligand to AgOTf ratio. This tunability has
been extended to catalysts based on electron-rich bipyridines,
including 3 and 4 (tBuBipy).13b The reasons for this
unexpected switch in reactivity were not clear; however,
The Schomaker group has reported silver-catalyzed nitrene
transfer reactions that override substrate control to achieve
Received: March 10, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.organomet.7b00190
Organometallics XXXX, XXX, XXX−XXX

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Organometallics
Article
resolving this issue is important to inform future efforts to
expand the scope of Ag-catalyzed chemoselective and site-
selective aminations. In this paper, we describe experimental
and computational studies that reveal mechanistic details of the
reactions in Scheme 1 by providing answers to the following
questions. (1) Are catalysts 1 and 2 capable of both
aziridination and C−H bond amination, and if so, how and
why do the rates differ? (2) Are the silver nitrenes formed from
1 and 2 electronically similar? (3) Do reactions catalyzed by 1
and 2 proceed through similar mechanistic pathways? (4) What
role does sterics play in controlling the tunable chemo-
selectivity?
Scheme 2. General Mechanisms of Metal-Catalyzed Nitrene
Transfer
RESULTS AND DISCUSSION
■
Factors That Influence the Selectivity and Mechanism
of Silver-Catalyzed Nitrene Transfer. One hypothesis for
the tunable chemoselectivity observed with silver catalysis is
that distinct catalytic species 1 and 2 form in solution when the
Ag:ligand ratio is changed. These species may have different
coordination environments and/or nuclearities, leading to the
observed divergence in chemoselectivity. Typical catalysts for
nitrene transfer tend to display similar coordination geometries
within specific classes. For example, dinuclear Rh and Ru
complexes employ bridging ligands to maintain a “paddle-
wheel”-type geometry in the complex,3,4,7,11,12a while porphyr-
in- and phthalocyanine-based ligands supporting monomeric
Co,9 Fe,10 and Mn12c complexes also tend to have similar
coordination geometries around the metal center. In contrast,
silver complexes that catalyze nitrene transfer display a diverse
array of coordination geometries and steric constraints in
response to changes in the silver counterion, Ag:ligand ratio,
solvent, temperature, and pH.14 For example, while an ∼1:1
tBubipy:AgOTf ratio yields the tricoordinate complex 3
(Scheme 1) with the OTf bound to the metal, a
tBuBipy:AgOTf ratio of 2:1 forms the tetracoordinate 4 with
an outer-sphere OTf.14a,h Other Ag:bipyridine ratios furnish
dimeric and oligomeric structures14a,i influenced by the nature
of the solvent, counterion, and stoichiometry.14 Though limited
analogies can be drawn between solid-state and solution
behaviors, these examples attest to the diversity of potential
bonding modes and nuclearities available to Ag(I) complexes.
Thus, correlations of the solution-state structure of the resting
state of Ag catalysts might be helpful to understand tunable
chemoselectivity.
A second possible reason for the observed tunable chemo-
selectivity is the existence of different mechanisms of nitrene
transfer for either the mono- or bis-ligated species (Scheme 1, 1
vs 2 and 3 vs 4). Two general mechanistic schemes for
intramolecular metal-catalyzed nitrene transfer tend to be
observed among diverse catalyst systems (Scheme 2),15 with
the basic features proposed in Kwart’s seminal report on Cu-
promoted decomposition of a sulfonyl azide.16 Reaction of a
nitrogen transfer reagent with an oxidant, such as PhIO,
generates the imidoiodinane 6, which is then transferred to the
metal to generate a metal−supported nitrene of the form 7 or
8. Variations in this general mechanistic scheme are often
attributed to differences in the electronic structures of the metal
nitrene intermediates.15f,i,k,17,18 These differences are often
presented as a simple binary scheme, wherein triplet metal
nitrenes 8 promote stepwise amination either through stepwise
addition of the nitrene to an alkene or by an H atom
abstraction/radical recombination process, while singlet metal
nitrenes 7 carry out concerted, asynchronous amination by
insertion into a reactive C−H or CC bond (Scheme 2, left).
While there are examples where this paradigm does not apply,
it is worth considering as a potential reason for the bifurcated
reactivity observed in our silver catalyst systems.
Common experimental probes of these competing mecha-
nistic pathways (Scheme 2, right) include isomerization of
alkene geometry, ring opening of radical clocks, the effects of
radical inhibitors, linear free energy relationship studies of
styrene aziridination and benzylic C−H insertion, and the
measurement of intrinsic kinetic isotope effects (KIE). Within
this mechanistic paradigm, the divergence in chemoselectivity
displayed by the two Ag complexes can be attributed to
electronically distinct nitrenes, each with a unique propensity
toward either aziridination or C−H amination.
A final proposal for our tunable chemoselectivity is that the
mono- and bis-ligated Ag complexes support nitrenes with the
same electronic structure, but the different steric environments
enforce divergent reaction pathways. In this scenario, both
catalysts favor similar mechanisms for the nitrene transfer event
but display different reaction rates for aziridination vs C−H
insertion. This is an intriguing scenario, but few detailed kinetic
studies of catalytic nitrene transfer reactions have been
reported. Jacobsen and co-workers obtained evidence for
ligand acceleration in Cu-catalyzed aziridination,15l while
Chang observed a second-order dependence on the copper
catalyst in aziridinations with 2-pyridylsulfonyl moieties.15m In
other selected studies, Du Bois’ initial rate kinetic studies of
Rh2-dicarboxylate catalysts revealed zero-order rate dependence
on the catalyst,15j while Warren’s stoichiometric kinetic study of
Cu nitrenes demonstrated an inverse dependence on added Cu,
a crucial piece of evidence supporting a pre-equilibrium
between Cu dimer nitrenes and the catalytically active
monomeric Cu nitrene intermediate.15i To achieve a better
understanding of our silver systems, including catalyst
activation, deactivation, product inhibition, and maximum
reaction rates, we undertook a closer examination of the
kinetic details of Ag-catalyzed nitrene transfer. Insight from
these studies can aid the development of more active and
selective catalysts and furnish standard design principles
applicable to other types of metal-catalyzed, chemoselective
oxidation protocols.
Dynamic Behavior of Ag(I) Complexes in Solution.
The effect of the AgOTf:ligand ratio on the population
distribution of Ag species in solution has been previously
explored by carrying out NMR studies at various AgOTf:tBu-
Bipy ratios.13b AgOTf was dissolved in CD2Cl2 and the ligand
added in 0.5 equiv portions up to 5.0 equiv ligand/equiv of
AgOTf. Unfortunately, the rapid rate of ligand exchange, even
at −80 °C, prevented the observation of discrete silver species
B
DOI: 10.1021/acs.organomet.7b00190
Organometallics XXXX, XXX, XXX−XXX

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Organometallics
Article
1
by H NMR spectroscopy. Efforts to isolate authentic samples
Scheme 4. Isomerization and Stereochemical Probes for
of 1 and 2 directly resulted only in the recovery of 2; however,
indirect evidence for dynamic exchange between 1 and 2 could
be obtained (Scheme 3).13b When 9 was treated with the
Nitrene Transfer
Scheme 3. Indirect Evidence for the Dynamic Behavior of
Silver Catalysts for Nitrene Transfer
AgOTf:tBuBipy under typical reaction conditions. Both
substrates gave only a single diastereomer of the product,
with no evidence for isomerization (Scheme 4, bot-
tom).1g,13b,15n
isolated complex 2 or with additional ligand, C−H bond
amination dominated to yield mainly 11. In contrast, when 9
was treated with both 2 and added AgOTf, the reaction
selectively gave aziridine 10.
Isomerization in the reactions reported in Scheme 4 would
have provided support for a stepwise mechanism, but the
observed retention of stereochemistry does not rule out a rapid
radical rebound pathway. Evans and Perez have demonstrated
that isomerization in stepwise aziridination processes can be
highly substrate dependent;8b,15h however, in our case, the
observed retention of stereochemistry under conditions that
select for either aziridination or C−H bond amination suggests
that both pathways are likely to proceed from electronically
similar metal nitrene complexes, whatever the actual mecha-
nism may be.
Dynamic ligand exchange with a bidentate nitrogenated
ligand can also favor dimeric or oligomeric Ag ligand clusters.
In their report on Ag-catalyzed aziridination, He and co-
workers obtained crystals of a disilver(I) complex and posited
this dimer as the active catalyst.5a Given this and the many
precedents for the existence of multinuclear Ag species in the
solid state,14 we considered that one or both of our active
catalysts might consist of dimers or higher aggregates, allowing
metal−metal interactions to play a central role in the observed
catalytic activity. DOSY-NMR aided in ascertaining the
nuclearity of the Ag complexes in solution. To improve
solubility, phen was replaced with a tBuBipy ligand, which
displayed tunable chemoselectivities similar to those observed
with phen. DOSY indicated that the primary resting state of the
complex in solution at a 1:1.25 AgOTf:tBuBipy ligand loading
(0.05 M) is monomeric Ag(tBuBipy)OTf, while higher ligand
loadings of up to 1:3 AgOTf:tBuBipy resulted in the formation
of Ag(tBuBipy)2OTf (Scheme 1, 3 and 4, respectively).13b
Addition of the PhIO oxidant did not affect the nuclearity of
either complex; both remained monomeric in solution. While
the possibility of Ag···Ag interactions in the reactive nitrene
intermediate cannot be ruled out, it appears that monomeric Ag
complexes with different coordination numbers are the most
likely species responsible for both nitrene transfer pathways.
Probing the Nature of the Metal Nitrene Intermedi-
ates. Differences in the electronic structure of the putative
nitrenes bound to the Ag atom in 3 and 4 might account for
differences in chemoselectivity, perhaps by promoting con-
certed vs stepwise nitrene transfer. Physical organic experi-
ments can help to elucidate the relationship among catalyst
composition, silver nitrene electronic structure, and the
observed chemoselectivity. However, there are caveats in that
mechanisms involving short-lived radical intermediates cannot
always be ruled out using experimental studies; in these cases,
computational studies are valuable for obtaining further insight
into reaction pathways (vide infra).12a,b,15h,j,17−19 Nonetheless,
independent reactions with 12-Z or 12-E yielded a single
stereoisomer of the expected products on treatment with
standard aziridination or C−H amination conditions (Scheme
4, top).13b Only cis-13a was obtained from 12-Z, while only the
trans-aziridine 13c was observed in the reaction of 12-E under
both sets of reaction conditions. In the C−H amination
pathway, the stereochemical probes 14a,c were treated with 1:3
The possibility of radical intermediates was further assessed
through reactions of 15 (Scheme 5, top) in the presence of
Scheme 5. Effect of Radical Inhibitors and Traps on the
Chemoselectivity
equimolar amounts of catalyst and the radical inhibitor
TEMPO (2,2,6,6-tetramethylpiperidine-1-oxy radical). Radical
inhibitors with exchangeable protons, such as BHT (2,6-di-tert-
butyl-4-methylphenol), do affect the rate of nitrene transfer, but
their ability to perturb the equilibrium between AgLOTf and
AgL2OTf leads to uncertainty as to whether this effect is due to
stepwise nitrene transfer.13b The differences between the yields
obtained under both sets of conditions were within
experimental error, indicating that, if radical intermediates are
formed, they have lifetimes too short to be intercepted by these
inhibitors. Silver-catalyzed nitrene transfer in the presence of 17
C
DOI: 10.1021/acs.organomet.7b00190
Organometallics XXXX, XXX, XXX−XXX

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Organometallics
Article
bearing an intramolecular radical trap (Scheme 5, bottom) was
also carried out. The putative radical generated from this
substrate should undergo cyclopropane ring opening with a rate
of approximately 4 × 1011 s−1.20 Performing the reaction under
conditions of both high and low Ag:ligand loadings gave only
18 as the product, along with remaining 17. No olefin product
19, indicative of cyclopropane ring opening, was noted in either
reaction.
facilitate aziridination or C−H amination via a concerted
mechanism (see Chart 1). Aziridination and C−H amination
Chart 1. Comparison of the Electronic Structures of Rh2-
and Ag-Nitrene Species
Intramolecular competition experiments using differentially
substituted bis-aryl carbamates 20a−e (Scheme 6)21 were
Scheme 6. Hammett Studies of Ag-Catalyzed C−H
Amination
mechanisms are therefore necessarily stepwise; however, Perez
has shown in the case of aziridination,15h and we have shown in
the case of C−H amination,25 that the recombination step can
occur without a barrier (i.e., radical species are not necessarily
stationary points on the potential energy surface). In the case of
C−H amination, a hydrogen atom transfer (HAT) transition
state is followed either by direct formation of products
(barrierless recombination) or by radical intermediates, with
different silver catalysts leading to different intermediate
lifetimes.26 The strength of the Ag−N bond in the nitrene
complex was important in distinguishing between complexes
that undergo barrierless vs nonbarrierless radical recombina-
tions. For example, a bidentate N,O-coordination mode for a
sulfamate-derived nitrene, similar to the proposed coordination
mode in a metastable copper nitrene complex described by Ray
and co-workers, leads to a barrierless radical recombination.26a
To provide computational insight into differences in the
electronic structures of nitrenes formed with differing Ag:ligand
ratios, nitrenes modeled on Ag(I) precursors 3 and 4 (Figure
1a) were studied. Density functional theory (DFT) methods
were used throughout, as we found these to be successful in
previous work and were consistent with the results of higher-
level CASSCF calculations.25 Optimized catalyst structures
were based on the crystallographic data27,28 for Ag(tBuBipy)-
OTf (3) or Ag(tBuBipy)2OTf (4) and used with the simplified
nitrene precursor 23. Imidoiodinane 23 adducts of 3 and 4
were optimized to Ag(tBuBipy)(OTf)(imidoiodinane) (24a,
not shown) and [Ag(tBuBipy)2(imidoiodinane)]OTf (25a, not
shown). Dissociation of PhI from these two intermediates gave
the complexes Ag(tBuBipy)(OTf) (nitrene) (24b) and [Ag-
(tBuBipy)2(nitrene)]+ (25b). Both structures were examined
using a relaxed surface scan that elongated the N−I bond
distance from the normal bond distance of ∼2.1 Å to a
nonbonding distance of ∼5 Å. The transformation 24a → 24b
+ PhI has ΔG = +13.0 kcal/mol, while 25a → 25b + PhI with
ΔG = +8.2 kcal/mol is more facile, as expected on the basis of
steric differences. Both processes are essentially barrierless.
Experimental support for the proposed intermediates 24b
and 25b was obtained by carrying out chemoselective
amination studies with a variety of different silver salts. If the
silver counteranion is bound to the metal under conditions
favoring aziridination, its identity is expected to influence the
chemoselectivity of the nitrene transfer. In contrast, the
selectivity under conditions favoring C−H amination should
show less of a response to the anion identity. Experiments
(Table S6 in the Supporting Information) supported this
hypothesis, as aziridination at low ligand loadings was
completely diverted to C−H amination when the silver
conducted, using analysis of the resulting mixtures of 21a−e
and 22a−e to infer differences in C−H amination rates. When
a 1:3 Ag:ligand ratio was employed, a linear correlation was
obtained using an equally weighted average of σ and σ+
parameters.22 Parameters that quantify radical involvement in
benzylic C−H functionalization, such as Jiang’s spin delocaliza-
tion constants, did not account for the strongly deactivating
effect of the p-CF3 group.23 The linear free energy correlation
revealed a ρ value of −0.58,24 which is consistent with
moderate positive charge buildup in the transition state during
C−H amination and would be expected for a concerted,
asynchronous process with a relatively early transition state. A
similar ρ value of −0.55 was obtained by Du Bois and co-
workers in their examination of Rh-catalyzed nitrene transfer,
which has also been attributed to concerted addition of a singlet
metal-nitrene.15j,1g An identical ρ value of −0.58 was obtained
when a 1:1.25 Ag:ligand ratio was employed (see Figure S1 in
the Supporting Information for complete details). However, the
possibility of a stepwise pathway still cannot be completely
ruled out, as these do not always result in Hammett plots
displaying concavity or nonlinear correlations.12a,c Nonetheless,
our experimental results again point to similar electronic
behavior between nitrenes formed from AgLOTf and AgL2OTf.
Computational Studies of Putative Nitrene Inter-
mediates. Silver nitrene species have been extensively
examined computationally by Per
́
ez15h and by us in previous
work.25 Related studies of copper nitrenes are also germane to
our work.26 The results of DFT and higher-level ab initio
(CASSCF) calculations implicate a triplet ground state for
silver nitrene species.25 In Rh2-catalyzed reactions, the viability
of a concerted mechanism has been traced to the requirement
for an empty N-centered orbital in the metal nitrene
intermediate, a requirement that is likely to be general and
thus necessitates a low-spin state.18b In contrast to Rh2
complexes, the electronic structure of the silver nitrene does
not provide the empty N-centered orbitals necessary to
D
DOI: 10.1021/acs.organomet.7b00190
Organometallics XXXX, XXX, XXX−XXX