Coordination-mode control of bound nitrile radical complex reactivity: Intercepting end-on nitrile-mo(III) radicals at low temperature

Article abstract of DOI:10.1021/ja905849a

Variable temperature equilibrium studies were used to derive thermodynamic data for formation of η¹ nitrile complexes with Mo(N[ ^tBu]Ar)₃, 1. (1-AdamantylCN = AdCN: ΔH^o = -6 ± 2 kcal mol^-1, ΔS^o = -20 ± 7 cal mol^-1 K^-1. C₆H₅CN = PhCN: ΔH^o = -14.5 ± 1.5 kcal mol^-1, ΔS ^o = -40 ± 5 cal mol^-1 K^-1. 2,4,6-(H ₃C)₃C₆H₂CN = MesCN: ΔH ^o = -15.4 ± 1.5 kcal mol^-1, ΔS^o = -52 ± 5 cal mol^-1 K^-1.) Solution calorimetric studies show that the enthalpy of formation of 1-[η²-NCNMe ₂] is more exothermic (ΔH^o = -22.0 ± 1.0 kcal mol^-1). Rate and activation parameters for η¹ binding of nitriles were measured by stopped flow kinetic studies (AdCN: ΔH _on^? = 5 ± 1 kcal mol^-1, ΔS_on^? = -28 ± 5 cal mol^-1 K^-1; PhCN: ΔH_on^? = 5.2 ± 0.2 kcal mol^-1, ΔS_on^? = -24 ± 1 cal mol^-1 K^-1; MesCN: ΔH _on^? = 5.0 ± 0.3 kcal mol^-1, ΔS_on^? = -26 ± 1 cal mol^-1 K^-1). Binding of Me₂NCN was observed to proceed by reversible formation of an intermediate complex 1-[η¹-NCNMe ₂] which subsequently forms 1-[η²-NCNMe₂]: ΔH^?_k1 = 6.4 ± 0.4 kcal mol ^-1, ΔS^?_k1 = -18 × 2 cal mol^-1 K^-1, and ΔH^?_k2 = 11.1 ± 0.2 kcal mol^-1, ΔS^? _k2 = -7.5 ± 0.8 cal mol^-1 K^-1. The oxidative addition of PhSSPh to 1-[η¹-NCPh] is a rapid second-order process with activation parameters: ΔH^? = 6.7 ± 0.6 kcal mol^-1, ΔS^? = -27 ± 4 cal mol^-1 K^-1. The oxidative addition of PhSSPh to 1-[η²-NCNMe₂] also followed a second-order rate law but was much slower: ΔH^? = 12.2 η 1.5 kcal mol^-1 and ΔS^? = -25.4 ± 5.0 cal mol^-1 K^-1. The crystal structure of 1-[η¹- NC(SPh)NMe₂] is reported. Trapping of in situ generated 1-[η¹-NCNMe₂] by PhSSPh was successful at low temperatures (-80 to -40°C) as studied by stopped flow methods. If 1-[η¹-NCNMe₂] is not intercepted before isomerization to 1-[η²-NCNMe₂] no oxidative addition occurs at low temperatures. The structures of key intermediates have been studied by density functional theory, confirming partial radical character of the carbon atom in η¹-bound nitriles. A complete reaction profile for reversible ligand binding, η¹ to η² isomerization, and oxidative addition of PhSSPh has been assembled and gives a clear picture of ligand reactivity as a function of hapticity in this system.

Full text of DOI:10.1021/ja905849a

Published on Web 10/01/2009
Coordination-Mode Control of Bound Nitrile Radical Complex
Reactivity: Intercepting End-on Nitrile-Mo(III) Radicals at Low
Temperature
Meaghan E. Germain,† Manuel Temprado,‡,§ Annie Castonguay,†
Olga P. Kryatova,† Elena V. Rybak-Akimova,*,† John J. Curley,‡ Arjun Mendiratta,‡
Yi-Chou Tsai,‡ Christopher C. Cummins,*,‡ Rajeev Prabhakar,*,§
James E. McDonough,§ and Carl D. Hoff*,§
Department of Chemistry, Tufts UniVersity, 62 Talbot AVenue, Medford, Massachusetts 02155,
Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts AVenue,
Cambridge, Massachusetts 02139, and Department of Chemistry, UniVersity of Miami, 1301
Memorial DriVe, Coral Gables, Florida 33146
Received July 20, 2009; E-mail: c.hoff@miami.edu
Abstract: Variable temperature equilibrium studies were used to derive thermodynamic data for formation
of η1 nitrile complexes with Mo(N[tBu]Ar)3, 1. (1-AdamantylCN ) AdCN: ∆Ho ) -6 ( 2 kcal mol-1, ∆So )
-20 ( 7 cal mol-1 K-1. C6H5CN ) PhCN: ∆Ho ) -14.5 ( 1.5 kcal mol-1, ∆So ) -40 ( 5 cal mol-1 K-1
.
2,4,6-(H3C)3C6H2CN ) MesCN: ∆Ho ) -15.4 ( 1.5 kcal mol-1, ∆So ) -52 ( 5 cal mol-1 K-1.) Solution
calorimetric studies show that the enthalpy of formation of 1-[η2-NCNMe2] is more exothermic (∆Ho ) -22.0
( 1.0 kcal mol-1). Rate and activation parameters for η1 binding of nitriles were measured by stopped flow
kinetic studies (AdCN: ∆Hon ) 5 ( 1 kcal mol-1, ∆Son ) -28 ( 5 cal mol-1 K-1; PhCN: ∆Hon ) 5.2 (
‡
‡
‡
0.2 kcal mol-1, ∆Son ) -24 ( 1 cal mol-1 K-1; MesCN: ∆Hon ) 5.0 ( 0.3 kcal mol-1, ∆Son ) -26 ( 1
‡
‡
‡
cal mol-1 K-1). Binding of Me2NCN was observed to proceed by reversible formation of an intermediate
complex 1-[η1-NCNMe2] which subsequently forms 1-[η2-NCNMe2]: ∆H‡k1 ) 6.4 ( 0.4 kcal mol-1, ∆S‡
)
k1
-18 ( 2 cal mol-1 K-1, and ∆H‡k2 ) 11.1 ( 0.2 kcal mol-1, ∆S‡k2 ) -7.5 ( 0.8 cal mol-1 K-1. The oxidative
addition of PhSSPh to 1-[η1-NCPh] is a rapid second-order process with activation parameters: ∆H‡ ) 6.7
( 0.6 kcal mol-1, ∆S‡ ) -27 ( 4 cal mol-1 K-1. The oxidative addition of PhSSPh to 1-[η2-NCNMe2] also
followed a second-order rate law but was much slower: ∆H‡ ) 12.2 ( 1.5 kcal mol-1 and ∆S‡ ) -25.4 (
5.0 cal mol-1 K-1. The crystal structure of 1-[η1-NC(SPh)NMe2] is reported. Trapping of in situ generated
1-[η1-NCNMe2] by PhSSPh was successful at low temperatures (-80 to -40 °C) as studied by stopped
flow methods. If 1-[η1-NCNMe2] is not intercepted before isomerization to 1-[η2-NCNMe2] no oxidative addition
occurs at low temperatures. The structures of key intermediates have been studied by density functional
theory, confirming partial radical character of the carbon atom in η1-bound nitriles. A complete reaction
profile for reversible ligand binding, η1 to η2 isomerization, and oxidative addition of PhSSPh has been
assembled and gives a clear picture of ligand reactivity as a function of hapticity in this system.
complished.2 Particularly for O2 as a ligand, it is recognized
Introduction
that both the coordination mode (e.g., η1 versus η2) and the
Coordination of a diatomic molecule to a single metal center
can be identified with two limiting modes: end-on (η1) or side-
on (η2), and the same is true for the coordination of certain
organic functional groups. In an early study of the coordination
chemistry of N2, Taube showed via isotopic labeling studies
that the end-on coordinated N2 ligand in [(η1-N2)Ru(NH3)5]2+
is able to exchange its terminal with its bound nitrogen, and
proposed that this process takes place via an unobserved side-
on intermediate, that is, [(η2-N2)Ru(NH3)5]2+.1 Characterization
using X-ray diffraction studies of metastable side-bound N2 and
NO ligands in mononuclear complexes has since been ac-
spin state (e.g., singlet versus triplet) may vary, giving rise to
four possibilities, each of which may be distinguished in terms
of the degree of metal to ligand charge transfer, as well as in
terms of characteristic ligand stretching frequency.3 Shan and
Que have detected a diiron(II,III) end-on bound superoxide
which rearranges to a diiron(III)-hydroperoxide.4 Shaik et al.
have investigated computationally the possibility of a single
oxidant with two spin states masquerading as two different
oxidants in the Cytochrome P-450 enzyme.5
In studies of O2 binding to biological copper sites, both CO
and MeCN have been used as models considered to be “redox-
(2) Coppens, P.; Novozhilova, I.; Kovalevsky, A. Chem. ReV. 2002, 102,
861.
† Tufts University.
‡ Massachusetts Institute of Technology.
§ University of Miami.
(3) de la Lande, A.; Ge´rard, H.; Moliner, V.; Izzet, G.; Reinaud, O.; Parisel,
O. J. Biol. Inorg. Chem. 2006, 11, 593.
(1) Armor, J. N.; Taube, H. J. Am. Chem. Soc. 1970, 92, 2560.
(4) Shan, X. P.; Que, L. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 5340.
9
15412 J. AM. CHEM. SOC. 2009, 131, 15412–15423
10.1021/ja905849a CCC: $40.75  2009 American Chemical Society

Intercepting End-on Nitrile-Mo(III) Radicals
A R T I C L E S
innocent”.3 In the particular case of MeCN, while η1 binding is
more common by far, description of an η2 binding mode was
documented by Wilkinson et al. in the classic organometallic
complex Cp2Mo(NCMe).6 Metal-activated organonitrile chem-
istry has received increasing study due to a myriad of catalytic
and stoichiometric reactions that have been discovered recently
in this area.7 Templeton has recently reported preparation8 and
subsequent reduction9 of η2 nitrile complexes of W(II) mono-
carbonyl bis(acetylacetonate). Parkin has reported the surpris-
ingly facile alkylation of (CptBu)2Mo(η2-NCMe).10 These and
other results have been taken to indicate possible increased
reactivity of nitriles in the η2 binding mode.11 The binding
properties of :NtN: are probably best approximated by
:NtC-R and thus the bond to nitriles provides a starting point
to understanding the bond to dinitrogen with the important
difference that variation of the R group in R-CtN: can provide
insight into the nature of the LnM-NtC-R bond. It should
be noted that some bonding motifs, such as “side-on-end-on”,12
that are available to N2 are probably not available to R-CtN:,
however for interrogating the nature of the M-N2 bond the
Mo-NtC-R analogy provides a tunable probe.
on complexes are more activated.18 However, opportunities to
examine reactivity in which both binding modalities can be
studied for the same metal complex are rare.
Despite the importance of the metal-nitrile bond there are
surprisingly few solution phase studies of bond energetics even
for bonding in the terminal mode to diamagnetic complexes.
We recently reported thermochemical and computational study
of bond energetics of a range of nitriles and other N donor
ligands to Mo(CO)3(P[iPr]3)2.16 While nitriles bound ∼6 kcal
mol-1 more strongly than N2, changes in R that caused
significant changes in IR frequencies and dramatic changes in
color resulted in no real difference in bond strength for the
nitriles in eq 1: ∆H ) 0.0 ( 0.4 kcal mol-1, AdCN )
1-adamantyl nitrile, Ar ) 4-NMe2C6H4; 2,4,6-Me3C6H2; 2,6-
F2C6H3; 4-CF3C6H4; F5C6.
The nature of the LnM-NtC-R bond is expected to depend
strongly on the nature of the metal complex and its spin state,
but changes in the R group of the nitrile can be equally important
in determining reactivity. Kukushkin and co-workers13 have
recently shown that changing from a conventional nitrile to a
“push-pull nitrile”,14 such as a dialkyl cyanamide, can com-
pletely alter the course of nucleophilic addition to Pt(II)
complexes. The lone pair of electrons on the dialkyl amino group
of R2N-CtN can become involved in the bonding and
reactivity of mixed dinitrogen organocyanamide complexes of
Mo(0) as reported also by Pombeiro and co-workers.15 Notably,
the crystal structures of both trans-[Mo(N2)(NCNEt2)(dppe)2]15
and W(P[iPr]3)2(CO)3(NCNMe2)16 have trigonal planar geometry
of the amide group.
It seems likely that the thermoneutral character of the Mo-
nitrile bond in Mo(CO)3(P[iPr]3)2(NCR) will not generalize to
other complexes. Paramagnetic nitrile complexes in which some
spin density is transferred to the nitrile ligand from the metal
might be expected to show more ligand dependent energetics.
To the authors’ knowledge, the first quantitative data for
binding of nitriles to a paramagnetic complex was reported by
Kovacs and co-workers19 for the iron complex
[FeIII(S2Me2N3(Et,Pr))]+ which is a model for nitrile hydratase.20
In that system, benzonitrile had a less favorable enthalpy of
binding (∆H ) -4.2 ( 0.6 kcal mol-1, ∆S ) -18 ( 3 cal
mol-1 K-1) than did acetonitrile (∆H ) -6.2 ( 0.2 kcal mol-1,
∆S ) -29.4 ( 0.8 cal mol-1 K-1). Grapperhaus has recently
characterized [LFe-NtC-Me]+, L ) N,N′-4,5-bis(2′-methyl-
2′-mercaptopropyl)-1-thia-4,7-diazacyclononane, as a new model
system for nitrile hydratase which binds acetonitrile with a Keq
comparable to that reported for [FeIII(S2Me2N3(Et,Pr))]+.21 Such
paramagnetic nitrile complexes are frequently observed to be
more labile than spin-paired analogues. Kinetic data on rates
of binding are scarce, as are structural data on complexes that
could be considered to be “bound radicals”.22
A final factor controlling binding and reactivity of nitriles is
the hapticity of the nitrile ligand itself. The conclusion discussed
earlier that some enhanced reactivity may be expected from η2
bound nitriles, certainly has precedent in N2 chemistry where
unique activation has been achieved in some systems utilizing
more complex binding motifs.17 Theoretical studies of switching
of end-on to side-on coordination in HCo(L)3 complexes predict
that hapticity can be controlled by ligand choice and that side
We have shown previously that nitriles may bind both
reversibly and in either an η1 or an η2 coordination mode to the
sterically hindered complex Mo(N[tBu]Ar)3 1, where Ar ) 3,5-
C6H3Me2 (Scheme 1).23 Furthermore, the same S ) /2 metal-
3
(5) Sharma, P. K.; de Visser, S. P.; Shaik, S. J. Am. Chem. Soc. 2003,
125, 8698.
loradical complex Mo(N[tBu]Ar)3 was shown to effect nitrile
reductive coupling via slow dimerization of the adduct (RCN)-
Mo(N[tBu]Ar)3, with C-C bond formation.23 In addition,
interception of the adduct (RCN)Mo(N[tBu]Ar)3 using X·
(6) Wright, T. C.; Wilkinson, G.; Motevalli, M.; Hursthouse, M. B.
J. Chem. Soc., Dalton Trans. 1986, 2017.
(7) Kukushkin, V. Y.; Pombeiro, J. L. Chem. ReV. 2002, 102, 1771.
(8) Jackson, A. B.; Schauer, C. K.; White, P. S.; Templeton, J. L. J. Am.
Chem. Soc. 2007, 129, 10628.
(9) Jackson, A. B.; Khosla, C.; Gaskins, H. E.; White, P. S.; Templeton,
J. L. Organometallics 2008, 27, 1322.
(17) Hidai, M.; Mizobe, Y. Chem. ReV. 1995, 95, 1115. Fryzuk, M. D.;
Johnson, S. A. Coord. Chem. ReV. 2000, 200-202, 379. MacKay,
B. A.; Fryzuk, M. D. Chem. ReV. 2004, 104, 385.
(10) Shin, J. H.; Savage, W.; Murphy, V. J.; Bonanno, J. B.; Churchill,
D. G.; Parkin, G. J. Chem. Soc., Dalton Trans. 2001, 1732.
(11) For additional discussion see ref 7, p 1792.
(18) Huo, C. F.; Zeng, T.; Li, Y. -W.; Beller, M.; Jiao, H. Organometallics
2005, 24, 6037.
(12) Fryzuk, M. D. Acc. Chem. Res. 2009, 42, 127.
(13) Gushchin, P. V.; Kuznetsov, M. L.; Haukka, M.; Wang, M. -J.;
Gribanov, A. V.; Kukushkin, V. Y. Inorg. Chem. 2009, 48, 2583.
(14) Le Questel, J. Y.; Berthelot, M.; Laurence, C. J. Phys. Org. Chem.
2000, 13, 347. Ziao, N.; Graton, J.; Laurence, C.; Le Questel, J. Y.
Acta Crystallogr., Sect. B 2001, B57, 850.
(19) Shearer, J.; Jackson, H. L.; Schweitzer, D.; Rittenberg, D. K.; Leavy,
T. M.; Kaminsky, W.; Scarrow, R. C.; Kovacs, J. A. J. Am. Chem.
Soc. 2002, 124, 11417.
(20) Kovacs, J. A. Chem. ReV. 2004, 104, 825.
(21) O’Toole, M. G.; Bennett, B.; Mashuta, M. S.; Grapperhaus, C. A.
Inorg. Chem. 2009, 48, 2300.
(15) Cunha, S. M. P. R. M.; Guedes da Silva, M. F. C.; Pombeiro, A. J. L.
Inorg. Chem. 2003, 42, 2157.
(22) Khusniyarov, M. M.; Weyhermür, T.; Bill, E.; Wieghardt, K. J. Am.
Chem. Soc. 2009, 131, 1208.
(16) Achord, P.; Fujita, E.; Muckerman, J. T.; Scott, B.; Fortman, G. C.;
Temprado, M.; Xiaochen, C.; Captain, B.; Isrow, D.; Weir, J. J.;
McDonough, J. E.; Hoff, C. D. Inorg. Chem. 2009, 48, 7891.
(23) Tsai, Y.-C.; Stephens, F. H.; Meyer, K.; Mendiratta, A.; Gheorghiu,
M. D.; Cummins, C. C. Organometallics 2003, 22, 2902.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15413

A R T I C L E S
Germain et al.
Scheme 1
Table 1. Thermodynamic (∆Ho in kcal mol-1 and ∆So in cal mol-1
K-1) and Spectroscopic Data (cm-1) for Binding of Nitriles to 1 in
Toluene Solution
donors such as Ar′EEAr′ (E ) S, Se, Te; Ar′ ) C6H5, C6F5) or
benzoyl peroxide allowed for generation of C-X bonds in
resultant ketimide complexes, R(X)CdN-Mo(N[tBu]Ar)3.24 It
also allowed for reductive cross-coupling reactions of benzoni-
trile with pyridine, and with CO2.25 Furthermore, when a
solution of PhCN-Mo(N[tBu]Ar)3 is treated with 1 equiv of
crystalline HMo(CO)3Cp or 0.5 equiv of H2Sn(n-Bu)2 isolation
of Ph(H)CN-Mo(N[tBu]Ar)3 can be achieved.26
1-[η1-NCPh]
1-[η1-NCMes]
1-[η1-NCAd]
∆Ho
-14.5 ( 1.5
-40 ( 5
2231
2035
-196
-15.4 ( 1.5
-52 ( 5
2220
2023
-197
-6 ( 2
-20 ( 7
2234
2079
-155
∆So
νCtN free
νCtNη1 bound
∆νCtN(η1)
Though originally formulated as an η2 complex based on its
EPR spectrum,23 the adduct between benzonitrile and 1 is now
known to bind as a terminal η1 complex.26,27 In contrast, the
crystal structure of the η2 adduct of 1 and dimethyl cyanamide
has been determined23 and leaves no doubt regarding its
hapticity. In the course of forming its η2 adduct, NtC-NMe2
would be expected to proceed via an initial η1 complex which
could be either more or less reactive than the final η2 adduct.
This work takes advantage of this unique situation to investigate
in detail the binding and reactivity of 1 with nitriles in which
different hapticities are accessible and can be kinetically resolved
for the same metal complex in reactions with radical coupling
partners, such as diphenyldisulfide (Scheme 1).
binding to 1 was made by Fourier transform-infrared (FT-IR)
spectroscopy. Spectroscopic data and van’t Hoff plots for
binding of these nitriles (Figures S1 and S2 in Supporting
Information) allow derivation of the thermodynamic and
spectroscopic parameters collected in Table 1.
Thermodynamics of Dimethylcyanamide Binding to 1. In
contrast to RCN (R ) Ph, Mes, Ad) it was observed previously23
that binding of dimethylcyanamide was quantitative at room
temperature and this allowed calorimetric determination of its
enthalpy of binding by two separate methods. The first was
direct measurement based on reaction 3.
-1
Results
∆H ) -21.6 ( 1.4 kcal mol
Mo(N[t-Bu]Ar)3 + Me2NCN
8
Spectroscopic and Equilibrium Studies of η1 Binding of
Nitriles to 1. Addition of excess R-C≡N (R ) Ph, C6H5; Mes,
3,5-(H3C)2C6H3; Ad, 1-adamantyl) to toluene solutions of 1 leads
to rapid establishment of the equilibrium shown in eq 2.
(η2-Me2NCN)Mo(N[t-Bu]Ar)3 (3)
The enthalpy of reaction 3 was measured using solid 1 as
limiting reagent and then corrected for the enthalpy of solution
of the complex to give the final derived value with all species
in toluene solution. Because this depended on the purity of the
highly air sensitive 1, a second approach was sought. Previously
the enthalpy of binding of adamantyl isonitrile was reported by
us (reaction 4).28
k1
Mo(N[t-Bu]Ar)3 + RCN y
\
z (η1-RCN)Mo(N[t-Bu]Ar)3
(2)
k-1
The equilibrium is characterized by a dramatic change in color
to deep purple and also produces a strong and broad infrared
absorption band in the range 2100-2000 cm-1 characteristic
of the η1 bound complexes. The νCtN band is shifted to
significantly lower wavenumbers (by 150-200 cm-1) upon
coordination to 1. Increasing the bulky nature of the nitrile shifts
the equilibrium to the reagents and thus, more RCN and lower
temperatures are required to achieve complete conversion to
1-[η1-NCR]. Because of the extremely weak binding in the case
of AdCN, concentrated solutions of nitrile ≈ 0.6 M were needed
to observe binding by 1 ≈ 0.02 M in toluene in the temperature
range 18 to -65 °C. Quantitative determination of Keq for RCN
-1
∆H ) -29.1 ( 0.4 kcal mol
Mo(N[t-Bu]Ar)3 + AdNC
8
(η1-AdNC)Mo(N[t-Bu]Ar)3 (4)
Investigation of displacement of Me2NCN by AdNC was
performed as shown in eq 5.
Subtraction of reaction 4 from reaction 5 yields reaction 3,
giving a value of ∆Ho ) -22.3 ( 1.0 kcal mol-1 for the
enthalpy of reaction 3. This is in good agreement with that
measured directly, and we adopt the average of these two
measurements, ∆Ho ) -22.0 ( 1.0 kcal mol-1, for the enthalpy
of binding of Me2NCN to 1 forming the η2-complex.
(24) Mendiratta, A.; Cummins, C. C.; Kryatova, O. P.; Rybak-Akimova,
E. V.; McDonough, J. E.; Hoff, C. D. Inorg. Chem. 2003, 42, 8621.
(25) Mendiratta, A.; Cummins, C. C. Inorg. Chem. 2005, 44, 7319.
(26) Temprado, M.; McDonough, J. E.; Mendiratta, A.; Tsai, Y.-C.;
Fortman, G. C.; Cummins, C. C.; Rybak-Akimova, E. V.; Hoff, C. D.
Inorg. Chem. 2008, 47, 9380.
(27) Mendiratta, A.; Cummins, C. C.; Kryatova, O. P.; Rybak-Akimova,
E. V.; McDonough, J. E.; Hoff, C. D. J. Am. Chem. Soc. 2006, 128,
4881.
(28) Stephens, F. H.; Figueroa, J. S.; Cummins, C. C.; Kryatova, O. P.;
Kryatov, S. V.; Rybak-Akimova, E. V.; McDonough, J. E.; Hoff, C. D.
Organometallics 2004, 23, 3126.
9
15414 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Intercepting End-on Nitrile-Mo(III) Radicals
(η2-Me2NCN)Mo(N[t-Bu]Ar)3 +
A R T I C L E S
-1
∆H ) -6.8 ( 0.6 kcal mol
AdNC
8
(η1-AdNC)Mo(N[t-Bu]Ar)3 + Me2NCN (5)
Kinetics of PhCN, MesCN, and AdCN Binding to 1. The rates
of binding of nitriles (PhCN, MesCN, AdCN) to 1 were
determined by stopped-flow kinetics by following the build-up
of the characteristic intense absorption bands of the nitrile
complexes in the visible region (Figure 1 and Figure S3). In
contrast to the more complex kinetic picture found for binding
of Me2NCN (discussed in a later section), the kinetic traces for
Figure 1. PhCN binding to 1 over t ) 1 s. [Mo(N[tBu]Ar)3] ) 0.15 mM,
[PhCN] ) 0.75 mM, -20 °C. Concentrations are listed after mixing in the
stopped-flow cell. For the absorption band of 1-[η1-NCPh], λ ) 560 nm,
ꢀ ) 3000 M-1 cm-1
.
binding could be fit to single exponential functions with kobs
)
k1[RCN] + k-1. Resolution of kobs into its k1 and k-1 components
was carried out graphically as shown in Figure 2 from plots of
kobs versus [RCN] which have slopes of k1 and intercepts of
k-1. Data for the slope (k1) is generally known to higher accuracy
than data for the intercept (k-1) as is typical in this type of
analysis.
Collected values derived from data in Figure 2, including
kinetic determination of Keq as a function of temperature by
this method are reported in Table S1. Activation parameters
were derived using Eyring plots as shown in Figure S4. Kinetic
data for these nitriles are summarized in Table 2.
Data in Table 2 can be used to calculate the enthalpies of
nitrile binding: PhCN (-8 kcal mol-1); MesCN (-4 kcal mol-1);
AdCN (-6 kcal mol-1) and entropies of nitrile binding PhCN
(-25 cal mol-1 K-1); MesCN (-10 cal mol-1 K-1); and AdCN
(-24 cal mol-1 K-1). These data, except for AdCN, are in only
poor agreement with FT-IR data reported earlier in Table 1
which are believed to be more accurate. The equilibrium
constant for binding of PhCN was independently determined
by UV-vis spectroscopy at room temperature by titration as
shown in Figure S5. The derived value of Keq ) 156 M-1 at 20
°C is in reasonable agreement with that determined by FT-IR
studies of 116 M-1. The value derived by the purely kinetic
method Keq ) k1/k-1 ) 45 M-1 is reasonable at this temperature
but diverges as the temperature decreases due to differences in
resolution of ∆H and ∆S. It should also be considered when
comparing data obtained by different techniques that the FT-
IR data was obtained generally in more concentrated solution
and at higher temperatures than the stopped flow data.
Kinetics of Dimethylcyanamide Binding to 1. Stopped-flow
kinetic studies of the reaction between 1 and Me2NCN gave
the first indication of the formation of a detectable η1-adduct
formed on route to the final η2-complex in keeping with the
general kinetic scheme shown in reactions 6 and 7:
Figure 2. Second-order rate plot for binding of PhCN to 1, at T ) -40
((),-20 (9), 5 (b), and 20 °C (2) with [Mo(N[tBu]Ar)3] ) 0.15 mM.
Table 2. Activation Parameters Derived from Eyring Plots, Rate
Constants, and Equilibrium Constants for RCN Adducts
PhCN
MesCN
AdCN
‡
∆Hon (kcal mol-1
)
5.2 ( 0.2
-24 ( 1
14 ( 1
5.0 ( 0.3
-26 ( 1
9 ( 1
-18 ( 5
193 ( 14
1.6 ( 0.1
121 ( 14
5 ( 1
-28 ( 5
11 ( 1
-4 ( 3
97b
‡
∆Son (cal mol-1 K-1
)
‡
∆Hoff (kcal mol-1
)
‡
∆Soff (cal mol-1 K-1
)
0
a
kon (M-1 s-1
)
468 ( 22
0.15 ( 0.3
3120 ( 22
a
koff (s-1
)
55b
a
Keq (M-1
)
2b
a Values reported for T ) -40 °C. b Values extrapolated for T )
-40 °C.
nm, ε ) 1000 M-1 cm-1). The absorbance spectrum of the η1-
coordinated complex was generated by computer subtraction
as shown in Figure S6.
Stopped-flow kinetic studies of the accumulation and decay
of the intermediate species were possible for temperatures
varying from -80 to 0 °C. Typical spectroscopic data and a
kinetic trace are shown in Figure 3 where the build-up and decay
of the intermediate complex at λ ) 456 nm is readily followed.
At 25 °C, the process of intermediate formation becomes too
fast and only its decay into the final η2-complex can be readily
observed. Also, the hydrolytic decomposition of complex 1 in
the stopped-flow cell became notable above 0 °C, and kinetic
data became noisy and poorly reproducible. Therefore, higher-
temperature results were excluded from the final analysis. The
decay of the absorption band with λ ) 456 nm and the growth
of the band with λ ) 650 nm occur on the same time scale. For
technical reasons, the kinetics of end-on to side-on conversion
of the cyanamide adduct were best followed at 456 nm.29
k1
Mo(N[t-Bu]Ar)3 + Me2NCN y\z
k-1
(η1-Me2NCN)Mo(N[t-Bu]Ar)3 (6)
k2
\z
(η1-Me2NCN)Mo(N[t-Bu]Ar)3 y
k-2
(η2-Me2NCN)Mo(N[t-Bu]Ar)3 (7)
The intermediate proposed to be 1-[η1-NCNMe2] with an intense
absorption at λ ) 456 nm (ε ) 5670 M-1 cm-1) rapidly
accumulates, and then undergoes a transformation into the final
product, with electronic transitions that are in agreement with
the previously characterized η2-coordinated complex (λ ) 650
(29) The main reason is that the absorbance was higher at this wavelength,
and the higher sensitivity of the detector allowed for dramatically better
signal-to-noise ratios on the kinetic traces. Analysis at the 650 nm
was found to be in agreement with the analysis at 456 nm-1
.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15415

A R T I C L E S
Germain et al.
Figure 4. Pseudo-first-order kinetics for 0.15 mM 1 and [Me2NCN] ranging
from 0.75 to 3.1 mM after mixing at -40 ((), -32 (b), and -28 °C (1).
Inset: 0 (9) and -80 °C (2).
Figure 3. Spectral changes and kinetic trace (at 456 nm) obtained upon
mixing at -60 °C toluene solutions of 1 (0.15 mM) and Me2NCN (1.5
mM). Concentrations are listed after mixing in a stopped-flow cell. Inset:
Kinetic trace at 456 nm.
Kinetic measurements were performed under pseudo-first-
order conditions (5-300-fold excess of Me2NCN) over a range
of temperatures and fit to the general two term rate law shown
in eq 8 which represents a general solution to the mechanism
in eqs 6 and 7.
1obst
2obst
rate ) ae-k + be-k
(8)
The exponential roots, k1obs and k2obs can be a complex function
of the three first-order rate constants (k-1, k2, and k-2) and the
pseudo-first-order rate constant k1[Me2NCN] in reactions 6 and
7.30 The change in relative magnitudes of the rate constants as
a function of temperature and nitrile concentration, as well as
the highly air and moisture sensitive nature of the reactants,
made this a challenging system to analyze. Room temperature
studies described later showed that ligand displacement and
oxidative addition reactions of 1-[η2-NCNMe2] occurred on a
time scale much slower than that observed in the stopped-flow.
From this it can be concluded that k-2 , k1, k-1, k2, and so
under stopped-flow conditions we assume k-2 ≈ 0.31
Figure 5. Saturation curves of k2obs dependence on [Me2NCN] (measured
from decay of 456 nm band) at -80 (2), -40 ((), -32 (b), and -28 °C
(1). Inset: 0 °C (9).
the increasing contribution of k-1 to k1obs as was seen earlier
for PhCN. As shown in the inset in Figure 4, at low temperature
the intercept can be taken as 0 within experimental error.
In addition, the observed rate constant for the second reaction
step, k2obs was found to be independent of Me2NCN concentra-
tion at lower temperatures. However, there is a dependence of
k2obs on nitrile concentration as shown in Figure 5 at tempera-
tures above -40 °C.33
Resolution of the data into the separate individual rate
constants at higher temperatures (T from -40 to 0 °C) was done
using the Specfit program.34,35 The combined rate data for both
low and high temperature regimes are combined in Table 3 and
show how the numerical variation of the rate data change as a
function of temperature. An important aspect of these data is
the increasing significance of k-1 relative to k2 with increasing
temperature. The greatest uncertainty in these data are the
estimates for k-1, particularly at -40 °C where its value is quite
small compared to k2. Owing to uncertainties in the exact
assessment of k-1, derivation of activation parameters as well
Fitting of kinetic data to general eq 8 was excellent and
showed negligible residual traces. Two limiting temperature
regimes became apparent during this work. At low temperature
(-80 to ca. -40 °C) the value of k-1, as discussed below, is
vanishingly small and no significant contribution to k1obs from
k-1 was observed.32 Thus, the reaction sequence in eqs 6 and 7
simplifies to two consecutive irreversible processes. In that case
the exponential roots k1obs and k2obs correspond simply to
k1[Me2NCN] and k2. At higher temperatures, (-40 to 0 °C) this
assumption is no longer valid since the reversible nature of the
first step was found to play a role in the kinetic sequence shown
in eqs 6 and 7, and more complex analysis was needed.
Separation of these two regimes is somewhat arbitrary; however,
the general behavior is illustrated in Figure 4 where plots of
k1obs versus nitrile concentration as a function of temperature
are displayed.
The increasing value of the intercept with increasing tem-
perature in the main plot and a 0 °C inset of Figure 4 is due to
(33) Despite the fact that k2obs depends on [NCNMe2], it was confirmed
separately by Specfit analysis that, as expected, k2 does not.
(34) SPECFIT Global Analysis System, version 2.11; Spectrum Software
Associates: Singapore.
(35) A number of other analytical methods were attempted, and they were
in qualitative agreement with the Specfit but the best overall quantita-
tive treatment of this data was done with this program. The qualitative
agreement between these methods indicated a relatively “soft” surface
for resolving the rate data, and the low temperature data is believed
to be most accurate.
(30) Capellos, C.; Bielski, B. H. J. Kinetic SystemssMathematical Descrip-
tion of Chemical Kinetics in Solution; R. E. Krieger: Huntington, NY,
1980.
(31) It is certainly true that k-2 , k2 since Keq strongly favors the η2 over
η1 binding mode in this complex.
(32) This division is arbitrary, but examination of data in Table 3 shows
that only above -40 °C does k-1 become significant relative to the
other rate constants.
9
15416 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Intercepting End-on Nitrile-Mo(III) Radicals
A R T I C L E S
Table 3. Directly Measured (-80 to -50 °C) and Derived (-40 to
0 °C) Rate Constantsa for Reactions 6 and 7
by AdNC (reaction 5) was determined calorimetrically to
indirectly derive the enthalpy of binding of Me2NCN to 1. In
addition, the rate of reaction 5 was conveniently followed by
FT-IR spectroscopy of either decay of free AdNC or build-up
of free Me2NCN as shown in Figure S9. In the presence of
excess AdNC excellent first-order kinetic plots were obtained
and the reaction was found to be first order in 1 and independent
of AdNC concentration provided it was present in modest
excess. Activation parameters, ∆H‡ ) 25.6 ( 0.6 kcal mol-1
and ∆S‡ ) 14.4 ( 2.1 cal mol-1 K-1, were derived from the
Eyring plot shown in Figure S10. These results are in agreement
with the enthalpy of nitrile binding of -22.0 ( 1.0 kcal mol-1
in 1-[η2-NCNMe2] (vide supra) and a mechanism in which
complete dissociation of Me2NCN is the rate determining step
followed by rapid trapping by AdNC.
temperature (°C)
k1 (M-1 s-1
)
k-1 (s-1
)
k2 (s-1
)
-80
-70
-60
-50
-40
-32
-28
0
52
55
0.03
0.1
0.5
1.7
4.2
11.4
16.1
179.9
145
361
708
879
1563
7606
0.001
0.08
4
110
a Data for k1 and k2 are accurate with errors of less than 10% of the
specific value. Derived values for k-1 are subject to greater errors and
considered as approximate as discussed in the text.
Rate of Oxidative Addition of PhSSPh to 1-[η1-NCPh]. The
rate of reaction 9 was investigated by stopped-flow kinetics using
techniques similar to those reported earlier for analogous
reactions of PhTeTePh.24
1
2
(η1-PhCN)Mo(N[t-Bu]Ar)3
+
PhSSPh
f
(η1-PhC(SPh)N)Mo(N[t-Bu]Ar)3 (9)
Figure 6. Eyring plot (-80 to 0 °C) for k1 and k2 of reactions 6 and 7
from which activation parameters ∆H‡k1 ) 6.4 ( 0.4 kcal mol-1, ∆S‡
-18 ( 2 cal mol-1 K-1, ∆H‡ ) 11.1 ( 0.2 kcal mol-1, and ∆S‡
)
)
Addition of PhSSPh to 1-[η1-NCPh] is a relatively rapid
process. Reaction of 1 with PhSSPh is slow as shown in Figure
S11 and was found to not interfere with the target oxidative
addition of PhSSPh to 1-[η1-NCPh]. Representative UV-vis
data for reaction 9 are shown in Figure S12.
k1
k2
k2
-7.5 ( 0.8 cal mol-1 K-1 are derived.
as estimates of Keq on a quantitative basis are not warranted. It
is clear from the data, however, that k-1 begins to play a kinetic
role above -40 °C.36
Formation of the ketimide was found to be a clean second-
order process (first order in 1-[η1-NCPh], and first order in
PhSSPh). The concentration dependence from [PhSSPh] at -10
°C under pseudo-first-order conditions gave a linear plot with
almost zero (4%) intercept as shown in Figure S13. Collected
rate constants in the temperature range -20 to +5 °C were used
to derive the activation parameters ∆H‡ ) 6.7 ( 0.6 kcal mol-1,
∆S‡ ) -27 ( 4 cal mol-1 K-1 as shown in Figure S14. The
enthalpy of activation of this reaction is significantly higher than
the value reported earlier for the Te-analogue, 1-[η1-NC-
In contrast to difficulties in quantitative assignment of k-1
,
the data for k1 and k2 over the entire range of -80 to 0 °C
present a consistent picture as summarized in Table 3. The
Eyring plot of the combined data shown in Figure 6 gives
additional confidence to the Specfit analysis of the high
temperature data since there is a seamless fit between the two
temperature regimes. From these data the following activation
‡
parameters are derived: ∆H1‡ ) 6.4 ( 0.4 kcal mol-1, ∆S1
)
‡
-18 ( 2 cal mol-1 K-1, ∆H2 ) 11.1 ( 0.2 kcal mol-1, and
‡
∆S2 ) -7.5 ( 0.8 cal mol-1 K-1
.
(TePh)Ph], ∆H‡ ) 3.8 kcal mol-1, ∆S‡ ) -29 cal mol-1 K-1
,
Low-Temperature FT-IR Study of Binding of Me2NCN to
1. To detect an η1 intermediate in the binding of Me2NCN to 1,
detected in stopped-flow UV-vis experiments described above,
the FT-IR spectrum of solutions of 1 and excess Me2NCN were
studied in the first 200 s of mixing at -75 ( 5 °C. As shown
in Figure S7, a weak band near 1570 cm-1 is formed initially
and assigned to the intermediate which is proposed to be 1-[η1-
NCNMe2] based on DFT calculations indicating a weak absor-
bance in this region as discussed later. As the 1570 cm-1 band
decreases, bands near 1650 and 1600 cm-1 increase and are
assigned to 1-[η2-NCNMe2]. A plot of ln[A] versus time for
these data give an apparent first order rate constant for this
interconversion of 0.022 s-1 as shown in Figure S8 in agreement
to the one determined by stopped-flow methods (Table S1)
confirming that the same process is being observed.
in agreement with higher S-S bond enthalpy.26 Solution
calorimetric data for reaction 9 were reported previously with
27
∆Ho ) -27.1 ( 0.3 kcal mol-1
.
Synthesis and X-ray Structure of 1-[η1-NC(SPh)NMe2]. Green
1-[η2-NCNMe2] reacts with PhSSPh as shown in eq 10 in an
analogous fashion to 1-[η1-NCPh],27 albeit at an attenuated rate.
1
2
(η2-Me2NCN)Mo(N[t-Bu]Ar)3
+
PhSSPh
f
(η1-Me2NC(SPh)N)Mo(N[t-Bu]Ar)3 (10)
The solid-state structure of 1-[η1-NC(SPh)NMe2] is shown
in Figure 7.
Rate of Ligand Displacement from 1-[η2-NCNMe2] by
AdNC Forming 1-[η1-CNAd]. As stated previously, the enthalpy
of the reaction of Me2NCN displacement from 1-[η2-NCNMe2]
The structure of 1-[η1-NC(SPh)NMe2] is similar to that of
1-[η1-NC(SPh)Ph]24 but the former contains a longer Mo-N
(1.8325 vs 1.801 Å) and a shorter NdC (1.281 vs 1.322 Å)
bond. The same differences were noticed for the calculated
geometries of 1-[η1-NCPh] and 1-[η1-Me2NCN] as discussed
later.
(36) At still higher temperatures, k-2 can be expected to contribute to the
complex rate law for equations 6 and 7 when all of the rate constants
are nontrivial.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15417

A R T I C L E S
Germain et al.
Figure 8. Spectral changes upon addition of PhSSPh to 1-[η1-NCNMe2]
at -80 °C over t ) 180 s. Concentrations after mixing: 15, 1.5, 0.15 mM
for PhSSPh, Me2NCN, and 1, respectively; growth and decay at λ ) 456
nm as well as growth at λ ) 500 and 636 nm. Inset: (A) kinetic traces at
λ ) 456 nm (solid line), λ ) 500 nm (dotted line), and 636 nm (dashed
line) with only Me2NCN added; (B) kinetic traces at λ ) 456 nm (solid
line), λ ) 500 nm (dotted line), and 636 nm (dashed line) with both PhSSPh
and Me2NCN added.
Figure 7. Thermal ellipsoid plot (50% probability) of 1-[η1-NC(SPh)NMe2].
Selected distances (Å) and angles (deg): Mo-N(4), 1.8325(15); N(4)-C(4),
1.281(2); C(4)-S(1), 1.775(2); C(4)-N(5), 1.410(3); Mo-Namide, 1.960
(average); Mo-N(4)-C(4), 161.30(15); N(4)-C(4)-N(5), 124.09(18).
Hydrogen atoms have been omitted for clarity.
of two peaks at λ ) 500 and 636 nm, which correspond to the
absorbance of the ketimide 1-[η1-NC(SPh)NMe2]. The kinetic
traces at these key wavelengths qualitatively describe the
differences with and without added PhSSPh. The competing
reaction, end-on to side-on isomerization of coordinated dim-
ethylcyanamide, results in the growth of the absorption band
with λ ) 650 nm (see above), which partially overlaps with
one of the absorption bands of the ketimide at 636 nm, but the
latter is narrower and more intense.
Rate and Thermochemistry of Oxidative Addition of
PhSSPh to 1-[η2-NCNMe2]. Owing to its slow rate, reaction 10
was investigated using conventional UV-vis techniques. The
reaction was run under pseudo-first-order conditions with a large
excess of PhSSPh in toluene solution at 15, 30, and 40 °C. Plots
of kobs versus [PhSSPh] shown in Figure S15 were found to be
linear with near zero intercept. An Eyring plot of these rate
data is shown in Figure S16 from which ∆H‡ ) 12.2 ( 1.5
kcal mol-1 and ∆S‡ ) -25.4 ( 5.0 cal mol-1 K-1 were derived.
The enthalpy of reaction 10 was measured using Calvet
calorimetry as ∆H ) -10.0 ( 0.5 kcal mol-1. The considerably
less exothermic nature of the oxidative addition of PhSSPh to
1-[η2-NCNMe2] compared to the PhCN derivative (∆Ho )
Trapping experiments follow the general reaction manifold
illustrated in Scheme 1. As expected, the initial rapid growth
of the absorbance at 456 nm does not depend on the presence
of PhSSPh, and its derived rate constant averaged over all
[PhSSPh], kobs ) 0.173 s-1, corresponds to the previously
characterized reaction between 1 and Me2NCN, kobs ) 0.176
-27.1 ( 0.3 kcal mol-1
)
27 is due in part to the greater stability
s-1
.
of 1-[η2-NCNMe2] compared to 1-[η1-NCPh] whose bond
strengths differ by 22.0 - 14.5 ) 7.5 kcal mol-1. However,
considerable difference in the stabilities of the ketimide
complexes formed must also exist based on the difference of
≈ 17 kcal mol-1 in enthalpies of formation.
A second-order rate plot for the growth of the ketimide
species was measured from the growth at λ ) 636 nm and yields
a rate constant of k ) 1.9 M-1 s-1, obtained from its slope as
shown at the bottom of Figure 9. Because both the η2 complex
and the ketimide product absorb at 636 nm, the intercept of the
line at the bottom of Figure 9 cannot be readily used to estimate
rate data owing to differences in the extinction coefficient for
the two products. The reaction could also be followed by the
decay of the absorption at λ ) 456 nm (Figure 9, top). The
second-order rate constant generated this way, k ) 1.9 ( 0.1
M-1 s-1 agrees well with the value measured at λ ) 636 nm
Low Temperature Stopped-Flow Study of Reaction of 1
and Me2NCN/PhEEPh Mixtures (E
) S, Se): In Situ
Trapping of 1-[η1-NCNMe2]. The observed build-up at -80 °C
of 1-[η1-NCNMe2] and its conversion in about 1 min to 1-[η2-
NCNMe2] suggested that it might be possible to trap the
proposed η1-intermediate prior to its conversion to the more
stable η2-isomer. Control experiments confirmed, as anticipated
from extrapolation to low temperature of the room temperature
experiments described in the previous section, that under
stopped-flow conditions no detectable reaction occurred between
1-[η2-NCNMe2] and PhSSPh over 5 min at -80 °C. In addition,
as stated before, the reaction of 1 with PhSSPh is slow and
was found to not interfere under the conditions employed with
the target trapping reaction between PhSSPh and 1-[η1-NC-
NMe2].
described above. The intercept on this plot (0.040 ( 0.002 s-1
)
corresponds to the rate of PhSSPh-independent conversion of
1-[η1-NCNMe2] into 1-[η2-NCNMe2] (0.04 s-1 at -80 °C, Table
S1).
Similar trapping experiments were performed under condi-
tions of excess PhSeSePh at -80 °C. This reaction proceeded
much more rapidly than with PhSSPh, and the appearance of
the 1-[η1-NCNMe2] intermediate was hard to detect by eye with
high excess of PhSeSePh. The final spectrum of 1-[η1-NC-
(SePh)NMe2] revealed two peaks that were slightly red-shifted
from those due to 1-[η1-NC(SPh)NMe2] at λ ) 510 and 642
nm (Figure S17). As predicted, the observed rates for the decay
of the intermediate at λ ) 456 nm as well as the formation of
the low energy band at λ ) 642 nm (kobs ) 0.17 s-1 with 7.5
A kinetic trace for the reaction of 1 and a mixture of
Me2NCN/PhSSPh at -80 °C over 180 s reaction time is shown
in Figure 8. Qualitatively, the spectral features for the overall
reaction include the growth of the 1-[η1-NCNMe2] intermediate
species at λ ) 456 nm, followed by its decay and the formation
9
15418 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Intercepting End-on Nitrile-Mo(III) Radicals
A R T I C L E S
Table 4. Computational Vibrational Data for Binding of Nitriles to
1. All Values in cm-1
calcuation level
νC≡N free
νC≡Nη1 bound
∆νC≡N(η1)
νC≡Nη2 bound
∆νC≡N(η2)
PhCN
2009
2118
BP86/LANL2Z
B3LYP/LANL2Z
2169
2262
160
144
1580
1678
589
584
Me2NCN
1618a
BP86/LANL2Z
B3LYP/LANL2Z
2168
2249
550
549
1585
1640
583
609
1700a
a The intensity of the band was computed to be 10 times weaker than
that for the corresponding PhCN derivative.
(MWB28 with an additional set of f functions for Mo) level of
the 1-[NCNMe2] and 1-[NCPh] adducts in the doublet state is
collected in Table S2. For 1-[η2-NCNMe2], the only compound
calculated for which the X-ray structure has been previously
determined, computed and experimental geometries are in good
agreement. The main difference between the computed structures
of 1-[η1-NCPh] and 1-[η1-NCNMe2] is the angle N-C-R.
Whereas the N-C-R moiety in 1-[η1-NCPh] is essentially
linear, the same arrangement in 1-[η1-NCNMe2] forms an angle
of 127.7°. In addition the Mo-N bonds are shorter in the
Me2NCN adducts. Since the Me2NCN ligand is smaller, it can
approach closer to the Mo and maximize the bonding interaction;
however, steric hindrance prohibits a closer approach of the
bulkier PhCN ligand. The amide N exhibits trigonal planar
geometry in 1-[NCNMe2] for both η1 and η2 conformers. This
behavior has been previously reported for other complexes
such as trans-[Mo(N2)(NCNEt2)(dppe)2],15 W(P[iPr]3)2(CO)3(Me2-
NCN),16 Cr(CO)5(NCNEt2),37 and trans-[Fe(Et2PCH2CH2PEt2)2-
(NCNEt2)2][BF4]2.38
DFT calculations at the BP86/6-311G(d,p) (MWB28 with an
additional set of f functions for Mo) level predict a ∆E0 ) -16.3
kcal mol-1 for formation of 1-[η1-NCPh] which compares to
the experimentally determined value of ∆H ) -14.5 ( 1.5 kcal
mol-1. For the formation of 1-[η2-NCPh], ∆E0 ) -10.3 kcal
mol-1, implying that for PhCN the η2 binding motif is less stable
than is the η1. For generation of 1-[η1-NCNMe2], ∆E0 ) -20.1
kcal mol-1 and for 1-[η2-NCNMe2], ∆E0 ) -23.2 kcal mol-1
for the formation of the η2-adduct. The computed energy for
1-[η1-NCPh] and 1-[η2-NCNMe2] is in good agreement with
the experimentally determined values. Calculations predict a
more stable η1 conformation for the PhCN adduct by 6 and 7.5
kcal mol-1 at the BP86 and B3LYP levels, respectively.
However, for the Me2NCN complex, the η2 was found to be
3.1 (BP86) and 2.3 (B3LYP) kcal mol-1 more stable than the
η1 isomer. The entropies were calculated to be -48, -57, -58,
and -61 cal mol-1 K-1 at the BP86/LANL2Z level for
formation of 1-[η1-NCPh], 1-[η2-NCPh], 1-[η1-NCNMe2], and
1-[η2-NCNMe2], respectively, from the metallic and ligand
fragments.
Figure 9. Differences in the formation and decay of the 1-[η1-NCNMe2]
complex as a function of [PhSSPh], top. Second-order rate plot for the
formation of 1-[η1-NC(SPh)NMe2], bottom.
mM PhSeSePh) were faster at higher concentration of Ph-
SeSePh. The reaction with PhSeSePh is similar to the reaction
with PhSSPh with respect to the spectral signature of the
resulting 1-[η1-NC(EPh)NMe2] species.
At -40 °C, the reaction proceeds with PhEEPh (E ) S, Se)
as seen at lower temperatures; however, the rates of the
formation of 1-[η1-NC(EPh)NMe2] and the conversion from
1-[η1-NCNMe2] to 1-[η2-NCNMe2] are more competitive.
Although the PhEEPh can trap the 1-[η1-NCNMe2] complex,
the conversion rate to the 1-[η2-NCNMe2] complex is more
rapid. To make trapping by PhEEPh more competitive its
concentration can be increased. This change can be qualitatively
observed as the λ of the low energy transition shifts from the
dominant 1-[η1-NC(SPh)NMe2] species at λ ) 636 nm to the
increasing appearance of the 1-[η2-NCNMe2] species at λ )
650 nm (Figure S18).
Finally, it was of interest to compare the reactivity of the
two in situ generated radicals 1-[η1-NCPh] and 1-[η1-NCNMe2]
toward oxidative addition by PhSSPh at low temperature. As
shown in Figure S19, under comparable conditions in which
1-[η1-NCNMe2] was found to react, no reaction of 1-[η1-NCPh]
and PhSSPh was observed (as predicted from the activation
parameters for this reaction described above) confirming the
much faster rate of oxidative addition of in situ generated 1-[η1-
NCNMe2] relative to 1-[η1-NCPh].
Computational Analysis of Binding of PhCN and Me2NCN
as both η1 and η2 Ligands to 1 in Both Doublet and Quartet
States. Density functional theory (DFT) calculations of the
1-[NCNMe2] and 1-[NCPh] adducts both as η1 and η2 species
in both doublet and quartet states were carried out in the current
study to gather knowledge about the structure and preferred
hapticity in these complexes and to further confirm the nature
of the observed intermediate in the stopped-flow experiments
as 1-[η1-NCNMe2]. Binding of nitriles in the quartet state was
found to be repulsive for all the species and leads to dissociation
of the ligand. Correspondingly, no minima were found in the
quartet state for any of the complexes explored. A summary of
selected bonds and angles calculated at the BP86/6-311G(d,p)
Table 4 collects the computational vibrational data for binding
of nitriles to 1 at the BP86/LANL2Z and B3LYP/LANL2Z
levels. The calculated IR frequencies are in good agreement to
those obtained experimentally (see Table 1).
The electronic transitions of 1-[η1-NCNMe2] and 1-[η2-
NCNMe2] were computed by time-dependent density functional
(37) Ohnet, M.-N.; Spasojevic-de Brie, A.; Dao, N. Q.; Schweiss, P.;
Braden, M.; Fischer, H.; Reindl, D. J. Chem. Soc., Dalton Trans. 1995,
665.
(38) Martins, L. M. R. D. S.; Frau´sto da Silva, J. J. R.; Pombeiro, A. J. L.;
Henderson, R. A.; Evans, D. J.; Benetollo, F.; Bombieri, G.; Michelin,
R. A. Inorg. Chim. Acta 1999, 291, 39.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15419

A R T I C L E S
Germain et al.
Table 5. Mulliken Spin Densities Calculated at the BP86/
6-311G(d,p) (MWB28 with an Additional Set of f Functions for Mo)
Level
atom
1-[η1-NCPh]
1-[η2-NCPh]
1-[η1-NCNMe2]
1-[η2-NCNMe2]
Mo
Nnitrile
Cnitrile
N(Me2)
+0.81
-0.08
+0.23
+0.86
+0.08
-0.06
+0.83
-0.04
+0.08
+0.09
+0.81
+0.10
-0.05
-0.01
Nanilides -0.02, -0.03, +0.04, +0.04, -0.03, <0.01, +0.06, +0.03,
+0.06 -0.3 <0.01 -0.01
Figure 11. SOMOs of 1-[η2-NCPh] (a) and 1-[η2-NCNMe2] (b) calculated
at the BP86/6-311G(d,p) (MWB28 with an additional set of f functions for
Mo) level.
Figure 10. SOMOs of 1-[η1-NCPh] (a) and 1-[η1-NCNMe2] (b) calculated
at the BP86/6-311G(d,p) (MWB28 with an additional set of f functions for
Mo) level.
theory (TD-DFT) calculations at the BP86/6-311G(d,p) (MWB28
with an additional set of f functions for Mo) level. The computed
transitions are red-shifted by ∼50 nm as values of λ ) 507 nm
(f ) 0.0190) and 706 nm (f ) 0.0060) were obtained for the η1
and η2 isomers, respectively. Whereas, the experimentally
determined values, by UV-vis spectroscopy, are λ ) 456 nm
Figure 12. HOMOs of 1-[η1-NCNMe2] (a) and 1-[η1-NCPh] (b) calculated
at the BP86/6-311G(d,p) (MWB28 with an additional set of f functions for
Mo) level.
(ε ) 5670 M-1 cm-1) and λ ) 650 nm (ε ) 1000 M-1 cm-1
)
for the η1 and η2 adducts. The agreement between the computed
and experimentally determined electronic transitions is reason-
ably good.
8% spin density on the C atom, and 9% spin density on the N
atom of the NMe2 group are more diffuse. Though the data in
Table 5 show that 1-[η1-NCNMe2] has a smaller Mulliken spin
density on C compared to 1-[η1-NCPh], the SOMO picture in
Figure 10 shows that for 1-[η1-NCNMe2] this reduced spin
density is in a more diffuse and sterically accessible orbital.
The calculated SOMOs for the η2 complexes shown in Figure
11 display similar shapes and electron distributions for both
1-[η2-NCPh] and 1-[η2-NCNMe2]. Aside from the large electron
density concentrated in a Mo d orbital, there is a significant
Mulliken spin density on the N atom; this appears to be in a
sterically more protected area. Most notable is the absence of
spin density on the C atom which is the reacting atom.
Finally, the computed contour diagrams for the HOMO
orbitals of both η1 complexes are shown in Figure 12. While
these are complex delocalized molecular orbitals, the lone pair
character of the rehybridized C atom of 1-[η1-NCNMe2] from
sp to sp2 is evident. This accounts for the differences in
computed vibrational spectra of the two complexes and severe
reduction to lower wavenumbers for 1-[η1-NCNMe2].
Mulliken spin densities calculated at the same level of theory
are collected in Table 5 and show that the spin densities are
mainly located in the metal for all complexes. For 1-[η1-NCPh]
a large contribution in the nitrile C (+0.23) was obtained. This
spin localization in the carbon accounts for the observed C
radical character and therefore the reactivity observed for the
complex.23-27 In addition, there is not a large delocalization of
the spin density in the ring in 1-[η1-NCPh] (<0.01 for all the C
but -0.03 (Cipso) and +0.01 (Cortho)). Furthermore, the Mulliken
spin densities calculated for C and N in CtN have opposite
signs in the η1 and η2 adducts.
The calculated singly occupied molecular orbitals (SOMOs)
for the η1 complexes shown in Figure 10 clearly contrast the
essentially linear nature of 1-[η1-NCPh] (which retains sp
hybridization at C) with the angular nature of 1-[η1-NCNMe2]
(which converts to sp2 hybridization at C). It should be noted
that the drawings in Figure 10 reflect the spatial extension of
orbital contours and not the net Mulliken spin density.
In Figure 10 the d-orbitals do not appear large due to their
relatively compact nature. In spite of that, the majority of the
spin density (>80%) resides in a relatively inaccessible Mo based
d orbital for both η1 complexes. For 1-[η1-NCPh] the 23% spin
density located on the C atom appears in a relatively compact
2p orbital and small residual contributions can be seen in the
arene ring and coordinated amides. For 1-[η1-NCNMe2], the
Discussion
The goal of this work was to determine the thermodynamic
and kinetic parameters distinguishing η1 and η2 binding of
nitriles to 1 and to characterize the radical-like reactivity of both
η1 and η2 nitrile complexes. Despite considerable previous
study23-27 of the reactivity of 1-[η1-NCPh] and report of the
structure of 1-[η2-NCNMe2]23 fundamental aspects of their
9
15420 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Intercepting End-on Nitrile-Mo(III) Radicals
A R T I C L E S
Table 6. Selected Activation Parameters and Thermodynamic
electronic rearrangements including spin pairing. A notable
exception to this ligand binding profile is •NO which binds at
a rate too fast to measure using conventional methods.42
Thermodynamic data in Table 6 span a much greater range
of binding enthalpies than are typically seen for nitrile binding
to low valent spin paired complexes such as Mo(CO)3(P[iPr]3)2
(see eq 1). While there is no doubt that steric factors play a
dominant role in ligand binding to 1, electronic factors may be
even more important. For the complexes 1-[η1-NCAd] and 1-[η1-
CNAd] the steric ligand profile must be similar, yet the
difference in enthalpies of binding, ≈ 24 kcal mol-1 is a much
larger difference than typically seen in comparative thermo-
chemistry of nitriles and isonitriles in binding to transition
metals.16,28,43
Computational studies played a key role in formulation of the
intermediate complex with λ ) 456 nm detected in stopped flow
studies as being 1-[η1-NCNMe2]. On the basis of the absence of a
nitrile band in the IR in the expected region 1800-2300 cm-1 in
spectroscopic studies done at -80 °C, it was considered possible
that this intermediate might be either a high-spin adduct or a
conformational isomer of the η2 complex. However, computations
indicated that none of the nitrile complexes were stable in high
spin (S ) 3/2) states and effectively ruled out that possibility. The
computed structure of 1-[η1-NCNMe2] was found to differ sig-
nificantly from that of 1-[η1-NCPh] in that for the “push-pull” nitrile
Me2NCN, involvement of the lone pair of electrons on the dimethyl
amine group cause rehybridization of the C atom to sp2 hybrid
and an even more drastic reduction in the bond order and stretching
frequency of the nitrile due to this effect abetted by donation of
electron density from 1. The computed UV-vis spectral data
support this as do computed vibrational data which are shifted close
to 1600 cm-1 to essentially a CdN area. The computed thermo-
dynamic bond strength of 1-[η1-NCNMe2] place it as being more
strongly held than PhCN. This surprising prediction was shown to
hold true in kinetic studies where the rate of dissociation of
Me2NCN from 1-[η1-NCNMe2] showed a smaller value for k-1
than that found for 1-[η1-NCPh]. The generally good agreement
between calculated and observed spectroscopic and thermodynamic
data for other complexes accessible to more extensive experimental
probing give confidence to assignment of the structure and
properties of the initial product of interaction as being 1-[η1-
NCNMe2].
Data for Nitrile Binding
1-[η1-NCPh]
1-[η1-NCMes] 1-[η1-NCAd] 1-[η1-NCNMe2] 1-[η1-CNAd]a
∆H‡
∆S‡
5.2 ( 0.2
-24 ( 1
5.0 ( 0.3
-26 ( 1
5 ( 1
6.4 ( 0.4
5.5 ( 0.5
-15
on
-28 ( 5 -18 ( 2
on
∆Ho -14.5 ( 1.5 -15.4 ( 1.5 -6 ( 2 -17 ( 3b -29.4 ( 0.4
∆So
-40 ( 5
-52 ( 5
-20 ( 7
a Data taken from ref 28. b This value is an estimate based on compu-
tational data and experimental observations, see text for discussion.
binding to form these two disparate bound radicals have not
been reported to date. In addition, one electron oxidation by
PhSSPh presents the opportunity to compare reaction mecha-
nisms for what in valence bond formalism might be viewed as
a Mo(IV)-bound nitrile radical complex in 1-[η1-NCPh] and a
Mo(V) metallocycle radical complex in 1-[η2-NCNMe2]. Ki-
netic, thermodynamic, and computational data for nitrile binding
and activation in initial η1 hapticity, reversible isomerization
to η2 hapticity, and oxidative addition by PhSSPh, according
to the general mechanism shown in Scheme 1, have been
determined and are discussed in separate sections below.
Formation and Properties of 1-[η1-NCR] Complexes. The
3
paramagnetic (S ) /2) complex 1 presents an unusual and
selective ligand binding pattern. Coordination of ligands to 1
generally involves pairing of the electrons on the metal, at least
partial electron transfer to the ligand, and geometrical rear-
rangements at the sterically demanding Mo(III) metal center.
As discussed ealier, there is a large shift to lower wavenumber
upon coordination of the coordinated nitrile stretching frequency.
Typical shifts of η1-coordinated nitriles are illustrated by the
work of Taube and co-workers39 who reported a shift to higher
wavenumber (∆ν ) +39 cm-1) for Ru(NH3)5(NCPh)3+ but to
lower wavenumber (∆ν ) -44 cm-1) for Ru(NH3)5(NCPh)2+
.
The observed shifts in nitrile stretching frequency for 1-[η1-
NCR] are in the same region as gas phase time-resolved (TR)-
IR spectroscopy assignments of νC≡N for photogenerated nitrile
radical anions40 and suggest significant electron transfer and
radical character associated with the bound nitrile ligand. The
arene nitriles, PhCN and MesCN, show more significant shifts
to lower energy than does AdCN; however, its shift is also
significant and implies occupation of the antibonding π* orbitals
of the nitrile.
Formation and Properties of the 1-[η2-NCR] Complexes.
Thermodynamic studies showed that the enthalpy of formation
of 1-[η2-NCNMe2] was approximately 7 kcal mol-1 more
exothermic than 1-[η1-NCPh]. Conventional kinetic studies
showed that ligand substitution reactions of 1-[η2-NCNMe2] by
AdNC had an enthalpy of activation of ∼25 kcal mol-1 and
occurred by a fully dissociative pathway, presumably through
formation first of 1-[η1-NCNMe2] followed by dissociation of
Me2NCN in this high activation enthalpy process. The enthalpy
Since catalytic reactions generally cannot occur faster than
rates of ligand binding and exchange, data in this area are always
essential to understanding chemical transformations. Combined
activation parameters and thermodynamic data for nitrile binding
are collected in Table 6.
The enthalpies of activation show surprisingly little variation.
Differences in rate constants appear to be controlled primarily
by entropies of activation which show larger variation than do
corresponding enthalpies of activation. Qualitatively this may
mean that the transition state for binding of AdNC for example,
the strongest ligand in this set, requires a less restrictive
arrangement of the ligand set on 1. Despite the fact that these
reactions are rapid and the enthalpy of activation is low, 5.5 (
0.5 kcal mol-1, it is not as low as those which can occur when
a donor ligand approaches a truly vacant site in a metal
complex.41 Contributions to the enthalpy of activation may be
due to steric rearrangements at the crowded metal center, or to
(41) Yang, H.; Kotz, K. T.; Asplund, M. C.; Wilkens, M. J.; Harris, C. B.
Acc. Chem. Res. 1999, 32, 551.
(42) The reason for the significantly faster rate of binding of this small,
paramagnetic ligand is not resolved. For rate data on binding of NO
see: Cherry, J. P. F.; Johnson, A. R.; Baraldo, L. M.; Tsai, Y. -C.;
Cummins, C. C.; Kryatov, S. V.; Rybak-Akimova, E. V.; Capps, K. B.;
Hoff, C. D.; Haar, C. M.; Nolan, S. P. J. Am. Chem. Soc. 2001, 123,
7271.
(43) These data imply that the isomerization reaction AdCN-1 f AdNC-1
should be thermodynamically favorable by ∼10 kcal mol-1, compared
to the same reaction for AdCN-Mo(CO)3(P[i-Pr]3)2 which is ap-
proximately thermoneutral,16 and the recently computed endothermic
enthalpies of this isomerization reported for main group compounds:
Timoshkin, A. Y.; Schaefer, H. F. J. Am. Chem. Soc. 2003, 125, 9998.
(39) Sekine, M.; Harman, W. D.; Taube, H. Inorg. Chem. 1988, 27, 3604.
(40) νCtN for the Me2NC6H4CN radical anion were located at 2096 cm-1
for the singlet state and 2040 cm-1 for the triplet state: Hashimoto,
M.; Hamaguchi, H J. Phys. Chem. 1995, 99, 7875.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15421

A R T I C L E S
Germain et al.
1-[η1-NCNMe2] and PhSSPh occurs relatively rapidly at -80
°C because 1-[η1-NCNMe2] can be trapped prior to its isomer-
ization to 1-[η2-NCNMe2] at low but not at high temperature.
The fact that isomerization, the k2 step in eq 7, has an enthalpy
of activation of 11 kcal mol-1 compared to the estimated 6 kcal
mol-1 enthalpy of activation for oxidative addition of PhSSPh
(which has a more unfavorable entropy of activation) has the
practical consequence that at high temperatures the rate of
isomerization will eclipse that of oxidative addition. Thus if
1-[η1-NCNMe2] is successfully trapped in its first pass through,
oxidative addition will be successful. However if this op-
portunity is missed and 1-[η1-NCNMe2] is allowed to isomerize
to 1-[η2-NCNMe2] then severe retardation of the oxidative
addition step will occur.
Comparing Reactivity of 1-[η1-NCR] and 1-[η2-NCR] Com-
plexes: Combined Reaction Profile. For the dimethyl cyanamide
system a combined potential energy diagram (enthalpy of reaction)
for all reactions studied here is assembled in Figure 13 which is
consistent with all experimental kinetic and computational data. It
is postulated that reactions of 1-[η2-NCNMe2] proceed through
initial formation of metastable 1-[η1-NCNMe2].
A key point in Figure 13 is “anchoring” of the data for the
enthalpy of binding of Me2NCN to 1 to form 1-[η1-NCNMe2].
The assigned enthalpy of 1-[η1-NCNMe2] at ∆H ) -17 ( 3
kcal mol-1 overlaps with kinetic and computational estimates
of this bond strength.46 All other data are derived from
experimental enthalpies of reaction and activation enthalpies.
The potential well stabilizing 1-[η1-NCNMe2], activation pa-
rameters for isomerization of 1-[η1-NCNMe2] to 1-[η2-NC-
NMe2], as well as binding and dissociation and oxidative
addition reactions are all derived and serve to present a
reasonably complete view of reactivity in this system. A
summary of key steps is presented below:
Figure 13. Combined potential energy diagram; enthalpies of reaction and
a
activation in kcal mol-1
.
Taken from ref 28.
of oxidative addition for PhSSPh reacting with 1-[η2-NCNMe2]
was lower than that (∼10 kcal mol-1) and is presumed to occur
by reversible formation of 1-[η1-NCNMe2] which then reacts
with PhSSPh.
Computational studies show remarkably similar structures for
1-[η2-NCR] for R ) Ph and NMe2. Presumably for steric
reasons, the η2 isomer is computed to be ∼6 kcal mol-1 less
stable than the η2 binding hapticity in the case R ) Ph. For the
smaller Me2NCN ligand the situation is reversed and the η2
isomer is computed to be ∼3 kcal mol-1 more stable. The fact
that this is probably a steric factor is supported by the
observation that for the complex between PhCN and “Mo(N[i-
Pr]Ar)3”44 the η2 binding hapticity is experimentally observed
to be the most stable structural isomer.
(i) Binding of Me2NCN, which involves a spin state change
3
1
from S ) /2 to S ) /2 and electron transfer from the metal to
the ligand occurs with an enthalpy of activation of 6.4 kcal mol-1
and is rapid even at low temperature.
(ii) Once formed, 1-[η1-NCNMe2] can convert to 1-[η2-
NCNMe2] which is a first order reaction with a (relatively) high
enthalpy of activation of 11.1 kcal mol-1. At low temperatures
this reaction is slow enough to allow trapping, but at higher
temperatures it gains in rate more rapidly than other processes
with lower enthalpies of activation. As a result the ability to
detect and/or trap 1-[η1-NCNMe2] is gone at high temperatures.
(iii) The rate of oxidative addition of PhSSPh to 1-[η1-
NCNMe2] (formed either in a low temperature trapping experi-
ment or a high temperature isomerization) is fast. It is faster
than the comparable reaction of 1-[η1-NCPh] in spite of the
latter complex having a higher spin density on the reactive C
atom. This is presumably due to the geometric change on
rehybridization which occurs for the “push-pull” nitrile.
Stopped flow kinetic studies allowed accurate assessment of
the rate of the isomerization reaction converting 1-[η1-NCNMe2]
to 1-[η2-NCNMe2]. Derived activation parameters for this
process are: ∆H‡ ) 11.1 ( 0.2 kcal mol-1, ∆S‡ ) -7.5 (
k2
k2
0.8 cal mol-1 K-1. The enthalpy of activation is a significant
fraction of the bond energy itself (assigned as 17 ( 3 kcal
mol-1) but does not have an extremely unfavorable entropy of
activation. Further study of the reaction mechanism for this
change in hapticity by computational methods is planned.45
Oxidative Addition of PhSSPh and Low-Temperature
Trapping of in Situ Generated 1-[η1-NCNMe2]. The most
surprising aspect of this work, that reaction of 1 with a mixture
of Me2NCN/PhSSPh will occur faster at -80 °C than at room
temperature, is fully consistent with the potential energy diagram
shown in Figure 13. The oxidative addition reaction between
(46) The kinetic estimates are qualitative and based on the observed input
of k-1 for dissociation of 1-[η1-NCNMe2] becoming significant in a
temperature range comparable to but somewhat higher than that of
PhCN. Assuming comparable entropies of binding, the enthalpy of
binding of 1-[η1-NCNMe2] would thus be expected to be somewhat
more favorable than -14.5 ( 1.5 kcal mol-1. It is clearly less favorable
than the 1-[η2-NCNMe2] complex at -22 kcal mol-1. Computational
estimates place 1-[η1-NCNMe2] approximately 3 kcal mol-1 above
1-[η2-NCNMe2] and hence would predict an enthalpy of binding of
1-[η1-NCNMe2] of -19 kcal mol-1. The assigned value of -17 ( 3
kcal mol-1 overlaps within experimental error both kinetic and
computational estimates.
(44) “Mo(N[i-Pr]Ar)3” is the isopropyl derivative which is actually a Mo(V)
hydride complex in which one C-H bond of an isopropyl group has
undergone oxidative addition at Mo. This complex serves as a
“masked” form of the Mo(III) complex, see: Tsai, Y.-C.; Johnson,
M. J. A.; Mindiola, D. J.; Cummins, C. C.; Klooster, W. T.; Koetzle,
T. F. J. Am. Chem. Soc. 1999, 121, 10426.
(45) Germain, M. E.; Temprado, M. Rybak-Akimova, E. V.; Cummins,
C. C.; Hoff, C. D, unpublished results.
9
15422 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Intercepting End-on Nitrile-Mo(III) Radicals
A R T I C L E S
Conclusions
density (>80%) remains located on the metal center for all the
complexes. Distribution of the remaining ∼20% spin density
to the other atoms depended strongly on the specific nitrile and
its hapticity. Despite having a lower spin density on the reactive
C atom, 1-[η1-NCNMe2] (8%) was found to react faster with
PhSSPh than 1-[η1-NCPh] (23%). The observed faster rate of
oxidation of 1-[η1-NCNMe2] compared to 1-[η1-NCPh] (despite
a lower spin density on the C atom) is ascribed to the
rehybridized nature of central carbon of the “push-pull” nitrile
bringing it closer to the transition state.
Successful detection, characterization, and interception of 1-[η1-
NCNMe2] has been achieved on a quantitative basis as shown in
Figure 13 which summarizes our best view of reactivity in this
system. The ability to intercept the η1 complex prior to its
isomerization was predicted in advance of observation and provides
hope that catalyst design by reaction engineering may be achieved
once a manifold of reactions are well understood on a quantitative
basis. A target of more interest would be the adduct formed between
dinitrogen and 1 ([(•)NN]δ-[Mo(N[tBu]Ar)3]δ+) which is a pre-
sumed intermediate on the reaction coordinate leading to
[µ-N2][Mo(N[tBu]Ar)3]2. The large differences in reactivity ob-
served for the simple nitrile complexes studied here suggest that
direct extrapolation of results to N2 is not justified, although general
principles of radical couplings might be applicable to dinitrogen
activation by transition metal complexes. For reactive paramagnetic
complexes no simple picture has emerged as yet. On the basis of
the large differences in reactivity of the bound nitrile radical
complexes studied, the extent to which results for nitriles can be
compared to dinitrogen presents a challenge. The work reported
here provides a solid basis for such comparison, and experimental
and computational investigation of dinitrogen and other N donor
ligands binding to Mo(N[tBu]Ar)3 is in progress.
This paper provides fundamental rate and energetic data for
binding of nitriles to a paramagnetic complex and reactivity of
the resulting “bound radical complex”. Nitrile binding to
Mo(N[tBu]Ar)3 does not correspond to simple association of
the lone pair of electrons with a vacant orbital on the metal
complex. The S ) 3/2 complex does not have a readily available
low energy orbital for binding and formation of an “inner
sphere” complex necessarily involves spin pairing at the metal
to form a vacant orbital. It also incorporates at least partial
electron transfer from the metal to nitrile to form what could
be viewed as
a bound radical nitrile anion complex
[R-C(•)N]δ-[Mo(N[tBu]Ar)3]δ+. Strong evidence for electron
transfer in these systems is found in the large shifts of the nitrile
stretch to lower frequency upon coordination (150-400 cm-1).
The large difference in the enthalpies of binding of AdCN
(-6 kcal/mol-1) and PhCN (-14.5 kcal mol-1) may be due to
steric and electronic factors related to more favorable charge
delocalization for the aromatic nitrile, PhCN. Surprisingly,
computations conclude that binding of Me2NCN to 1 would be
even stronger than binding of PhCN in a η1 mode. This is
attributed to active participation of the lone pair of electrons
on the Me2N group in the complex 1-[η1-NCNMe2] and
concomitant rehybridization of the C atom of the nitrile.
The rates and associated activation parameters were found
to be quite similar for the different nitriles studied in terms of
rates of ligand binding (k1 ) kon). Differences in Keq (k1/k-1
)
values for binding are due primarily to the large differences
found in rates of nitrile dissociation (k-1 ) koff). The rate of
conversion of the η1 and η2 complexes between 1 and Me2NCN
is slow enough that at low temperatures it may be detected and
also used as an in situ reagent which is orders of magnitude
more reactive than its η2 isomer.
Binding in an η2 mode is generally viewed as resulting in
“greater activation” of a ligand compared to η1 hapticity. In
agreement with that, computational structural data indicate that
the C-N bond of the coordinated nitrile does in fact undergo
more bond lengthening in the η2 mode of binding than for η1
for both benzonitrile and dimethyl cyanamide. In terms of one
electron oxidation by PhSSPh, however, it is the η1 complex
that is most reactive. The complex with the more lengthened
C-N bond will be more reactiVe only if it is closer to the
transition state for the desired reaction. In the one electron
oxidation reaction studied here, the 1-[η1-NCNMe2] was found
to be sufficiently reactive that, provided it could be intercepted
before forming the more stable η2 derivative, oxidative addition
occurred within seconds at -80 °C. If allowed to isomerize to
the more thermodynamically stable η2 isomer, reactivity ef-
fectively ceases.
Acknowledgment. Support of this work by the National Science
Foundation (Grant No. CHE 0615743, (CDH) CHE 0750140
(ERA); CHE 0719157 (CCC) is gratefully acknowledged. M.T.
thanks the Spanish Ministry of Science and Innovation and the
Fulbright commission under Fellowship FU-2006-1738. The rapid
kinetics instrumentation at Tufts was supported by the NSF (CRIF
program, Grant CHE 0639138). The authors also thank Prof. Obis
D. Castaño (Universidad de Alcalá de Henares) for help in
computational chemistry and Jared Silvia (MIT) for help in
structural data analysis.
Supporting Information Available: Information on prepara-
tion of complexes and standard techniques used in physical
measurements; experimental procedure description, spectro-
scopic, kinetic, thermodynamic and computational Figures and
Tables, and CIF file for crystal structure determination of 1-[η1-
NC(SPh)NMe2]. This material is available free of charge via
the Internet at http://pubs.acs.org.
Differences in reactivity cannot be assigned in a simple way
to differences in the amount of spin density on the reactive C
atom of the bound nitrile. Computational results showed that
for the highly delocalized SOMOs the majority of the spin
JA905849A
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15423