M=Nα Cycloaddition and Nα-N β Insertion in the reactions of titanium hydrazido compounds with alkynes: A combined experimental and computational study

Article abstract of DOI:10.1021/ja1036513

A combined experimental and DFT study of the reactions of diamide-amine supported titanium hydrazides with alkynes is presented. Reaction of Ti(N ₂N^py)(NNPh₂)(py) (1, N₂N ^py = (2-NC₅H₄)CMe(CH₂NSiMe ₃)₂) with terminal and internal aryl alkynes ArCCR (Ar = Ph or substituted phenyl, R = Me or H) at room temperature gave the fully authenticated azatitanacyclobutenes Ti(N₂N^py){N(NPh ₂)C(R)CAr} via ArCCR [2 + 2] cycloaddition to the Ti=N _α bond of the hydrazide ligand. In contrast, reaction of 1 with PhCCMe at 60 °C, or of Ti(N₂N^Me)(NNPh ₂)(py) (11, N₂N^Me = MeN(CH₂CH ₂NSiMe₃)₂) with RCCMe (R = Me, Ph or substituted phenyl) at room temperature or below, gave vinyl imido compounds of the type Ti(N₂N^R′){NC(R)C(Me)NPh₂}(py), in which RCCMe had undergone net insertion into the N_α-N _β bond. These are the first examples of this type of reaction for any metal hydrazide. The reaction of 11 with PhCCMe had the activation parameters ΔH^? = 18.8(4) kcal mol^-1, ΔS^? = 1(1) cal mol^-1 K^-1 and ΔG298? = 18.5(7) kcal mol^-1. Mechanistic and DFT studies for 1 and 11 found that the N_α-N_β insertion event is preceded by alkyne cycloaddition to Ti=N_α, and that N_α-N_β bond insertion is really an intramolecular N_α atom migration process within the azatitanacyclobutenes following intramolecular chelation of NPh₂ of the hydrazide ligand. Electron-withdrawing aryl groups on ArCCMe stabilize the azatitanacyclobutenes and also promote a specific regiochemistry (ArC carbon bound to Ti). This in turn defines the regiochemistry of the overall N _α-N_β insertion reaction (ArC carbon bound to N_α). In contrast, electron-releasing aryl groups promote the final N_α migration stage of the mechanism, and a Hammett analysis of the rates of insertion of (4-C₆H₄X)CCMe into the N_α-N_β bond of 11 found a reaction constant, ρ, of -0.74(5), consistent with NPA charge changes of ArC along the DFT reaction coordinate.

Full text of DOI:10.1021/ja1036513

Published on Web 07/09/2010
MdNr Cycloaddition and Nr-Nꢀ Insertion in the Reactions of
Titanium Hydrazido Compounds with Alkynes: A Combined
Experimental and Computational Study
A. Daniel Schofield,† Ainara Nova,‡ Jonathan D. Selby,† Catherine D. Manley,†
Andrew D. Schwarz,† Eric Clot,*,‡ and Philip Mountford*,†
Chemistry Research Laboratory, Department of Chemistry, UniVersity of Oxford, Mansfield
Road, Oxford OX1 3TA, U.K., and Institut Charles Gerhardt, UniVersite´ Montpellier 2, CNRS
5253, cc 1501, Place Euge`ne Bataillon, 34095 Montpellier Cedex 5, France
Received April 29, 2010; E-mail: eric.clot@univ-montp2.fr; philip.mountford@chem.ox.ac.uk
Abstract: A combined experimental and DFT study of the reactions of diamide-amine supported titanium
hydrazides with alkynes is presented. Reaction of Ti(N2Npy)(NNPh2)(py) (1, N2Npy
)
(2-
NC5H4)CMe(CH2NSiMe3)2) with terminal and internal aryl alkynes ArCCR (Ar ) Ph or substituted phenyl,
Me or H) at room temperature gave the fully authenticated azatitanacyclobutenes
Ti(N2Npy){N(NPh2)C(R)CAr} via ArCCR [2 + 2] cycloaddition to the TidNR bond of the hydrazide ligand. In
contrast, reaction of
with PhCCMe at 60 °C, or of Ti(N2NMe)(NNPh2)(py) (11, N2NMe
R
)
1
)
MeN(CH2CH2NSiMe3)2) with RCCMe (R ) Me, Ph or substituted phenyl) at room temperature or below,
gave vinyl imido compounds of the type Ti(N2NR′){NC(R)C(Me)NPh2}(py), in which RCCMe had undergone
net insertion into the NR-Nꢀ bond. These are the first examples of this type of reaction for any metal
hydrazide. The reaction of 11 with PhCCMe had the activation parameters ∆Hq ) 18.8(4) kcal mol-1, ∆Sq
) 1(1) cal mol-1 K-1 and ∆G2q98 ) 18.5(7) kcal mol-1. Mechanistic and DFT studies for 1 and 11 found that
the NR-Nꢀ insertion event is preceded by alkyne cycloaddition to TidNR, and that NR-Nꢀ bond “insertion”
is really an intramolecular NR atom migration process within the azatitanacyclobutenes following intramo-
lecular chelation of NPh2 of the hydrazide ligand. Electron-withdrawing aryl groups on ArCCMe stabilize
the azatitanacyclobutenes and also promote a specific regiochemistry (ArC carbon bound to Ti). This in
turn defines the regiochemistry of the overall NR-Nꢀ insertion reaction (ArC carbon bound to NR). In contrast,
electron-releasing aryl groups promote the final NR migration stage of the mechanism, and a Hammett
analysis of the rates of insertion of (4-C6H4X)CCMe into the NR-Nꢀ bond of 11 found a reaction constant,
F, of -0.74(5), consistent with NPA charge changes of ArC along the DFT reaction coordinate.
Introduction
systems involve (L)MdNNH2 intermediates immediately prior
to the key NR-Nꢀ bond-cleavage step, in agreement with
Terminal transition metal hydrazide(2-) or isodiazene(0)
compounds (L)MdNNR2 (R ) H, alkyl or aryl) have been of
ongoing interest for over 30 years.1-3 Initial studies focused
on molybdenum and tungsten derivatives in the context of the
biological conversion of dinitrogen to ammonia.4-6 Both the
stoichiometric Chatt7-10 and catalytic Schrock11,12 model
computational studies.13-17 Considerable effort has also been
spent on trying to use midtransition metal complexes of the type
(L)MdNNR2 as reagents or intermediates for the synthesis of
organo-nitrogen products.4,18,19 However, MdNNR2 functional
group reactivity is minimal for these metals, being characterized
only by transformations involving the Nꢀ atom, namely pro-
tonation or reactions involving Nꢀ-H.18,20,21 In particular,
reactions at the MdNR bond of the NNR2 group are not seen,
and NR-Nꢀ bond cleavage only occurs under forcing conditions
using external reductants, for example LiAlH4. Structural,22,23
† University of Oxford.
‡ Universite´ Montpellier 2.
(1) Johnson, B. F. G.; Haymore, B. L.; Dilworth, J. D. In ComprehensiVe
Coordination Chemistry; Wilkinson, G., Gillard, R. D., McCleverty,
J. A., Eds.; Pergamon Press: Oxford, 1987; Vol. 2, p 100.
(2) Nugent, W. A.; Mayer, J. M. Metal-Ligand Multiple Bonds; Wiley-
Interscience: New York, 1988.
(13) Stephan, G. C.; Sivasankar, C.; Studt, F.; Tuczek, F. Chem.sEur. J.
2008, 14, 644.
(3) Mindiola, D. J. Angew. Chem., Int. Ed. 2008, 47, 1557.
(4) Fryzuk, M. D. Chem. Rec. 2003, 3, 2.
(14) Studt, F.; Tuczek, F. Angew. Chem., Int. Ed. 2005, 44, 5639.
(15) Schrock, R. R. Angew. Chem., Int. Ed. 2008, 47, 5512.
(16) Schenk, S.; Kirchner, B.; Reiher, M. Chem.sEur. J. 2009, 15, 5073.
(17) Schenk, S.; Le Guennic, B.; Kirchner, B.; Reiher, M. Inorg. Chem.
2008, 47, 3634.
(5) MacKay, B. A.; Fryzuk, M. D. Chem. ReV. 2004, 104, 385.
(6) Kozak, C. M.; Mountford, P. Angew. Chem., Int. Ed. 2004, 43, 1186.
(7) Chatt, J.; Pearman, A. J.; Richards, R. L. Nature 1975, 253, 39.
(8) Pickett, C. J.; Leigh, G. J. J. Chem. Soc., Chem. Commun. 1981, 1033.
(9) Hussain, W.; Leigh, G. J.; Pickett, C. J. J. Chem. Soc., Chem. Commun.
1982, 747.
(18) Hidai, M. Coord. Chem. ReV. 1999, 185-186, 99.
(19) Ohki, Y.; Fryzuk, M. D. Angew. Chem., Int. Ed. 2007, 46, 3180.
(20) Fryzuk, M. D.; Johnson, S. A. Coord. Chem. ReV. 2000, 200-202,
379.
(10) Pickett, C. J.; Talamarin, J. Nature 1985, 317, 652.
(11) Yandulov, D. V.; Schrock, R. R. Science 2003, 76.
(12) Schrock, R. R. Acc. Chem. Res. 2005, 38, 955.
(21) Hidai, M. Chem. ReV. 1995, 95, 1115.
9
10484 J. AM. CHEM. SOC. 2010, 132, 10484–10497
10.1021/ja1036513  2010 American Chemical Society

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
Figure 1. Examples of products formed between terminal Group 4 hydrazides and organic substrates. The atoms of the original MdNNR2 group are shown
in red; those of the substrate are in blue.
spectroscopic,24 and DFT computational13,14,25,26 data explain
this: for mid to later metals “NNR2” is best viewed as a neutral
isodiazene :NdNR2 (formally (L)Mr:NdNR2). The more
dative (donor/acceptor) nature of the Mr:NR bond and multiple
bond character of NRdNꢀ reduces the intrinsic reactivity at these
two sites.
reaction products. We note that N-N bond cleavage in a
hafnium complex has been shown to occur upon addition of
CO.59
Given the emerging richness and significance of early
transition metal hydrazide chemistry, it is now important to
understand the relationship between MdNR cycloaddition and
NR-Nꢀ cleavage/insertion, and in particular the mechanism(s)
of the latter. Although there has been some speculation regarding
Although a handful of reports of terminal hydrazides of
titanium27,28 and zirconium29 appeared some time ago, the first
crystallographically characterized titanium example only
appeared in 2004.30 Until this time, Bergman’s
Cp2Zr(NNPh2)(DMAP)29 was the sole structurally authenticated
example. In contrast to their later metal counterparts, Group 4
MdNNR2 functional groups show very rich and diverse
reactivity.3 Recent DFT calculations31-34 and a growing pool
of structural data22,23 show that the NNR2 ligands for these
metals should be viewed as hydrazide(2-) moieties. The MdNR
bonding (formally a σ2π4 triple bond except for metallocene
and related “π-loaded” systems) is comparable to that in the
corresponding Group 4 imides (L)MdNR (R ) alkyl, aryl),35-37
except that one of the M-NR π bonds is considerably
destabilized by interaction with the Nꢀ atom lone pair (in
principle making it more reactive). The hydrazide MdNR bond
lengths are comparable to those in Group 4 aryl imides. On the
other hand, the NR-Nꢀ bond is substantially lengthened
compared to Group 6 (L)MdNNR2 complexes as expected from
the [N-NR2]2- (hydrazide(2-)) vs NdNR2 (isodiazene) rep-
resentations, respectively. One therefore expects quite different
reactivity patterns between early and later metal hydrazide/
isodiazene systems, in much same way that early and later metal
alkylidene/carbene systems differ.2,38
While synthetic routes to isolable Group 4 hydrazides have
steadily become established,27–34,39–45 their stoichiometric and
catalytic reaction chemistry has only been developed during the
past few years, and is still very much in its infancy. This is in
stark contrast to the position for Group 4 imides, themselves
pivotal in the development of metal-heteroatom multiple bonds
in general, and which have been extensively studied in terms
of reactions, applications, mechanisms and bonding.46–53 Above
all, the combination of an unsaturated, polar MdNR bond and
a lengthened and reduced NR-Nꢀ bond confers divergent and
unique reactivity on the Group 4 MdN-NR2 functional group
which undergoes two distinct types of reaction. These apparently
involve either reaction at the MdNR28,44,45,54–58 bond (cycload-
dition or insertion), or cleavage or insertion reactions at the
NR-Nꢀ bond.29,41,42,54,55 Figure 1 shows examples of these
(22) Allen, F. H.; Kennard, O. Chem. Des. Autom. News 1993, 8, 1&31.
(23) Fletcher, D. A.; McMeeking, R. F.; Parkin, D. The United Kingdom
Chemical Database Service. J. Chem. Inf. Comput. Sci. 1996, 36, 746;
The UK Chemical Database Service: CSD version 5.31, updated May
2010.
(24) Lehnert, N.; Tuczek, F. Inorg. Chem. 1999, 38, 1659.
(25) Mersmann, K.; Horn, K. H.; Bores, N.; Lehnert, N.; Studt, F.; Paulat,
F.; Peters, G.; Ivanovic-Burmazovic, I.; van Eldik, R.; Tuczek, F.
Inorg. Chem. 2005, 44, 3031.
(26) Lehnert, N.; Tuczek, F. Inorg. Chem. 1999, 38, 1671.
(27) Wiberg, N.; Haring, H.-W.; Huttner, G.; Friedrich, P. Chem. Ber. 1978,
111, 2708.
(28) Blake, A. J.; McInnes, J. M.; Mountford, P.; Nikonov, G. I.; Swallow,
D.; Watkin, D. J. J. Chem. Soc., Dalton Trans. 1999, 379.
(29) Walsh, P. J.; Carney, M. J.; Bergman, R. G. J. Am. Chem. Soc. 1991,
113, 6343.
(30) Li, Y.; Shi, Y.; Odom, A. L. J. Am. Chem. Soc. 2004, 126, 1794.
(31) Parsons, T. B.; Hazari, N.; Cowley, A. R.; Green, J. C.; Mountford,
P. Inorg. Chem. 2005, 44, 8442.
(32) Patel, S.; Li, Y.; Odom, A. L. Inorg. Chem. 2007, 46, 6373.
(33) Herrmann, H.; Fillol, J. L.; Gehrmann, T.; Enders, M.; Wadepohl,
H.; Gade, L. H. Chem.sEur. J. 2008, 14, 8131.
(34) Selby, J. D.; Manley, C. D.; Schwarz, A. D.; Clot, E.; Mountford, P.
Organometallics 2008, 27, 6479.
(35) Kaltsoyannis, N.; Mountford, P. J. Chem. Soc., Dalton Trans. 1999,
781.
(36) Cundari, T. R. Chem. ReV. 2000, 100, 807.
(37) Bolton, P. D.; Clot, E.; Cowley, A. R.; Mountford, P. J. Am. Chem.
Soc. 2006, 128, 15005.
(38) Schrock, R. R. J. Chem. Soc., Dalton Trans. 2001, 2541.
(39) Thorman, J. L.; Woo, L. K. Inorg. Chem. 2000, 39, 1301.
(40) Banerjee, S.; Odom, A. L. Organometallics 2006, 25, 3099.
(41) Selby, J. D.; Manley, C. D.; Feliz, M.; Schwarz, A. D.; Clot, E.;
Mountford, P. Chem. Commun. 2007, 4937.
(42) Herrmann, H.; Fillol, J. L.; Wadepohl, H.; Gade, L. H. Angew. Chem.,
Int. Ed. 2007, 46, 8426.
(43) Herrmann, H.; Wadepohl, H.; Gade, L. H. Dalton Trans. 2008, 2111.
(44) Clulow, A. J.; Selby, J. D.; Cushion, M. G.; Schwarz, A. D.;
Mountford, P. Inorg. Chem. 2008, 47, 12049.
(45) Weitershaus, K.; Wadepohl, H.; Gade, L. H. Organometallics 2009,
28, 3381.
(46) Wigley, D. E. Prog. Inorg. Chem. 1994, 42, 239.
(47) Mountford, P. Chem. Commun. 1997, 2127.
(48) Gade, L. H.; Mountford, P. Coord. Chem. ReV. 2001, 216-217, 65.
(49) Duncan, A. P.; Bergman, R. G. Chem. Rec. 2002, 2, 431.
(50) Odom, A. L. Dalton Trans. 2005, 225.
(51) Hazari, N.; Mountford, P. Acc. Chem. Res. 2005, 38, 839.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10485

A R T I C L E S
Schofield et al.
the mechanism of the NR-Nꢀ bond cleavage/insertion reactions
of Group 4 hydrazido3,42,55 and related60 compounds, no detailed
study has yet emerged. We have used a combination of
synthetic, mechanistic and DFT methods to understand the
intriguing dichotomy between MdNR and NR-Nꢀ bond reactiv-
ity of certain titanium hydrazides with alkynes as representative
substrates. The conclusions drawn regarding these specific
compounds and substrates can be extrapolated to early transition
metal hydrazide reactivity in general.
Synthesis and Mechanistic Studies
The metal-catalyzed reactions of alkynes with hydrazines in
general are important and topical from an organic synthesis point
of view, as variations of the well-established hydroamination
reaction with amines RNH2 (R ) alkyl or aryl).50,61–63 Berg-
man,64 Odom30,65,66 and others45,67–69 have found that a range
of titanium compounds catalyze the hydrohydrazination of
allenes or alkynes with hydrazines (N-H bond addition across
CtC). It is proposed50 that the key step in these reactions (and
in the related iminohydrazination reaction40,70) involves [2 +
2] cycloaddition to a transient hydrazido intermediate forming
azametallacycles, but as yet nothing is known about their
electronic and molecular structures, and they have only been
characterized spectroscopically in the last 2 years.45,54 Since
the principal synthetic aspects of this work have been
communicated,41,54 only a brief discussion is given here,
focusing in particular on the main features of scope, regio-
chemistry, and stability.
Figure 2. Displacement ellipsoid plot (25% probability) of
Ti(N2Npy){N(NPh2)C(Me)C(C6F5)} (4). H atoms omitted for clarity. Selected
distances (Å): Ti(1)-N(1) 2.119(2), Ti(1)-N(3) 1.895(2), Ti(1)-N(4)
1.902(2), Ti(1)-N(5) 2.232(2), Ti(1)-C(1) 2.065(2), N(1)-N(2) 1.418(2),
N(1)-C(2) 1.382(3), N(2)-C(10) 1.398(2), N(2)-C(16) 1.416(2), C(1)-C(2)
1.378(3), C(1)-C(3) 1.448(3).
aryl alkynes ArCCH, forming analogous metallacycles
Ti(N2Npy){N(NPh2)C(H)CAr} (Ar ) Ph (5), Tol (6), 3-C6H4F
(7), 4-C6H4CF3 (8) or C6F5 (9)). Single crystals of 7 were grown
from hexanes. Although these were of rather poor quality, they
nonetheless allowed the connectivity to be established (see the
Supporting Information), confirming that shown in eq 1.
Compounds 2, 5, and 6 (containing Ar ) Ph or Tol substituents
on the metallacyclic ring) were unstable in solution, in the case
of 5 and 6 decomposing over ∼10 h at room temperature to
unknown mixtures. Furthermore, reaction of 1 with (4-
C6H4OMe)CCMe formed a complex mixture of products, and
with MeCCMe no reaction occurred. In contrast, the other
compounds Ti(N2Npy){N(NPh2)C(R)CAr} (Ar ) 4-C6H4CF3,
3-C6H4F or C6F5; R ) H or Me) with electron-withdrawing aryl
groups were stable for at least one week. When followed by 1H
NMR in C6D6, the reaction between 1 and PhCCMe formed a
∼55:45 equilibrium mixture of starting hydrazide 1 and product
Ti(N2Npy){N(NPh2)C(Ph)CMe} (2). In contrast, reaction of 1
in C6D6 with ArCCMe (Ar ) 4-C6H4CF3 or C6F5) or any of the
Alkyne
Cycloaddition
to
TidNr.
Reaction of
Ti(N2Npy)(NNPh2)(py) (1) with the internal alkynes ArCCMe
(Ar ) Ph, 4-C6H4CF3 or C6F5) in benzene at room temperature
for 2-6 h gave the [2 + 2] cycloaddition products
Ti(N2Npy){N(NPh2)C(Me)CAr} (Ar ) Ph (2), 4-C6H4CF3 (3),
or C6F5 (4)) as dark-red microcrystalline solids (eq 1). The
regioisomer shown in eq 1 (i.e., with ArC bound to Ti and MeC
bound to NR) were the only isomers observed.
(55) Gehrmann, T.; Fillol, J. L.; Wadepohl, H.; Gade, L. H. Angew. Chem.,
Int. Ed. 2009, 48, 2152.
(56) Weitershaus, K.; Fillol, J. L.; Wadepohl, H.; Gade, L. H. Organome-
tallics 2009, 28, 4747.
(57) Gehrmann, T.; Fillol, J. L.; Wadepohl, H.; Gade, L. H. Organometallics
2010, 29, 28.
(58) Tiong, P.-J.; Schofield, A. D.; Selby, J. D.; Nova, A.; Clot, E.;
Mountford, P. Chem. Commun. 2010, 46, 85.
(59) Knobloch, D. J.; Lobkovsky, E.; Chirik, P. J. Nat. Chem. 2010, 2.
(60) Kaplan, A. W.; Polse, J. L.; Ball, G. E.; Andersen, R. A.; Bergman,
R. G. J. Am. Chem. Soc. 1998, 120, 11649.
(61) Bytschkov, I.; Doye, S. Eur. J. Org. Chem. 2003, 935.
(62) Doye, S. Synlett 2004, 1653.
(63) Hultzsch, K. C. AdV. Synth. Catal. 2005, 347, 367.
(64) Johnson, J. S.; Bergman, R. G. J. Am. Chem. Soc. 2001, 123, 2923.
(65) Cao, C.; Shi, Y.; Odom, A. L. Org. Lett. 2002, 4, 2853.
(66) Banerjee, S.; Barnea, E.; Odom, A. L. Organometallics 2008, 27, 1005.
(67) Ackermann, L.; Born, R. Tetrahedron Lett. 2004, 45, 9541.
(68) Khedkar, V.; Tillack, A.; Michalik, M.; Beller, M. Tetrahedron Lett.
2004, 45, 3123.
The solid-state structure of 4 (Figure 2) confirms the proposed
structures. Compound 1 also reacted cleanly with terminal
(52) Bolton, P. D.; Mountford, P. AdV. Synth. Catal. 2005, 347, 355.
(53) Fout, A. R.; Kilgore, U. J.; Mindiola, D. J. Chem.sEur. J. 2007, 13,
9428.
(69) Tillack, A.; Jiao, H.; Garcia Castro, I.; Hartung, C. G.; Beller, M.
Chem.sEur. J. 2004, 10, 2410.
(54) Selby, J. D.; Schulten, C.; Schwarz, A. D.; Stasch, A.; Clot, E.; Jones,
C.; Mountford, P. Chem. Commun. 2008, 5101.
(70) Banerjee, S.; Shi, Y.; Cao, C.; Odom, A. L. J. Organomet. Chem.
2005, 690, 5066.
9
10486 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
Figure 3. Displacement ellipsoid plots (20% probability) of (a) Ti(N2Npy){NC(Ph)C(Me)NPh2}(py) (10) and (b) Ti(N2NMe){NC(Ph)C(Me)NPh2}(py) (13).
H atoms omitted for clarity. Distances and angles are given in the Supporting Information (SI).
terminal alkynes in eq 1 gave quantitative formation of the
corresponding metallacycles and free pyridine. Taken together,
these results show that electron-releasing substituents on the
alkynes ArCCR destabilize the azatitanacyclobutene products,
whereas electron-deficient Ar groups or terminal alkynes tend
to give more stable species.
Although [2 + 2] cycloaddition reactions of Group 4 imido
compounds with terminal and internal alkynes forming metal-
lacycles are well-established,46–51,71–77 and hydrazide-derived
metallacycles of the type 2-9 have been implicated as
intermediates in the catalytic hydrohydrazination of alkynes50
(and in one instance spectroscopically observed45), none other
than those described herein have yet been isolated or structurally
authenticated for any transition metal.
Ti(N2NMe){NC(R)C(Me)NPh2}(py) (12-18, R ) Me, Ph or
substituted phenyl, eq 3). All of the reactions proceeded readily
at room temperature and were quantitative when followed by
1H NMR in C6D6. In general, the more electron-rich alkynes
t
ArCCMe (Ar ) 4-C6H4 Bu or 4-C6H4OMe) gave the fastest
insertion reactions. In contrast to the corresponding reactions
of 1 (eq 1), no cycloaddition products were observed at room
temperature. A detailed experimental and DFT mechanistic study
is presented later on.
Alkyne Insertion into Nr-Nꢀ. When the equilibrium mixture
formed from 1 and PhCCMe in C6D6 was allowed to stand for
three days at room temperature the new complex
Ti(N2Npy){NC(Ph)C(Me)NPh2}(py) (10) was formed as the sole
product (eq 2). Heating a mixture of 1 and PhCCMe at 60 °C
accelerated the reaction forming 10 within 3 h. Compound 10
is a vinyl imido species formally formed by insertion of
PhCCMe into the NR-Nꢀ bond of 1. The solid state structure
has been determined (Figure 3) and is discussed below. On scale
up in toluene, 10 was obtained in 71% isolated yield. Unfor-
tunately, attempts to extend this unique reaction to other alkynes
using 1 were unsuccessful. Heating solutions of 1 and ArCCMe
(Ar ) 4-C6H4CF3 or C6F5) in C6D6 gave only the metallacyclic
products 3 and 4, whereas with ArCCH (Ar ) Ph, Tol,
4-C6H4CF3 or C6F5) mixtures of products were formed.
Extending the reactions of 11 to the dialkyne MeCC(2,3,5,6-
C6F4)CCMe (19) in a 2:1 (11:19) ratio gave the dinuclear
NR-Nꢀ insertion product 20 (eq 4) in 36% isolated yield.
(71) Polse, J. L.; Andersen, R. A.; Bergman, R. G. J. Am. Chem. Soc. 1998,
120, 13405.
(72) Ward, B. D.; Maisse-Francois, A.; Mountford, P.; Gade, L. H. Chem.
Commun. 2004, 704.
(73) Wang, H.; Chan, H.; Xie, Z. Organometallics 2005, 24, 3772.
(74) Bolton, P. D.; Feliz, M.; Cowley, A. H.; Clot, E.; Mountford, P.
Organometallics 2008, 27, 6096.
Encouraged by the formation of the TidNR cycloaddition
products 2-9 and the NR-Nꢀ bond insertion reaction giving
10, we screened a number of other diamide-amine supported
titanium hydrazides34 for their reactions with terminal and
internal alkynes. Gratifyingly, Ti(N2NMe)(NNPh2)(py) (11)
readily formed a range of well-defined NR-Nꢀ insertion products
(75) Vujkovic, N.; Ward, B. D.; Maisse-Francois, A.; Wadepohl, H.;
Mountford, P.; Gade, L. H. Organometallics 2007, 26, 5522.
(76) Walsh, P. J.; Baranger, A. M.; Bergman, R. G. J. Am. Chem. Soc.
1992, 114, 1708.
(77) Basuli, F.; Aneetha, H.; Huffman, J. C.; Mindiola, D. J. J. Am. Chem.
Soc. 2005, 127, 17992.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10487

A R T I C L E S
Schofield et al.
Reaction of 11 with 19 in a 1:1 ratio gave a mixture of species
containing 11 and 20. However, unidentified mixtures were
formed between 11 and terminal alkynes RCCH (R ) Ph,
SiMe3, 4-C6H4CF3), EtCCEt or PhCCPh. No reaction occurred
with Me3SiCCMe either at room temperature or upon heating.
The solid state structures of Ti(N2Npy){NC(Ph)C(Me)-
NPh2}(py) (10) and Ti(N2NMe){NC(Ph)C(Me)NPh2}(py) (13)
are shown in Figure 3, and that of [Ti(N2NMe){NCC(Me)NPh2}-
(py)]2(µ-2,3,5,6-C6F4) (20) is given in the SI. Single crystals
of 18 were also analyzed but found to be poor diffractors.
Nonetheless, its connectivity could be confidently established
(see the SI) which was analogous to that for 10 and 13.
All three compounds 10, 13 and 20 possess trigonal bipyra-
midal Ti atoms with the Namide atoms and NR of the new vinyl
imido ligand occupying the equatorial coordination sites. The
TidNR bond distances (avg. 1.760 Å) are equivalent within error
and consistent with a formal triple bond as previously established
for compounds of the general type Ti(N2NR)(NR′)(py) (R′ )
alkyl, aryl, NPh2; R ) Me, SiMe3 and py).34,75,78,79 Overall,
the new vinyl imides 10, 13 and 20 have very similar structures,
and the key distances and angles around titanium are comparable
to the starting hydrazides themselves.
Titanium vinyl imido compounds have previously been
prepared by reaction of sources of alkylidene compounds such
as Cp2Ti)CH2 with organic nitriles.80,81 These vinyl imides can
have a rich reaction chemistry with other unsaturated substrates
Via [4 + 2] cycloaddition reactions. Although cleavage reactions
of the NR-Nꢀ bond of terminal Group 4 hydrazide and related
ligands with a number of unsaturated substrates (including
alkynes) have been previously reported (see Introduction), the
reactions in eqs 2, 3, and 4 are unique examples of net substrate
insertion into the NR-Nꢀ bond of a terminal hydrazide ligand.
Mechanistic Investigations of the Nr-Nꢀ Insertion Reac-
tion. The reaction between Ti(N2NMe)(NNPh2)(py) (11) and
PhCCMe in toluene-d8 was monitored by 1H NMR spectroscopy
between -50 and 0 °C in the presence of 1,4-dimethoxybenzene
as an internal standard. Between -50 and -40 °C there was
no reaction between 11 and the alkyne. Above -40 to -35 °C,
the starting materials were quantitatively converted to the
Figure 4. (Top) Decay of Ti(N2NMe)(NNPh2)(py) (11, filled diamonds)
and growth of Ti(N2NMe){NC(Ph)C(Me)NPh2} (13, hollow diamonds) with
time for the reaction between 11 and PhCCMe (30 equiv) at -26 °C.
(Bottom) Pseudo-first-order log plots for the reaction of
Ti(N2NMe)(NNPh2)(py) (11) with PhCCMe (10, 30, 50 equiv) at -26 °C.
The best-fit lines show r2 ) 0.999 in all cases.
NR-Nꢀ insertion product Ti(N2NMe){NC(Ph)C(Me)NPh2}(py)
(13). No intermediates or other product regioisomers were
observed, and the reaction was complete within two hours at 0
°C. The corresponding reaction of Ti(N2Npy)(NNPh2)(py) (1)
and PhCCMe was also followed by 1H NMR. The first-
formed equilibrium mixture of 1, PhCCMe, metallacycle
Ti(N2Npy){N(NPh2)C(Me)CPh} (2) and pyridine was quantita-
tively converted to Ti(N2Npy){NC(Ph)C(Me)NPh2}(py) (10) at
room temperature over three days (or at 60 °C over 3 h). Again
no other regioisomer of 10 was observed. Since the reaction
between 11 and PhCCMe is significantly faster than for 1, and
tolerates a wider range of substrates RCCR′ (cf. eq 3), it was
the one selected for detailed further study.
When the reaction between 11 and an excess of PhCCMe
(10-50 equiv) was followed at -26 °C in toluene-d8 the decay
of hydrazide starting material (and evolution of 13) followed
well-behaved pseudo-first-order kinetics with respect to Ti as
illustrated in Figure 4. However, as shown in Figure 5 (top),
the observed rate constants (kobs) obtained from the linear log
plots do not scale linearly with the equivalents of PhCCMe used.
At high values of [PhCCMe]:[11] the onset of saturation kinetics
is very evident. In contrast, the corresponding double-reciprocal
plot (Figure 5, bottom) is linear over the range of PhCCMe
equivalents used. At first sight, the data in Figures 4 and 5 are
consistent with the general process outlined in Scheme 1 in
which dissociation of py from 11 forms a reactive pyridine-
free intermediate 11′ prior to the rate-limiting step (or steps),
the rate constant for which is denoted as “k2”. Scheme 1 is
(78) Pugh, S. M.; Clark, H. S. C.; Love, J. B.; Blake, A. J.; Cloke, F. G. N.;
Mountford, P. Inorg. Chem. 2000, 39, 2001.
(79) Blake, A. J.; Collier, P. E.; Gade, L. H.; Mountford, P.; Lloyd, J.;
Pugh, S. M.; Schubart, M.; Skinner, M. E. G.; Tro¨sch, J. M. Inorg.
Chem. 2001, 40, 870.
(80) Doxsee, K. M.; Farahi, J. B.; Hope, H. J. Am. Chem. Soc. 1991, 113,
8889.
(81) Bailey, B. C.; Fout, A. R.; Fan, H.; Tomaszewski, J.; Huffman, J. C.;
Gary, J. B.; Johnson, M. J. A.; Mindiola, D. J. J. Am. Chem. Soc.
2007, 129, 2234.
9
10488 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
k1k2[PhCCMe]
kobs
)
(6)
k2[PhCCMe] + k-1[py]
k-1[py]
1
kobs
1
k1
1
)
+
×
(7)
k1k2
[PhCCMe]
As a test of eq 6 the DMAP analogue of 11, namely
Ti(N2NMe)(NNPh2)(DMAP) (21) was prepared. DMAP is a
stronger donor than pyridine and, in agreement with Scheme 1
and eq 6, compound 21 reacted with PhCCMe significantly more
slowly than 11 to form the corresponding insertion product
Ti(N2NMe){NC(Ph)C(Me)NPh2}(DMAP) (22, eq 3). When fol-
1
lowed by H NMR spectroscopy in toluene-d8, the reaction
between 21 and PhCCMe (20 equiv) had t1/2 ) 3 h at 0 °C
whereas for 11 t1/2 ) 41 min at -26 °C under otherwise identical
conditions.
Addition of 1 or 2 equiv of pyridine to a mixture of 11 and
PhCCMe in toluene-d8 also significantly decreased the rate of
the insertion reaction, consistent with Scheme 1 and eq 6.
However, a control experiment in toluene-d8 showed that 11
on its own reacts with added pyridine to form a fluxional new
species (23) with apparent Cs symmetry, the amount of which
increases with decreasing temperature. Compound 23 is probably
the bis(pyridine) adduct Ti(N2NMe)(NNPh2)(py)2 (Scheme 2).
Although the decreased rate of reaction of 11 with PhCCMe in
the presence of added pyridine is consistent with a reciprocal
dependence on k-1[py] (eqs 5 and 6), it is also consistent with
partial trapping (to some unknown extent) of 11 to form
unreactive 23 which exists in a rapid dynamic equilibrium with
11 at -30 °C (separate resonances for 11, 23 and py are only
observed below -50 °C).
Although 23 is labile and cannot be isolated, the 2,2′-bipyridyl
analogue Ti(N2NMe)(NNPh2)(bipy) (24) was readily prepared
and structurally characterized (see the SI). The geometry at Ti(1)
is approximately octahedral with the NNPh2 positioned trans
to the NMe donor, as indicated in Scheme 2. Compound 24
did not react at all with PhCCMe or C6F5CCMe at room
temperature over the course of several days.
Figure 5. Top: plot of kobs vs equivalents of alkyne for the reaction of
Ti(N2NMe)(NNPh2)(py) (11) with PhCCMe; the dotted line is that expected
for a linear relationship between kobs and equivalents based on extrapolation
from the first data point (10 equiv). Bottom: the corresponding double-
reciprocal plot. The linear fit (r2 ) 0.995) gives an intercept of 1/kobs
1.15(8) × 103 s.
)
Scheme 1. Experimentally Implicated Steps in the Formation of 13
from 11 and PhCCMe
The reaction between 11 and PhCCMe (20 equiv) was
followed at five different temperatures between -16 and -35
°C. The linear pseudo-first-order log plots and corresponding
Eyring analysis (ln(kobs/T) vs 1/T) are shown in the SI. The
derived activation enthalpy and entropy values for the rate-
limiting barrier on the reaction coordinate 11 + PhCCMe f
13 are ∆Hq ) 18.8(4) kcal mol-1 and ∆Sq ) 1(1) cal mol-1
K-1 (∆G2q98 ) 18.5(7) kcal mol-1). The negligible ∆Sq term
implies that the rate limiting step involves neither ligand loss
(e.g., 11 f 11′ + py (k1, Scheme 1)) nor ligand gain.83 The
highest activation barrier within the “k2” process in Scheme 1
is therefore likely to be an intramolecular rearrangement.
As shown in eq 3, compound 11 also reacted with a series of
analogous to mechanisms previously established75,82 for the
cycloaddition reactions of the base-stabilized imides
Cp2Zr(NR)(L) or Ti(N2R′Npy)(NR)(py) (R ) SiMe3 or Xyl) with
alkynes. Scheme 1 predicts that, in the presence of an excess
of alkyne, the rates of reaction should follow pseudo-first-order
kinetics according to eq 5 where kobs is given by eq 6. The linear
double-reciprocal plot is consistent with eq 7.
t
p-substituted alkynes (4-C6H4X)CCMe (X ) Me, Bu, OMe,
CF3) to form the corresponding NR-Nꢀ insertion products
14-17. To gain insight into the rate-limiting “k2” step, all four
reactions were followed at -26 °C in toluene-d8 with 20 equiv
t
alkyne. The reactions for X ) Me, Bu, and OMe showed the
same behavior as for PhCCMe (i.e., X ) H, Figure 4, top),
whereas (4-C6H4CF3)CCMe entered a different kinetic regime
which is discussed below. The pseudo-first-order rate constants
t
(kobs) for X ) H, Me, Bu, OMe are listed in Table 1 along
Rate ) kobs[11]
(5)
(83) Jordan, R. B. Reaction Mechanisms of Inorganic and Organometallic
Systems; Oxford University Press: Oxford, 1998.
(82) Lee, S. Y.; Bergman, R. G. Tetrahedron 1995, 51, 4255.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10489

A R T I C L E S
Schofield et al.
Scheme 2. Reactions of 11 with Pyridine and bipy
Table 1. Rate Constants (kobs) for the NR-Nꢀ Insertion Reactions
of Ti(N2NMe)(NNPh2)(py) (11) with (4-C6H4X)CCMe (20 equiv, X )
As mentioned above, the more electron-deficient alkynes
ArFCCMe (ArF ) 4-C6H4CF3 or C6F5) also undergo NR-Nꢀ
insertion reactions with 11, forming 17 and 18 (eq 3). However,
when these reactions were monitored in toluene-d8 at low
temperature, a different kinetic regime was observed. In each
instance, at temperatures between -40 and -30 °C, equilibrium
mixtures were formed (Scheme 3) consisting of 11 (and/or its
bis(pyridine) homologue, 23), ArFCCMe, free pyridine, and new
species characterized as the cycloaddition products
Ti(N2NMe){N(NPh2)C(Me)CArF} (ArF ) 4-C6H4CF3 (25) or
C6F5 (26)) on the basis of the similarity of their NMR data to
those of the isolated and fully authenticated metallacycles
Ti(N2Npy){N(NPh2)C(Me)CArF} (ArF ) 4-C6H4CF3 (3) or C6F5
(4)). On warming, these first-formed equilibrium mixtures
quantitatively converted to the final insertion products 17 and
18, respectively (Scheme 3). Single crystals of 18 were of poor
quality, but allowed the connectivity to be established (see the
SI), confirming that shown in Scheme 3.
t
H, Me, Bu, or OMe) at -26 °C in Toluene-d8; σp is the Substituent
Constant; ∆Gq is the Gibbs Free Energy (kcal mol-1) of the
Transition State for NR-Nꢀ Cleavage According to DFT Using the
Model Hydrazide A1 (See the Computational Studies Section for
Further Details)
X
kobs × 104 (s-1
)
σp
∆Gq (kcal mol-1
)
H
2.79(1)
3.85(2)
3.90(4)
4.42(2)
0.00
-0.17
-0.20
-0.27
20.7
20.4
20.0
19.6
Me
tBu
OMe
Due to the experimental uncertainties concerning the extent
to which 11 (or indeed 17 or 18) trap (or are trapped by) the
free pyridine generated by formation of the intermediates 25 or
26 (whose relative concentration varies throughout the reaction),
we have not quantitatively modeled the new kinetic regime.
However, it is qualitatively clear from the rate of evolution of
17 (Figure 7) that the overall reaction 11 with (4-
C6H4CF3)CCMe takes significantly longer than with (4-
Figure 6. Plot of log(kobs/kH) vs σp for the reactions of
Ti(N2NMe)(NNPh2)(py) (11) with (4-C6H4X)CCMe (20 equiv, X ) H, Me,
tBu, OMe) at -26 °C in toluene-d8. The best-fit line shown had r2 ) 0.992
giving F ) -0.74(5) and kH is kobs for X ) H.
t
C6H4X)CCMe (X ) H, Me, Bu, OMe; t1/2 ) ∼26-41 min)
under otherwise identical conditions.
with the corresponding Hammett substituent constants, σp.84
Table 1 also gives the Gibbs free energy (∆Gq, kcal mol-1) of
the DFT calculated transition state for NR-Nꢀ cleavage which
will be discussed in the Computational Studies section below.
The kobs data clearly show that the rate of reaction increases
as the para-X substituent becomes more electron-releasing. The
Hammett plot of log(kobs/kH) vs σp (Figure 6) gave a good
agreement and yielded a reaction constant, F, of -0.74(5). A
The reactions of 11 with ArFCCMe are reminiscent of that
between Ti(N2Npy)(NNPh2)(py) (1) and PhCCMe. In each case
a first-formed equilibrium mixture of starting hydrazide and [2
+ 2] cycloaddition product goes on to form a NR-Nꢀ insertion
product. For both 1 and 11, it appears that the electron-
withdrawing aryl ring substituents in ArFCCMe stabilize the
resultant metallacycles, whereas electron-donating groups (cf.
Figure 6) promote NR-Nꢀ bond insertion. Although Figure 7
suggests that metallacycles might lie on the reaction coordinate
from 1 and 11 to the insertion products, such species are not
observed in the majority of instances (cf. Figure 4 and the
reactions of 11 with (4-C6H4X)CCMe (X ) H, Me, tBu, OMe)).
To gain further insight into this chemistry we carried out a
comprehensive series of DFT studies as described below.
+
plot of log(kobs/kH) vs σp gave a poorer fit (r2 ) 0.778). The
negative value of F is indicative of the development of positive
charge (or disappearance of negative charge) at the ArC alkyne
carbon in the transition state for alkyne insertion into NR-Nꢀ.
We will return to this point later on in the Computational Studies
section.
Scheme 3. Reaction of Ti(N2NMe)(NNPh2)(py) (11) with ArFCCMe (ArF ) 4-C6H4CF3 or C6F5)
9
10490 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
Scheme 5. Gibbs Free Energy (kcal mol-1) Profile for MeCCMe
SiH3
Insertion into the NR-Nꢀ Bond of Ti(N2
N
Me)(NNMe2)(py) (A1)
SiH3 Me
along Pathway A; [Ti] ) Ti(N2
N )
Figure 7. Decay and/or evolution with time of titanium species involved
in the reaction of Ti(N2NMe)(NNPh2)(py) (11, filled diamonds) with (4-
C6H4CF3)CCMe (20 equiv) at -26 °C. Other species are: Ti(N2NMe){NC(4-
C6H4CF3)C(Me)NPh2} (17, hollow diamonds) and Ti(N2NMe){N(NPh2)C-
(Me)C(4-C6H4CF3)} (25, hollow circles).
Computational Studies
Mechanism of Alkyne Insertion into Nr-Nꢀ. DFT(B3PW91)
SiH3
calculations were carried out using Ti(N2
N
Me)(NNR2)(py)
(R ) Me (A1) or Ph (B1)) as models for 11, and
SiH3 py
Ti(N2 N )(NNMe2)(py) (C1) as a model for 1. In all three,
following our initial communication on the insertion reaction
of 11 with PhCCMe.3 In this mechanism, loss of pyridine from
I forms an η2-hydrazide III Via base-free II. This intermediate
was proposed to then undergo [2 + 1] cycloaddition at NR with
RCCR′ forming an N-metalated azirine intermediate (IV) with
concomitant NR-Nꢀ bond breaking. Ring-opening of the azirine
Via NPh2 group transfer from Ti back to CR′ affords the four-
coordinate insertion product V which is then trapped by pyridine
giving VI. Gade has also proposed a similar η2-hydrazido
intermediate in the NR-Nꢀ cleavage reaction of
Zr(N2*Npy)(NNPh2)(py) with isocyanides (N2*Npy ) (2-
the SiMe3 groups of N2NR (R ) Me or py) have been replaced
by SiH3, and in A1 and C1 TidNNPh2 is modeled by
TidNNMe2 for computational efficiency. The results obtained
with B1 are similar to those with A1 and are presented in the
SI. To establish the basic electronic features of the mechanism,
MeCCMe (2-butyne) was considered as the reacting alkyne.
42
t
NC5H4)CMe(CH2NSiMe2 Bu)2). Mindiola’s mechanism is
consistent with our experimental evidence for a pyridine-free
intermediate being formed before reaction with the alkyne.
Furthermore, if the intramolecular transformation IV f V were
the rate-limiting step, then the experimental near-zero entropy
of activation for 11 + PhCCMe f 13 could likewise be
consistent with Pathway A. Finally, Gade has described how a
version of III derived from M(N2*Npy)(NNPh2)(py) (M ) Zr,
Hf) can be trapped by electrophilic addition of B(C6F5)3 to NR.33
Therefore, DFT calculations were used to determine the details
of the mechanism along Pathway A and the nature of the rate-
limiting step.
Scheme 4. Two Potential Mechanisms for Alkyne Insertion into the
NR-Nꢀ Bond of the Hydrazide Ligand in 1 or 11. [Ti] ) Ti(N2Npy)
or Ti(N2NMe
)
The DFT Gibbs free energy profile along Pathway A for A1
reacting with MeCCMe is shown in Scheme 5. Formation of a
vacant site (in A2) by pyridine dissociation is unfavorable by
∆G ) 9.4 kcal mol-1. Dissociation of DMAP was calculated
to be even more difficult (∆G ) 11.8 kcal mol-1), as expected.
The calculated ∆G value for pyridine loss from A1 compares
well with those measured experimentally for some imidotitanium
complexes.85 The activation barrier for dissociation of a ligand
is not easy to estimate computationally as often no transition
state (TS) can be located for such a process. However, on the
assumption86 that the activation barrier ∆Gq is essentially the
enthalpy cost of breaking the metal-ligand bond (i.e., ∆Gq ≈
∆H), the barriers (i.e., ∆Gq) for pyridine and DMAP dissociation
are 21.9 and 24.6 kcal mol-1, respectively. These values
(84) Hansch, C.; Leo, A.; Taft, R. W. Chem. ReV. 1991, 91, 165.
(85) Stewart, P. J.; Blake, A. J.; Mountford, P. Inorg. Chem. 1997, 36,
3616.
Scheme 4 shows the two mechanisms (designated Pathways
A and B) we have investigated for the NR-Nꢀ insertion reactions
of 1 and 11. Pathway A has recently been proposed by Mindiola
(86) Hartwig, J. F.; Cook, K. S.; Hapke, M.; Incarvito, C. D.; Fan, Y.;
Webster, C. E.; Hall, M. B. J. Am. Chem. Soc. 2005, 127, 2538.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10491

A R T I C L E S
Schofield et al.
N
Me)(NNMe2)(py) (A1) along
SiH3
Scheme 6. Gibbs Free Energy (kcal mol-1) Profile for MeCCMe Insertion into the NR-Nꢀ Bond of Ti(N2
Pathway Ba)
a
SiH3
The bond shown in bold for the alkyne is to help the reader trace the fate of the Ti-bound CMe group in A8. [Ti] ) Ti(N2
N
Me).
constitute an upper estimate of the actual activation barriers.
The larger value for DMAP is in agreement with the lower rate
of reaction observed with 21 compared to that with 11.
Experimentally determined ∆Gq values for pyridine loss from
certain 5- and 6-coordinate imidotitanium complexes (including
a homologue of 1) have been reported to be ∼15 kcal mol-1.75,85
From A2 (Scheme 5) coordination of Nꢀ yields two potential
intermediates: A3, where NR shifts to the vacant site and Nꢀ
coordinates in the equatorial plane; and A3b, where Nꢀ
coordinates directly to the vacant site as indicated for species
III, Scheme 4. A3 is 8.7 kcal mol-1 more stable than A3b, and
only for the former could a TS for NR-Nꢀ bond breaking be
located. Formation of A3 from A2 is exergonic (-4.7 kcal
mol-1) and proceeds with a low activation barrier (9.1 kcal
mol-1 relative to A2).
The crucial step in Pathway A is the formation of the
N-metalated azirine IV. This is modeled as A4 in Scheme 5,
formed by reaction of A3 with MeCCMe. As proposed by
Mindiola, this reaction is a [2 + 1] cycloaddition between the
alkyne and NR of a bent hydrazido ligand. The cycloaddition
occurs with concomitant NR-Nꢀ bond cleavage as illustrated
by the lengthening of the NR-Nꢀ bond from 1.41 Å in A3 to
2.03 Å in the transition state A3-4. However, even though the
transformation from A3 to A4 is slightly exergonic (-2.8 kcal
mol-1), the associated activation barrier is prohibitively high
(53.7 kcal mol-1 at 298 K). Frontier orbital analysis shows that
this is because NR in this type of η2-hydrazide species is
nucleophilic in nature, not electrophilic.33 A3-4 is a late
transition state for NR-Nꢀ cleavage with little C-NR bond-
forming character (C-NR ) 1.96 and 2.48 Å in A3-4), thus
explaining the high activation barrier (see Figure S4 in the SI
for the geometry of A3-4).
and coordination of pyridine yields finally the insertion product
(not shown in Scheme 5).
The high computed barrier A3-4 for Pathway A and the
bimolecular nature of the rate-determining step suggest that the
NR-Nꢀ insertion reactions of 1 and 11 follow another mecha-
nism. This is designated Pathway B in Scheme 4 and assumes
that azatitanacyclobutenes of the type VII play an actiVe role
in the formation of V and VI, rather than being separate species
unconnected with NR-Nꢀ insertion. While they are not always
observed, metallacycles (e.g., 2 (eq 1) and 25 and 26 (Scheme
3)) have sometimes been seen in reaction mixtures that also
give rise to NR-Nꢀ insertion. DFT calculations of Pathway B
were therefore carried out also on the model system A1 (Scheme
6). This pathway would constitute a way to insert an alkyne
into the NR-Nꢀ bond after pyridine dissociation and so could
be in agreement with the experimental kinetic studies, providing
the rate-determining step occurs after the [2 + 2] cycloaddition
of the alkyne. A pathway starting by coordination of the alkyne
before the pyridine dissociates was searched for, but no stable
intermediate could be located on the potential energy surface.
Therefore, only Pathway B, initiated by pyridine dissociation,
will be described.
The vacant site in A2, created by pyridine decoordination, is
a potential site for alkyne coordination. Coordination of MeC-
CMe (forming A7) is less favorable than formation of the η2-
hydrazide A3 (Scheme 5), partly due to the loss of translational
entropy in the former case. Nonetheless, [2 + 2] cycloaddition
from A7 to form the azatitanacyclobutene A8 occurs readily
with a low barrier (3.3 kcal mol-1) and is thermodynamically
competitive with A1 (∆G ) 1.9 kcal mol-1). From A8,
intramolecular coordination of NꢀMe2 is facile and proceeds with
a low barrier (6.1 kcal mol-1) to yield the key intermediate A9,
lying only 4.2 kcal mol-1 above the separated reactants A1 and
MeCCMe. The NR-Nꢀ bond-breaking step proceeds through
A9-10, yielding the C,N-metalated azirine A10 in a thermo-
neutral reaction. The activation barrier is only 17.8 kcal mol-1
above A9, and is 36.4 kcal mol-1 less than the NR-Nꢀ bond
From A4, nucleophilic transfer of the Ti-bound NMe2 group
to an azirine ring carbon atom affords A5, and subsequent ring-
opening leads to the Z aza-allyl intermediate A6 in an energeti-
cally favorable process. Upon decoordination of the chelating
NMe2 group, the vinyl imido shifts to the equatorial position,
9
10492 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
Figure 8. (a) HOMO of A8. (b) Schematic MO diagram for the π interactions within the 4-membered metallacycle
cleavage through A3-4 (Scheme 5). This is because NR-Nꢀ
bond breaking in A9-10 is coupled with C-NR bond forming,
thus lowering the activation barrier of the process. A detailed
discussion of the sequence A8 f A9 f A9-10 and the
electronic structures of these species is given later on.
To fully understand the origins of this critical regiochemical
preference, we have analyzed the electronic structure of the
azatitanacyclobutene A8. The HOMO of this complex (Figure
8a) results from the stabilization of the 4-electron repulsion
between the lone pair on NR (LPR) and the filled πcc MO by the
vacant π*cc MO (Figure 8b). As a consequence of the mixing
between these three orbitals, the HOMO is essentially based
on NR and the carbon atom bonded to Ti (denoted CTi).
With regard to regiochemical preferences, substitution of a
methyl group in A8 by any group “X” that would decrease the
4-electron repulsion would result in a more stable metallacycle.
For example, an electron-withdrawing group X would increase
the effective electronegativity of the carbon to which it is
attached. This would in turn result in a stabilization of both πcc
and π*cc and also in a polarization of πcc toward the X-
substituted carbon atom. The schematic MO diagram in Figure
8b indicates that the best situation corresponds to the regioisomer
with the electron-withdrawing X bonded to CTi. In that case,
the destabilizing overlap between LPR and πcc would be reduced
and the favorable overlap between LPR and π*cc would be
increased; both factors combine to lower the energy of the
HOMO.
These arguments, based on the effective electronegativity of
the alkyne carbon atoms, essentially involve the σ-acceptor
properties of the X group. However, π-effects must also be
considered. For example, it is well-known87 that the HOMO of
an alkene moiety XC(R)dCR2 is more strongly developed on
the CR2 carbon when X is a π-donor. In the case of an
azatitanacyclobutene of the type A8, the best arrangement to
minimize repulsion between πcc and LPR would be the regioi-
somer with X bonded to CN. For σ-accepting but π-donating
groups the preferred product will come from a balance of the
two effects as each favors a different regioisomer. For X groups
which are both σ- and π-accepting, the two contributions favor
the same regioisomer, with X bonded to CTi.
From A10 the reaction proceeds Via exergonic nucleophilic
attack of the Ti-bound NMe2 group on the azirine ring to form
A11 with a low activation barrier. Two other isomers (A12 and
A13), differing in the relative orientations of NR and Nꢀ with
SiH3 Me
respect to the N2
N
ligand, have been located on the
potential energy surface. The activation barrier for the isomer-
ization of A11 to A12 (through A11-12, not shown in Scheme
6) is only 5.5 kcal mol-1. Each of these three intermediates can
lead to the cleavage of the C-NR bond to form A6. However,
the TS with the lowest energy is A13-6 originating from A13,
and the corresponding activation barrier is 7.5 kcal mol-1
.
Overall, the net NR-Nꢀ insertion reaction is mechanistically a
TidNR cycloaddition followed by an NR atom migration process.
Comparison of the DFT computed Pathways A and B (Schemes
5 and 6) shows clearly that the latter is preferred. From A1, the
activation barrier to overcome along Pathway A (A3-4) is 58.4
kcal mol-1, whereas the highest point along Pathway B (A9-10)
is only 22.0 kcal mol-1. The latter value is in much better agreement
with the experimental activation barrier determined for PhCCMe
reacting with 11 (∆Gq298 ) 18.5(7) kcal mol-1). Moreover, whereas
in Pathway A the rate-determining step is bimolecular, in Pathway
B it is intramolecular. The experimental low value for ∆Sq (1(1)
cal mol-1 K-1) is in better agreement with an intramolecular rate-
determining step, lending further support to Pathway B as the most
likely reaction mechanism. The relevant calculated ∆Sq values are
-41.3 cal mol-1 K-1 (A3-4) and 4 cal mol-1 K-1 (A9-10), the
latter being consistent with experimental measurement.
Origins of Regioselectivity for Nr-Nꢀ Insertion. When 1 and
11 undergo NR-Nꢀ insertion with unsymmetrical alkynes
ArCCMe, the products always have CAr bound to TidNR of
the originating hydrazide and CMe bound to NꢀPh2 (eqs 2-4).
In the corresponding metallacycles, however, it is always CMe
which is bound to NR (eqs 1 and 3). This is fully consistent
with Pathway B. As illustrated by a bold bond for one of the
alkyne C-Me groups in Scheme 6, it is the regiochemistry of
the cycloaddition product which determines that of the insertion
product. That is to say, the alkyne carbon initially bonded to Ti
in the metallacycle A8 ends up bonded to NR in the insertion
product A6, and the alkyne substituents are mutually cis as found
exclusively by experiment.
Equation 8 shows the possible products Ar(Ti)_R and
Ar(N)_R that can be formed between A1 and ArCCR (Ar )
Ph or substituted phenyl; R ) Me or H). These are homologues
of A8, formed from A1 and MeCCMe for which ∆G ) 1.9
kcal mol-1 (Scheme 6). Changing from MeCCMe to PhCCMe
gives two possible regioisomers, Ph(Ti)_Me and Ph(N)_Me
for which ∆G ) 1.3 kcal mol-1 and 6.7 kcal mol-1, respectively.
The increased stability of Ph(Ti)_Me compared to A8 (∆∆G
(87) Nguyen, T. A. Frontier Orbitals: A Practical Manual; John Wiley &
Sons: Chichester, 2007.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10493

A R T I C L E S
Schofield et al.
Table 2. Gibbs Free Energies (∆G, kcal mol-1) for the
Cycloaddition Products Ar(Ti)_R (see eq 8) Expressed Relative to
Separated A1 and ArCCRa
Ar
R
∆G
Ph
Ph
H
-7.5
1.3
-0.7
-1.6
1.3
Me
Me
Me
Me
Me
4-C6H4CF3
C6F5
4-C6H4 Bu
t
4-C6H4OMe
1.4
a The cycloaddition product formed with MeCCMe (A8) has ∆G )
1.9 kcal mol-1
Figure 9. Optimized geometries for the azatitanacyclobutene A8, the
NꢀMe2-chelated intermediate A9, and the TS for NR-Nꢀ cleavage (A9-10).
H atoms have been omitted for clarity.
.
) -0.6 kcal mol-1) is in agreement with a larger effective
electronegativity at an sp2 hybridized carbon atom (ipso-Ph)
compared to an sp3 hybridized one (Me). The decreased stability
of Ph(N)_Me with respect to both A8 (∆∆G ) +4.8 kcal
mol-1) and Ph(Ti)_Me (∆∆G ) +5.4 kcal mol-1) indicates
that Ph behaves as a π-accepting group in the azametallacycles.
Consistent with this, the C-Ph bond in Ph(Ti)_Me (1.459 Å)
is shorter than that in Ph(N)_Me (1.495 Å) as a result of better
π conjugation. Shorter C-NR and longer CdC bonds in
Ph(Ti)_Me compared to Ph(N)_Me are also consistent with
this model.
The DFT results are fully consistent with the experimental
data. For example, 1 forms more stable [2 + 2] cycloaddition
products with PhCCH than with PhCCMe, and none at all with
MeCCMe; the metallacycles are more stable with electron-
withdrawing Ar groups for both 1 and 11; the only regioisomers
observed are with Ar bound to CTi. Furthermore, when Ar is
bound to CTi in Ar(Ti)_Me (i.e., at the bold bond in A8 in
Scheme 6) the product of the insertion reaction has precisely
the geometry of the compounds observed experimentally.
Understanding the Nr-Nꢀ Bond-Cleavage Step. The ground-
state electronic preference of the azatitanacyclobutenes, which
ultimately control the regiochemistry of the NR-Nꢀ insertion
reaction, is for electron-withdrawing alkyne substituents, ideally
attached to CTi. However, the experimental kinetic data (Table
1, Figure 6) suggest that oVerall rates of the insertion reaction
are favored by electron-releasing alkyne substituents. At first
sight, the factors that control the stability of the metallacycle
are not necessarily the same as those governing the transition
state for NR-Nꢀ cleavage. We therefore examined in further
detail the rate-determining NR-Nꢀ breaking process for RCCMe
(R ) Me or Ar).
In the reaction of A1 with MeCCMe (Scheme 6), cleavage
of the NR-Nꢀ bond is effective in two steps from the
azatitanacyclobutene A8. First, coordination of NꢀMe2 to Ti
forms A9, and then NR-Nꢀ cleavage occurs through the
transition state A9-10 to form A10. This last step constitutes
the rate-determining step of the entire alkyne insertion reaction.
The geometries of the three critical extrema (A8, A9, and
A9-10) are shown in Figure 9.
The coordination of NꢀMe2 induces a deformation of the
4-membered ring in the metallacycle A8. In A9, Nꢀ is coplanar
with Ti and the two carbon atoms (Nꢀ-Ti-CTi-CN ) 1.3°),
while NR is situated above that plane. Ti-Nꢀ is longer than
Ti-CTi (2.193 Å vs 2.068 Å). The Nꢀ-Ti-CTi angle is 101°,
and NR is situated between CN and Nꢀ (NR-CN ) 1.444 Å,
NR-Nꢀ ) 1.454 Å). The coordination of NꢀMe2 results in a
slight activation of the NR-Nꢀ bond (1.401 Å in A8, 1.454 Å
in A9), and has positioned the NR-Nꢀ axis above the plane
containing the CdC π bond.
The geometry of A9 is perfectly adapted for nucleophilic
attack from the π density polarized toward CTi into the
σ*(NR-Nꢀ) antibonding MO. Indeed, the geometry of the
transition state A9-10 confirms the description of the NR-Nꢀ
cleavage as effectively a nucleophilic attack of CTi on NR. The
CTi · · ·NR distance decreases by ∼0.2 Å (from 2.268 Å in A9
to 2.077 Å in A9-10) while the NR · · ·Nꢀ distance increases by
∼0.7 Å (from 1.454 Å in A9 to 2.185 Å in A9-10). These
variations are accompanied by an increase of the Nꢀ-Ti-CTi
angle to 117°, a lengthening of Ti-CTi to 2.127 Å, and a
shortening of Ti-Nꢀ to 2.068 Å. To increase the nucleophilic
Changing from PhCCMe to the terminal alkyne PhCCH
stabilizes both regioisomers relative to PhCCMe and MeCCMe.
For Ph(Ti)_H ∆G ) -7.5 kcal mol-1 and for Ph(N)_H ∆G )
-1.9 kcal mol-1. This is because H is less electron-releasing
than Me, but Ph is more electron-accepting. Again the regioi-
somer with Ph bonded to CTi is very strongly favored by ∼104
in terms of an equilibrium between the two. Focusing on this
regioisomer, we also calculated ∆G for forming Ar(Ti)_Me
from A1 and experimentally relevant para-substituted alkynes,
namely ArCCMe (Ar ) C6F5 or 4-C6H4X with X ) tBu, OMe
or CF3). The results are given in Table 2 together with those
for the other alkynes and are globally consistent with the
bonding model introduced above. Strongly σ-accepting groups
(Ar ) 4-C6H4CF3 (∆G ) -0.7 kcal mol-1) or C6F5 (∆G )
-1.6 kcal mol-1)) stabilize Ar(Ti)_Me compared to the
t
compounds where Ar ) Ph or 4-C6H4 Bu (∆G ) 1.3 kcal
mol-1). Interestingly, although the strongly π-donating para
substituent OMe should in principle destabilize Ar(Ti)_Me (see
above), the competing inductive and mesomeric influences of
σ acceptor (-I) and π donor (+M) effects, respectively, lead
to a product of only slightly lower stability (∆G ) 1.4 kcal
t
mol-1) than when Ar ) Ph or 4-C6H4 Bu.
9
10494 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
SiH3 py
Scheme 7. Gibbs Free Energy (kcal mol-1) Profile for MeCCMe Insertion into the NR-Nꢀ Bond of Ti(N2 N )(NNMe2)(py) (C1) along
Pathway B
The bond shown in bold for the alkyne is to help the reader trace the fate of the Ti-bound CMe group in C8. [Ti] ) Ti(N2SiH Npy).
3
character at CTi, the CdC π bond is partially lost. This is
illustrated by the increase in the Me-CTi-CN-Me dihedral
angle from 4.1° in A9 to 33.2° in A9-10 and lengthening of
CTi-CN from 1.355 Å in A9 to 1.383 Å in A9-10. The twist
respectively). An analogous trend was found for the corre-
sponding species formed with PhCCMe (-0.506, -0.387,
-0.232, respectively) but with the Ph group stabilizing a more
negative charge at CTi at each stage as expected. The overall
NPA charge of the Ph group also becomes more positive on
going from the metallacycle Ph(Ti)_Me (-0.060) to the NR-Nꢀ
breaking TS (+0.030), again consistent with the effects of
electron-releasing ring substituents.
around the CTi-CN axis allows the formation of a NRdCN
π
bond, while preserving some lone pair character at CTi. The
NR-CN bond distance shortens significantly in the transition
state A9-10 (1.444 Å in A9, 1.348 Å in A9-10). The relative
low energy of A9-10, and hence the facile NR-Nꢀ bond
cleavage, is the result of several stabilizing effects working
cooperatively. The nucleophilic substitution at NR of Nꢀ by CTi
is facilitated by the formation of stabilizing Ti-NMe2 and
NRdCN interactions. Any factor increasing the nucleophilicity
at CTi is likely to lower the energy of the transition state.
The TS Gibbs free energies (∆Gq) for the NR-Nꢀ cleavage
step (analogous to A9-10 with MeCCMe) for the alkynes (4-
Taken together, the trends in the computed ∆Gq and
experimental kobs values, and the negative Hammett reaction
constant F are consistent with NR-Nꢀ bond cleavage being
caused by intramolecular nucleophilic attack of CTi on NR. The
form of the conversion vs time plot in Figure 7 is a consequence
of enhanced ground-state stabilization of the metallacycle (cf.
Table 2) and concomitant destabilization of the NR-Nꢀ cleavage
TS by the electron-withdrawing 4-C6H4CF3 group. The observed
product regiochemistry with ArCCMe comes from both the
relative stability of the precursor metallacycles and the associ-
ated transition states.
t
C6H4X)CCMe (X ) H, Me, Bu, OMe) used in the Hammett
study are listed in Table 1 alongside the experimental rate
constants. The ∆Gq values for ArCCMe are all less than for
MeCCMe itself (22.0 kcal mol-1), in keeping with experimental
observations that the reaction of 11 with MeCCMe is much
slower than with PhCCMe. The ∆Gq values decrease in the order
N2Npy vs N2NMe: Influence of the Diamide-Amine Ligand
Sets. Experimentally, there is a significant difference between
the rates of the NR-Nꢀ insertion reaction for
Ti(N2Npy)(NNPh2)(py) (1) and Ti(N2NMe)(NNPh2)(py) (11).
While the latter leads to PhCCMe insertion within ∼5-10 min
at room temperature, the former requires ∼3 days (or heating
at 60 °C for 3 h). In addition, alkyne cycloaddition products
are readily isolated with 1, whereas these were barely observable
with 11, and only at -30 °C or below (Scheme 3).
t
para-X ) H > Me > Bu > OMe, which is consistent with the
ordering of the experimental rate constants (kobs for para-X )
H < Me < tBu < OMe) and nucleophilic attack of CTi on NR in
the NR-Nꢀ breaking TS being promoted by electron-releasing
para-substituents. ∆Gq for the reaction of (4-C6H4CF3)CCMe
with A1 was calculated to be 21.7 kcal mol-1, higher than for
PhCCMe (20.7 kcal mol-1) because the electron-withdrawing
para-CF3 group diminished the nucleophilicity of CTi. Note that
∆Gq for NR-Nꢀ cleavage starting from the less stable, alternative
To probe these differences, the Gibbs free energy profile for
the NR-Nꢀ insertion reaction along Pathway B (Scheme 4) for
SiH3 py
the model system Ti(N2 N )(NNMe2)(py) (C1) has been
cycloaddition regioisomer Ph(N)_Me was 24.1 kcal mol-1
.
calculated (Scheme 7). From C1, formation of a vacant site is
almost thermoneutral (∆G ) 0.3 kcal mol-1), and the activation
barrier (12.1 kcal mol-1) for pyridine dissociation is significantly
reduced compared to that in A1 (21.9 kcal mol-1). The relative
Consistent with the para substituent effects, we find that the
NPA charge of CTi in A8, A9, and A9-10 becomes more
positive along the reaction coordinate (-0.473, -0.363, -0.187,
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10495

A R T I C L E S
Schofield et al.
stability of the base-free complex C2 compared to that of A2
observed regiochemistry. TidNR [2 + 2] cycloaddition products
were also observed for 11 but were only stable below ∼-30
°C and with electron-deficient alkynes ArFCCMe (ArF)
4-C6H4CF3 or C6F5).
SiH3 py
is due to the higher trans influence of pyridyl (in N2 N )
SiH3
compared to that of NMe (in N2
N
Me) in the ground states of
the adducts C1 and A1 (cf. Ti-Npy in 1 and 11 ) 2.259(2) Å
and 2.215(5) Å, respectively34) and the larger chelate ring size
Whereas reaction of 1 with PhCCMe at room temperature
initially gives equilibrium mixtures of 1 and the cycloaddition
product 2, longer reaction times or gentle heating forms the
vinyl imido product 10 in which PhCCMe has undergone net
insertion into the NR-Nꢀ bond of 1. The corresponding reactions
of 11 with a range of alkynes RCCMe proceed smoothly at
temperatures as low as -35 °C and follow pseudo-first-order
kinetics with respect to 11. DFT studies showed that the
previously proposed mechanism3 involving attack of the alkyne
at Nꢀ of an η2-bound NNR2 ligand is highly unfavorable due to
the nucleophilic nature of Nꢀ and the late nature of the C-N
bond-forming transition state. Instead, a low-energy pathway
involving initial alkyne cycloaddition to TidNR followed by
intramolecular coordination of NPh2 precedes the actual NR-Nꢀ
bond-breaking event. The calculated activation parameters for
this process (modest ∆G2q98 and near-zero ∆Sq) for realistic
model systems were also fully consistent with experiment. The
rate-limiting step in the NR-Nꢀ insertion process, nucleophilic
attack of CTi at NR (displacing NPh2), is supported by a Hammett
analysis of the rate constants and NPA charge analyses.
We reported in our preliminary communication41 that 11
catalyzes (10 mol % loading) the 1,2-diamination of PhCCMe
with Ph2NNH2 at room temperature Via 13, rather than undergo-
ing the more widely established hydrohydrazination reaction.
It is generally accepted that Group 4 hydrohydrazination
catalysis proceeds Via azatitanacyclobutenes.50,61,62 The results
presented here have shown that such metallacycles can also be
intermediates en route to NR-Nꢀ insertion and hence can lead
to alkyne diamination rather than hydrohydrazination. It was
beyond the scope of this paper to develop these catalytic aspects;
they will be the topic of future contributions from our
laboratories. However, in terms of general future develop-
ment of this reaction, the DFT calculations for
SiH3
(6-membered vs 5-membered) of N2
N
py between the amide
nitrogens and the apical donor (Npyridyl or NMe). This allows
better relative orientation of the various Namide and NR π-donors
in C2 leading to overall more effective stabilization of the
unsaturation at Ti.34 As a result, the metal center is less
electrophilic in C2 compared to that in A2, resulting in an
endoergic coordination of MeCCMe (∆G ) 9.5 kcal mol-1),
whereas the coordination is exoergic in A2 (∆G ) -2.3 kcal
mol-1).
From C7, [2 + 2] cycloaddition of MeCCMe to form C8 is
thermodynamically downhill (-8.7 kcal mol-1) with a low
activation barrier (4.2 kcal mol-1). The metallacycle C8 is only
1.1 kcal mol-1 above the starting materials and is ∼1 kcal mol-1
more stable than A8, in qualitative agreement with the relative
ease of experimental observation of metallacycles for 1 vs 11.
From C8, the NR-Nꢀ bond cleavage is effective in two steps
as observed for the reaction of A1 with MeCCMe. Thus,
coordination of NꢀMe2 to Ti forms C9, and then NR-Nꢀ
SiH3 py
cleavage proceeds through the TS C9-10. Because the N2
N
ligand is less compact (6-membered chelate rings) than
SiH3
N2
N
Me and the metal center is less electrophilic, the intramo-
lecular coordination of NMe2 is less effective, and this results
in an overall higher energy for both C9 (6.4 kcal mol-1) and
the crucial transition state C9-10 (25.5 kcal mol-1) compared
to A9 (4.2 kcal mol-1) and A9-10 (22.0 kcal mol-1).
Once C10 is formed, NMe2 transfer from Ti to C is easier
than for A10 (∆Gq ) 4.8 kcal mol-1 and ∆G ) -3.1 kcal
mol-1), and the three isomers C11, C12, and C13 have similar
energies. The final step (TS C13-6, ∆Gq ) 13.1 kcal mol-1
)
is more difficult (relative to C13) than was the case for A13-6,
but as C13 is overall more stable than A13, the TS C13-6 lies
at a lower energy with respect to C1 than does A13-6 with
respect to A1. Coordination of pyridine to C6 (not shown in
Scheme 7) affords the final insertion product. As in the
mechanism starting from A1, the rate-determining step is the
NR-Nꢀ bond cleavage step (C9-10).
SiH3
SiH3
Ti(N2
N
Me)(NNMe2)(py) and Ti(N2
N
Me)(NNPh2)(py)
(NNMe2 vs NNPh2) show that insertion is not restricted only
to hydrazides with phenyl substituents at Nꢀ, thus pointing to
the potential for considerable development of its scope and
applications.
Taking the Gibbs free energy of A9-10 or C9-10 with
respect to separated reactants (A1 or C1 and MeCCMe) as an
estimate of the activation energy, then the ∆∆Gq value of 3.5
kcal mol-1 between the two model systems translates into a
ratio between the rate constants of between 2 and 3 orders of
magnitude at room temperature, with the system A being the
faster. Qualitatively, this is in good agreement with the relative
rates of the real NR-Nꢀ insertion reactions 1 and 11 with
PhCCMe (∼3 days and 5-10 min, respectively).
The supporting ligand also has a major role to play as it must
allow NꢀR2 coordination along the reaction pathway but
nonetheless stabilize terminal hydrazides in the first instance.
Even for apparently similar ligands such as N2Npy and N2NMe
there can be several orders of magnitude difference in the
reaction rate for the same alkyne. The electronic properties of
the alkyne substrates are also critical. These govern the
regiochemistry of the [2 + 2] cycloaddition intermediates which
in turn define those in the arising NR-Nꢀ insertion products.
Importantly, although electron-withdrawing aryl substituents in
ArCCMe thermodynamically favor initial metallacycle formation
(with ArC bound to Ti), electron-releasing Ar groups promote
insertion (CTi-NR bond formation). For reasons that are not yet
clear (but which probably relate to the Brønsted acidity of sp
C-H bonds), terminal alkynes do not undergo clean insertion
reactions, although [2 + 2] cycloaddition products can be
isolated in certain cases.
Conclusions
The closely related compounds Ti(N2Npy)(NNPh2)(py) (1) and
Ti(N2NMe)(NNPh2)(py) (11) have provided an excellent base
for a detailed synthetic, mechanistic, and computational study
of the reactions of titanium hydrazides with alkynes. Reactions
of 1 with terminal and internal aryl alkynes ArCCR (Ar ) Ph
or substituted phenyl, R ) Me or H) at room temperature
give isolable TidNR [2
+ 2] cycloaddition products
Ti(N2Npy){N(NPh2)C(R)CAr} with a specific regiochemistry
(ArC bound to Ti). These are the first fully authenticated
azametallacyclobutenes for any metal hydrazide. Electron-
withdrawing aryl groups stabilize the metallacycles in the
Our detailed mechanistic and theoretical studies of 1 and 11
explicitly connect MdNR bond (“imide-like”) and NR-Nꢀ bond
reactivity along the same reaction coordinate. Given the rich
chemistry of Group 4 metal-ligand multiple bonds in general,
9
10496 J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010

Reactions of Titanium Hydrazido Compounds with Alkynes
A R T I C L E S
ylsilyl groups relative to the internal standard. Pseudo-first-order
rate contants were obtained from linear plots of ln([Ti]/[Ti]0) vs
time.
it seems possible that all examples of NR-Nꢀ cleavage may
proceed Via initial addition to TidNR. As mentioned, Bergman
has reported that reaction of in situ generated Cp2Zr(NNPh2)
with alkynes gave concomitant NR-Nꢀ cleavage (as opposed
to insertion) and C-H activation reactions (Figure 1).29 It is
not apparent how (or whether) these products are related to those
formed with 1 or 11, but given the rich [2 + 2] cycloaddition
chemistry of imido-zirconocenes with alkynes and other sub-
strates in general,49 it is likely that the complex sequence of
steps leading from Cp2Zr(NNPh2) involve azazirconacy-
clobutenes. This is consistent with Gade’s very recent report
Computational Details. All calculations were performed with
the Gaussian03 package90 of programs with the hybrid B3PW91
functional.91,92 The Ti atom was represented by the relativistic
effective core potential (RECP) from the Stuttgart group and the
associated basis set.93 The remaining atoms (C, H, N, O, F) were
represented by a 6-31G(d,p) basis set.94 The Si atom was
represented by RECP from the Stuttgart group and the associated
basis set,95 augmented by a d polarization function.96 Full optimiza-
tion of geometry was performed without any symmetry constraint,
followed by analytical computation of the Hessian matrix to identify
the nature of the located extrema as minima or transition states.
Connection between reactant and product through a given transition
state was checked by optimization of slightly altered geometries
of the transition state along the two directions of the transition state
vector associated to the imaginary frequency. All energies given
in the text are Gibbs free energies (T ) 298 K, P ) 1 atm) in the
gas phase
of [2
+ 3] cycloaddition products formed between
Zr(N2*Npy)(NNPh2)(py) and organic azides.55 In this system,
an intermediate featuring a nitrene fragment (formed by N2 loss
from the first-formed [2 + 3] cycloaddition product) coupled
across ZrdNR is proposed to precede NR-Nꢀ cleavage, although
no mechanistic studies or transition state calculations ac-
companied the experimental and DFT ground state results.
Finally we note that the diazoalkane compound Cp*2Ti{η2-
NNC(H)Tol} (homologues of which have been shown to have
imido-like reactivity patterns88,89) undergoes NR-Nꢀ cleavage
Acknowledgment. We thank the EPSRC for studentships to
A.D. Schofield and J.D.S., a Ph.D plus postdoctoral award to J.D.S.,
and a postdoctoral fellowship to A.D. Schwarz. We also thank the
Spanish MICINN for a MEC postdoctoral fellowship to A.N.
with CO to form
a product with new NR-CO and
Ti-NꢀdC(H)Tol bonds60 in a reaction reminiscent of that of
Cp2Zr(NNPh2)(DMAP) with CO forming Cp2Zr(NPh2)(NCO)
(Figure 1). It was postulated that attack of CO at the MdNR
bond of a terminal diazoalkane intermediate (without Nꢀ
coordination) precedes NR-Nꢀ cleavage. Our results are also
consistent with this suggestion.
Supporting Information Available: General experimental
methods and synthesis and characterization of the new com-
pounds; details of the crystal structure determinations and full
crystallographic data in CIF format for the structure determina-
tions of 4, 10, 13, 19, 20, and 24; displacement ellipsoid plots
of 19, 20, and 24; preliminary structure determinations for 7
and 18; variable temperature studies of the reaction of 11 with
PhCCMe; optimized geometries for A3-4, A12, and A13;
discussion of the influence of the hydrazide substituent (NNMe2
vs NNPh2) on the reaction mechanism and comparison of Gibbs
free energy profiles for MeCCMe insertion into the NR-Nꢀ bond
of A1 and B1; complete reference 90 for Gaussian 03; DFT
computed Cartesian coordinates for the optimized molecules
and their electronic energies. This material is available free of
charge Via the Internet at http://pubs.acs.org.
Experimental Section
Kinetic Measurements. A standard solution of p-methoxyben-
zene in toluene-d8 was used for all kinetic experiments. The general
procedure is as follows. In a glovebox, compound 11 (15.1 mg,
26.6 µmol) was weighed into a vial and dissolved in the standard
solution (0.35 mL) and cooled to -45 °C. In a separate vial,
RCCMe was dissolved in the standard solution (0.30 mL) and also
cooled to -45 °C. The two vials were placed on a precooled tile
and the solutions combined using a precooled pipet, agitated
quickly, and transferred to a precooled J. Young NMR tube. The
NMR tube was taken out of the glovebox and placed in a dry ice/
isopropanol cold bath. The NMR probe was cooled to -35 °C and
the sample added before being locked and shimmed. It was then
warmed to -26 °C (other reaction temperatures were also used as
appropriate as noted in the main text) and allowed to thermally
equilibrate; the shimming was checked, and an array was set up
recording a spectrum of 4 scans every 140 s. The ratios of the
species were calculated by measuring the integrals of the trimeth-
JA1036513
(90) Frisch, M. J.; et al. Gaussian 03, Revision D.01; Gaussian, Inc.:
Wallingford, CT, 2004.
(91) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(92) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244.
(93) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
Chim. Acta 1990, 77, 123.
(94) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213.
(95) Bergner, A.; Dolg, M.; Kuchle, W.; Stoll, H.; Preuss, H. Mol. Phys.
1993, 30, 1431.
(96) Hollwarth, A.; Bohme, H.; Dapprich, S.; Ehlers, A. W.; Gobbi, A.;
Jonas, V.; Kohler, K. F.; Stagmann, R.; Veldkamp, A.; Frenking, G.
Chem. Phys. Lett. 1993, 203, 237.
(88) Polse, J. L.; Kaplan, A. W.; Andersen, R. A.; Bergman, R. G. J. Am.
Chem. Soc. 1998, 120, 6316.
(89) Hanna, T. E.; Keresztes, I.; Lobkovsky, E.; Bernskoetter, W. H.; Chirik,
P. J. Organometallics 2004, 23, 3448.
9
J. AM. CHEM. SOC. VOL. 132, NO. 30, 2010 10497