MM quadruply bonded complexes supported by vinylbenzoate ligands: synthesis, characterization, photophysical properties and application as synthons

MM complexes as potential synthons for the development of higher order extended structures via Heck coupling reactions, exhibiting interesting photophysical properties.


Introduction
The eld of organic electronics based on conducting organic polymers is a rapidly developing eld embracing both chemistry and materials science. Poly-p-phenylene vinylene (PPV) is one of the most studied conducting organic polymers to date and has been at the center of the development of organic electronics. [1][2][3][4][5][6][7] Friend and coworkers discovered the polymer light emitting diode with PPV showing yellow-green electroluminescence in 1990. 1 The rst bulk heterojunction photovoltaic cell was made with PPV derivatives as the light absorbing layer by Heeger 3 and Friend 8 in 1995. The incorporation of transition metals into PPV derivatives offers the opportunity to couple the properties of the transition metal with those of the polymer and was rst achieved with ruthenium and osmium via the Heck reaction and the properties of these complexes have been explored in detail. 9 They possess the p-p* transition of the parent polymer and also an MLCT transition from the metal to the bipyridine that it is bound to in the polymer. The luminescence of these systems is dominated by the 3 MLCT state traditionally present in metalpyridyl systems. 10 Ruthenium and osmium PPV derivatives were also examined as light emitting layers in polymer light emitting diodes (PLEDs) and it was shown the emission was largely MLCT based. The metal can also be used to toggle between conjugated and nonconjugated organic polymers, where coordination of the metal forces the planarization of the polymer and enhances the conjugation. 11 With this in mind, it seems appropriate to explore the incorporation of MM quadruply bonded complexes into PPV derivatives by way of carbon-carbon cross coupling reactions, such as the Heck reaction. The use of carbon-carbon cross coupling reactions on the periphery of transition metal complexes has been previously explored; 12 however, there is a limited scope of these reactions on MM paddlewheel complexes. Ren and company have shown the Heck reaction can be successfully employed on Ru 2 formamidinate compounds to add a styrene functionality. Reactions on the quadruply bonded Mo 2 tetracarboxylate complexes have proven more difficult due to the kinetic lability of the carboxylate ligands. The initial example of this has shown that the Sonogashira reaction can be employed on the periphery of MM compounds successfully without compromising the quadruply bonded core. 13 We have previously shown that the attachment of p-conjugated organic units such as oligothiophenes via carboxylate units to the MM quadruply bonded centers have intense 1 MLCT transitions that may be tuned to span the entire visible region of the spectrum. 14 Moreover, the 1 MLCT states have relatively long lifetimes, typically in the range of 1-20 ps. In this paper, a series of Mo 2 and W 2 bis-bis complexes supported by vinylbenzoate ligands have been prepared and their ground state and excited state photophysical properties are explored. It is also shown that the Mo 2 complexes are viable synthons for the preparation of higher order extended structures that possess very interesting photophysical properties.

Synthesis
The synthesis of the new compounds is outlined in Scheme 1. The trans-M 2 compounds 1A-2B were prepared by the ligand exchange reaction between M 2 T i PB 4 with the respective carboxylic acid, either 3-vinylbenzoic acid (1) or 4-vinylbenzoic acid (2), where M ¼ Mo (A) or W (B) and T i PB ¼ 2,4,6-triisopropylbenzoate. The reactions were performed in toluene, where the desired product was slightly soluble. The desired product was then isolated by reducing the volume of the solvent in vacuo and precipitation by the addition of hexanes. They are highly colored solids that are yellow (1A), orange (2A), maroon (1B) and blue (2B). The new compounds are air sensitive and soluble in toluene, THF, and DCM. They gave molecular ions in the MALDI-TOF MS and 1 H NMR data are consistent with their formulations. Further characterization details are given in the Experimental. The new compounds 1A and 2A were then subjected to further reactions. A Heck cross coupling reaction was successfully performed on both Mo 2 complexes with excess iodobenzene in the presence of palladium(II) acetate, Et 3 N and heat. For 1A, there was no visible color change; however, for 2A, the solution went from orange to red. The products were isolated by ltration of the reaction mixture through a bed of celite. The solvent was then reduced to a minimal amount and hexanes were added to precipitate the product. The desired products were puried further by recrystallization. 3A and 4A gave molecular ions in the MALDI-TOF MS and 1 H NMR consistent with their formulations. Further characterization details are given in the Experimental. MALDI-TOF MS and 1 H NMR plots can be seen in the ESI. † Reactions were attempted on the W 2 compounds; however, they are more sensitive to oxidation than their Mo 2 counterparts and the W 2 desired coupled products were not obtained.
Further studies pertaining to C-C coupling reactions involving W 2 -quadruply bonded complexes are underway.

Single crystal X-ray structures
Single crystals suitable for X-ray analysis for 1A and 2A were grown from vapor diffusion of hexanes into a concentrated THF solution. The molecular structures of 1A and 2A are shown in Fig. 1 and 2, respectively. Both structures have a center of inversion and the T i PB units are twisted such that the phenyl rings are out of conjugation with the -CO 2 units with dihedral angles of 79.6 and 83.3 , respectively. The vinylphenyl unit, however, is coplanar with its -CO 2 unit making it in conjugation with the Mo 2 center with dihedral angles of 11.3 and <1 in 1A and 2A, respectively. The angle between the vinyl and phenyl units in 1A is 4. 5

Electronic structure calculations
In order to assist in the interpretation of the spectral data, electronic structure calculations have been performed on model complexes of the form trans-M 2 (O 2 CH) 2 (vinylbenzoate) 2 , where T i PB is substituted for formate. Density functional theory (DFT) and time-dependent DFT (TD-DFT) have been employed to this end. The TD-DFT calculations predict these compounds to possess several intense, fully-allowed transitions. There is a metal-to-ligand charge transfer (MLCT) from the M 2 d to vinylbenzoate p*, a MLCT from the M 2 d to the formate p* and a p-p* transition based on the vinylbenzoate ligands.
Calculations were performed on model complexes 1A 0 , 1B 0 , 2A 0 , 2B 0 , 3A 0 and 4A 0 where the T i PB moiety has been replaced by a formate to save on computational resources. This is a reasonable assumption because the phenyl ring of the T i PB is twisted $90 out of conjugation with respect to the M 2 center which effectively removes its contribution to the electronic structure.
The frontier MO energy level diagram for the meta-vinylbenzoate M 2 complexes 1A 0 and 1B 0 is shown in Fig. 3. The HOMO for both compounds is the M 2 d orbital with a signicant degree of mixing with the ligands. The LUMO for the Mo 2 compound is the M 2 d*. To higher energy, the LUMO + 1 and LUMO + 2 are the symmetric and anti-symmetric combinations of the ligand p* orbitals, where the anti-symmetric   combination has the correct symmetry to mix with the Mo 2 d orbital. For the W 2 complex, the LUMO is the symmetric p* combination and to higher energy, the anti-symmetric p* orbital is the LUMO + 1 and the M 2 d* is the LUMO + 2.
Below the M 2 d HOMO lie two ligand-p orbitals then two M 2 p and nally the M 2 s for 1A 0 . For the W 2 complex 1B 0 , the order of the ligand-p and M 2 p orbitals are reversed. The difference in ordering of the Lp and M 2 p orbitals follows from the higher W 2vs. Mo 2 -based orbitals by $0.5 eV. The HOMO-LUMO gap for the tungsten compound tracks with the difference in M 2 d orbital energy and is $0.5 eV smaller than that in the Mo 2 analog.
A similar comparison of the frontier orbitals of 2A 0 and 2B 0 can be made and is shown in Fig. 4. Qualitatively, the orbitals are similar except for the ligand p orbitals. For this series of complexes, the vinyl substituent is in conjugation with the M 2 d and the carboxylate core raises the energy of the occupied ligand p orbitals and lowers the energy of the ligand p* orbitals. In the Mo 2 complex, the symmetric p* orbital is lowered below the M 2 d* making it the LUMO. This is responsible for the reduced HOMO-LUMO gap in this series of compounds relative to 1A 0 and 1B 0 .
The splitting between the two combinations of p* orbitals is a representation of the electronic coupling present and the values for these compounds are listed in Table 1. Between the complexes of the same metal, the splitting does not change signicantly; however, upon going from Mo 2 to W 2 , the magnitude of the splitting increases due to the increased electronic coupling from greater mixing of the W 2 d orbital with the ligand p*.
For compounds 3A 0 and 4A 0 , the frontier MO energy level diagrams are shown in Fig. 5. The MO diagram for compound 4A 0 is very similar to its precursor 2A 0 . The addition of the phenyl unit extends the p-conjugation, raises the energy of the p-orbitals, and lowers the energy of the p* orbitals. This additionally causes the secondary effect of raising the energy of the Mo 2 d orbital. Overall, this results in a reduction of the HOMO-LUMO gap to 2.89 eV. The splitting between the two p* combinations is reduced to 0.18 eV.
For compound 3A 0 , the orbital diagram is more complicated. The HOMO of 3A 0 is the Mo 2 d with some admixture from the ligands and the LUMO is the Mo 2 d* orbital. The p system of the vinylbenzoate ligand has been perturbed relative to the parent compound 1A 0 . The symmetric and anti-symmetric combinations of the Lp* are the LUMO + 1 and LUMO + 2, respectively. Upon addition of the phenyl group, there is little energetic modication of the p* orbitals; however, there is a mixing of the Mo 2 d* orbital with the symmetric p* combination that is not normally present in M 2 tetracarboxylates. Further, there is character from the outer phenyl ring in this orbital, but this does not result in a change in the energy of the orbital. In addition, there are two new p* orbitals to higher energy as a result of the extended ring system in 3A 0 . Below the HOMO, there is a degenerate pair of ligand p orbitals that are shied signicantly higher relative to 1A 0 , À5.7 eV vs. À6.5 eV.

Electronic absorption and emission spectroscopy
The electronic absorption spectra of 1A, 1B, 2A, and 2B in THF are shown in Fig. 6. To higher energy, the electronic  Table 1 Calculated energy splitting between the in-phase and out-ofphase combinations of p* orbitals spectra of all compounds feature a ligand centered transition that is not sensitive to the metal, but is sensitive to the position of the vinyl group, with the para-substitution shiing the absorption to lower energy relative to the meta-substitution. In the Mo 2 compounds 1A and 2A, there is a metal-toligand charge transfer from the Mo 2 d orbital to the T i PB carboxylate at 330 nm. For the W 2 compounds 1B and 2B, this transition occurs at 405 nm due to the raised energy of the d orbital. Further to lower energy is the fully allowed 1 MLCT transition from the M 2 d orbital to the vinylbenzoate p* orbital. The observed energetic trends in the absorption maxima match those predicted by the TD-DFT calculations for the two ligand sets and two different metal centers. A table comparing the experimental and calculated absorptions can be found in the ESI. † This MLCT transition is sensitive to both the selection of metal and the substitution pattern of the vinyl group on the benzoate ligand. For 1A, 2A, 1B, and 2B it occurs at 437 nm, 468 nm, 601 nm, and 651 nm, respectively. This trend can be easily understood by the locations of the M 2 d and vinylbenzoate p* orbitals from the electronic structure calculations. The 1 MLCT energy is lower for the W 2 compounds (1B and 2B) relative to Mo 2 compounds (1A and 2A) due to the raised energy of the W 2 d orbital. The compounds with 4-vinylbenzoate (2A and 2B) as the ligand occur at lower energy due to the lower p* orbital relative to 3-vinylbenzoate (1A and 1B).
Also shown in Fig. 6 are the UV-visible emission spectra in THF for the four compounds described above. Upon irradiation into the MLCT absorption, these compounds show uorescence from their 1 MLCT states and the energy of the emission correlates with the energy of the 1 MLCT absorption described above. The Mo 2 compounds 1A and 2A also show phosphorescence from their lowest energy triplet state centered at 1100 nm, which is typical for Mo 2 tetracarboxylates, and corresponds to emission from the 3 MoModd* state. Upon cooling to 77 K, the emission notably increases in intensity and a vibronic progression corresponding to the Mo-Mo stretch ($350 cm À1 ) becomes visible. Compounds 1B and 2B show broad 1 MLCT emissions in the visible and NIR extending to 1200 nm. There is no evidence of any phosphorescence from either W 2 complex. The NIR emission data can be seen in the ESI Fig. S1. † The electronic absorption and emission of 3A and 4A were taken in THF and can be seen in Fig. 7. These compounds show the same gross features as their parent Mo 2 compounds 1A and 2A. For 3A, the addition of the phenyl substituent shis the pp* absorption to lower energy (295 nm) compared to the parent complex 1A. However, the energy of the 1 MLCT absorption remains at almost the same wavelength (430 nm). For 4A, the p-p* absorption (319 nm) is shied to lower energy than that in 2A (267 nm) and in 3A (295 nm). The 1 MLCT energy shis to lower energy (487 nm) relative to the parent 2A (467 nm) and to the meta-substituted 3A (430 nm).
The change in the energy of the p-p* and MLCT absorptions in 3A and 4A can be understood in terms of their substitution pattern. For 3A, the meta substitution pattern determines the outer phenyl-vinyl moiety is not in conjugation with the carboxylate or the Mo 2 center. The p-p* is then located principally on the phenyl-vinyl-phenyl subunit and has no contribution from the carboxylate. The meta-substitution has no effect on the MLCT and therefore it behaves as if the ligand is simply a benzoate. For 4A, the entire ligand phenyl-vinyl-phenyl-CO 2 is involved in the p-p* transition making it occur at slightly lower energy than in 3A. Since the substitution pattern is para in 4A, the MLCT spans the entire ligand and not just the simple benzoate as in 3A. This is shown pictorially in Fig. 8 with colored boxes denoting the subunits of the molecule involved in each transition for 3A and 4A.
Both 3A and 4A show emission from their 1 MLCT states upon irradiation in THF that correlates energetically with the MLCT absorption (Fig. 7), similar to the emission of the parent compounds 1A and 2A. Both compounds also show emission in the NIR centered around 1100 nm. Upon cooling,

Transient absorption spectroscopy
The new compounds 1A, 1B, 2A, 2B, 3A, and 4A were all investigated by nanosecond (ns) and femtosecond (fs) transient absorption (TA) spectroscopy in THF solution to monitor the singlet and triplet states. The fsTA spectra of 4A are shown as representative in Fig. 9. Here, there is a strong absorption centered on 540 nm present from the singlet state that decays within $9 ps to the triplet state represented by a persistent bleach at 480 nm that lasts beyond 3 ns. The kinetics of the singlet states were measured for all compounds and the lifetimes fall within 0.5-10 ps, typical for 1 MLCT S 1 states, and are listed in Table 2 Fig. S22 and S23. † Previously, the T 1 state of W 2 T i PB 4 was determined to be 3 WWdd* with a lifetime of $1 ms and phosphorescence from this state was observed around 860 nm. 16 The much faster triplet lifetimes of 1B and 2B coupled with the lack of phosphorescence lead to the assignment of a 3 MLCT T 1 state in the W 2 complexes.
Upon examination of the S 1 excited state lifetimes, some trends are revealed. The Mo 2 complexes, 1A and 2A, show  longer-lived S 1 states when compared with their W 2 counterparts, 1B and 2B. This may be attributed to the heavy atom effect and the larger spin orbit coupling of tungsten as compared to molybdenum. The shorter S 1 lifetime of 3A compared with 4A may be a result of the mixing of the Mo 2 d* as evidenced by the electronic structure calculations, vida supra, suggesting a more efficient pathway for conversion to the dd* state.

Time-resolved infrared spectroscopy
In order to examine charge distribution in the excited states of these molecules, compounds 2A, 2B, and 4A have been studied by fs time resolved infrared (TRIR) spectroscopy. In principle, the vibrational frequency of the vinyl group should be sensitive to the excited state electron distribution and therefore be able to provide insight into charge localization; however, this has not been shown to be the case. With the assistance of the electronic structure calculations employing time-dependent DFT, the dominant features are attributable to the phenyl ring stretches and the carboxylate stretching modes.
For 2A, the TRIR spectra are shown in Fig. 10 and the kinetics can be seen in Fig. S24. † Present is a large transient at 1525 cm À1 that has been assigned to phenyl ring stretching mode. This feature decays with a lifetime of 3.5 ps, which is similar to the singlet lifetime found in the fsTA experiment for this compound. At long times, there are features that are associated with the carboxylate stretches and have previously been assigned as characteristic of the 3 MoModd* state. Specifically, the band at 1525 cm À1 is assignable to n as (CO 2 ) of the T 1 state that, relative to the ground state, has one less electron in the d orbital which is involved in back-bonding to the CO 2 p*, thus causing the vibration to increase in energy. 17 For 2B, the phenyl stretching mode is present as a transient in the singlet state at 1546 cm À1 and carboxylate stretches appear at 1481 cm À1 and 1405 cm À1 (ESI Fig. S25 †). From the decay of these features, the singlet lifetime is estimated to be 2 ps, which is effectively the same lifetime that was determined by fsTA spectroscopy. There are several ligand vibrations at 1532 cm À1 , 1503 cm À1 , and 1405 cm À1 indicative of a long-lived triplet state as seen in the fsTA. Apparent is the lack of features associated with the dd* state. This, coupled with the lack of phosphorescence is consistent with a 3 MLCT T 1 state.
The ground state IR and fsTRIR of compound 4A were taken in THF and are largely dominated by -CO 2 and phenyl ring stretching modes, see Fig. 11. In the ground state, the bands at 1607 cm À1 and 1570 cm À1 are assigned to phenyl ring stretching modes. The band at 1495 cm À1 is assigned to the n as (CO 2 ) and the band at 1380 cm À1 to the n s (CO 2 ). Upon excitation, there are four transients, which correspond to the two n(CO 2 ) stretching modes and the two phenyl ring stretching modes. From these features, the lifetime of the 1 MLCT is estimated to be 10 ps, which is in good agreement with the fsTA data (ESI Fig. S26 †). At longer times, the characteristic modes of the 3 MoModd* state are present indicating this as the T 1 state.
The lack of clear vinyl features in the TRIR spectra preclude the assignment of excited state distribution. Instead of a discrete vinyl mode, the C]C stretch couples to the breathing modes of the phenyl rings as would be expected in the quinoidal form of the ligand in 4A, Fig. 12.
Previously, metal complexes bearing trans-stilbene or derivatives of trans-stilbene as ligands have been shown to undergo cis/trans isomerization upon photoexcitation. The complexes, fac-Re(CO) 3 (bpy)(X), where X ¼ 1,2-bis(4-pyridyl) ethylene [bpe] or 4-styrylpyridine [stpy], were shown to undergo photoisomerization. 18 These complexes were excited at 313 nm into a transition with IL and MLCT character, where photoisomerization then took place with quantum yields in the range of 0.003-0.46 depending upon the solvent. It was proposed that this isomerization occurs through a 3 p* intermediate, where the olen is twisted. 19 In a related case, the bpe ligand was used to bridge two Re centers or Re and Os together in complexes of the form, [(CO) 3 (phen)Re(trans-bpe)Re(phen)(CO) 3 ] 2+ and [(phen)Re(CO) 3 (trans-bpe)Os(trpy)(bpy)] 3+ respectively. 20 In the bis Re case, photoisomerization occurred despite the bulky groups on either end of the bpe ligand. Conversely, no photoisomerization was observed for the mixed metal Re/Os which was attributed to a low lying MLCT state of the osmium unit that quenches the chemical reaction.
In this regard, it is worth exploring the potential for isomerization in 4A. Aer photoexcitation, we observe rapid internal conversion (<1 ps) to the 1 MLCT S 1 state, where the excited electron resides in a p* orbital and the hole remains on the Mo 2 d orbital. This MLCT state lies lower in energy than the pp* states responsible for isomerization thereby preventing photoisomerization from occurring similar to that seen in the mixed Re/Os case discussed above.

Conclusions
Several new vinylbenzoate supported M 2 bis-bis complexes were prepared including two compounds, 3A and 4A, by the Heck cross coupling reactions involving 1A and 2A, respectively. Based on this work the prospect of performing carbon-carbon cross coupling reactions on the periphery of M 2 complexes in the preparation of extended structures containing M 2 quadruply bonded units looks promising. The photophysical properties of the new compounds have been explored, but the vinyl groups did not prove to be viable IR reporters of excited localization or delocalization across the M 2 center. Alternatively, the use of C^X functionalities, where X ¼ N, O, or CR, has proved largely successful in the determination of charge localization or delocalization in M 2 complexes. 13,21,22 Experimental section

General methods
All reactions were performed under 1 atm of UHP argon using standard Schlenk or glovebox techniques. Solvents were dried and distilled from the appropriate drying agents and then degassed prior to use. Solvents were stored over 4Å molecular  sieves in asks with Kontes tops. Manipulations of the studied compounds were performed under nitrogen inside a glovebox or under argon on a Schlenk line. NMR spectra were recorded on a 400 MHz Bruker DPX Advance 400 spectrometer or a 500 MHz DPX Advance 500 spectrometer. All 1 H NMR chemical shis are in parts per million (ppm) relative to the protio impurity in THF-d 8 at 1.73 ppm.
High resolution matrix-assisted laser desorption ionization time-of-ight (MALDI-TOF) mass spectra were obtained on a Bruker Microex mass spectrometer for 1A and 2A and calibrated using an internal standard. All other samples were analyzed on a Bruker ultraeXtreme mass spectrometer run in positive mode. The measurements for 1B, 2B, 3A, and 4A were standardized using 9 peptides with molecular weights ranging from 750 g mol À1 to 3100 g mol À1 . These peptides were supported in a matrix of HCCA. Dithranol was used as the matrix. Samples were prepared by adding a solution of the matrix to the solid sample. This was then spotted on the plate for analysis. Also, the observed isotope pattern for each sample has been correlated to the predicted pattern based on the given formulation.
3-Vinylbenzoic acid and 4-vinylbenzoic acid were purchased from Alfa Aesar and used as received. Mo 2 T i PB 4 and W 2 T i PB 4 were synthesized from known procedures. 16,23 Iodobenzene and palladium(II) acetate were purchased from Sigma Aldrich. Iodobenzene was degassed by freeze-pump-thaw and stored over 4Å molecular sieves in a Kontes top ask under argon.

Electronic structure calculations
Density functional theory was used to perform electronic structure calculations on model complexes utilizing the Gaussian 09 suite of programs. 24 The Becke three parameter correlation functional combined with the Lee, Yang, and Parr exchange functional was employed (B3LYP). 25,26 Model complexes were optimized in the gas phase to a minimum as conrmed by the lack of imaginary frequencies in a vibrational analysis. For C, H, and O the 6-31g* basis set was used. For Mo and W, the Stuttgart/Dresden energy-consistent pseudopotential (SDD) was used to represent the core orbitals and the SDD basis set. 27 Formate was used instead of T i PB in order to simplify the calculations and save on computational resources. This substitution allows the complexes to obtain an idealized symmetry higher than that with T i PB. The complexes were idealized to the highest symmetry possible while still containing an inversion center. A correction factor of 0.96 was applied to the calculated vibrational frequencies to correlate them with experimentally observed frequencies.
Visualizations of the molecular orbitals was performed with GaussView 5.0 with an isovalue of 0.02.

Electronic absorption and emission
Steady-state electronic absorption and UV-visible emission measurements were carried out with 1.0 Â 1.0 cm quartz cuvettes equipped with Kontes stopcocks. Electronic absorption spectra at room temperature were recorded using a Perkin-Elmer Lambda 900 spectrometer in THF solution.
Fluorescence measurements were made on a SPEX Fluoromax-2 spectrouorometer in the UV-visible region in THF solution. The compounds were irradiated into their MLCT absorption bands. Emission measurements in the near-infrared region were performed in J. Young NMR tubes at room temperature and 77 K in 2-methyl THF. Spectra were recorded on a home-built instrument equipped with a germanium detector. Samples were irradiated into their MLCT bands. On the NIR setup, this was done with either 405 nm (1A-4A) or 658 nm (1B, 2B). Samples were prepared with an absorbance of <0.3.

Crystallographic information
Single crystals of 1A and 2A were isolated as yellow and orange blocks, respectively, and handled under a pool of uorinated oil. Examination of the diffraction pattern was done on a Nonius Kappa CCD diffractometer with Mo Ka radiation. All work was done at 150 K using an Oxford Cryosystems Cryostream Cooler. Data integration was done with Denzo, and scaling and merging of the data was done with Scalepack. 28 The structures were solved by the direct methods program in SHELXS-13. 29 Full-matrix least-squares renements based on F 2 were performed in SHELXL-13, 29 as incorporated in the WinGX package. 30 For each methyl group, the hydrogen atoms were added at calculated positions using a riding model with U(H) ¼ 1.5U eq (bonded carbon atom). The rest of the hydrogen atoms were included in the model at calculated positions using a riding model with U(H) ¼ 1.2U eq (bonded atom). Neutral atom scattering factors were used and include terms for anomalous dispersion. 31 One isopropyl group in the T i PB moiety of 1A was disordered over two locations. This disorder was modeled using similarity restraints as well as rigid bond restraints to ensure reasonable bond lengths and anisotropic displacement parameters. The occupancy of these two parts was allowed to rene on a free variable leading to an occupancy of $54% in the major component.
Compound 2A had disorder in the p-vinyl moiety. This disorder was handled similarly to that in 1A and the occupancy of the major component rened to be $50%. Aer nal renement of 2A several residual 'Q' peaks remained in the difference map and were unable to be adequately modeled as solvent. To remove this electron density the SQUEEZE 32 protocol of the PLATON 33 suite of programs was used to exclude 45 electrons from a void of 250Å 3 . This electron density likely corresponds to disordered hexanes used in crystallization of 2A. †

Time-resolved spectroscopy
Nanosecond TA was performed on samples in 1 Â 1 cm square quartz cuvettes with Kontes stopcocks. 34 Measurements were made on a home-built instrument pumped by a frequencydoubled (532 nm) or frequency tripled (355 nm) Spectra-Physics GCR-150 Nd:YAG laser (fwhm z 8 ns, $5 mJ per pulse). The power at the sample was set to 100 mW Signal from a Hamamatsu R928 photomultiplier tube was processed with a Tektronics 400 MHz oscilloscope (TDS 380). For 1A, 1B, 2B, and 3A the samples were excited with 355 nm and for 2A and 4A the samples were excited at 532 nm with laser power at the sample of 100 mW.
Femtosecond transient absorption experiments were performed with a Ti:sapphire and regenerative amplier combination (1 kHz, 50 fs full width at half-maximum) that has been previously described. 35 Samples were prepared with absorbances of $0.3-0.8 in a 1.0 mm quartz cuvette with Kontes top. Excitation power at the sample was 1-2 mJ. Spectra collected underwent wavelength calibration and group velocity dispersion corrections.
Time-resolved infrared 36 spectroscopy experiments, previously described, were performed with a Ti:sapphire and regenerative amplier combination (1 kHz, 50 fs fwhm). Samples of 2A, 2B and 4A were prepared with an absorbance of $1 at l max in THF solution. A PerkinElmer semi-demountable cell with a 0.1 mm Teon spacer between 4 mm CaF 2 windows was used to provide an airtight sample chamber. The compounds 2A, 2B, and 4A were excited with a laser power of $1 mJ at 515 nm, 515 nm and 675 nm, respectively.
In general, kinetics were t to a sum of exponential decay terms, S(t) ¼ P i A i exp(À1/s i ) + C, using Igor Pro 6.0 or SigmaPlot 12.0, where A i is the amplitude, s is the lifetime, and C is an offset. Error bars for the lifetimes are reported as standard errors of the exponential ts.