Controlling the hydrogenolysis of silica-supported tungsten pentamethyl leads to a class of highly electron deficient partially alkylated metal hydrides

Accessing highly electron deficient partially alkylated tungsten hydrides on silica via controlled hydrogenolysis of surface organometallic complex ( 
<svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="16.000000pt" height="16.000000pt" viewBox="0 0 16.000000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata>
Created by potrace 1.16, written by Peter Selinger 2001-2019
</metadata><g transform="translate(1.000000,15.000000) scale(0.005147,-0.005147)" fill="currentColor" stroke="none"><path d="M0 1760 l0 -80 1360 0 1360 0 0 80 0 80 -1360 0 -1360 0 0 -80z M0 1280 l0 -80 1360 0 1360 0 0 80 0 80 -1360 0 -1360 0 0 -80z M0 800 l0 -80 1360 0 1360 0 0 80 0 80 -1360 0 -1360 0 0 -80z"/></g></svg>
 Si–O–)W(Me)5.

The energy of the absorption edge is defined as that at the inflection point of the first absorption peak at the W LIII edge. This value was compared with the calibrated edge energy determined for the tungsten reference foil. ATHENA was used for edge calibration and deglitching. XDAP was used for data normalization, background subtraction, and EXAFS data fitting, which was done with a "difference-file" technique 38,39 to determine a best fit based on a comparison of data with overall fits as well as fits of individual shell contributions. In the XDAP code, the disorder term (Debye-Waller factor) and inner potential correction (ΔE0) are with respect to the corresponding reference files.
Fitting was an iterative process that continued until the parameters characterizing each of the proposed shells and the overall fit were in good agreement with the k 1 -and k 3 -weighted EXAFS data (k is the wave vector) and the Fourier-transformed data. Fitting was done iteratively for each of the individual shells, and all the fitting parameters were required to have physically appropriate S7 values for an acceptable fit. The number of usable parameters in the fitting was limited by the Nyquist theorem (a statistical criterion): n = 2ΔkΔr/π + 2, where Δk and Δr are the ranges in k space used in the fitting and the range in R (distance) space fit in the Fourier Transform.

EXAFS investigation of 1
XDAP was used to generate EXAFS models that could be compared to the EXAFS data of 1.
The quality of the fit is determined based upon the goodness of fit (ΔΧ) 2 , indicating the deviation between the model and the data, and if the values of the fitted parameters make good chemical sense. As stated in the main text, 1 was changing under the influence of the beam requiring analysis of the first and fourth scans independently.
The starting material was speculated to be [(Si-O-)W(CH3)5]. Multiple models were compared to the data of scan 1. Table S2 and Figure S10 show the best fit model, that within error, agrees with the starting material [(Si-O-)W(CH3)5]. Alternate models were also compared to the data. Models varied the coordination of the W-O and W-C contributions.
These models were deemed unfit due to Debeye Waller factors and inner potentials that were deemed to be outside of range of acceptable. When comparing the plots of the models to the data the overall k space appeared adequate but when comparing the k 1 and k 3 weighted R space it was evident that the models did not match the data.

Analysis of the fourth scan of the EXAFS was fit to the form of [(Si-(O)Y)-W(CH3)X] with
the anticipation that the overall podality of the tungsten was changing under the influence of the beam, which is similar to long term storage at room temperature. Due to the complexity of the silica surface it is expected that multiple species now exist as postulated by the NMR, resulting in a greater uncertainty in the optimal fit. Multiple models were tested varying the oxygen and carbon contributions. Models were also tested that included a =CH2 contribution. All models except the one presented in Table S3 and Figure S11 were rejected due to unrealistic parameters, mainly Debeye Waller factors and inner potentials, or due to the model not aligning well with the data when comparing the k space, R space, and individual shell fits.

Additional details pertaining to the computation of hydrogen magnetic shielding and 1 H chemical shift values.
As specified in the main text, representative systems were selected to establish the applicability of the density functional theory (DFT) method that was chosen to calculate the various magnetic shielding (and ultimately chemical shift) values. The systems chosen supplement the other W-containing systems chosen as part of an earlier study, 1 with the key difference in the present study being the inclusion of molecules which contain WH groups. In many other respects, the general approach remains the same as specified in the earlier account.
The new systems under consideration are depicted in Figure S12. appropriate COSMO settings can be found in the coordinates disclosure sections below.
Subsequent to a full geometry optimization including non-explicit solvent effects via the COSMO, hydrogen magnetic shielding calculations were performed. By merging the data generated at as part of the present study, with that which was published in the recent study on surface-supported W-containing species (as disclosed above), we arrive at the following plot which correlates the computed hydrogen magnetic shielding values with experimental 1 H chemical shift values ( Figure S13):

Fig. S13
Correlation plots relating isotropic hydrogen magnetic shielding and experimental 1 H isotropic chemical shift values (in ppm) for the calibration molecules featured in Figure S12 (green data points), in combination with those specified in the earlier account of (red data points). 1 Linear regression fit to all data and Pearson R 2 value appears in the upper right the plot. This equation was subsequently used to transform calculated isotropic hydrogen magnetic shielding values for the tungsten hydride model systems into calculated isotropic 1 H chemical shift values.
With the 1 H magnetic shielding/chemical shift correlation curve established using the data summarized in Figure S13, we now discuss the models chosen to represent the possible surfacesupported species found in 2, 3, and 4. As noted in the main text, and corroborated with additional DFT computations that highlighted a variety of possible products and their relative free S21 energies, it is believed that one of the major products created as a result of hydrogenolysis at 150  Table S4. a As depicted in Figure S14. b Paramagnetic shielding contribution. c Diamagnetic shielding contribution. d Spin-orbit shielding contribution (relativistic contribution). e The isotropic magnetic shielding values in this column are a simple linear combination of the three preceeding columns (i.e., σiso = σd + σp + σSO). f The isotropic chemical shift values were determined using the relationship depicted in Figure S13.
Importantly, it is evident that it is the relativistic spin-orbit mechanism which appears to be responsible for much of the difference in the various computed shifts for these model structures For the other surface-supported species, additional model systems were also considered. In Figure S15, pictures of these models are provided, and the hydrogen magnetic shielding and 1 H S23 chemical shift values are given in Table S5. Atomic coordinates for all systems are presented in this SI.  Table S5). Note that in (B), the structure is identical as provided in structure 'XII' in Scheme 2 of the main paper. Note also that the H3 group faces away from the silica surface in (A), whereas it points towards the surface in (B). As expected, this has a significant influence on the computed 1 H shift values (Table S5). In (C), we provide the model used for . Note the almost bridging nature of one of the hydrogen atoms in the alkylidene moiety. However, it is important that, in each case, the calculated shift value is predicted to be greater than 10 ppm, which is consistent with the region to which it is assigned in the experimental 1 H solid-state NMR spectra. Furthermore, we note that the computational method appeared to not be quite as successful at predicting the 1 H chemical shift value for the W=CH2 group, as well as the group of three hydrides. The geometry-optimized structure displayed in Figure S15C, which is consistent with that also provided in Structure XIV (Scheme 2 of the main text), shows that the S25 methylidene group hydrogen atoms appear to be able to bridge the tungsten atom. It is appropriate to note that, in the literature, there are a very large number of examples whereby a hydrogen atom bridges metal centres in polymetallic systems (included those containing tungsten).                              ::::::::::::::