β-SiH-Containing Tris(silazido) Rare-Earth Complexes as Homogeneous and Grafted Single- Site Catalyst Precursors for Hydroamination

A series of homoleptic rare-earth silazido compounds and their silica-grafted derivatives were prepared to compare spectroscopic and catalytic features under homogeneous and interfacial conditions. Trivalent tris(silazido) compounds Ln{N(SiHMe2)tBu}3 (Ln = Sc (1), Y (2), Lu (3)) are prepared in high yield by salt metathesis reactions. Solution-phase and solid-state characterization of 1− 3 by NMR and IR spectroscopy and X-ray diffraction reveals Ln↼H−Si interactions. These features are retained in solventcoordinated 2·Et2O, 2·THF, and 3·THF. The change in spectroscopic features characterizing the secondary interactions (νSiH, JSiH) from the unactivated SiH in the silazane HN(SiHMe2)tBu follows the trend 3 > 2 > 1 ≈ 2·Et2O > 2·THF ≈ 3· THF. Ligand lability follows the same pattern, with Et2O readily dissociating from 2·Et2O while THF is displaced only during surface grafting reactions. 1 and 2·THF graft onto mesoporous silica nanoparticles (MSN) to give Ln{N(SiHMe2)tBu}n@MSN (Ln = Sc (1@MSN), Y (2@MSN)) along with THF and protonated silazido as HN(SiHMe2)tBu and H2NtBu. The surface species are characterized by multinuclear and multidimensional solid-state (SS) NMR spectroscopic techniques, as well as diffuse reflectance FTIR, elemental analysis, and reaction stoichiometry. A key JSiH SSNMR measurement reveals that the grafted sites most closely resemble Ln·THF adducts, suggesting that siloxane coordination occurs in grafted compounds. These species catalyze the hydroamination/bicyclization of aminodialkenes, and both solution-phase and interfacial conditions provide the bicyclized product with equivalent cis:trans ratios. Similar diastereoselectivities mediated by catalytic sites under the two conditions suggest similar effective environments. ■ INTRODUCTION Complexes containing only one type of ligand (MXn), known as homoleptic compounds, represent the simplest systems for characterizing the nature of metal−ligand interactions because all ligands equivalently contribute electronic and steric effects. The resulting complexes often have intriguing structural and spectroscopic features that are associated with secondary metal−ligand interactions and non-VSEPR geometries. In addition, homoleptic compounds in high oxidation states are often electronically and/or coordinatively unsaturated, giving highly electrophilic metal centers and geometric distortions to counterbalance low electron counts. The nature of the M−X bond is also important to their reactivity; for example, selective substitution of these groups with ancillary ligands (LX) provides routes to reactive complexes, including catalysts. While the rich chemistry of surface-supported organometallic compounds indicates that alkyl species are desirable, work with grafted early-metal amides suggests their emerging potential in catalysis. In rare-earth chemistry, homoleptic organometallic and pseudo-organometallic compounds are particularly important starting materials, but their large ionic radii and low numbers of X-type ligands (either two or three) add to the challenge of preparing reactive monometallic species. As a result, disilazido groups, such as hexamethyldisilazide and tetramethyldisilazide, are the primary N-based ligand types to support monometallic homoleptic rare-earth compounds. Trivalent Ln{N(SiMe3)2}3 15−17 and Ln{N(SiHMe2)2}3 18,19 and divalent compounds are prevalent starting materials for a range of rare-earth chemistries, including as homogeneous catalysts and as precursors for single-site supported rareearth catalysts. Such surface-grafted materials catalyze alkyne dimerization, Tishchenko aldehyde dimerization, hydroamination (the addition of amines and olefins), and polymerization. However, a downside of disilazido complexes as catalyst precursors is that HN(SiMe3)2 and especially HN(SiHMe2)2 can be poor leaving groups due to their relatively high acidity, with pKa values of 25.7 and 22.6, Received: December 23, 2016 Article pubs.acs.org/Organometallics © XXXX American Chemical Society A DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX respectively. Disilazanes are effective silylating agents, and grafting of Ln{N(SiHMe2)2}3 on silica results in significant surface silylation. In addition, compounds containing the smaller N(SiHMe2)2 ligand are often multimetallic (e.g., [La{N(SiHMe2)2}3]2). 31 Thus, the basicity of the bulkier silazido ligand N(SiMe3)tBu was invoked for reactions of Ln{N(SiMe3)tBu}3 (Ln = Y, La) as a precursor to homogeneous hydroamination catalysts. Still, the SiH group in N(SiHMe2)2 provides a valuable spectroscopic handle for both NMR and IR analysis, stabilization of coordinatively and electronically unsaturated metal centers through labile secondary interactions, and a site for reactivity. The silazide N(SiHMe2)tBu incorporates a number of these desired features: enhanced steric protection, a more basic amide group, and the SiH moiety. This silazido ligand has been underutilized as a supporting ligand in homoleptic compounds in comparison to the disilazido ligands, despite the early promise of the only trivalent homoleptic Er{N(SiHMe2) tBu}3 39 and the rich chemistry of Cp2Zr{N(SiHMe2)tBu}X (X = hydride, halide, alkyl). Both of these systems, as well as the main-group compound [Mg{N(SiHMe2)tBu}2]2, 43 exhibit structural and spectroscopic features associated with multicenter M↼H−Si interactions, including short M···H distances and small ∠M−N−Si angles in X-ray diffraction studies, low-energy νSiH bands in infrared spectra, upfield δSiH signals in H NMR spectra, and low JSiH values in Si and H NMR spectra. The NMR properties, however, have not been evaluated for Er{N(SiHMe2)tBu}3 because of its paramagnetism, although the solid-state structure and infrared spectra established that all three SiH groups interact with the rare-earth center. Thus, the N(SiHMe2)tBu ligand could provide useful precursors for catalysis, such as hydroamination. Despite the high reactivity of disilazido rare-earth compounds as precatalysts for this process, examples of grafted single-site rare-earth hydroamination catalysts are limited. Moreover, those examples suggested that silica-supported catalysts are diminished in activity in comparison to the homogeneous analogues. A number of additional challenges face the development of the catalytic hydroamination reaction, including functional group tolerance, catalytic efficiency for intermolecular additions, and control over selectivity. The selectivity and activity in catalytic conversions of aminodialkenes could provide a means for examining the effect of surface and pore environment on hydroamination processes; because both monoand bicyclization products are possible, each product has cis and trans diastereomers (Scheme 1) and the diastereoselectivity is sensitive to reaction conditions. For example, we recently reported that substrate concentration affected the cis:trans ratio in an enantioselective Zr-catalyzed monocyclization reaction of aminodialkenes and aminodialkynes to give optically active pyrrolidines. In addition, Marks and co-workers showed in their seminal study that diastereoselectivity in Cp*2LaCH(SiMe3)2-catalyzed hydroaminations of chiral aminoalkenes is also influenced by concentration. Rare-earth compounds and a few zirconium catalysts give hydroamination/bicyclization products through a two-step process in which the second cyclization requires conditions more forcing than those in the first. Ligand−metal or surface−metal center interactions might provide control over selectivity in the hydroamination of aminodialkenes. Experiments are needed to test for surface effects on (a) monocyclization vs bicyclization of aminodialkenes and (b) cis:trans selectivity of the products to elucidate factors that ultimately control selectivity in such C−N bond forming reactions. Controlling selectivity in these reactions has synthetic value in terms of additional transformations of the olefin-substituted heterocycles. In addition, the azabicyclo[2.2.1]heptane product contains motifs found in natural products and biologically active substances, and the exo and endo selectivity is also important for their applications. Thus, homoleptic monometallic compounds of the type Ln{N(SiHMe2)tBu}3 may be effective precatalysts and precursors for single-site heterogeneous catalysts. The present study describes our efforts to synthesize a series of homoleptic rare-earth silazido compounds. The NMR, IR, and structural properties of Ln{N(SiHMe2)tBu}3 and their ethereal solvent adducts were analyzed to provide molecular models for surfacebonded species. Such surface-supported analogues are obtained by grafting on mesoporous silica nanoparticles (MSN). Characterization of MSN-supported rare-earth silazido materials via IR and solid-state NMR spectroscopy provides a molecular picture of the surface sites. With this picture and the spectroscopic comparison between solution-phase homoleptic vs grafted species in hand, we studied their catalytic reactivity (activity and selectivity) in hydroamination/cyclization of aminoalkenes and aminodialkenes. ■ RESULTS AND DISCUSSION Synthesis and Spectroscopic Characterization of Ln{N(SiHMe2)tBu}3 and Ln{N(SiHMe2)tBu}3L. Reactions of three equiv of [LiN(SiHMe2)tBu] and LnCl3 (Ln = Y, Lu) or LnCl3THF3 (Ln = Sc, Lu) in THF or Et2O provide Ln{N(SiHMe2)tBu}3 or Ln{N(SiHMe2)tBu}3L, as outlined in Scheme 2 (Ln = Sc (1), Y (2, 2·Et2O, 2·THF), Lu (3, 3· THF)). Compound 1 is isolated as a light yellow sticky solid, and neither Et2O nor THF is retained in the scandium’s coordination sphere. The solvent-free Y compound may be obtained by subliming 2·Et2O or by performing the synthesis under concentrated conditions (0.46 M). The solvent-free Lu compound is obtained from the reaction of LuCl3 in Et2O. The complexes 2·Et2O, 2·THF, and 3·THF are isolated as white sticky solids from pentane crystallization or precipitation. Sublimation of the sticky solids of 2·THF and 3·THF affords analytically pure powders while retaining the THF ligand. While the H NMR spectrum of 2·THF is not altered by sublimation, the νSiH region of the infrared spectra is slightly sharper after this treatment (see the Supporting Information). The infrared spectra of the series of compounds contained bands attributable to Si−H stretching modes, ranging from 2019 to 1849 cm−1 (Figure 1 and Table 1; see the Supporting Information for full IR spectra). Spectra for 1−3 and 2·Et2O Scheme 1. Cis and Trans Diastereomers Accessible from Monocyclization and Bicyclization of Aminodialkenes Organometallics Article DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX B revealed a single strong band assigned to bridging Ln↼H−Si groups, with the tricoordinated scandium complex’s peak appearing at higher energy than the signal for the ethercoordinated yttrium species. The signal for the νSiH band of 3 (1849 cm−1) appeared with the lowest energy of the series, which follows the trend 3 < Er{N(SiHMe2)tBu}3 (1858 cm−1) < 2 < 1. In contrast, the νSiH region for nonsublimed 2· THF contained two peaks at 2019 and 1967 cm−1 at notably higher energy; once it was sublimed, a signal at 2117 cm−1 was detected due to sharpening of the broad 2019 cm−1 band. The signals for the SiH group in the lutetium THF adduct were observed around 2000 cm−1. For comparison, the SiH stretching frequencies of the silazane HN(SiHMe2)tBu and lithium silazido [LiN(SiHMe2)tBu] appeared at higher energy in comparison to those of the rare-earth silazido compounds. Room-temperature H NMR spectra suggested that the homoleptic rare-earth species are C3v symmetric on the basis of three resonances, which were assigned to the SiH, SiMe2, and tBu groups in equivalent silazido ligands. The JHH coupling in SiHMe2 is small and resolved clearly as doublets (3 Hz) for the Me in 2, 2·THF, 3, and 3·THF. The JSiH values (Table 1) vary depending on the rare-earth element and the coordinated THF or Et2O ligands but are generally low and suggest Ln↼H−Si bonding motifs. The JSiH values in HN(SiHMe2)tBu (192 Hz) and [LiN(SiHMe2)tBu] (168 Hz) are larger than those in the rare-earth compounds. At low temperature (190 K), the SiMe2 signals in the H NMR spectra of 1, 2·THF, and 3·THF Scheme 2. Synthesis of Homoleptic Silazido Rare-Earth Compounds Figure 1. Infrared spectra of HN(SiHMe2)tBu, [LiN(SiHMe2)tBu], 1, 2·Et2O, and 2·THF (before sublimation) corresponding to Si−H stretching modes. Full spectra are shown in the Supporting Information. Table 1. Spectroscopic Data for Silazido Compounds compound δSiH, ppm JSiH, Hz δSi, ppm δN, ppm νSiH, cm −1 Sc{N(SiHMe2)tBu}3 (1) 4.18 125 −22.9 −208 1893 Y{N(SiHMe2)tBu}3 (2) 4.26 124 −24.8 −221 1860 2·Et2O 4.30 126 −25.9 −222 1864 2·THF 4.59 143 −30.5 −231 2019, 1967 Lu{N(SiHMe2)tBu}3 (3) 4.42 121 −20.5 −221 1849 3·THF 4.63 137 −28.7 −232 1988 HN(SiHMe2)tBu 4.83 192 −18.8 −329 2135, 2104 [LiN(SiHMe2)tBu] 4.87 168 −23.1 −301 2055 Organometallics Article DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX C appeared as two signals of equal intensity, implying lowtemperature C3-symmetric structures. Interestingly, the SiH chemical shift and coupling constants were identical in spectra acquired down to 190 K, and this observation suggested that the fluxional process did not involve disruption of the secondary Ln↼H−Si interactions. In contrast, the H NMR spectra of 2, 2·Et2O, and 3 merely broadened at 190 K in toluene-d8. The Si NMR spectra vary from −20.5 to −30.5 ppm depending on the identity of the rare-earth element, and these were slightly upfield in comparison to HN(SiHMe2)tBu. A similar trend was observed in the Si NMR spectra of the rareearth disilazido compounds Ln{N(SiHMe2)2}3THFn, which are ca. 10 ppm upfield in comparison to the disilazane HN(SiHMe2)2 (−11.1 ppm). 18 In addition, H−N HMQC experiments (at natural abundance) revealed cross-peaks between N and tBu signals but not with the SiHMe2 group. The N NMR chemical shifts were downfield in comparison to those of HN(SiHMe2)tBu and [LiN(SiHMe2)tBu] (see Table 1). We also noticed the same trend in the N NMR chemical shifts for Sc{N(SiHMe2)2}3THF (−253 ppm) and Y{N(SiMe3)2}3 (−243.1 ppm), which are downfield of those for HN(SiHMe2)2 (−365 ppm) and HN(SiMe3)2 (−354 ppm). Likewise, the N NMR chemical shifts for Cp2Zr{N(SiHMe2) tBu}H (−260 ppm) and Cp2Zr{N(SiHMe2)2}H (−292 ppm) are downfield with respect to those of HN(SiHMe2)tBu (−329 ppm) and HN(SiHMe2)2 (−365.3 ppm). 36 X-ray Crystallography. Single-crystal X-ray diffraction studies provided solid-state structures of 1 (Figure 2; see the Supporting Information for other structures), 2, 2·Et2O, 2·THF (Figure 3), 3, and 3·THF for comparison to Er{N(SiHMe2) tBu}3. 39 The molecular structures of 1−3 and the erbium analogue are similarly pseudo-C3 with the N(SiHMe2)tBu ligands adopting a propeller-like conformation. All three SiH groups are located (identified objectively on the difference Fourier map) on the same face of the LnN3 core, and these groups are directed toward the rare-earth center. The methyl groups in the SiMe2 are inequivalent in these structures, and this is consistent with the low-temperature H NMR spectrum described above. The LnN3 cores of 1−3 adopt similar trigonal geometries distorted by pyramidalization (∑NLnN: Sc, 348.62(9)°; Y, 351.3(2)°; Lu, 349.0(2)° (vs Er, 350.42°)). There are three short Ln···H and three short Ln···Si distances. Remarkably, the scandium−silicon distances in 1 (Sc1−Si1, 2.8603(3) Å; Sc1− Si2, 2.8343(4) Å; Sc1−Si3, 2.8557(4) Å) are similar to the distance in the scandium silyl compound Cp2ScSi(SiMe3)3THF (2.863(2) Å) and only slightly longer than that in Cp*2ScSiH2SiPh3 (2.797(1) Å), 57 both of which contain bona fide two-center−two-electron Sc−Si bonds. These short distances are complemented by the Sc−N−Si angles, which are much smaller than the 120° expected for a trigonal-planar N center (98.47(4), 97.98(4), and 97.93(4)°). Taking into consideration the short distances to N and Si, the N3Si3 atoms form a trigonal prism, with the smaller N3 end-capping triangle twisted from the triangular face composed of Si3 vertices. The Sc center is 0.41 and 1.02 Å from the N3 and Si3 planes, respectively. The scandium−hydrogen distances (Sc1−H1s, 2.26(1) Å; Sc1−H2s, 2.20(2) Å; Sc1−H3s, 2.23(1) Å), however, are significantly longer than the calculated distance in ScH3 (1.82 Å). 58 For comparison, the homoleptic, solvent-free tris(amido)scandium compound Sc{N(SiMe3)2}3 is pyramidal in the solid state (∑NScN = 346.5°) but planar in the gas phase. In that compound, the solid-state and gasphase Sc−N distances (2.047(2) and 2.02(3) Å, respectively) are slightly shorter than those in 1 (Sc1−N1, 2.0656(6) Å; Sc1−N2, 2.063(1) Å; Sc1−N3, 2.071(2) Å). The Sc−N distances in 1, however, are similar to those in the fourcoordinate THF adduct Sc{N(SiHMe2)2}3THF. 18 As expected, the Ln−N, Ln···Si, and Ln···H distances in 2 and 3 are longer than those in 1 (see the Supporting Information). The Y−N distances (2.223(2), 2.223(2), and 2.227(2) Å) are slightly Figure 2. Rendered thermal ellipsoid plot of Sc{N(SiHMe2)tBu}3 (1) at 50% probability. See the Supporting Information for isostructural yttrium (2) and lutetium (3). H atoms bonded to Si were located objectively in the Fourier difference map and are included in the rendition; all other H atoms are not included for clarity. Figure 3. Rendered thermal ellipsoid plot (50% probability) of Y{N(SiHMe2)2}3THF (2·THF). See the Supporting Information for solid-state structures of 2·Et2O and 3·THF. H atoms bonded to Si were located objectively in the Fourier difference map and are included in the rendition; all other H atoms are not included for clarity. Organometallics Article DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX D longer than the Er−N distances (2.206(2) Å), while the Lu− N distances (2.174(2), 2.177(2), and 2.178(2) Å) are shorter. The yttrium species 2, 2·Et2O, and 2·THF are further compared because the last two compounds have distinct νSiH IR bands and JSiH values that suggest inequivalent structures. At the same time, the coordination numbers of 2 and 2·Et2O are inequivalent, although their spectroscopic features are similar. 2·THF contains two crystallographically unique molecules (Z = 8) per unit cell, whereas 2 and 2·Et2O crystallize with only one (Z = 4). All three crystallographically unique molecules of 2·L are four-coordinate on the basis of the YN3O core, with the YN3 part flattened (∑NYN = 344.7(3)° (2·Et2O), 346.4(3) and 347.9(3)° (2·THF)) in comparison to the sum of three angles of an ideal tetrahedron (∑ = 327°). In addition, one of the N− Y−O angles is ca. 90° in each of the structures (i.e., the molecules lack even a pseudo-C3 axis). All three Si−H groups point toward the Y center, and each of these H atoms is pseudo-trans to either a silazide or ether ligand (e.g., in 2·Et2O H1s−Y1−N2 is 175(1)°, H2s−Y1−N3 is 152(1)°, and H3s− Y1−O1 is 153(1)°). The Y−O distances for diethyl ether and THF are nearly identical (2·Et2O, 2.377(3) Å; 2·THF, 2.385(2) and 2.376(2) Å), even though Et2O is removed during sublimation while THF is not. The similar conformations, as well as the interatomic angles and distances associated with Y−N−Si−H structural motifs of the 2·L yttrium species, are in contrast with the distinguishing νSiH and JSiH spectroscopic features noted above. The geometry of the lutetium analogue Lu{N(SiHMe2)tBu}3THF (3·THF; Figure S24) is similar to that of 2·Et2O and 2·THF, yet the SiHcentered spectroscopic features are between those of the two yttrium species (Table 1). Thus, the trends of Ln···H and Ln··· Si distances do not correlate one to one with energies and coupling constants indicated by the spectroscopic signatures, although the features are consistently present in all the compounds. Synthesis and Characterization of Ln{N(SiHMe2) tBu}n@MSN. Compounds 1 and 2·THF were stirred with SBA-type mesoporous silica nanoparticles (MSN) previously heated under vacuum either at 550 °C (MSN550, 1.5 mmol of OH/g) or 700 °C (MSN700, 0.9 mmol of OH/g) to graft the rare-earth species on the material, as depicted in Scheme 3. These rare-earth elements were initially chosen for study because Anwander and co-workers showed that grafted yttrium complexes are more active in hydroamination/cyclization than the corresponding grafted lanthanide catalysts, and we also wished to compare mild conditions for cyclization with diastereoselective Zr-catalyzed hydroamination (see below). Micromole-scale grafting reactions were performed in benzened6 and monitored by H NMR spectroscopy, while preparativescale syntheses were performed in pentane at room temperature for 20 h. The former experiments provided an initial estimate of loading and possible surface species on the basis of reaction stoichiometry (Table 2). For example, a micromolescale reaction in benzene-d6 consumed 0.48 mmol of 2·THF and produced 0.33 mmol of tBuNH2 and 0.35 mmol of HN(SiHMe2)tBu per gram of MSN550. In addition, the intensities of THF H NMR signals are not diminished in the spectroscopically monitored grafting experiments, implying that THF dissociates from surface-bonded Y sites and coordinates to or rapidly exchanges with the solvent in 2· Scheme 3. Surface Grafting Reactions and Proposed Surface-Supported Homoleptic Silazido Rare-Earth Compounds Formed from MSN550 or MSN700 and Sc{N(SiHMe2)tBu}3 (1) or Y{N(SiHMe2)tBu}3THF (2·THF) Table 2. Stoichiometry of Surface Grafting Reactions preparation amt of Ln consumed (mmol) amt of tBuNH2 measd (mmol) amt of silazane measd (mmol) 1 + MSN550 0.54 0.37 0.46 1 + MSN700 0.35 0.14 0.21 2·THF + MSN550 0.48 0.33 0.35 2·THF + MSN700 0.22 0.05 0.15 Organometallics Article DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX E THF. These experiments provide a rough estimate of the yttrium loading (see Table 2, 0.48 mmol/g in this example), the average podality (∼1:1 monopodal and bipodal in this example), and the quantity of surface silylation in the grafting experiments. Notably, less rare-earth amide is consumed and less tBuNH2 and HN(SiHMe2)tBu are formed in reactions with MSN700, while the tBuNH2:HN(SiHMe2)tBu ratio also decreased in experiments with the high-temperature-treated MSN. The ratio of consumed rare-earth silazide to silazane and amine produced in reactions with MSN700 suggests that the grafted species are primarily monopodal in those cases. Systematic and corroborative quantitative analysis with inductively coupled plasma optical emission spectroscopy (ICP-OES) and CHN combustion analysis supports the initial estimates. The loading of grafted metal species was quantified by ICPOES, while the loading of N(SiHMe2)tBu ligands was measured by nitrogen mass percentage using CHN combustion analysis (Table 3). The N:Sc ratio of 1.3:1 for 1@MSN550 suggested a mixture of bipodal mono(silazido)scandium ( SiO−)2ScN(SiHMe2)tBu and monopodal bis(silazido)scandium SiO−Sc{N(SiHMe2)tBu}2 surface species. As in the above experiments that measure stoichiometry, these values average the composition of the surface species rather than provide a precise structure. Alternatively, the N:Sc ratio of 1.9 for 1@MSN700 implies that bis(silazido) scandium SiO− Sc{N(SiHMe2)tBu}2 is the dominant surface species. In addition, an excess of carbon is present on the surface. The C:N ratio in a N(SiHMe2)tBu ligand is 5.15:1, whereas the measured C:N ratios are higher for grafted materials (e.g., 1@ MSN550, C:N = 8.1:1). This higher carbon loading is readily rationalized by a silylation of the surface silanols, as reported for disilazanes HN(SiMe3)2 and HN(SiHMe2)2 and supported by solid-state NMR experiments and IR spectroscopy (see below). The observation of tBuNH2 in the supernatant is also consistent with such a process. Less surface silylation occurs in grafting reactions involving MSN700 in comparison with MSN550. Diffuse reflectance IR spectra of the rare earth silazide treated materials (Figure 4), in comparison to pristine MSN550 and MSN700, revealed that isolated silanols are consumed in the grafting reactions and the new surface species contain CH and SiH groups. In all reactions of MSN and rare-earth silazides, the absorption band at 3747 cm−1 assigned to isolated silanol groups disappeared upon grafting; however a broad signal from 3740 to 3280 cm−1 assigned to hydrogen-bonded silanols was apparent in the grafted materials’ spectra. These remaining SiOH groups were not readily accessible for reactivity, as demonstrated by the trace amounts of toluene detected upon addition of Mg(CH2Ph)2(O2C4H8)2. The SiH region of the diffuse reflectance IR spectra of 2@MSN contained a sharp signal at 2149 cm−1 and a broad signal from 2080 to 1780 cm−1 with a maximum at 1924 cm−1. The former signal was assigned to SiO−SiHMe2 surface groups on the basis of comparison with MSN independently treated with HN(SiHMe2)tBu or HN(SiHMe2)2 (at 2152 cm −1; see Figure 4D) and literature reports. The SiO−SiHMe2 functionality arises from the reaction of silanols and HN(SiHMe2)tBu, the byproduct from grafting of 1 or 2·THF. The broad signal contained features around 2000 and 1900 cm−1 assigned to terminal Si−H and bridging Y↼H−Si groups in surface-grafted 2@MSN. The diffuse reflectance IR spectra of the scandium material 1@MSN suggest the surface scandium silazido species also contain bridging Sc↼H−Si groups (see Figure S25). Characterization by Solid-State NMR. The atomic-scale structures of Ln{N(SiHMe2)tBu}n@MSNs (Ln = Y, Sc) were further probed by solid-state (SS) NMR spectroscopy. The experimental parameters are given in the figure captions using the following symbols: νR denotes the magic angle spinning (MAS) rate, νRF(X) is the magnitude of the RF magnetic field at the resonance frequency of the X nucleus, τCP is the crosspolarization (CP) contact time, Δt1 is the increment of t1 during 2D acquisition, τRD is the recycle delay, and NS is the number of scans (see the Supporting Information for more experimental details). Here, we will report the spectra obtained for 2@MSN550 and 2@MSN700; the Sc-containing analogues 1@MSN550 and 1@MSN700 yielded very similar results (Figures S29−S33), to which we will refer when appropriate. The 2D C−H heteronuclear correlation (HETCOR) spectrum of 2@MSN550, acquired using the indirectly (or H) detected scheme, referred herein as H{C} idHETCOR, is shown in Figure 5A. The spectrum is consistent with the presence of MSN-bound Y{N(SiHMe2)tBu}n species Table 3. Quantification of Ln, N, and C using ICP-OES and CHN (Combustion) Analysis preparation amt of Ln (wt %) amt of Ln (mmol/g) amt of N (mmol/g) N:Ln amt of C (mmol/ g) C:N Sc{N(SiHMe2)tBu}3 + MSN550 (1@MSN550) 2.5 0.56 0.75 1.3 6.1 8.1 Sc{N(SiHMe2)tBu}3 + MSN700 (1@MSN700) 1.5 0.33 0.64 1.9 4.3 6.8 Y{N(SiHMe2)tBu}3THF + MSN550 (2@ MSN550) 4.2 ± 0.1 0.47 ± 0.01 0.86 ± 0.01 1.83 ± 0.05 6.4 ± 0.1 7.5 ± 0.2 Y{N(SiHMe2)tBu}3THF + MSN700 (2@ MSN700) 2.3 0.26 0.54 2.1 3.4 6.4 Error is given on the basis of standard deviation determined from four measurements on 2@MSN550. Figure 4. Diffuse reflectance infrared spectra of (A) MSN550, (B) Y{N(SiHMe2)tBu}3THF, (C) Y{N(SiHMe2)tBu}3THF + MSN550, and (D) HN(SiHMe2)tBu + MSN550. See the Supporting Information for Sc and MSN700 analogues. Organometallics Article DOI: 10.1021/acs.organomet.6b00956 Organometallics XXXX, XXX, XXX−XXX F shown in Figure 5B. The assignments of H and C resonances (Table 4) are based on the δC values reported for Y{N(SiHMe2)tBu}3 (2; Figure S10) and tert-butylamine (H2NCMe3). 67 The correlations involving Si−H(H3) are suppressed in Figure 5A because the H{C} idHETCOR scheme uses two CP transfers (H → C and C → H); however, these protons are quantitatively observed in the H MAS SSNMR spectra of 2@MSNs (Figure S27). Note that the carbon intensity in dimethylsilyl groups (C1) relative to C2 decreases in the MSN700 sample (Figure S28), suggesting that C1 also includes surface-bound −SiHMe2 groups whose population is reduced on MSNs pretreated at higher temperature (see also the discussion of Si spectra below). Importantly, we do not detect any THF ligands in the 2@MSN550 sample, which would yield C resonances at 26 and 68 ppm. The 1D Si{H} CPMAS spectrum of 2@MSN550 (Figure 5C) shows a resonance centered at ∼−105 ppm attributable to the so-called Q sites forming the MSN framework and described by the general formula (SiO)nSi(OR)4−n; here mainly with n = 4, 3, where R = Ln, H (Figure 5B). The Q sites are underrepresented in the CPMAS spectrum and are almost invisible in the H{Si} idHETCOR spectra, due to the lack of H nuclei suitable for cross-polarization of these sites. The signal at −3 ppm is assigned to −OSiHMe2 groups directly bound to the silica surface (typically denoted as M sites), on the basis of the H{Si} idHETCOR spectrum of an MSN sample grafted with HN(SiHMe2)2, which shows a dominant cross-peak at δSi −3 ppm and δH 4.5 ppm (Figure 5D). The Si1 peak in Figure 5E,F correlates strongly to H3 and weakly to H1 and H2, and thus represents the silicon site in yttrium complexes. Importantly, there is a difference of ~0.2 ppm between δH values for H3 correlated to M and Si1 (this is clearly seen in Figure 5F where there is a shoulder on the upfield side of H3). This shift may suggest the presence of an Y↼H-Si structure in the silica-bound complex, which we unambiguously confirmed by the 2D J-resolved experiment (vide infra). Again, the intensity of M sites relative to Si1 is lower in the sample prepared using MSNs pretreated at higher temperature (compare Figure 5E and 5F). The remaining weak peaks Si2′ and Si2′′, which are only observed in samples grafted with metal species, most likely represent silica-bound (−O)2SiMe2 and (−O)2SiHMe sites formed through processes mediated by the rare earth center (Table 4). The latter Figure 5. (A) H{C} idHETCOR spectrum of Y{N(SiHMe2)tBu}n@MSN550 (2@MSN550), obtained at 14.1 T using νR = 36 kHz, νRF( H 90°, CP) = 100 kHz, νRF( C 90°) = 100 kHz, νRF( C CP) = 64 kHz, τCP1 = τCP2 = 6 ms, νRF( H HORROR) = 18 kHz, νRF( H SPINAL-64) = 10 kHz, νRF( C SPINAL-64) = 100 kHz, τRD = 1.5 s, Δt1 = 27.8 μs (256 rows), and NS = 64. (B) Postulated constitution of MSN-bound surface species. (C) Si{H} CPMAS spectrum of 2@MSN550. (D−F) H{Si} idHETCOR spectra of SiHMe2@MSN550, 2@MSN550, and 2@MSN700, respectively. Spectra C−F were obtained at 9.4 T using νR = 18 kHz, νRF(H CP) = 83 kHz, νRF(Si CP) = 65 kHz, τCP = 3 ms (C), τCP1 = τCP2 = 4 ms (D), τCP1 = τCP2 = 2 ms (E, F), νRF( H SPINAL-64) = 83 kHz, νRF( Si SPINAL-64) = 68 kHz (D−F), νRF(H HORROR) = 18 kHz (D−F), τRD = 1.3 s, NS = 10000 (C), Δt1 = 55.5 μs (D−F), 128 rows (D−F), and NS = 256 (D−F). Table 4. Summary of Peak Assignments in the SSNMR Spectra of Y{N(SiHMe2)tBu}n@MSN (2@MSN) resonance chemical shift (δ, ppm) assignment