New Molybdenum(II) Complexes with α-Diimine Ligands: Synthesis, Structure, and Catalytic Activity in Olefin Epoxidation

Three new complexes [Mo(η3-C3H5)Br(CO)2{iPrN=C(R)C5H4N}], where R = H (IMP = N-isopropyl 2-iminomethylpyridine), Me, and Ph, were synthesized and characterized, and were fluxional in solution. The most interesting feature was the presence, in the crystal structure of the IMP derivative, of the two main isomers (allyl and carbonyls exo), namely the equatorial isomer with the Br trans to the allyl and the equatorial with the Br trans to one carbonyl, the position trans to the allyl being occupied by the imine nitrogen atom. For the R = Me complex, the less common axial isomer was observed in the crystal. These complexes were immobilized in MCM-41 (MCM), following functionalization of the diimine ligands with Si(OEt)3, in order to study the catalytic activity in olefin epoxidation of similar complexes as homogeneous and heterogeneous catalysts. FTIR, 13C- and 29Si-NMR, elemental analysis, and adsorption isotherms showed that the complexes were covalently bound to the MCM walls. The epoxidation activity was very good in both catalysts for the cis-cyclooctene and cis-hex-3-en-1-ol, but modest for the other substrates tested, and no relevant differences were found between the complexes and the Mo-containing materials as catalysts.

[Mo(CO) 4 (IMP)] was also immobilized in MCM-41 (MCM) with a spacer, and both the complex and the porous material were tested in the catalytic epoxidation of cis-cyclooctene [5]. More interestingly, W(IV) complexes, [WBr 4 (R 1 ,R 2 -IMP)] (R 1 = H, R 2 = Me, Et, n Pr, tBu), were also synthesized and their magnetic properties studied by 1 H-NMR [6]. Other metals were chosen to prepare complexes, namely Cu(I) with R 1 = H and several R 2 chains, which were active in MMA polymerization [7]. Palladium nanoparticles with IMP supported in functionalized silica were characterized and shown to be active in Suzuki-Miyaura cross coupling reactions and hydrogenation in water [8]. Another recent report addressed the selective oxidative cleavage of 1-octene catalyzed by arene Ru(II) complexes of R 1 ,R 2 -IMP (R 1 = H, R 2 = functionalized chains) [9]. Besides catalysis, IMP ligands have also been studied owing to the photochemical properties exhibited by some of their derivatives, namely [Os 3 (CO) 19 (IMP)] clusters [10].
We synthesized Mo(II) complexes [Mo(X) 2 (CO) 3 (IMP)] (X = I, Br) and studied their catalytic activity in olefin epoxidation in the homogeneous phase and after their immobilization in MCM, having found an improved efficiency in the immobilized catalysts [11]. Since the complexes [Mo(η 3 -C 3 H 5 )X(CO) 2 (L 2 )] (X = halide, L 2 = bidentate ligand) are often more active catalysts than [Mo(X) 2 (CO) 3 (L 2 )], we decided to prepare several derivatives of IMP with R 1 = Me or Ph, as shown in Scheme 1 and designated as Me-IMP and Ph-IMP for facility. They were also immobilized in MCM, and the activity of the complexes and obtained materials was studied.
We synthesized Mo(II) complexes [Mo(X)2(CO)3(IMP)] (X = I, Br) and studied their catalytic activity in olefin epoxidation in the homogeneous phase and after their immobilization in MCM, having found an improved efficiency in the immobilized catalysts [11]. Since the complexes [Mo(η 3 -C3H5)X(CO)2(L2)] (X = halide, L2 = bidentate ligand) are often more active catalysts than [Mo(X)2(CO)3(L2)], we decided to prepare several derivatives of IMP with R1 = Me or Ph, as shown in Scheme 1 and designated as Me-IMP and Ph-IMP for facility. They were also immobilized in MCM, and the activity of the complexes and obtained materials was studied.
The 1 H-NMR spectrum of 1 shows at least two isomers in solution. Two sets of peaks for H anti protons at 1.25 and 1.44 ppm, two sets for the H syn protons at 3.95 and 4.00 ppm, and one large peak at 4.46 assigned to H meso were associated to the η 3 -C 3 H 5 group. Three peaks at 1.09, 2.31, and 4.24 ppm were assigned to the hydrogens of the propyl chain of the coordinated IMP ligand, and those at 7.69 (H6), 7.86 (H7), 7.48 (H8), and 8.48 (H9) ppm, shifted from their position relative to the free ligand (7.87 (H6), 7.61 (H7), 7.98 (H8), and 8.70 (H9) ppm), belong to the pyridine ring. The signal of the H3 atom was observed at 8.24 ppm. Other signals were weaker and could not be completely assigned. Selected spectra for this and the other complexes are shown in Figures S1-S5 and Table S1 in Supporting Information (SI).
The isomers observed in the 1 H-NMR spectrum arose from the fluxional behavior of the allyl ligand, as sketched in Scheme 3. The two isomers present in higher amounts should be the equatorial and the axial ones (exo) and their interconversion proceeded by allyl rotation [12,13]. Their ratio depended on the steric and electronic effects of the substituents. coordination to the metal. They vary from 1655 to 1647 cm −1 (1), 1638 to 1593 cm −1 (2), and 1629 and 1586 cm −1 (3). The 1 H-NMR spectrum of 1 shows at least two isomers in solution. Two sets of peaks for Hanti protons at 1.25 and 1.44 ppm, two sets for the Hsyn protons at 3.95 and 4.00 ppm, and one large peak at 4.46 assigned to Hmeso were associated to the η 3 -C3H5 group. Three peaks at 1.09, 2.31, and 4.24 ppm were assigned to the hydrogens of the propyl chain of the coordinated IMP ligand, and those at 7.69 (H6), 7.86 (H7), 7.48 (H8), and 8.48 (H9) ppm, shifted from their position relative to the free ligand (7.87 (H6), 7.61 (H7), 7.98 (H8), and 8.70 (H9) ppm), belong to the pyridine ring. The signal of the H3 atom was observed at 8.24 ppm. Other signals were weaker and could not be completely assigned. Selected spectra for this and the other complexes are shown in Figure S1-S5 and Table S1 in Supporting Information (SI).
The isomers observed in the 1 H-NMR spectrum arose from the fluxional behavior of the allyl ligand, as sketched in Scheme 3. The two isomers present in higher amounts should be the equatorial and the axial ones (exo) and their interconversion proceeded by allyl rotation [12,13]. Their ratio depended on the steric and electronic effects of the substituents. The 1 H-NMR spectrum of 2 is very similar-with two sets of three signals corresponding to the allyl protons at 4.41 and 4.80 ppm (two multiplets), 4.26 and 4.11 ppm (two doublets), and 1.61 and 1.49 ppm (two doublets), which were assigned to Hmeso, Hsyn, and Hanti protons-, indicating the presence of two isomers in solution. The signal of H4 was replaced by the signal of the three methyl hydrogens.
The 1 H and 13 C{1H} NMR were studied in more detail using COSY and HMQC experiments (SI, Figure S3-S5). The COSY spectrum allowed the identification of three sets of signals, from the four doublets (1.16, 1.25, 1.34, and 1.48 ppm Hanti), three signals (2.21, 3.14, and 3.28 ppm Hsyn), and three multiplets (3.58, 3.76, and 3.95 ppm Hmeso). From the correlation, it was possible to assign one set for the equatorial isomer (exo), another for the equatorial isomer (endo), and the third-comprising two doublets, for the Hanti-for the axial isomer (Scheme 2). Indeed, the axial isomer was less symmetric and the two Hanti were more inequivalent [14].
The crystal structures of [MoBr(η 3 -C3H5)(CO)2(IMP)] (1) and [MoBr(η 3 -C3H5)(CO)2(Me-IMP)] (2) were determined by single crystal X-ray diffraction, and are shown in Figure 1 and Figure 2, respectively. Selected bond distances and angles for 1 and 2 are given together in Table 1. The 1 H-NMR spectrum of 2 is very similar-with two sets of three signals corresponding to the allyl protons at 4.41 and 4.80 ppm (two multiplets), 4.26 and 4.11 ppm (two doublets), and 1.61 and 1.49 ppm (two doublets), which were assigned to H meso , H syn , and H anti protons-, indicating the presence of two isomers in solution. The signal of H4 was replaced by the signal of the three methyl hydrogens.
The 1 H and 13 C{1H} NMR were studied in more detail using COSY and HMQC experiments (SI, Figures S3-S5). The COSY spectrum allowed the identification of three sets of signals, from the four doublets (1.16, 1.25, 1.34, and 1.48 ppm H anti ), three signals (2.21, 3.14, and 3.28 ppm H syn ), and three multiplets (3.58, 3.76, and 3.95 ppm H meso ). From the correlation, it was possible to assign one set for the equatorial isomer (exo), another for the equatorial isomer (endo), and the third-comprising two doublets, for the H anti -for the axial isomer (Scheme 2). Indeed, the axial isomer was less symmetric and the two H anti were more inequivalent [14].

Complex Isomer 1 (Equatorial) 1 (Axial) 2
The selected bond distances and angles for 1 and 2 are given together in Table 1. In both complexes, the allyl ligand adopted an exo conformation, with the terminal allylic carbons over the carbonyls, as sketched in Scheme 3. The coordination geometry around the molybdenum center can be described as pseudo octahedral, with the centroid of the allyl ligand and the two carbons from the cis-carbonyl groups determining a fac arrangement, as previously found in many structures [14]. Moreover, the asymmetric unit of 1 is composed of two geometric isomers with the ligands exhibiting different spatial dispositions in the metal coordination sphere. Thus, in the axial isomer, a nitrogen donor of IMP ligand occupied an axial coordination site (Figure 1, left side), while in the equatorial isomer, the two nitrogen donors of this chelating ligand defined the equatorial coordination plane together with the two carbonyl ligands (Figure 1, right side). The existence of these two isomers in the solid state were in agreement with the 1 H-NMR structural findings observed in the solution.

Chemical Studies: Materials
The three new complexes, 1, 2 and 3, were immobilized in MCM-41 (MCM) according to the reported procedure [16] described in Scheme 4. The MCM-41 was synthesized by a template approach [17].
In the first step, the silylated ligands (IMP-Si, Me-IMP-Si, and Ph-IMP-Si) reacted with the MCM surface OH groups to afford the ligand functionalized materials MCM-IMP-Si, MCM-Me-IMP-Si, and MCM-Ph-IMP-Si. The second step consisted of the reaction between these materials and the precursor complex [Mo(η 3 -C 3 H 5 )Br(CO) 2 (MeCN) 2 ] with formation of three Mo-containing materials (MCM-1, MCM-2, and MCM-3).
CHN analyses of MCM-1, MCM-2, and MCM-3 materials showed 6.01% C, 1.92% H, and 1.64% N for MCM-1; 5.08% C, 1.90% H, and 2.03% N for MCM-2; and 14.16% C, 2.26% H, and 2.68% N for MCM-3. Based on the N content, these results indicated that the loading of the Y-IMP ligand inside the pores of the different materials was found to be 0.59, 0.72, and 0.96 mmol·g −1 , for the MCM-1, MCM-2, and MCM-3 materials, respectively. In addition, the Mo contents were determined to be 4.90, 4.79, and 3.45% for MCM-1, MCM-2, and MCM-3, corresponding to loadings of 0.51, 0.50, and 0.36 mmol Mo ·g −1 , respectively. The latter results showed that all Mo cores were coordinated to the surface-grafted ligands, and not directly to the MCM wall, as the loading of ligands was higher in all materials, as shown in Scheme 4.
All materials were characterized by powder XRD, Diffuse Reflectance Infrared (DRIFT), and 29 Si and 13 C CP MAS NMR. Sorption/desorption N 2 isotherms were also determined for estimation of textural parameters. All characterization features are consistent with those reported in related mesoporous materials [17]. CHN analyses of MCM-1, MCM-2, and MCM-3 materials showed 6.01% C, 1.92% H, and 1.64% N for MCM-1; 5.08% C, 1.90% H, and 2.03% N for MCM-2; and 14.16% C, 2.26% H, and 2.68% N for MCM-3. Based on the N content, these results indicated that the loading of the Y-IMP ligand inside the pores of the different materials was found to be 0.59, 0.72, and 0.96 mmol·g −1 , for the MCM-1, MCM-2, and MCM-3 materials, respectively. In addition, the Mo contents were determined to be 4.90, 4.79, and 3.45% for MCM-1, MCM-2, and MCM-3, corresponding to loadings of 0.51, 0.50, and 0.36 mmolMo·g −1 , respectively. The latter results showed that all Mo cores were coordinated to the surfacegrafted ligands, and not directly to the MCM wall, as the loading of ligands was higher in all materials, as shown in Scheme 4.
All materials were characterized by powder XRD, Diffuse Reflectance Infrared (DRIFT), and 29 Si and 13 C CP MAS NMR. Sorption/desorption N2 isotherms were also determined for estimation of textural parameters. All characterization features are consistent with those reported in related mesoporous materials [17].
The powder X-ray patterns of the materials containing the ligands IMP-Si, Me-IMP-Si, and Ph-IMP-Si affording materials MCM-IMP-Si, MCM-Me-IMP-Si, and MCM-Ph-IMP-Si, respectively, are presented in Figure 3    There was a slight deviation of the maxima toward lower 2θ values, as compared to the parent MCM. In material MCM-IMP-Si, the d 100 value and the related a parameter increased with incorporation of the ligand (d 100 = 36.5 Å and a = 42.1 Å for MCM-IMP-Si) and the metal fragment (d 100 = 36.2 Å and a = 41.8 Å for MCM-1). Similar results were obtained for the materials MCM-Me-IMP-Si, MCM-Ph-IMP-Si, MCM-2, and MCM-3. Other data characterizing the materials are included in Table 2, summarizing the evolution of their relevant textural properties until the immobilization of Mo(II) precursor complex. The observed peak intensity reduction is common to all materials, being even more significant in the materials with the Mo cores. This is not due to a crystallinity loss, but rather to an X-ray scattering contrast reduction between the silica walls and the pore filling material. This has been observed for other types of materials and has been well described in the literature [19,20].
Nitrogen sorption/desorption studies at 77 K were performed and revealed that all materials exhibited a reversible type IV isotherm (Figure 4), typical of mesoporous solids (pore width between 2 nm and 50 nm, according to IUPAC) [21]. The calculated textural parameters (S BET and V P ) of the synthesized materials (Table 2) agreed with the literature data [22,23]. The nitrogen adsorption curves displayed a steep and sharp capillary condensation/evaporation step in the parent MCM-41 at 0.35-0.4 relative pressure range, indicating a uniform pore size distribution (PSD). After functionalization with the ligand, the isotherm of the new MCM-IMP-Si showed a much smaller step. The lower N 2 uptake reflected a decrease (Table 2) in both the volume V P (57%) and surface area S BET (46%). In the case of MCM-Me-IMP-Si, the obtained decrease was similar to that achieved for the previous material, with changes in S BET and V P of 35% and 57%, respectively. For MCM-Ph-IMP-Si, similar results were also obtained. These findings indicated that the ligand immobilization on the internal silica surface was accomplished ( Figure 4, Table 2). For the materials with the Mo(II) core, the S BET and V P decrease relative to MCM was 47% and 63%, 50% and 74%, and 47% and 65% for MCM-1, MCM-2, and MCM-3, respectively.    These results agreed with the p/p 0 coordinate decrease in the isotherm inflection points after post-synthesis treatments [24]. Furthermore, the maximum of the PSD curve ( Figure 4) determined by the BJH method, d BJH , for MCM materials, decreased from 36.0 Å to 28.5, 26.7 Å, and from 32.8 Å to 26.6 Å ( Table 2) for MCM-1, MCM-2, and MCM-3, respectively. These values were in line with those described above for the other set of materials. Textural parameters for MCM materials ( Table 2) were also found to match values reported in the literature for related systems [24].
The DRIFT spectrum ( Figure 5) of the parent MCM shows a broad band at 3600-3000 cm −1 , which can be assigned to the O-H stretching vibrations of hydrogen bonded silanol groups. Other important features comprised the band at ca. 1634 cm −1 due to OH bending modes and an intense broad band at 1240-950 cm −1 associated with asymmetric stretching vibration modes of the mesoporous framework (νSi-O-Si) [25]. Table 2. Textural parameters of host and composite materials from powder XRD and N 2 isotherms at 77 K, for all prepared materials. The introduction of the IMP-SI ligand to obtain the functionalized material MCM-IMP-Si did not significantly change the absorption profile of the mesoporous host material, but new bands characteristic of the ligand present inside the pores were now detected. Grafting of the IMP ligand was monitored by probing its νC=N mode, observed in the pyridine moiety of the free IMP ligand at 1655 cm −1 . After grafting, this mode was blue shifted to 1652 cm −1 , as expected after the formation of silyl esters and documented in the literature [26]. After binding the molybdenum complex [MoBr(η 3 -C 3 H 5 )(CO) 2 (MeCN) 2 ] to afford material MCM-1, two new bands corresponding to the νC≡O stretching modes were observed in the DRIFT spectra, at 2031 and 1934 cm −1 , having shifted from 1942 and 1854 cm −1 in 1.  Additionally, the absence of the bands due to the νC≡N vibrational modes from the acetonitrile (MeCN) ligands showed that they had been replaced by the immobilized ligand. Similar results were obtained for materials with ligands Me-IMP-Si and Ph-IMP-Si, after the functionalization with the Mo(II) core, as well as for the new materials MCM-2 and MCM-3. 13 C and 29 Si solid-state NMR spectroscopy studies were also performed on the materials and the resonances observed were assigned according to the literature. [26]. 13   Additionally, the absence of the bands due to the νC≡N vibrational modes from the acetonitrile (MeCN) ligands showed that they had been replaced by the immobilized ligand. Similar results were obtained for materials with ligands Me-IMP-Si and Ph-IMP-Si, after the functionalization with the Mo(II) core, as well as for the new materials MCM-2 and MCM-3. 13 C and 29 Si solid-state NMR spectroscopy studies were also performed on the materials and the resonances observed were assigned according to the literature [26]. 13   13 C and 29 Si solid-state NMR spectroscopy studies were also performed on the materials and the resonances observed were assigned according to the literature. [26]. 13 C CP MAS solid-state NMR spectra for MCM-IMP-Si and MCM-1 are presented in Figure 6. 13 C CP MAS solid-state NMR spectra for MCM-Me-IMP-Si, MCM-2, MCM-Ph-IMP-Si, and MCM-3 are not presented, nevertheless the results were similar to those obtained for the previous ones and were in accordance with results described in the literature.  In the 13 C solid-state spectra of the IMP-Si ligand (Figure 6), three strong signals were observed in the 0-100 ppm range, corresponding to the aliphatic carbons closer to the SiOR group of IMP-Si. The signal at the lower δ was attributed to the carbon linked to Si, at 9.7 ppm, while those at 21.2 and 42.6 ppm were assigned to the other two methylene carbon atoms of the aliphatic chain. Three signals observed at 125.6, 141.2, and 148.1 ppm correspond to the aromatic carbon atoms of the IMP-Si ligand. In general, the same signals were observed in the 13 C solid-state NMR spectra of the material after the reaction of the Mo(II) core with precursor materials. It was possible to assign the signals at 10.  Figure 7 shows the 29 Si CP MAS NMR spectra for pristine MCM and the derivatized MCM-IMP-Si and MCM-1 materials. The parent MCM material displayed three broad convoluted resonances in the 29 Si CP MAS NMR spectrum at −99.9 and −109.8ppm, assigned to Q 3 and Q 4 species of the silica framework, respectively [Q n = -Si(OSi) n (OH) 4−n ]. A weak shoulder was also observed at −90.4 ppm, due to the Q 2 species. The 29 Si CP MAS spectrum of MCM-IMP-Si material also displayed two broad signals at −101.0 and −109.0 ppm, assigned to Q 3 and Q 4 organosilicon species, respectively, and it was also possible to observe a small shoulder at −92.1 assigned to the Q 2 species. New signals corresponding to the T 1 , T 2 , and T 3 species of the Si(OR) 3 groups of the IMP-Si ligand were also observed at −49.7, −57.8, and −65.4 ppm, respectively. In the 13 C solid-state spectra of the IMP-Si ligand (Figure 6), three strong signals were observed in the 0-100 ppm range, corresponding to the aliphatic carbons closer to the SiOR group of IMP-Si. The signal at the lower δ was attributed to the carbon linked to Si, at 9.7 ppm, while those at 21.2 and 42.6 ppm were assigned to the other two methylene carbon atoms of the aliphatic chain. Three signals observed at 125.6, 141.2, and 148.1 ppm correspond to the aromatic carbon atoms of the IMP-Si ligand. In general, the same signals were observed in the 13 C solid-state NMR spectra of the material after the reaction of the Mo(II) core with precursor materials. It was possible to assign the signals at 10.  Figure 7 shows the 29 Si CP MAS NMR spectra for pristine MCM and the derivatized MCM-IMP-Si and MCM-1 materials. The parent MCM material displayed three broad convoluted resonances in the 29 Si CP MAS NMR spectrum at −99.9 and −109.8ppm, assigned to Q 3 and Q 4 species of the silica framework, respectively [Q n = -Si(OSi)n(OH)4-n]. A weak shoulder was also observed at −90.4 ppm, due to the Q 2 species. The 29 Si CP MAS spectrum of MCM-IMP-Si material also displayed two broad signals at −101.0 and −109.0 ppm, assigned to Q 3 and Q 4 organosilicon species, respectively, and it was also possible to observe a small shoulder at −92.1 assigned to the Q 2 species. New signals corresponding to the T 1 , T 2 , and T 3 species of the Si(OR)3 groups of the IMP-Si ligand were also observed at −49.7, −57.8, and −65.4 ppm, respectively. The 29 Si CP MAS spectra in Figure 7 show that reaction of MCM-IMP-Si with the organometallic core complex Mo(II) did not significantly change the environment of Si, as expected, indicating that The 29 Si CP MAS spectra in Figure 7 show that reaction of MCM-IMP-Si with the organometallic core complex Mo(II) did not significantly change the environment of Si, as expected, indicating that the metal fragment reacted with the immobilized ligand and did not interact with the wall surface. It was possible to assign signals at −106.5, −107.8, −102.6, −67.0, −58.8, and −47.8 ppm to the Q 2 , Q 3 , Q 4 , T 1 , T 2 and T 3 species, respectively.
The previously described characterization techniques indicated that the surface of the MCM porous material was functionalized with the three ligands, IMP-Si, Me-IMP-Si, and Ph-IMP-Si, which in turn reacted with the Mo(II) metal fragments to yield metal-containing materials.
All of the complexes were able to oxidize cy8 and cis with 99-100% conversion, and selectivity for the respective epoxide and yield (entries 1-3, 7-9), as had been observed for many related Mo(II) complex precursors. Very high conversions (98%) were determined for ger, but two products were formed in~1:1 ration, the epoxide (2,3-oxyrane) and 6,7-oxyrane (entries 13-15). Relatively high conversions (90 and 82%) with 1 and 2, and medium (50%) with 3 were found in the oxidation of trans (entries [10][11][12]. In all cases, the selectivity for the epoxide was 86-89%, but a second product (aldehyde) was also formed. 1-oct was not easily oxidized, and conversions reached 55, 43, and 72% with 1, 2, and 3, respectively, with epoxide selectivity in the range of 96-100% (entries [16][17][18]. The second by-product (1-oct-2-ketone) was formed in very small amounts. Sty conversions did not go beyond 40%, and two products, the epoxide and benzaldehyde, were formed in~3:2 ratios (entries 4-6). The differences between the activity of the three complexes were very small, suggesting that the change in substituent (H, Me, and Ph) had negligible effect. Only for sty was complex 1 less active than the others. The complexes oxidized cis more efficiently than trans, with higher conversion and selectivity for the epoxide, although the differences were small. On the other hand, we showed in a recent work [26] that, for a series of α-diimine (derivatives of 2,2 -bipyridyl and 1,10-phenantroline) complexes of MoBr(η 3 -C 3 H 5 )(CO) 2 , the selectivities were higher for cis but the conversions were lower. These ligands were more symmetrical than Y-IMP, and the approach of the trans substrate to the active species may have been more restricted than that of cis. Table 3. Olefin epoxidation of cis-cyclooctene (cy8), styrene (sty), cis-hex-3-en-1-ol (cis), trans-hex-3en-1-ol (trans), geraniol (ger), and 1-octene (1-oct) catalyzed by complexes 1-3.

Entry
Reaction No conversion of any substrate was observed in blank runs without a catalyst and in the presence of an oxidizing agent. All of the complexes were able to oxidize cy8 and cis with 99-100% conversion, and selectivity for the respective epoxide and yield (entries 1-3, 7-9), as had been observed for many related Mo(II) complex precursors. Very high conversions (98%) were determined for ger, but two products were formed in ~1:1 ration, the epoxide (2,3-oxyrane) and 6,7-oxyrane (entries 13-15). Relatively high conversions (90 and 82%) with 1 and 2, and medium (50%) with 3 were found in the oxidation of trans (entries [10][11][12]. In all cases, the selectivity for the epoxide was 86-89%, but a second product (aldehyde) was also formed. 1-oct was not easily oxidized, and conversions reached 55, 43, and 72% with 1, 2, and 3, respectively, with epoxide selectivity in the range of 96-100% (entries [16][17][18]. The second by-product (1-oct-2-ketone) was formed in very small amounts. Sty conversions did not go beyond 40%, and two products, the epoxide and benzaldehyde, were formed in ~3:2 ratios (entries 4-6). The differences between the activity of the three complexes were very small, suggesting that the change in substituent (H, Me, and Ph) had negligible effect. Only for sty was complex 1 less active than the others. The complexes oxidized cis more efficiently than trans, with higher conversion and selectivity for the epoxide, although the differences were small. On the other hand, we showed in a recent work [26] that, for a series of α-diimine (derivatives of 2,2′-bipyridyl and 1,10-phenantroline) complexes of MoBr(η 3 -C3H5)(CO)2, the selectivities were higher for cis but the conversions were lower. These ligands were more symmetrical than Y-IMP, and the approach of the trans substrate to the active species may have been more restricted than that of cis. Table 3. Olefin epoxidation of cis-cyclooctene (cy8), styrene (sty), cis-hex-3-en-1-ol (cis), trans-hex-3en-1-ol (trans), geraniol (ger), and 1-octene (1-oct) catalyzed by complexes 1-3.

Entry
Reaction geraniol (ger), 1-octene (1-oct), and S-(+)-limonene (S-lim). No conversion of any substrate was observed in blank runs without a catalyst and in the presence of an oxidizing agent. All of the complexes were able to oxidize cy8 and cis with 99-100% conversion, and selectivity for the respective epoxide and yield (entries 1-3, 7-9), as had been observed for many related Mo(II) complex precursors. Very high conversions (98%) were determined for ger, but two products were formed in ~1:1 ration, the epoxide (2,3-oxyrane) and 6,7-oxyrane (entries [13][14][15]. Relatively high conversions (90 and 82%) with 1 and 2, and medium (50%) with 3 were found in the oxidation of trans (entries [10][11][12]. In all cases, the selectivity for the epoxide was 86-89%, but a second product (aldehyde) was also formed. 1-oct was not easily oxidized, and conversions reached 55, 43, and 72% with 1, 2, and 3, respectively, with epoxide selectivity in the range of 96-100% (entries [16][17][18]. The second by-product (1-oct-2-ketone) was formed in very small amounts. Sty conversions did not go beyond 40%, and two products, the epoxide and benzaldehyde, were formed in ~3:2 ratios (entries 4-6). The differences between the activity of the three complexes were very small, suggesting that the change in substituent (H, Me, and Ph) had negligible effect. Only for sty was complex 1 less active than the others. The complexes oxidized cis more efficiently than trans, with higher conversion and selectivity for the epoxide, although the differences were small. On the other hand, we showed in a recent work [26] that, for a series of α-diimine (derivatives of 2,2′-bipyridyl and 1,10-phenantroline) complexes of MoBr(η 3 -C3H5)(CO)2, the selectivities were higher for cis but the conversions were lower. These ligands were more symmetrical than Y-IMP, and the approach of the trans substrate to the active species may have been more restricted than that of cis. Table 3. Olefin epoxidation of cis-cyclooctene (cy8), styrene (sty), cis-hex-3-en-1-ol (cis), trans-hex-3en-1-ol (trans), geraniol (ger), and 1-octene (1-oct) catalyzed by complexes 1-3.

Entry
Reaction geraniol (ger), 1-octene (1-oct), and S-(+)-limonene (S-lim). No conversion of any substrate was observed in blank runs without a catalyst and in the presence of an oxidizing agent. All of the complexes were able to oxidize cy8 and cis with 99-100% conversion, and selectivity for the respective epoxide and yield (entries 1-3, 7-9), as had been observed for many related Mo(II) complex precursors. Very high conversions (98%) were determined for ger, but two products were formed in ~1:1 ration, the epoxide (2,3-oxyrane) and 6,7-oxyrane (entries [13][14][15]. Relatively high conversions (90 and 82%) with 1 and 2, and medium (50%) with 3 were found in the oxidation of trans (entries [10][11][12]. In all cases, the selectivity for the epoxide was 86-89%, but a second product (aldehyde) was also formed. 1-oct was not easily oxidized, and conversions reached 55, 43, and 72% with 1, 2, and 3, respectively, with epoxide selectivity in the range of 96-100% (entries [16][17][18]. The second by-product (1-oct-2-ketone) was formed in very small amounts. Sty conversions did not go beyond 40%, and two products, the epoxide and benzaldehyde, were formed in ~3:2 ratios (entries 4-6). The differences between the activity of the three complexes were very small, suggesting that the change in substituent (H, Me, and Ph) had negligible effect. Only for sty was complex 1 less active than the others. The complexes oxidized cis more efficiently than trans, with higher conversion and selectivity for the epoxide, although the differences were small. On the other hand, we showed in a recent work [26] that, for a series of α-diimine (derivatives of 2,2′-bipyridyl and 1,10-phenantroline) complexes of MoBr(η 3 -C3H5)(CO)2, the selectivities were higher for cis but the conversions were lower. These ligands were more symmetrical than Y-IMP, and the approach of the trans substrate to the active species may have been more restricted than that of cis. Table 3. Olefin epoxidation of cis-cyclooctene (cy8), styrene (sty), cis-hex-3-en-1-ol (cis), trans-hex-3en-1-ol (trans), geraniol (ger), and 1-octene (1-oct) catalyzed by complexes 1-3.

Entry
Reaction  a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product. a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product. Table 4. Olefin epoxidation of cis-cyclooctene (cy8), styrene (sty), cis-hex-3-en-1-ol (cis), trans-hex-3en-1-ol (trans), geraniol (ger), 1-octene (1-oct), and S-(+)-limonene (S-lim) catalyzed by materials MCM-1, MCM-2, and MCM-3. a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product.    All reactions were carried in dichloromethane in the presence of 2 eq. of oxidant (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f yield of 6,7-oxyrane product; g yield of Zepoxide, E-epoxide, Z-limonene alcohol, and E-limonene alcohol products; and h yield of 1-oct-2ketone product. a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product. a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product.   a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product. a All reactions were carried out in dichloromethane in the presence of 2 eq. of oxidant tert-Butyl hydroperoxide (tbhp) and 175 mg of catalyst at 328 K; b calculated after 24 h; c calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f Yield of 6,7-oxyrane product; and g yield of 1-oct-2-ketone product. calculated as "Yield of epoxide"/"Conversion" × 100%; d benzaldehyde as by-product; e aldehyde as by-product; f yield of 6,7-oxyrane product; g yield of Z-epoxide, E-epoxide, Z-limonene alcohol, and E-limonene alcohol products; and h yield of 1-oct-2-ketone product.

MCM
The immobilization of the three complexes in MCM did not improve their capability as catalysts. This result was surprising, since in a previous work dealing with another Mo(II) catalyst precursor with the same ligand ([MoX 2 (CO) 3 (IMP)]), activity increased when the metal complex was supported in MCM [26]. However, the allylic Mo(II) system described in the present work was usually a better catalyst [27].
Epoxidation of cy8 remained very efficient when any of the three materials acted as catalysts (entries 1-3), but conversions of cis into its epoxide became lower, even though the selectivity was kept (entries 4-6). The conversions of ger (entries 13-15) decreased from 98% to 56, 66, and 23% for MCM-1, MCM-2, and MCM-3, respectively, and the relative amount of the 2,3-oxyrane also was tendentially lower for the two more active catalysts. The oxidation of trans was interesting, in the sense that no epoxide was formed and the reaction was 100% selective for the aldehyde (entries 10-12), with conversions slightly higher than in the homogeneous catalysts 1 and 2, and significantly higher for 3. The heterogeneous catalysts, except MCM-2, were less active than the homogeneous ones in the oxidation of 1-oct, and the selectivity decreased to very modest values (less than half, entries [16][17][18]. On the other hand, higher conversions took place for the three catalysts with sty, and the selectivity toward the epoxide increased from~2:1 to~3:1(entries 4-6), with MCM-3 behaving in a different way than the others (much less epoxide formed). Another substrate, S-lim, was tested with the materials as catalysts. The conversions were high, in the range of 83-89% (entries [19][20][21], but the yield of epoxide was very low, even considering both Z-and E-epoxides. The E-and Z-alcohols were obtained in much higher proportion. The selectivity toward the epoxide was extremely low. The analogous complex with 8-aminoquinoline [27], for instance, displayed a much lower conversion in the oxidation of cis upon immobilization (14 compared with 69%), but the opposite was observed with trans (56 and 75%), with the selectivity always at 100%. The lower activity of the heterogeneous catalysts in general may be explained by the steric hindrance of the side chain of the Y-IMP complexes, which seemed to be more effective in the confined pores of the material.
The catalytic performance of the heterogeneous systems was also assessed in terms of leaching and recycling. To study the leaching, a catalytic experiment, using cy8 and MCM-1, was started and the catalyst was filtered from the mixture after 2 h. The conversion at this point was 82% and after 24 h only 86% ( Figure S6), while it reached 100% in the reference experiment with the catalyst present the entire time. This showed that the reaction practically stopped when the catalyst was removed, indicating a small amount of leaching to the homogeneous phase. Therefore, the material acted as a heterogeneous catalyst. The recycling experiments were performed for the same system (cy8 and MCM-1). After one 24 h cycle, the solid catalyst was separated; another load of substrate, solvent, and oxidant was added; and the reaction was followed for another 24 h. The conversions were 100%, 79%, and 60% in the first, second, and third cycle, respectively, reflecting a loss of activity of~20% between cycles.
The Mo(II) complexes and the Mo(II)-containing materials were catalyst precursors that had to be oxidized to Mo(VI) by tbhp in the first step to afford the catalysis active species. In this process, the allyl and the carbonyl ligands were lost and the Mo=O and Mo-O-Mo bonds were formed, with the bidentate ligand-here Y-IMP-remaining bound to the metal [27]. The complexes 1-3 were fluxional in solution, as reflected in their NMR spectra, and, more surprisingly, the two main isomers even coexisted in the solid for 1. This suggested that several isomers of the active species (see Scheme 3) were more likely to exist in the solution than with sterically crowded ligands, such as 2,9-R 2 -1,10-phenanthroline (R = Me, tBu) [27]. This could be a possible explanation for the lack of selectivity observed in general. The immobilization procedure may have also ended up with several supported species, which would then be oxidized.

Computational Studies
Density functional theory (DFT) calculations [28] were performed using the Amsterdam Density Functional (ADF) program [29] (more details in Experimental) in order to explore the energy difference between the axial and the equatorial isomers. As described above, for the IMP ligand, the two isomers could coexist not only in solution but also in the solid state, which was an unprecedented finding. Indeed, the Me-IMP complex was observed in the solid state as the axial isomers. It was found that the axial isomer was the most stable for the three complexes by small amounts, namely 1.20 for 1, 1.38 for 2, and 1.13 kcal·mol −1 for 3. The meaning of these values is that the two isomers had essentially the same energy and the crystal packing had a determining effect on the species present in the solid state. A view of relevant packing motifs for 1 and 2 can be seen in Figure S7 (SI). A slight change to the methodology, for instance by optimizing without considering the solvent, led to the equatorial isomer of 1 being more stable by only 0.13 kcal·mol −1 . A close look at the molecular structures ( Figure S7 in SI) suggested that the addition of the Me or Ph substituent would influence the solid-state structure.
FTIR spectra were obtained on a Nicolet 6700 in the 400-4000 cm −1 range with 4 cm −1 resolution, using KBr pellets for complexes and DRIFT for materials. Powder XRD measurements were recorded with a Philips Analytical PW 3050/60 X'Pert PRO (θ/2θ) equipped with X'Celerator detector and with automatic data acquisition (X'Pert Data Collector (v2.0b) software, Philips, Eindhoven, Netherlands), using a monochromatic Cu Kα radiation as the incident beam, operating at 40 kV-30 mA. XRD diffraction patterns were obtained by continuous scanning in a 2θ-range of 2 to 10 • with 2θ-step size of 0.017 • and a scan step time of 99.695 s. A Bruker Avance 400 spectrometer with frequencies of 400.13 MHz for 1 H and 100.61 MHz for 13 C was used to obtain 1 H and 13 C solution NMR spectra. The solid-state NMR spectra were obtained on a (9.4 T) Bruker Avance 400P spectrometer at the University of Aveiro by Dr. Paula Ferreira (Bruker, Billerica, MA, USA). The 29 Si spectra were recorded with a frequency of 79.49 MHz. 29 Si MAS NMR spectra were recorded with 40 • pulses, spinning rates of 5.0-5.5 kHz, and 60 s recycle delays and 29 Si CP MAS NMR spectra with 5.5 µs 1H 90 • pulse, 2 ms contact time, a spinning rate of 8 kHz, and 4 s recycle delays. Chemical shifts are given in ppm with tetramethylsilane (TMS) as reference. 13 C solid-state NMR spectra were recorded at 125.76 MHz on a Bruker Avance 500 spectrometer (Bruker, Billerica, MA, USA). The N 2 sorption was measured in an automatic apparatus (ASAP 2010; Micrometrics). BET specific surface areas (S BET , p/p 0 from 0.03 to 0.13) and specific total pore volume, V p , were estimated from N 2 adsorption isotherms measured at 77 K. The PSD were calculated by the BJH method using the modified Kelvin equation, with correction for the statistical film thickness on the pore walls [24]. The statistical film thickness was calculated using the Harkins-Jura equation in the p/p 0 range from 0.1 to 0.95. Microanalyses (C, N, S, H) and atomic absorption for determination of molybdenum and tungsten were performed at the Institute of Chemical and Biological Technology (Instituto de Tecnologia Química e Biológica, or ITQB) and the University of Vigo.