Selective hydrosilylation of olefins by a two-dimensional Rh(I) low-valent metal–organic framework

Abstract

Multitopic phosphine ligands combined with Rh(I) or Ir(I) nodes are shown to assemble into low-valent metal–organic frameworks (LVMOFs) with a two-dimensional (2D) topology. The 2D Rh(I) LVMOF exhibits catalytic activity for the hydrosilylation of olefins, functioning in a heterogeneous and selective manner.

---

Low-valent metal–organic frameworks (LVMOFs) represent an emerging class of materials within the broader field of MOFs.¹ Unlike conventional MOFs, LVMOFs employ secondary building units (SBUs) composed of low-valent metal ions or metal ion clusters coordinated with soft Lewis basic ligands, instead of the high oxidation state metal ions and hard Lewis basic ligands of conventional MOFs.² Early examples of LVMOFs were reported by Robson et al. in 1989, featuring coordination networks constructed from tetratopic cyano linkers and CuI nodes.^3,4 More recent advancements include reports by Figueroa et al. on the use of sterically encumbered isocyanide ligands to construct the LVMOFs [Cu(CNAr^Mes2)]_n[PF₆] with Cu^I nodes and [Ni(CNAr^Mes2)]_n (Mes = 2,4,6-Me₃-C₆H₃) with Ni⁰ nodes.^5–7 Pederson et al. have employed pyrazine/pyridyl ligands in conjunction with group 6 metal carbonyl complexes M(CO)₆ (M = Cr⁰, Mo⁰, W⁰) to prepare LVMOFs [fac-M(CO)₃(pyz)_3/2]_n (pyz = pyrazine; M = Cr⁰, Mo⁰, W⁰), and [fac-M(CO)₃(bpy)_3/2]_n (bpy = bipyridine; M = Mo⁰, W⁰).^8,9 Pederson, Brozek, Bejger, and others have also recently described the use of preformed, low-valent metal clusters for the preparation of LVMOFs.^10–12 These examples underscore the potential of LVMOFs to facilitate the design of innovative materials with tuneable metal nodes and distinctive reactivity. Additionally, stabilizing low-valent metals in rigid frameworks may reduce deleterious metal leaching, thereby generating more stable, recyclable heterogeneous catalyst systems. It is also possible that embedding catalytically inefficient homogeneous species into these robust, crystalline heterogeneous platforms may allow for enhanced activity and stability.

Low-valent metals can form 1D and 3D LVMOFs using tetratopic phosphine ligands (Fig. 1).¹³ Tetratopic ligands Si1 and Sn1 (Fig. 1) produced 3-dimensional (3D), crystalline LVMOFs with Pd⁰ and Pt⁰ nodes when combined with appropriate metal sources in the presence of a PPh₃ modulator. Similarly, 1-dimensional (1D) LVMOFs (coordination polymers) have been formed with Rh^I and Ir^I using these tetratopic phosphine ligands.^13,14 In these examples, the formation of 1D vs. 3D phases can be explained by the differences in the coordination geometry preferences of the metal centres. Herein, it is shown that a seemingly minor change to the central atom of the tetratopic phosphine ligand can induce the formation of a novel 2D LVMOF phase. Furthermore, the Rh^I derivative is shown to exhibit catalytic activity in the hydrosilylation of olefins.

Fig. 1 Tetratopic phosphine ligands C1, Si1, and Sn1 (E = C, Si, Sn). Tetratopic phosphine ligands Si1 and Sn1 have been shown to form 3D LVMOFs with Pd⁰ and Pt⁰ (top), but Sn1 has been shown to form 1D LVMOFs with Rh^I and Ir^I (bottom).

Using the same tetratopic ligand design as Si1 and Sn1, a new ligand with a carbon core (designated C1) was synthesized (see ESI† for details). It was expected that C1 would readily generate materials similar to Si1 and Sn1; however, synthesis conditions similar to those reported for Sn1–Rh failed to give any crystalline material.¹³ Exploration of alternative reaction conditions revealed that a THF solution containing 2 equiv. of [RhCl(CO)(PPh₃)₂] and C1 (1 equiv.) at 80 °C for 5–7 d under inert atmosphere (Scheme 1) afforded yellow irregularly shaped crystals suitable for single crystal X-ray diffraction (SCXRD, Fig. 2). [IrCl(CO)(PPh₃)₂] (Vaska's complex) with C1 similarly yielded yellow crystals suitable for SCXRD (under similar conditions, but at more elevated temperature, 100 °C, see ESI† for details). Notably, these initial reaction conditions for producing materials (designated as C1–Rh and C1–Ir) excluded the use of PPh₃ as a modulator, which was previously found to be essential for the formation of LVMOFs with Si1 and Sn1. However, after these initial conditions were identified, it was determined that incorporation of a modulator could significantly shorten the reaction time; specifically, a mixture of 2.0 equiv. [RhCl(CO)(PPh₃)₂] or [IrCl(CO)(PPh₃)₂], C1 (0.014 mmol), and P(o-Tol)₃ (4.0 equiv.) in THF at elevated temperature yielded identical crystalline material in <48 h (see ESI† for details).

Scheme 1 Synthesis of LVMOFs C1–Rh and C1–Ir.

Fig. 2 Top Left: View of extended repeat unit in C1–Rh showing full metal and ligand connectivity. Color scheme: C = gray, Cl = green, O = red, P = orange, Rh = teal; hydrogens omitted for clarity. Top Right: Simplified diagram of the 4,2-connected C1–Rh 2D network topology. Bottom: Packing diagram of C1–Rh.

Initial SCXRD unit cell parameters of C1–Rh and C1–Ir showed that these materials were different from the 1D networks of Sn1–Rh and Sn1–Ir. The structures of C1–Rh and C1–Ir were orthorhombic in the Fddd space group (unlike tetragonal I4/m for Sn1–Rh and Sn1–Ir). In C1–Rh and C1–Ir, the asymmetric unit {M(CO)Cl(C1)_1/4} (M = Rh, Ir) is assembled into a tetrahedral {(M(CO)Cl)₄-C1} (M = Rh, Ir) arrangement where four separate metal nodes are connected to two distinct C1 linkers resulting in an extended 2D network (Fig. 2). These constitute a 4,2-connected 2D network (Fig. 2). The metal centres retain a square planar geometry, and there are no counter ions in the lattice, indicating the metals remain in the 1+ oxidation state. The metal oxidation state was confirmed by FTIR spectroscopy of C1–Rh, which exhibits a ν_CO (1971 cm⁻¹) comparable to that of Sn1–Rh (1967 cm⁻¹) and the analogous molecular complex [Rh(CO)Cl(PPh₃)₂] (1961 cm⁻¹) (Fig. S1, ESI†), indicating that the electronic environments of these metal centres are the same. Similarly, the FTIR spectrum analysis of C1–Ir shows a ν_CO of 1968 cm⁻¹ akin to that of the molecular [Ir(CO)Cl(PPh₃)₂] (1966 cm⁻¹) complex, corroborating the preservation of the oxidation state at the iridium centre (Fig. S2, ESI†). The differences in framework structures of C1–Rh/Irvs.Sn1–Rh/Ir is attributed to the smaller size of the central carbon atom of the tetratopic linker. In C1–Rh, the mean bond length between the central carbon (C_cent) atom and the aromatic carbon atoms (C_Ar) is ca. 1.52 Å, with the corresponding C_Ar–C_cent–C_Ar bond angles spanning from ca. 103.7–115.2°. Sn1–Rh exhibits a notably elongated Sn–C_Ar bond distance of 2.14 Å with a geometry around the Sn centre of C_Ar–Sn–C_Ar bond angles ranging from 107.5–113.4°. In C1–Ir, the average bond length between atoms C_cent and C_Ar is ca. 1.66 Å, with the corresponding C_Ar–C_cent–C_Ar bond angles of ca. 97.7–127.2°, while Sn1–Ir displays an elongated Sn–C_Ar bond distance of 2.14 Å and corresponding C_Ar–Sn–C_Ar bond angles of 108.0–112.4°. These bond metrics underscore the structural differences between the central C and Sn atoms, which manifests as substantial changes in the topology of each material. Hence, the small structural change of the central atom greatly influences the structure of the LVMOF.

Powder X-ray Diffraction (PXRD) confirmed crystallinity of the bulk materials and showed good agreement with the simulated PXRD pattern of C1–Rh and C1–Ir (Fig. S3, ESI†). C1–Rh retains excellent crystallinity under ambient conditions (>59 d) as confirmed by PXRD and FTIR analysis (Fig. S5 and S6, ESI†). The thermal stability of C1–Rh was assessed through thermogravimetric analysis (TGA), which indicated a decomposition temperature of ca. ∼368 °C. This represents an increase of ∼28 °C compared to Sn1–Rh and ∼134 °C higher than molecular complex [Rh(CO)Cl(PPh₃)₂] (Fig. S8, ESI†). Similarly, TGA analysis of C1–Ir displayed a decomposition temperature of ca. ∼377 °C, exceeding that of the Sn1–Ir material by ∼27 °C and surpassing the molecular complex [Ir(CO)Cl(PPh₃)₂] by ∼102 °C.¹⁴ Notably, activated samples of C1–Rh displayed no uptake of N₂ at cryogenic temperatures (Fig. S12, ESI†).

The application of LVMOFs in heterogeneous olefin hydrosilylation was investigated. Hydrosilylation is a pivotal process for synthesizing organosilicon compounds through the catalytic addition of Si–H bonds to olefins. The resulting products find extensive use in the preparation of silicone-based polymers, rubbers, oils, and various coupling reagents.^15–17 Among the numerous Rh^I complexes employed in homogeneous hydrosilylation catalysis, Wilkinson's catalyst [RhCl(PPh₃)₃] has been widely reported to facilitate hydrosilylation reactions.^18,19

As a model substrate, 1-octene was combined with triethyl silane (Et₃SiH) as a silylating reagent. Starting with a control reaction of 1-octene (2.0 equiv.) with Et₃SiH (1.0 equiv.) in 1 mL toluene at 80 °C for 24 h without catalyst yielded no hydrosilylation product (Table 1, entry 1). By contrast, the same reaction performed in presence of 0.28 mol% C1–Rh in toluene at 25 °C for 24 h gave <1% yield of triethyl(octyl)silane hydrosilylation (HS) and <1% of the dehydrogenative silylation (DS) byproduct (Table 1, entry 2). When the reaction temperature was raised to 80 °C either in toluene or benzene under otherwise similar conditions, the reaction produced 39 ± 1% (n = 3) of the HS product and <4 ± 1% (n = 3) of DS (Table 1, entries 3 and 4). During the hydrosilylation process, isomerization of 1-octene to internal octenes was observed. This was confirmed by ¹H NMR analysis of the crude reaction mixtures, which showed changes in the position of double bonds within the 1-octene structure (Fig. S22, ESI†). This isomerization typically occurs due to the action of the catalyst under the reaction conditions and is a common side reaction in hydrosilylation.²⁰ Different solvents were tested including THF and CDCl₃ but these yielded HS products in only 22% and 23%, respectively, showing no significant improvement in selectivity compared to toluene (Table 1, entries 5 and 6). The employment of C1–Ir LVMOF as a heterogeneous catalyst yielded no detectable hydrosilylation product under the same reaction conditions (Table 1, entry 7). Importantly, the analogous molecular complex, [Rh(CO)Cl(PPh₃)₂], showed reduced catalytic activity relative to C1–Rh, yielding just 4 ± 2% (n = 3) HS and 3 ± 1% (n = 3) DS (Table 1, entry 8). Additionally, Wilkinson's catalyst [RhCl(PPh₃)₃] was tested in the hydrosilylation of 1-octene with Et₃SiH, producing 28 ± 3% (n = 3) of HS and 2 ± 1% (n = 3) of DS products (Table 1, entry 9).

Table 1 Summary of the optimization of hydrosilylation (HS) product and dehydrogenative silylation (DS) byproduct of 1-octene with HSiEt₃

Entry	Cat.	Sol.	Temp. (°C)	Yield HS^a (%)	Yield DS^a (%)
a Conditions: 0.6 mmol of alkene, 0.3 mmol of silane, toluene (1.0 mL), 80 °C, 24 h (see ESI for details). Yields determined by GCMS with n-dodecane internal standard. b Yields are based on three independent experiments (n = 3). c Use of PhMe2SiH instead of Et3SiH. d Yields are based on two independent experiments (n = 2).
1	—	Toluene	80	0	0
2	C1–Rh	Toluene	25	<1	<1
3^b	C1–Rh	Toluene	80	39 ± 1	3 ± 1
4	C1–Rh	Benzene	80	40	3
5	C1–Rh	THF	70	22	2
6	C1–Rh	CDCl3	70	23	3
7	C1–Ir	Toluene	80	0	0
8^b	Rh(CO)Cl(PPh3)2	Toluene	80	4 ± 2	3 ± 1
9^b	RhCl(PPh3)3	Toluene	80	28 ± 3	2 ± 1
10	Sn1–Rh	Toluene	80	30	5
11	Sn1–Ir	Toluene	80	0	0
12^c^d	C1–Rh	Toluene	80	>99 ± 0	0
13^c	Sn1–Rh	Toluene	80	>99	0

---

To evaluate a material with identical active sites but a different topology, the previously reported 1D coordination polymer Sn1–Rh was evaluated under identical catalytic conditions. Hydrosilylation of 1-octene with Sn1–Rh and Et₃SiH in toluene produced 30% HS and 5% DS yield (Table 1, entry 10). The structurally analogous Sn1–Ir displayed no catalytic activity in the silylation of 1-octene under identical conditions (Table 1, entry 11).

The use of an alternative silane, dimethylphenylsilane (PhMe₂SiH), due to its combination of lower steric hindrance and enhanced reactivity from the phenyl group, resulted in a significant improvement in both yield and selectivity of the hydrosilylation reaction. When the reaction was conducted with either C1–Rh or Sn1–Rh as the catalyst at 80 °C for 24 h in toluene, the yield of the HS product was >99%, with no DS byproduct detected (Table 1, entries 12 and 13). These findings underscore the importance of both the catalyst and the choice of silane in achieving high efficiency and selectivity in silylation reactions.

To ensure that catalysis was occurring in a heterogenous manner, filtration tests were conducted. Inductively coupled plasma mass spectrometry (ICP-MS) analysis of the filtrate revealed no dissolution of metal species (see ESI† for details). Furthermore, the recyclability of the C1–Rh catalyst was demonstrated over three reaction cycles (Cycle I: 35 ± 6% HS, 3 ± 1% DS (n = 2); Cycle II: 37 ± 8% HS, 3 ± 1% DS (n = 2); Cycle III: 45 ± 5% HS, 4 ± 0% DS (n = 2)) with 1-octene and Et₃SiH under optimized conditions, without compromising the integrity of the crystalline network and with no dissolution of metal species (Fig. S23, see ESI† for details). Compared to the molecular [Rh(CO)Cl(PPh₃)₂] complex, C1–Rh offers superior stability, ease of recovery, and reusability.

Having established optimized conditions for 1-octene, a focused investigation with a range of other olefins and silanes was subsequently conducted. Employing C1–Rh (0.28 mol%) at 80 °C for 24 h in toluene, long-chain alkenes showed good selectivity: 1-pentene (31 ± 8% HS; 0% DS), 1-hexene (28 ± 5% HS; 0% DS), 1-heptene (37 ± 3% HS; 2 ± 1% DS), and 4-phenyl-1-butene (85 ± 9% HS; 0% DS) yielded exclusively silylation products (based on three independent experiments, n = 3, Fig. 3). By contrast, allyl-pentafluorobenzene gave 11 ± 2% of HS and 9 ± 1% DS (n = 2) byproduct. Styrene derivatives, including 4-chlorostyrene, demonstrated significantly lower yields, producing 9 ± 1% HS and 29 ± 2% DS (n = 2). This suggests that the electronic nature of the substrate plays a crucial role in influencing the selectivity and efficiency of these hydrosilylation reactions. In contrast, 2-methyl-1-heptene yielded 2 ± 1% HS (n = 3) product, indicating that the increasing steric hindrance of the olefin markedly impedes the reaction kinetics. When the silane source was changed to PhMe₂SiH, a large enhancement of yield was observed for 1-hexene (>99% HS; 0% DS), 1-heptene (>99% HS; 0% DS), 1-octene (>99 ± 0% HS; 0% DS) and 4-phenyl-1-butene (>99% HS; 0% DS) with the exclusive formation of HS products (Fig. 3). Furthermore, the natural product methyl eugenol was successfully silylated with excellent selectivity and yield utilizing either C1–Rh (>99% HS; 0% DS) or Sn1–Rh (>99% HS; 0% DS) as the catalyst. Conversely, silylation of 1-octene with the use of sterically hindered and electron deficient triethoxysilane ((EtO)₃SiH), yielded no detectable HS products. These findings underscore the importance of both silane reactivity and substrate electronic characteristics in determining the outcomes of these hydrosilylation reactions.

Fig. 3 Reaction scope of olefins and silanes for HS reaction. Reaction conditions: 0.6 mmol of alkene, 0.3 mmol of silane, toluene (1.0 mL), 80 °C, 24 h (see ESI† for details). Yields were determined by GCMS using an internal n-dodecane standard.

In conclusion, this study has elucidated that the topology of LVMOFs exhibits a remarkable sensitivity to even subtle modifications of tetratopic phosphine ligands. While the tetratopic linkers C1 and Sn1 are topologically similar, the stereo-electronic influence of employing carbon as a core atom induced the formation of a 2D framework for C1–Rh, which is distinct from the 1D coordination polymer of Sn1–Rh. C1–Rh, C1–Ir, Sn1–Rh, and Sn1–Ir were investigated for their catalytic activity in silylation reactions. C1–Rh and Sn1–Rh were found to catalyze the silylation of olefins with varying selectivity and efficiency depending on the substrates. This study highlights the new structures and new catalysts that can be discovered within the realm of LVMOFs.

The data supporting this article have been included as part of the ESI.† Crystallographic data for C1–Rh and C1–Ir have been deposited at the CCDC under deposition numbers 2419327 and 2419328.

This research was supported by the National Science Foundation (Award No CHE-2153240). M. R. E. was supported in part by a San Diego Fellowship awarded by the UC San Diego Materials Research Science and Engineering Center (UCSD MRSEC, DMR-2011924). S. E. G. was supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC) Postdoctoral Fellowships (PDF) program. We thank Prof. Joshua Figueroa (UCSD) who provided generous access to an inert atmosphere glove box to support these studies. We thank Dr. Jake Bailey (UCSD) for assistance with X-ray crystallography and Dr. Yongxuan Su (Molecular Mass Spectrometry Facility, UCSD) for assistance with GCMS analysis.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the ESI.†

Notes and references

1. R. E. Sikma, K. P. Balto, J. S. Figueroa and S. M. Cohen, Angew. Chem., Int. Ed., 2022, 61, e202206353.
2. N. L. Rosi, J. Eckert, M. Eddaoudi, D. T. Vodak, J. Kim, M. O’Keeffe and O. M. Yaghi, Science, 2003, 300, 1127–1129.
3. B. F. Hoskins and R. Robson, J. Am. Chem. Soc., 1989, 111, 5962–5964.
4. R. Robson, Chem. Rec., 2024, 24, e202400038.
5. D. W. Agnew, M. Gembicky, C. E. Moore, A. L. Rheingold and J. S. Figueroa, J. Am. Chem. Soc., 2016, 138, 15138–15141.
6. D. W. Agnew, I. M. DiMucci, A. Arroyave, M. Gembicky, C. E. Moore, S. N. MacMillan, A. L. Rheingold, K. M. Lancaster and J. S. Figueroa, J. Am. Chem. Soc., 2017, 139, 17257–17260.
7. A. Arroyave, M. Gembicky, A. L. Rheingold and J. S. Figueroa, Inorg. Chem., 2020, 59, 11868–11878.
8. L. Voigt, R. W. Larsen, M. Kubus and K. S. Pedersen, Chem. Commun., 2021, 57, 3861–3864.
9. C. E. Andersen, J. N. McPherson, M. G.-Marqués, J. Li, M. Kubus, S. Ito, C. R. Göb, S. Ott, R. W. Larsen, G. Minguez Espallargas and K. S. Pedersen, J. Mater. Chem. C, 2023, 11, 11460–11465.
10. J. H. Gillen, C. A. Moore, M. Vuong, J. Shajahan, M. R. Anstey, J. R. Alston and C. M. Bejger, Chem. Commun., 2022, 58, 4885–4888.
11. K. Kadota, T. Chen, E. L. Gormley, C. H. Hendon, M. Dinca and C. K. Brozek, Chem. Sci., 2023, 14, 11410–11416.
12. X. Liu, J. N. McPherson, C. E. Andersen, M. S. B. Jorgensen, R. W. Larsen, N. J. Yutronkie, F. Wilhelm, A. Rogalev, M. Gimenez-Marques, G. Minguez Espallargas, C. R. Gob and K. S. Pedersen, Nat. Commun., 2024, 15, 1177.
13. R. E. Sikma and S. M. Cohen, Angew. Chem., Int. Ed., 2022, 61, e202115454.
14. S. E. Griffin, G. P. Domecus, C. E. Flores, R. E. Sikma, L. Benz and S. M. Cohen, Angew. Chem., Int. Ed., 2023, 62, e202301611.
15. B. Marciniec, Hydrosilylation: A comprehensive review on recent advances, Springer, 2009.
16. L. N. Lewis, J. Stein, Y. Gao, R. E. Colborn and G. Hutchins, Platinum Met. Rev., 1997, 41, 66–75.
17. T. Komiyama, Y. Minami and T. Hiyama, ACS Catal., 2017, 7, 631–651.
18. B. Marciniec, Silicon Chem., 2002, 1, 155–174.
19. J. B. Baruah, K. Osakada and T. Yamamoto, J. Mol. Catal., 1995, 101, 17–24.
20. B. B. Sarma, J. Kim, J. Amsler, G. Agostini, C. Weidenthaler, N. Pfander, R. Arenal, P. Concepcion, P. Plessow, F. Studt and G. Prieto, Angew. Chem., Int. Ed., 2020, 59, 5806–5815.

---

Footnote

† Electronic supplementary information (ESI) available: Synthesis and characterization details, experimental details for catalysis and recyclability of the catalysts, GCMS chromato-grams for optimizations and reactions scope. CCDC 2419327 (C1-Rh) and 2419328 (C1-Ir). For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d5cc02201k

---