Heterogenized Molecular Rhodium Phosphine Catalyst within Metal-Organic Framework for Ethylene Hydroformylation

: Molecularly-defined organometallic rhodium phosphine complexes were efficiently heterogenized within a MOF structure without affecting neither their molecular nature nor their catalytic behavior. Phosphine-functionalized MOF-808 served as solid ligand in a series of eight rhodium phosphine catalysts. These MOF-heterogenized molecular catalysts showed activity up to 2100 h -1 for ethylene hydroformyla-tion towards propionaldehyde as sole carbon-containing product. Combined experimental and computational methods applied to this unique MOF-based molecular system allowed unravelling structure and evolution of the Rh active species within the MOF under catalytic conditions, in line with molecular mechanisms at play during the hydroformylation reaction. The MOF-808 designed as a porous crystalline macroligand for well-defined molecular catalysts allows benefiting from molecular-scale understanding of interactions and mechanisms as well as from stabilization through site-isolation and recycling ability.


INTRODUCTION
Homogeneous and heterogeneous catalysis are two essential keys to our industrialized society.These two processes reign supreme over the production of essential commodities while remaining impervious to each other.2][3][4] The design of novel heterogenized molecular catalysts will meet the needs of increased productivity compared to parent molecular analogues, resulting from the combination of high site density per catalyst mass or volume unit, high accessibility towards active sites, enhanced stability allowing recycling or continuous processes while avoiding leaching of the active phase. 5 this perspective, siliceous, purely organic or hybrid porous materials as supports have attracted special attentions for the design of single site heterogeneous catalysts by immobilizing molecular species. 6,7Among these aforementioned porous solids, metal-organic frameworks (MOFs) have drawn intense research interest over the last few years due to their hybrid organic-inorganic nature associated with a high textural and chemical modularity. 8Moreover, the MOF extended crystalline network allows for combining crystallographic techniques with computational studies to elucidate catalytic site nature and molecular catalytic mechanisms at play inside the porosity.9-19   In molecular catalysis, phosphines are widely used ligands whose both electronic and steric properties strongly influence the catalytic activity of the bound metal cation. 20Phosphines have become essential to drive the catalytic activity and selectivity in applications ranging from fine chemistry with asymmetry and C-C coupling reactions, to petrochemistry with valorization of olefins using catalyzed oligomerization, hydroformylation, hydrogenation and metathesis. 21n the case of hydroformylation reactions, the known deactivation of rhodium phosphine molecular catalysts by the formation of dinuclear species would be circumvented thanks to their irreversible single site isolation onto solid supports. 22though MOFs have been employed as solid porous supports for molecular complexes in various catalytic applications, [23][24][25] MOFs based heterogeneous catalysts with well-defined and accessible free phosphine groups are still scarce. 26,279][30] Without spectroscopic evidences, strictly inert atmosphere for both synthesis and metal coordination has been claimed to be mandatory for the successful preparation of oxide-free P-MOF catalysts, 31,32 leaching of the active metal being observed in some cases upon recycling. 335][36] Wright, Clarke and coworkers already reported the post-synthetic grafting of sulfonated phosphine at the surface of the inorganic nodes of Hf-MOF, without oxidation, and its use as support for molecular iridium and rhodium catalysts, respectively for the reductive amination reaction and for the hydroaminomethylation of alkenes. 37spite some achievements using non-covalently bonded Rh-based catalysts within MOF, [38][39][40][41] the use of covalently functionalized structures for the hydroformylation reaction remains restricted so far to N-coordinated Rh and Co single sites into a pyrazolyl-based MOF for the transformation of styrene. 42is irreversible heterogenization of well-defined molecular catalysts is a prerequisite to fully heterogeneous single-site catalysis.Such catalyst would benefit from recyclability, thus increasing the sustainability of the catalytic process, as well as molecular understanding of the catalyst behavior at the MOF interface, both remaining still to be demonstrated so far.
In this work, we report well-defined rhodium phosphine-functionalized MOFs able to promote the hydroformylation of ethylene.We demonstrate the efficient and controlled phosphine ligand installation on the backbone of the MOF, to further fabricate Rh-based heterogeneous catalysts, by grafting the 4-(diphenylphosphino)benzoic acid (DPPB) on the secondary building unit (SBU) of a highly stable Zr-MOF without significant undesired oxidation (Scheme 1).Thanks to a combined experimental-computational approach, this well-defined Rh(DPPB) complex anchored inside the MOF cavity is demonstrated to retain its molecular nature and subsequently its molecular catalyst behavior while taking advantage of its heterogenization within a porous solid.

RESULTS AND DISCUSSIONS
Synthesis and characterizations of Rh@MOF-808 catalysts series.Zirconium-based MOF-808 was chosen here as the targeted solid support for the synthesis of heterogeneous catalyst for hydroformylation reaction.][46][47] MOF-808 is built around Zr6O4(OH)6 inorganic nodes connected by 1,3,5-benzenetricarboxylate (BTC) organic linkers to form a network of one-dimensional channels with pore size of 1.8 nm. 43[50] In this work, MOF-808 was synthesized by following a previously reported protocol by Yaghi and co-workers. 48After the synthesis and solvent molecules removal by exchange and activation at 120°C under vacuum, the MOF was characterized by powder X-ray diffraction (PXRD), N2 adsorption, elemental analysis, Fourier transform infrared (FT-IR), thermogravimetric analysis (TGA), transmission electron microscopy (TEM) microscopy etc. (see Supporting Information for details).The PXRD pattern of the synthesized MOF-808 was found to give a good match with those previously reported (Figure 1a).A Brunauer-Emmett-Teller (BET) surface area of 2282 m 2 g -1 was determined for MOF-808 from N2 adsorption at 77 K (Table 1 and Figure S2).Also, a combined study of scanning electron microscopy (SEM) and transmission electron microscopy (TEM) revealed octahedral morphology for the MOF particles with an average size of 500-600 nm (Figure S3-S4).
The grafting of DPPB moiety inside MOF-808 was first carried out with desolvated MOF (250 mg) using 0.25 mmole of the aforementioned phosphine linker in dimethylformamide (DMF, see supporting Information for details).The phosphine-grafted material was washed with DMF and ethanol, and further activated at 120 °C under vacuum.The amount of phosphorus was measured by ICP-OES to be around 2 wt% in the DPPB-functionalized MOF-808, which was named accordingly MOF-808-2.0DPPB.
The structural integrity of MOF-808-2.0DPPBwas confirmed by PXRD (Figure 1a).The grafting of the DPPB occurs via linker exchange at the ZrO4(OH)6 node by replacing a formate with the carboxylate group of the DPPB molecule in an equimolar ratio (Figure S13).Moreover, the presence of phosphorus in the resulting material was confirmed by solid state 31 P NMR analysis, with a peak at around -6.1 ppm corresponding to the dangling free phosphine, ie.
without significant oxidation and non-coordinated to any metal (Figure 1b).
The FT-IR spectrum of MOF-808-2.0DPPBexhibits an additional peak at 1186 cm -1 , when compared to that of the parent MOF-808, which can be assigned to the stretching of P-phenyl rings (Figure S9), supporting its successful grafting.The octahedral morphology of the MOF-808 crystals was found to be retained from SEM and TEM analyses upon DPPB grafting (Figure S10-S11).
The ratio between the two organic units, BTC and DPPB, was also calculated from 1 H NMR analysis of the MOF-808-2.0DPPBdigested in HF and DMSO-d6 solution (Figure S13).Both ICP-OES and 1 H NMR analyses were found in line regarding the number of DPPB linkers, ca.1.3 linkers grafted per node of MOF-808.A decrease of the surface area for MOF-808-2.0DPPB(1247 m 2 g -1 ) when compared to that of the parent MOF was observed and can be attributed to the inclusion of bulky DPPB linkers inside the porous network of MOF-808.(Table 1).The postulated molecular formula of the MOF-808-2.0DPPBwas calculated based on elemental analysis, ICP-OES, and NMR spectroscopy to give Zr6O4(OH)6(C9H3O6)2(HCOO)2.7 (C19H15O2P)1.3(DMF)1.5.
To gain more insight on the influence of the phosphine concentration inside the framework on both the active metal coordination at the phosphine site and the subsequent catalytic activity of the respective materials, we varied the amount of DPPB linkers grafted inside MOF-808.A series of DPPB-grafted MOF materials have been synthesized by varying the amount of DPPB from 0.1 to 2 mmol (in total 0.1, 0.25, 0.5, 1.0 and 2.0 mmol, respectively) during the grafting process.The structural integrity of these post-synthetically modified materials was found to be retained according to PXRD analysis (Figure S14).From ICP-OES measurements, the loading of phosphorus for these additional materials was found to be 0.9 wt%, 3.1wt%, 3.1 wt% and 3.7 wt% of phosphorus for the 0.1, 0.5, 1.0 and 2.0 mmol of linkers used for the grafting, respectively.Since the use of 0.5 and 1.0 mmol of linkers during post-modification have led to a similar amount of phosphine grafted into the MOF-808, the material synthesized from 1.0 mmol of phosphine was excluded from further studies.In addition to MOF-808-2.0DPPB,according to the loading in mass of phosphorus, the three additional aforementioned materials were named as MOF-808-0.9DPPB,MOF-808-3.1DPPBand MOF-808-3.7DPPB,respectively.Accordingly, the number of DPPB linkers per SBU was estimated for MOF-808-0.9DPPB,MOF-808-3.1DPPBand MOF-808-3.7DPPBto be 0.4, 2.5 and 3.3 respectively (Table 1).To further confirm the number of DPPB grafted in MOF-808, liquid state 1 H NMR analyses have been performed for the aforementioned compounds after digestion in HF and DMSO-d6 (Figure S16-S18).The number of grafted phosphines in the respective materials calculated from 1 H NMR was found to be in line with the number of DPPB per SBU obtained from ICP-OES.Nitrogen physisorption measurements were carried out at 77 K to assess the porosity of the DPPB-grafted materials.(Figure S15).Again, with increasing amount of bulky DPPB dangling moieties inside MOF-808, the N2 uptake was found to decrease along with the BET surface area and the total pore volume (Table 1).Moreover, solid state 31 P-NMR measurements for all these materials have revealed a negligible amount of oxidation on the phosphorus sites which thus remain readily available for further metalation (Figure S19-S21).
Finally, according to SEM and TEM analysis, the octahedron morphology of the MOF-808 crystallites was found to be retained after Rh precursor infiltration and no Rh aggregates have been observed (Figure S28-S31).
The actual coordination of the rhodium towards dangling phosphine atoms within the MOF cavity as well as the molecular nature of the Rh-P catalyst being assessed, the same methodology has been applied to the whole DPPB-grafted MOF-808 series.
The obtained new six catalysts were named according to the mass loading (wt%) of phosphorus and the nature of the rhodium precursor: RhH(CO)(PPh3)3 has led to MOF-808-0.9DPPB-RhH,MOF-808-3.1DPPB-RhHand MOF-808-3.7DPPB-RhH;while MOF-808-0.9DPPB-Rh,MOF-808-3.1DPPB-Rhand MOF-808-3.7DPPB-Rhwere obtained using Rh(CO)2(acac).According to PXRD analyses, the structural integrity of these catalysts has been maintained similar to that of the parent MOF-808 (Figure S34-S35).All the aforementioned six MOF-808 based catalysts have been found to be retained substantial amount of porosity even after the infiltration of Rh metal inside the porous architecture of the MOF-808.The magnitude in BET surface areas in these materials have followed the amount of the linkers grafted in MOF-808 through postsynthetic modification (Table 1).ICP-OES analyses have provided the amount of Rh infiltrated in respective catalysts (Table 1).It can be stated that with the higher amount of DPPB linkers inside MOF-808 the extent of Rh infiltration has got reduced slightly.This result can be attributed to the pore blocking effect of bulky DPPB linkers which have opposed the effective diffusion of the respective Rh precursors into the porous channels of MOF-808.For the set of catalysts obtained from RhH(CO)(PPh3)3 precursors, the wt % of Rh have been found to be 0.79, 0.62, 0.56 and 0.48 for MOF-808-0.9DPPB-RhH,MOF-808-2.0DPPB-RhH,MOF-808-3.1DPPB-RhHand MOF-808-3.7DPPB-RhH,respectively (Table 1).The ratio of linker to Rh (DPPB:Rh) have been calculated from aforementioned data and which have been 4:1, 11:1, 18:1 and 25:1 for MOF-808-0.9DPPB-RhH,MOF-808-2.0DPPB-RhH,MOF-808-3.1DPPB-RhHand MOF-808-3.7DPPB-RhH,respectively.On the other hand, catalysts obtained from Rh(CO)2(acac) precursors, the wt % of Rh have been found to be 0.83, 0.72, 0.63 and 0.54 for MOF-808-0.9DPPB-Rh,MOF-808-2.0DPPB-Rh,MOF-808-3.1DPPB-Rhand MOF-808-3.7DPPB-Rh,respectively (Table 1).Again, the linker to Rh ratio (DPPB:Rh) have been calculated to be 4:1, 9:1, 16:1 and 23:1 for MOF-808-0.9DPPB-Rh,MOF-808-2.0DPPB-Rh,MOF-808-3.1DPPB-Rhand MOF-808-3.7DPPB-Rh,respectively.Molecular structure of heterogenized catalysts.In order to get further insight into the atomic-level structure of the phosphinegrafted MOF-808 and the heterogenized Rh complexes, force-field based simulations and density functional theory (DFT)-level calculations were performed (see Supporting Information for computational details).In a first step, relying on the experimental phosphorus content in MOF-808-2.0DPPBand the reported X-ray crystal structure of MOF-808 , 43 we realized the in silico functionalization of the cubic unit cell of MOF-808 with 21 DPPB molecules by means of sequential force-field-based Monte Carlo (MC) docking and Molecular Dynamics (MD) simulations (see Supporting Information for computational details).Either in the docking steps or in subsequent MD equilibration steps, all phosphines were indeed found to simultaneously interact with two Zr neighboring atoms of a Zr6-oxocluster through their carboxylate groups (Figure S72).In a second step, using a similar MC/MD computational strategy than above and in line with the experimental Rh-loading in MOF-808-2.0DPPB-RhH(P : Rh = 21 : 2), two molecular Rh precursors were installed in the unit cell of the DPPB-functionalized MOF in the form of RhH(CO)(PPh)2 species, the Rh center lacking one of its PPh3 ligands thus allowing for a possible additional coordination to a dangling MOF-grafted phosphine (Figure 2a).In the MOF-808-2.0DPPB-RhH,the two Rh centers were indeed found to sit on one MOF-grafted phosphine, leading to RhH(CO)(PPh)2(DPPB) species.The Rh molecular complexes are not only attached to the MOF through direct Rh-P bond but also stabilized via π-π stacking with other free neighboring grafted phosphines (Figure 2b).
The same type of interactions was found when refining the proposed structure of MOF-808-2.0DPPB-RhHwith periodic DFT-D3 level calculations, further validating the employed methodology (Figure S73).
Prompted by the observed proximity between the fully coordinated RhH(CO)(PPh)2(DPPB) species and other grafted DPPB within the same pore, we explored whether more than one PPh3 ligand remaining at the Rh center could be further exchanged with one MOFgrafted DPPB.To this end, DFT calculations at the ωB97X-D level and including solvent effects were conducted on cluster models extracted from the periodic systems described above.These calculations predict that the conformation in which the Rh keeps two initial PPh3 ligands and coordinates to only one MOF-grafted DPPB is slightly more thermodynamically favorable (by 0.8 kJ/mol) than the binding to two DPPB from the MOF (Scheme S5).However, both of them might coexist in the MOF due to their small free-energy difference.The same approach was used to study the grafting of the Rh(acac) precursor into the DPPB-grafted MOF while exploring four different binding modes.More specifically, we analyzed the possible replacement of a CO or one of the oxygens of the acac ligand by a DPPB, exploring for the latter its binding to both an axial or an equatorial position at the Rh site (Scheme 2 and see Supporting Information for computational details).In the MOF-808-2.0DPPB-Rh,the most stable complex conformation involved a square-planar Rh center coordinated to the pristine acac ligand, one CO and one MOF-grafted DPPB (green frame in Scheme 2).Scheme 2. Various binding modes for the Rh-acac precursor studied inside the pores of the phosphine-modified MOF-808.Relative Gibbs free energies and relative enthalpies (italics, in parentheses) are given in kJ/mol.Again here, the coordination of the Rh center by two DPPB at MOF-808 was found to be less energetically favorable.Overall, the above theoretical calculations strongly support that the Rh phosphine catalysts can be efficiently heterogenized within the phosphine-grafted MOF-808, whereby the key molecular features of their coordination environment required to be catalytically active are preserved.The MOF-808-DDPB acts as a solid reservoir of phosphines whose site isolation allows for the generation of active carbonyl rhodium phosphine complexes without reaching the saturation of the Rh coordination sites with excess phosphines in its environment, in contrast to phenomena observed in solution where excess phosphines led to inactive Rh species. 51ir distribution function (PDF) analysis was used to further get experimental insight into MOF-based catalyst structure by ways of inter-atomic distances, relying on the above computed models of MOF-808-2.0DPPBand MOF-808-2.0DPPB-Rh(see Supporting Information for details).This technique has been already shown to be applicable even to amorphous MOF materials 52 and has been recently successfully used to assess the structural integrity of MOFsupported polyoxometalate catalysts 53 as well as Rh active sites in MOP-based materials. 54It has also been recently used to detect highly loaded single site metals trapped by catechol-benzoate ligands into MOF-808 (approx.0.3-0.4Fe or Cu per Zr). 55Here, our PDF analysis of MOF-808-funtionalized solids showed, as seen by XRD, that the MOF-808 long range ordered structure is preserved upon both DPPB and Rh species functionalization (see Supporting Information).Despite a low complex loading within the MOF (approx.0.2 Rh per Zr), small differences are observed between experimental G(r) curves at low range order (from MOF-808 to MOF-808-2.0DPPB,MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rh,see Figure S69-S70).We interpreted those small modifications as a signature of the grafting of the phosphine ligand (with P-C distances responsible of a shoulder at 2.75 Å) and Rh-P coordination.Moreover, a slight Zr6 node local modification can be observed and attributed to a small distortion of the Zr6(µ3-O)4(µ3-OH)4(µ1-OH)2(µ1-H2O)2(HCOO)4 node geometry as a consequence of the grafting of DPPB, and subsequent Rh coordination, replacing the more flexible formate moieties, while maintaining the Zr6 node's size and connectivity to BTC linkers.MOF-catalyzed ethylene hydroformylation.The obtained series of in-depth characterized heterogeneous catalysts were then employed in the ethylene hydroformylation reaction towards propionaldehyde.The increasing availability of ethylene from shale gas 56 and bioethanol 57 makes appealing its selective conversion into valueadded chemicals and the propionaldehyde produced from hydroformylation reaction can be further converted into propylene, 58 having a pivotal role in polymer industry.
In order to verify the role of grafted DPPB ligands in MOF-808 in catalysis, MOF based catalysts without DPPB and impregnated with respective Rh precursors were synthesized as well.MOF-808-RhH and MOF-808-Rh were synthesized from RhH(CO)(PPh3)3 or Rh(CO)2(acac) respectively, and further characterized with PXRD, ICP-OES and nitrogen physisorption.PXRD patterns of the respective catalysts confirmed the retention of structural integrity after the infiltration of respective Rh salts (Figure S40 and S43).The ICP-OES measurements showed the presence of physisorbed Rh species with loading of 0.73 and 0.82 wt% in MOF-808-RhH and MOF-808-Rh, respectively.
The early reaction conditions have been defined to be 20 bars of a 1:1:1 mixture of C2H4:CO:H2 using toluene as solvent at 110 °C in order to give reasonably high catalytic activity in triphasic batch reactor (Table 2).After completion of the catalysis, reactions were analyzed by gas chromatography (GC) of the liquid phase to calculate the turn over number (TON) with respect to internal standard (dodecane).It is worthy to note that all catalytic systems selectively produced propionaldehyde as sole carbon-containing product.Indeed, GC analysis of the gaseous phase, using head-space sampler, confirmed the absence of ethane or CO2 in the reaction mixture, whereas the formation of propanol was ruled out from the GC analysis of the reaction solution.

Table 2. Catalytic activity of MOF
To validate the role of grafted DPPB in MOF-808-DPPB-based catalysts, phosphine-free impregnated MOF-808-Rh and MOF-808-RhH were employed under the same catalytic conditions (Table 2, entries 3-4).The formation of propionaldehyde from ethylene was found to take place with TON of 887 and 1175 for MOF-808-RhH and MOF-808-Rh, respectively.The higher efficiency of MOF-808-2.0DPPB-Rhand MOF-808-2.0DPPB-RhHcan be attributed to the well-organized active catalytic sites which are coordinated by DPPB inside pores of MOF-808.Furthermore, ICP-OES analysis revealed a massive leaching of Rh in the case of DPPB-free MOF-808-RhH and MOF-808-Rh catalysts (Table S4).
To assess the stability of these heterogeneous catalyst, thorough characterizations of both MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rhhave been carried out.Retention of the structural integrity for both catalysts have been affirmed from PXRD analysis (Figure S54 and S60).BET surface area for the respective spent cathave been found to be maintained as similar to the parent catalysts (Figure S61).Also, TEM and SEM analysis have depicted that octahedral morphology of the catalyst crystals have been retained after catalysis and without formation of Rh particles (Figure S56-S57 and S62-S63).Retention of Rh inside the MOF materials (0.55 and 0.67 wt% for MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rhrespectively) was confirmed by ICP-OES analysis with a minimal leaching of ca.10% during the catalytic process.
The kinetic profile obtained for the best performing heterogeneous catalyst, MOF-808-2.0DPPB-RhH,showed a constant production of propionaldehyde for 3 hours which then slowed down slightly (Figure S53).This modification of reaction kinetics might be caused by an inhibition of active site in line with the increased propionaldehyde concentration, as previously reported for homogeneous counterparts, 59,60 and a change in gaseous substrates concentration in the solvent.A change of the reactor head-space for fresh gaseous feed allowed the reaction to restart with similar initial activity, highlighting the limitation due to change in reactant concentration in the closed vessel upon propionaldehyde production after 3 hours at 110°C and 20 bars (Figure S53).This evidenced the long-term stability of the catalyst under reaction conditions and allowed us to determine the intrinsic activity of the MOF-heterogenized Rh catalysts as turn over frequencies (TOF) defined as moles of propionaldehyde produced per moles of Rh per hours after 3 hours of reaction (Table 2).
Being demonstrated that heterogeneous MOF-808-DPPB-Rh systems were efficient catalysts for the ethylene hydroformylation, with comparable productivity than molecular homogeneous systems, the optimization of the DPPB loading within MOF-808, and the subsequent phosphine-to-rhodium ratio, was then investigated.All the eight catalysts obtained from two different Rh precursors and varied DPPB ligand have been tested for hydroformylation of ethylene using the same condition as earlier at 110 °C and 20 bar pressure for 3 hours (Figure 2).As shown in Figure 2a, In the case of the four catalysts obtained from RhH(CO)(PPh3)3, an initial increase in catalytic activity was observed with increasing DPPB loading from 0.4 linkers per SBU (MOF-808-0.9DPPB-RhH) to 1.3 linkers per SBU (MOF-808-2.0DPPB-RhH).The corresponding TON for MOF-808-0.9DPPB-RhHcatalyst has been found to be 875, while it raised up to 1685 for MOF-808-2.0DPPB-RhH.Upon further increase in the DPPB linkers loading, a decrease in the TON was observed (Table S3).As shown in Figure 2b, a less pronounced but similar trend was found for the other set of catalysts obtained from Rh(CO)2(acac).Thus, a maximum efficiency for the catalytic conversion of ethylene to propionaldehyde has been obtained with an optimum 1.3 DPPB linkers per SBU of MOF-808 for the both two series of catalysts.PXRD, physisorption, microscopy, XPS and ICP-OES analyses of the spent catalysts showed no change neither in the structure nor in the composition of both MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rhcatalysts (Figures S54 to S65).The PDF analysis further showed that long range ordered structure of the catalysts remains the same for the two Rh precursors after catalysis (Figure S71).
Having the MOF-808-2.0DPPBas the best performing platform for heterogenizing Rh-P species in the ethylene hydroformylation reaction, we scrutinized the effect of both the pressure and the temperature on the catalytic activity using the two MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rhcatalysts (Figure 3).As expected, in the two cases, the activity increased with both the pressure of the gas feed and the temperature.Under the conditions used here, ethylene was still selectively converted into propionaldehyde.From temperature studies, it was observed that there was no catalytic activity at 30 °C with any of the catalysts even at 20 bar pressure.At 50°C and above, with increasing pressure from 5 bar to 20 bar at a fixed temperature the catalytic activity increased significantly.The catalytic activity reached a maximum for both catalysts at 125 °C and 20 bar pressure of gas feed with TON of 6342 and 5483 and initial TOF of 2114 h -1 and 1828 h -1 for MOF-808-2.0DPPB-RhHand MOF-808-2.0DPPB-Rh,respectively.The recyclability of the MOF-808-2.0DPPB-RhH,selected as best performing catalyst, was then evaluated for 3 hours reactions (Figure 4).In this case, the reactor has been pressurized with the gas mixture at 20 bars and the pressure maintained constant using nitrogen flow upon CO/H2/C2H4 consumption.The MOF-808-2.0DPPB-RhH was found to reach the same catalytic activity for three consecutive runs, with a negligible Rh leaching, as determined by ICP-OES analysis of the supernatant solution.Thus MOF-808-2.0DPPB-RhHallowed to reach a cumulative TON of ca.13000 for the production of propionaldehyde after three catalytic runs.In contrast, the use of MOF-808-RhH, in which Rh sites are most likely physisorbed at the surface and pores of the MOF in the absence of phosphine ligand, led to a drastic decrease in catalytic activity upon reuse and a massive leaching of rhodium in solution (Figure 4 and Table S4).To the best of our knowledge, we report here the first example of ethylene hydroformylation catalyzed by well-defined MOFsupported molecular catalysts.For comparison, heterogenous organic polymer-supported Rh catalysts have been reported for the hydroformylation of ethylene.Under gas phase conditions, a phosphine-functionalized polystyrene loaded with Rh precursors show TON of 109 at 110 °C and 5 bar pressure of an equimolar mixture of CO/H2/C2H4 after 20 hours. 61More recently, another phosphine-based polymer along with the same Rh precursors has demonstrated very high activity under similar fixed-bed conditions with TOF up to 10000 h -1 for a Rh loading as low as 0.063 wt% within the polymer. 62[65][66] Molecular-level comparison of the two MOF-808-DPPB-Rh systems.To get insight into the evolution of the molecular Rh active site within the MOF, diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy and PDF analysis, in combination with computational models as structural references, evidence that the two catalysts from the two Rh precursors show similar spectroscopic fingerprints after exposure to CO/H2/C2H4 mixture.
In the case of MOF-808-2.0DPPB-RhH,DRIFT spectroscopy, performed under a CO/H2/C2H4 gas stream at atmospheric pressure and 70°C, showed two strong signals at 2080 and 2009 cm -1 (Fig. S66) which were attributed to geminal-dicarbonyls at the rhodium atom as previously reported for homogeneous Rh-P catalytic number of catalytic runs intermediate. 67The signal due to adsorbed hydrocarbons was essentially nil.After removal of CO from the gas feed, the signal at 2080 cm -1 was found to slowly vanish while the signal at 2009 cm -1 was shifted to 2017 cm -1, and no CH signal was formed.We concluded that ethylene was reacting with one of the two geminal CO-Rh, with an initial TOF of ca.4.7 h -1 (Figure S67, Left) and that no product accumulated at the surface.The same geminal-dicarbonyl signal was almost instantly recovered when CO was reintroduced in the feed. 68hese data show that CO adsorbed much more strongly than ethylene, which probably reacted from the gas-phase (or a weakly bound state) with one of the CO from the rhodium geminal-dicarbonyl species to form the reaction product that readily desorbed.This process is in line with the reported first catalytic steps for molecular Rh-catalyzed hydroformylation. 69The TOF measured in the DRIFTS experiment is smaller than that reported in Table 2, as lower temperatures (70 °C instead of 110 °C) and pressures (1 bar instead of 20 bars) were used.
In the case of MOF-808-2.0DPPB-Rh,the Rh(acac) cannot be considered as a catalytically active species, in contrast to RhH, according to reported hydroformylation catalytic mechanism. 70,71hus, under gas mixture stream at 1 atm and 70°C, the removal of the acac group took place leading to Rh gem-dicarbonyl species similar to those observed for the MOF-808-2.0DPPB-RhH.This species then reacts similarly with ethylene as in the case of the RhH precursor, with somewhat lower TOF (2.0 h -1 ) (see Figure S67, Right)).
Similarly, the PDF analysis performed on MOF-808-2.0DPPB-RhHcatalyst showed that the experimental G(r) curve remains the same at both small and long-range order before and after catalysis, highlighting the stability of the Rh active species within the MOF, as well as the integrity of the Zr6 nodes.In contrast, MOF-808-2.0DPPB-Rhexperimental G(r) curves, before and after catalysis, showed that the Rh(acac) precursor, as well as the Zr6 nodes, slightly evolved during catalysis to give the same structure after catalysis than that obtained for MOF-808-2.0DPPB-RhH(Figure S71).Thus, under catalytic conditions, the two MOF catalysts seem converging towards the same structure which is that based on the rhodium hydrido precursor, as expected from reaction mechanism. 70,71cordingly, from the experimental catalytic activities at 10 and 20 bars reported on Figure 3, the apparent energy of activation was calculated to be very similar for the two catalysts, with 30 kJ/mol for MOF-808-2.0DPPB-RhHand 32 kJ/mol for MOF-808-2.0DPPB-Rh[74]

CONCLUSION
We exploited the unique features of MOF materials to efficiently assemble molecular catalysts on a heterogeneous fashion following a stepwise methodology based on molecular chemistry principles.The ability of MOF-808 to undergo carboxylate ligand substitution allowed for the grafting of dangling phosphine groups, analogous to triphenylphosphines, without significant oxidation.The subsequent coordination with molecular rhodium complexes, using either hydrido or acetylacetonato precursors, gave rise to the heterogenization of organometallic Rh(CO) catalysts whose molecular nature was preserved within the porosity of the MOF.The obtained Rh-functionalized MOF were shown to heterogeneously catalyze the hydroformylation of ethylene towards propionaldehyde selectively under triphasic batch conditions.The best catalyst MOF-808-2.0DPPB-RhHshowed a TOF of 2114 h -1 and a TON of 6342 under 20 bars of a mixture C2H4:H2:CO (1:1:1) at 125 °C, also stable after recycling.Experimental PDF analysis supported by classical and DFT-level calculations allowed unravelling the atomic-level structure of the active site.Together, calculations and experimental data highlighted the evolution of the two Rh precursors within the MOF converging towards the same species upon catalysis.Furthermore, the MOF-heterogenized catalyst followed the reaction mechanisms established for homogeneous counterparts.Thus MOF, conceived as porous solid macroligands in molecular complexes, allow for the access of both the ultrafine knowledge of molecular catalysis and the operability of heterogeneous catalysis.
by1 H NMR analysis, b determined by ICP-OES analysis, c determined from nitrogen physisorption measurement at 77K using BET method.

Figure 2 .
Figure 2. Computed model of MOF-808-2.0DPPB-RhHcatalyst.a) 3D-periodic cubic cell of MOF-808 containing 21 DPPB and 2 Rh-hydrido precursors coordinated to one grafted phosphine each.The two Rh centers installed in the unit-cell are highlighted as cyan spheres.b) Detailed view of one of the grafted Rh-hydrido catalyst precursors, where the π-π stacking between phenyl rings is tracked by measuring distances between their centroids (red dashed lines).Distances are given in Å and C-bound H atoms are omitted for clarity.Color code: Zr (green), Rh (cyan), P (orange), C (gray), O (red), H (white).

a
-808-derived catalysts in ethylene hydroformylation compared to their homogeneous ana-Reaction conditions: 10 mg of solid catalysts (heterogenous) were used with Rh amount of 0.6-0.7 µmol; 110°C temperature, 20 bar pressure of a 1:1:1 mixture of C2H4:CO:H2, 20 mL toluene, 80 µL dodecane as internal standard; reaction time 3 hours; b turn over number (TON) was calculated from GC analysis and defined as moles of propionaldehyde produced per mole of Rh; c turn over frequency (TOF) defined as moles of propionaldehyde formed per moles of Rh per hour determined after 3 hours of reaction; d 0.65 µmol of Rh; e 0.7 µmol of Rh; f Homogeneous analogous reactions; g 1:1 mixture of Rh precursors and respective linkers (0.6-0.7 µmol).