A New Metal‐Ester Bonding Motif for the Synthesis of Hybrid Molecular Catalysts on Metal Oxide Supports Leads to Tunable Reactivity

ABSTRACT

The synthesis, characterization, and structure‐function properties for a new class of hybrid catalysts comprised of molecular catalysts with ligands covalently bound to metal oxide (MOx) supports via metal‐ester bonds is reported. Anchoring molecular catalysts to metal‐oxides through the ligand structure is an important motif for dye‐sensitized solar cells, electrocatalysis, and thermally driven catalysis. This new motif is amenable to a wide variety of metal oxide supports. Data show this new surface binding motif is able to achieve catalyst surface loadings of up to two molecules per nm² which is up to 60x greater than previously reported binding motifs. Catalytic reactivity trends show that the MOx support influences the molecular catalyst properties and reactivity through inductive effects. The influence of the support on catalytic properties has been correlated to the point of zero charge (PZC) of the support and shown to be predictable. Thus, this class of hybrid catalysts can be tuned by the choice of oxide support without making chemical changes to the catalyst, which will allow for reactivity beyond traditional Hammett parameter substituent changes. This allows for precise control of the catalyst nucleation, coordination environment, and accessible oxidation states, enabling highly tailored and controllable catalytic properties.

This work reports a new metal–ester bonding strategy for anchoring molecular catalysts to metal oxide supports, achieving exceptionally high surface loadings. Catalyst reactivity is predictably tuned by the support's properties, enabling control over electronic behavior and reaction pathways, thus opening new possibilities for designing highly customizable and efficient hybrid catalytic systems.

1. Introduction

Hybrid molecular/heterogeneous catalysts, which bind molecular catalysts to solid supports through support/ligand interactions, offer the ability to generate a heterogeneous catalytic system with molecular catalysts in controllable steric environments [1]. In a general sense, immobilization of molecular catalysts on solid supports offers the high selectivity of molecular catalysts with the practicality of heterogeneous catalysts, such as ease of catalyst separation from reaction solutions. Reports have shown additional advantages of hybrid catalysts, which include controlling the speciation (such as monomers vs. dimers), which can alleviate catalyst lifetime issues observed in some molecular homogeneous catalysis [2]. Hybrid catalysts can utilize a variety of solid supports [3, 4, 5, 6, 7, 8, 9, 10], but metal oxide (MOx) supports have the advantage of high thermal stability and, in some cases, semiconducting properties [11]. Figure 1A illustrates different approaches to anchor molecular catalysts to solid MOx supports that have been previously reported. Ligands containing acid/base groups, such as carboxylate or phosphonate, are synthetically simple and widely utilized to bind molecular catalysts to MOx supports through H‐bonding or ionic interactions [12, 13, 14, 15, 16]. This anchoring of molecular catalysts to the support through the ligand has the advantage of spatially separating the active transition metal from the support. This arrangement allows designing the steric environment and reactivity of the catalyst through traditional ligand design [14]. It also allows for a motif that is suitable for dye‐sensitized solar cells and electrocatalysis. However, the non‐covalent nature of the anchoring for this class of hybrid catalysts can lead to catalyst leaching from the surface under mild conditions [2, 9, 14, 17]. More stable, covalent bonding of molecular catalysts to MOx supports, however, can be achieved through ligands bearing silanol groups (Figure 1A) [18, 19, 20]. This approach, however, involves sterically bulky surface anchoring groups that require multi‐step syntheses and are highly susceptible to oligomerization.

FIGURE 1.

(A) Previously reported anchoring strategies for hybrid catalysts on metal oxide (MOx) supports. (B) Reported strategy for electrochemically inducing inductive effects on molecular catalysts anchored onto electrodes [23]. (C) This work depicts the new MOx‐ester hybrid catalyst reported here.

Recent reports have also shown that the reactivity of molecular catalysts anchored to solid electrode supports can be tuned through inductive effects by applying potentials without causing redox state changes, as shown in Figure 1B [21, 22]. The hydricity, oxidative addition, reductive elimination, and cross‐coupling reactivity of transition metal molecular catalysts can be tuned through inductive effects under electrochemical reaction conditions [23, 24, 25]. These reports show that hybrid molecular catalysts can break traditional Hammet parameter constraints and achieve highly tuned and targeted reactivity through inductive effects from the support. Here, we show this inductive effect can be achieved without applying electrochemical potentials, but instead by simply changing the identity of the solid support. Figure 1C shows this newly developed binding motif, which anchors molecular transition metal catalysts to MOx supports through ester linkages. When the ester linkages are bound to an aromatic ligand, the bulk properties of the support exhibit a direct and predictable effect on the reactivity of the molecular catalyst.

Support‐dependent reactivity trends have been observed for nanoparticle heterogeneous catalyst. In addition, surface organometallic chemistry (SOMC) has developed an approach to anchoring molecular catalysts to metal oxide supports through the catalytically active metal center, as shown in Figure 1A [26, 27, 28]. This approach generates uniform catalytic active sites with properties and catalytic activities that can be directly influenced by the MOx support [29, 30]. However, for both traditional nanoparticle catalysts and SOMC hybrid catalysts, direct binding of the catalytically active metal centers to the support limits the ability to sterically tune the active site in the same way that spatially separated hybrid catalysts offer [31].

Developing a new hybrid catalyst motif that: (1) covalently bonds the ligand structure to the support in a synthetically simple procedure, (2) achieves monolayer surface catalyst loadings, and (3) allows for highly tunable catalyst properties would be advantageous [31, 32]. Furthermore, a hybrid catalyst motif where the MOx support could influence the molecular catalyst properties through inductive effects without the need for an electrochemical system would result in hybrid catalysts that are greater than the sum of their parts and exhibit catalytic activity distinctly different from the molecular catalytic activity observed in homogeneous solution. Herein, we report on the design, characterization, and catalytic testing of a new hybrid catalyst motif that involves the covalent bonding of molecular catalysts to MOx supports through metal‐ester bonds (MOx‐ester). These MOx‐ester bonds are analogous to metal‐ester bonds achieved by Young through the high‐temperature reaction between acetic acid derivatives and metal oxides [33]. This motif, represented in Figure 1C, is distinctly different from previously studied hybrid catalysts as evidenced by infrared and x‐ray spectroscopy. The synthesis approach proved amenable to a variety of metal oxide supports that span a wide range in metal oxide properties. In direct comparison, the MOx‐ester loading achieved up to 60x greater catalyst surface loadings compared to acid/base binding hybrid catalysts. Catalytic testing of the MOx‐ester hybrid catalysts was used to highlight how the identity of the support could alter the electronic properties of the anchored hybrid catalysts. Suzuki–Miyaura cross‐coupling, starting from a Ni(II) pre‐catalyst, was specifically chosen as the test reaction due to the alternating reduction and oxidation steps in the catalytic cycle [34, 35, 36, 37]. Therefore, the influence the MOx support imparts on the molecular nickel catalyst should lead to discernible changes in the rate‐determining step of the catalytic mechanism. Differences in reactivity can then be correlated to measurable properties of the MOx support, such as the point of zero charge (PZC), to generate predictable reactivity trends and rational design of future hybrid catalysts.

2. Results and Discussion

2.1. Synthesis and Characterization

Covalent binding of the molecular catalyst to metal oxide supports through the newly developed transesterification reaction, shown in Figure 2, generated hybrid catalysts that will be labeled as MOx‐ester catalysts, where MOx represents the identity of the metal oxide support. The molecular complex [4,4′‐bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ was used to generate MOx‐ester catalysts. The reaction conditions were initially optimized using Aerosil 300 silica (SiO₂) powers as the oxide support and were later shown to be amenable to a variety of MOx supports as discussed below. General reaction conditions involved dissolving the [4,4′‐bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ complex in a 16:4:1 ratio of hexane:toluene:methanol. Metal oxide powders were then added to the catalyst solution to generate a slurry. The weighed amount of metal oxide was 10x the weight of nickel in the solution. The slurry was then heated to 105°C until the collection of methanol and hexanes in the reaction trap was concluded. As shown in Figure 2, the reaction converts the methyl ester on the ligand of the molecular catalyst into an MOx‐ester bond on the surface of the support, with the concurrent release of methanol, and the reaction equilibrium is driven towards MOx‐ester products by distilling off the methanol. The reaction was monitored by observing the collection of methanol in the trap. After the synthesis, the white metal oxide powders obtained the yellow‐green color of the molecular catalysts, and the solid hybrid catalysts were collected via vacuum filtration and washed with 50 mL of hexane and 20 mL of diethyl ether and dried in a 120°C oven for 30 min. The MOx‐ester catalysts were then stored on the benchtop with no further precautions.

FIGURE 2.

Reaction scheme for the anchoring of molecular nickel catalysts to metal oxide supports through the newly developed MOx‐ester motif.

Confirmation for the anchoring of the ligand to the SiO₂ support and the formation of silyl ester bonds was achieved with infrared (IR) spectroscopy. The C═O stretching frequency for the methyl ester group of the molecular [4,4′‐bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ complex in the absence of a support is at 1739 cm⁻¹, which is a common stretching frequency for an organic ester group. After synthesis of the hybrid SiO₂‐ester catalysts, the IR spectrum in Figure 3A showed a new C═O stretching frequency at 1747 cm⁻¹. Figure 3B also shows that the stretching frequency of 1747 cm⁻¹ is distinctly different than the C═O stretching frequency observed for noncovalent carboxylate binding to SiO₂, which occurs at 1720 cm⁻¹ for the end‐on binding mode and 1630 cm⁻¹ for the bridging binding mode as has been previously reported [33, 38, 39]. The lower C═O stretching frequency for the carboxylate group is the result of a delocalized negative charge leading to a weak, or partial C═O in the molecule [40]. Thus, the newly observed stretching frequency for the SiO₂‐ester catalyst indicates a new covalent silyl‐ester bond has formed between the molecular catalyst and the SiO₂ support. As can be seen in Figure 3A, the parent C═O stretching frequency for the methyl ester at 1739 cm⁻¹ was also observed for the SiO₂‐ester catalyst, indicating that the binding of the ligand to the support can occur through just one or both ester groups of the ligand, as illustrated in Figure 2.

FIGURE 3.

(A) IR spectral overlay of (blue) hybrid SiO₂‐ester catalyst and (yellow) molecular [4,4′‐bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst. (B) IR spectral overlay of (blue) hybrid SiO₂‐ester catalyst and (yellow) 4,4′‐dicarboxylate‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst bound to an SiO₂ support through acid/base binding. (C) IR spectral overlays of (blue) SiO₂‐ester, (yellow) ZrO₂‐ester, and (purple) Al₂O₃‐ester. D.) XPS spectrum of SiO₂‐ester as‐synthesized. E.) XPS spectrum of SiO₂‐ester after a 24‐hour cross‐coupling reaction performed at 105°C in dioxane solvent. (F) XPS spectral fit overlays of (blue) SiO₂‐ester, (yellow) ZrO₂‐ester, and (purple) Al₂O₃‐ester.

IR spectroscopy was also used to confirm binding of the molecular nickel complex to a variety of MOx supports (Figures 3C and S1–S7). The MOx supports for this study (SiO₂, TiO₂, ITO, ZrO₂, γ‐Al₂O₃, and MgO) were chosen for the range of point of zero charge (PZC) values of the metal oxides that vary from 3 to 13, which spans ten pH units [41]. For each of the oxides studied, a new C = O stretching frequency is observed with a higher wavenumber value than the parent C═O methyl ester stretching frequency. Thus, the IR spectra show that this approach to forming hybrid catalysts is amenable to a range of metal oxides, and the binding occurs through covalent bonds to metal oxide supports. This covalent bonding is unique in comparison to the widely utilized carboxylate binding to oxide supports and synthetically simpler than siloxane binding of molecular catalysts to SiO₂ supports [18].

The molecular structures of the MOx‐ester hybrid catalysts were examined with x‐ray photoelectron spectroscopy (XPS) due to the paramagnetic nature of the Ni catalysts, excluding NMR analysis. Figure 3D shows the XPS spectrum of the SiO₂‐ester, and Figure 3F shows the spectra overlays of the SiO₂‐ester, ZrO₂‐ester, and Al₂O₃‐ester catalysts. Figures S9–S16 show the XPS spectra for the hybrid catalysts on each of the SiO₂, TiO₂, ITO, ZrO₂, and γ‐Al₂O₃ supports. When these hybrid catalyst spectra are compared to the XPS spectrum for the unsupported molecular [4,4ʹ‐bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst (Figure S8), it can be seen that the splitting patterns and binding energies are very similar, which confirms the molecular structure of the hybrid catalysts. Furthermore, for each of the MOx‐ester hybrid catalysts, the electron binding energy of the Ni 2p_3/2 (855 eV) and the corresponding satellite and shake‐up peaks around 860 eV, along with the binding energy of the Ni 2p_1/2 (872 eV) and the satellite peaks near 878 eV, agree well with previously reported XPS spectra of a molecular nickel polypyridyl complexe [42]. Furthermore, diffuse reflectance spectroscopy on the SiO₂‐ester catalysts was compared to the UV‐visible spectrum of the molecular bis(methylester)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst. The molecular catalyst exhibited a strong absorbance with a λ_max at 330 nm assigned to the π‐π* charge transfer in the bipyridine ligand and a broad absorbance feature with a lower molar absorptivity near 430 nm likely due to a d‐π* metal to ligand charge transfer (Figure S17). The diffuse reflectance spectrum of SiO₂‐ester also exhibited a strong absorbance in the UV with a λ_max at 326 nm in addition to a broad and weak absorbance in the visible region from 380–440 nm (Figure S18). Thus, the combination of the IR, UV‐visible, and XPS spectra supports the hybrid structure of the MOx‐ester catalysts shown in Figure 2.

To determine the extent of catalyst loading on the supports, ICP‐MS analysis of the powder MOx‐ester catalysts was performed. For the SiO₂‐ester catalyst, the Aerosil 300 support has an average surface area of 300 m²/g (3x10²⁰ nm²/g) and contains an average of 4.6 surface silanol groups per nm² [43]. The computed van der Waal radii size of bipyridine molecular nickel catalysts is roughly 0.5 nm², thus a full monolayer of the catalyst should equate to 6 x 10²⁰ catalyst molecules per m² (per gram) of support, which is equivalent to two catalyst molecules per nm² of support. This equates to a maximum theoretical loading of 6.7 wt% nickel, taking into account the weight of the ligand, nickel, and the support. When this new MOx‐ester bonding approach was used with the Aerosil 300 support, a 6.0 wt% nickel loading was achieved. This wt% loading is higher when compared to previously reported single‐stie catalysts [44] and is a 60x greater catalyst loading compared to the maximum wt% loading we were able to achieve with the Ni carboxylate catalyst loaded onto the A300 SiO₂ support (0.1 wt%). This 6.0 wt% nickel loading occurred using the standard synthesis procedure described in the Supporting Information; however, by varying the ratio of molecular nickel catalyst to Aerosil 300, various wt% loadings of the catalyst on A300 could be achieved (Figure S20), showing that this catalyst synthesis technique can easily achieve targeted catalyst support loadings. Furthermore, when a large excess of the molecular nickel catalyst was used during synthesis, the determined surface loading never exceeded the calculated monolayer coverage, indicating that multilayers of the catalyst do not form. The higher catalyst loading for the MOx‐ester catalysts compared to the carboxylate‐loaded catalysts is attributed to the covalent metal‐ester bond formed during the synthesis of the MOx‐ester catalysts. Table S1 summarizes the PZC values, surface areas, and maximum achieved catalyst loadings for all of the MOx supports in this study, and the table shows that catalyst loadings in units of molecules/nm² scale with the surface hydroxyl group densities on the support. The results show that this new MOx‐ester motif is able to achieve high aerial surface loadings of molecular catalysts for multiple different metal oxides.

While this MOx‐ester hybrid catalyst offers a simple synthetic strategy for achieving targeted high catalyst loadings, it is worth noting that metal‐ester bonds are not infinitely stable. Methanol and ethanol drive the reaction in Figure 2 back to the organic esters, which results in the complete loss of the catalyst from the surface of the support. This highlights the importance of distillation step of the designed synthesis to ensure removal of the methanol and formation of the metal‐ester bond. However, soaking the SiO₂‐ester catalyst in solutions of sterically hindered alcohols, such as tert‐butanol, resulted in no observed loss of the molecular catalyst from the surface of the support (Figure S26). Aqueous solutions also lead to the hydrolysis of the metal ester bonds to form carboxylate groups that are no longer covalently bound to the MOx support. However, all reactions performed for this report were not meticulously dried, and added bases to the reactions were not anhydrous, yet the MOx‐ester catalysts exhibited bonding stability to the surface under these conditions, as evidenced by the XPS spectrum of the used catalyst shown in Figure 3E. These results show that this hybrid catalyst motif is stable to perform catalytic reactions in organic solvents prepared on the bench top, open to air and moisture.

2.2. Structure‐Function Properties

The Suzuki cross‐coupling between an aryl halide and an aryl boronic acid was chosen as the test reaction to elucidate the general reactivity trends for this class of hybrid catalysts. Complete optimization of the reaction conditions to maximum product yield was not performed; instead, each reaction was performed under standard and identical conditions to elucidate the effect the MOx support may have on the reactivity of the molecular catalysts. The Ni‐MOx‐ester catalysts in this report are synthesized in the Ni(II) oxidation state. Starting from a Ni(II) precatalyst is advantageous because the stability of most molecular Ni(II) complexes allows for benchtop reaction preparation [45]. When starting from a Ni(II) oxidation state, the catalyst must first be reduced to enter into the catalytic cycle, as shown in step 1 of Figure 4. Boronic acids in the presence of bases have been reported as weak one‐electron reductants for Ni(II) catalysts, thus the cross‐coupling mechanism would arise from a Ni(I/III) cycle as previously reported [46, 47, 48, 49, 50, 51]. Oxidative addition of the carbon electrophile takes place after formation of Ni(I), followed by transmetalation and subsequent reductive elimination to form the cross‐coupled product [17]. Mechanisms for Ni cross‐coupling have also proposed initial transmetalation followed by oxidative addition [46]. In either case, the redox step that follows the initial reduction of the catalyst is an oxidative addition step. The alternating reduction and oxidation steps of the mechanism should be sensitive to the electron density of the nickel catalyst. Electron‐deficient Ni(II) catalysts should undergo thermodynamically favored reductions to Ni(I) but exhibit rate‐determining oxidative addition. Whereas electron‐rich Ni(II) catalysts are expected to undergo facile oxidative addition but be inhibited by thermodynamic barriers to the initial Ni(II/I) reduction shown in step 1 of Figure 4.

FIGURE 4.

Left: proposed Suzuki coupling mechanism with MOx‐ester hybrid catalysts. Ph represents a phenyl group, tol represents a toluene group, and Ni represents a hybrid MOx‐ester catalyst composed of a molecular nickel catalyst covalently bound to a metal oxide support. Right: comparison of MOx‐ester catalyst reactivity towards varying the halogen on the carbon electrophile, highlighting the changes in reactivity with changes in support while keeping all other conditions constant.

To set a baseline of reactivity in comparison to the MOx‐ester catalysts, a series of control reactions was performed, which are summarized in Table S2. First, nickel sources in the absence of the bipyridine ligand did not exhibit catalytic cross‐coupling activity, indicating that nickel coordination to the ligand is necessary to achieve catalytic activity under the standard reaction conditions. Next, molecular nickel bipyridine catalysts in homogeneous solutions led to low cross‐coupled product yields ( < 10%) due to catalyst deactivation through dimerization of the molecular catalysts in solution, as has been previously shown [2, 52]. Homogeneous catalysts containing electron‐withdrawing moieties on the bipyridine ligand also led to a small amount of the homocoupled iodotoluene reactant in addition to the expected cross‐coupled product. This homocoupling has been previously seen and is attributed to the Ni(I) catalyst intermediate performing I• abstraction from the aryl iodide substrate [53, 54, 55]. A 4,4′‐dicarboxylate‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst was also loaded onto oxide supports through acid/base binding following a previously reported procedure [2]. This acid/base‐loaded hybrid catalyst produced a low yield of cross‐coupled product for both SiO₂ and ZrO₂‐supported catalysts. Acid/base hybrid catalysts achieve low catalyst surface loadings, thus, require large amounts of solid catalysts in the reaction, resulting in diffusion‐limited reaction [9]. Furthermore, loading the 4,4′‐dicarboxylate‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst onto a ZrO₂ support led to no observed change in reactivity per nickel catalyst compared to the SiO₂ supported catalyst. Lastly, the metal oxide supports did not exhibit any cross‐coupling reactivity in the absence of a nickel catalyst.

With baseline reactivity established, the new SiO₂‐ester hybrid catalyst was then tested under the standard reaction conditions. The point of zero charge (PZC) of the SiO₂ is reported as 3.5 [41]. Unexpectedly, the reaction solution turned purple during the reaction, and the solid SiO₂‐ester catalyst transformed from yellow‐green to orange in color (Figure S23). The reaction resulted in producing a 26% yield of cross‐coupled product and a considerable yield of the homocoupled iodotoluene and boronic acid reactants (Figure S28). A SiO₂‐ester catalyst with only 3 wt% loading (50% of the maximum catalyst surface loading) was also used to perform the selected cross‐coupling reaction, and a similar distribution of products was generated (Figure S41), which indicates the observed reactivity is inherent to the SiO₂‐ester catalyst and does not result from catalyst‐catalyst interactions occurring on fully loaded supports. A MOx‐ester catalyst on a non‐porous SiO₂ support (Ox50) was also tested, and the Ox50‐ester catalyst (Figure S42) showed similar product distributions to the (A300)SiO₂‐ester catalyst, which indicates that catalyst confinement and mass transport limitations do not have a direct effect on the observed catalytic activity.

The product yields and distributions achieved with the SiO₂‐ester catalysts are also distinctly different from the reactivity observed from the acid/base hybrid catalysts. The roughly even distribution of cross‐coupled to homocoupled products for the SiO₂‐ester catalyst indicates radical reactivity occurred during the reaction. Base‐activated boronic acids have been reported as weak one‐electron reductants and could explain the formation of radicals during the reaction [46, 56]. An EPR spectrum of the purple reaction solution (Figure S21A) exhibited a single signal with a g‐value near 2.0, indicating the formation of an organic radical [57], and the EPR spectrum of the orange, solid catalyst showed an EPR pattern indicative of a molecular Ni(I) complex (Figure S21B) [58, 59]. Thus, the Ni(II) hybrid catalyst can be reduced by the base‐activated boronic acid to generate an organic radical and a Ni(I) molecular catalyst bound to an SiO₂ support. For the case of the SiO₂‐ester catalyst, the Ni(I) intermediate may be too electron‐deficient to efficiently undergo two‐electron oxidative addition reactions with aryl halides, thus leading to the observed radical distribution of products. Homocoupling of the iodotoluene was also observed when using the molecular [4,4′‐bis(trifluromethyl)‐2,2′‐bipyridine Ni(II)]•2Cl⁻ catalyst in homogeneous solution (Table S2), which indicates that electron‐deficient nickel catalysts favor radical reactivity and a general Ni(I/II) oxidation state cycling compared to the Ni(I/III) cross‐coupling cycle shown in Figure 4.

To show further evidence that the SiO₂‐ester catalyst is electron‐deficient and therefore is rate‐limited by the Ni(I/III) oxidative addition step, the 4‐iodotoluene substrate was replaced with benzylic chloride because benzylic halides are activated carbon electrophiles and should undergo more facile oxidative additions. A GC‐MS analysis (Figure S31) showed that cross‐coupled products were achieved without observed homocoupling of the benzylic chloride. Furthermore, an attempted cross‐coupling reaction involving bromobenzene, which is thermodynamically more difficult to activate compared to iodobenzene, showed no formation of cross‐coupled product or homocoupled bromobenzene (Figure 4). This result is attributed to the inability of the catalyst or organic radicals in solution to perform Br• abstraction in addition to unfavorable oxidative addition involving the aryl bromide to the Ni(I) intermediate [60, 61]. The mechanistic studies on the SiO₂‐ester catalyst thus indicate that the low PZC SiO₂ support acts as an electron‐withdrawing group on the molecular nickel catalyst, which leads to preferential formation of a Ni(I) SiO₂‐ester species and rate‐determining two‐electron oxidative addition with aryl halides.

The standard cross‐coupling reaction was then performed using the ZrO₂‐ester hybrid catalyst with the PZC of ZrO₂ reported as 7.0 [41]. The ZrO₂‐ester catalyst generated the cross‐coupled product at a 60% yield. In addition, the purple color was not observed in the reaction solution, and only a minimal amount of homocoupled iodotoluene was observed (Figure S29). The reactivity exhibited by the ZrO₂‐ester is considerably different compared to the SiO₂‐ester, which indicates that the identity of the support influences the reactivity of the surface‐bound molecular catalyst in this new class of MOx‐ester hybrid catalysts. Compared to SiO₂, the less electron‐withdrawing nature of the ZrO₂ support generated a surface‐bound molecular nickel catalyst that can undergo two‐electron Ni(I/III) oxidative addition reactions with aryl halides (step 2 Figure 4), therefore avoiding the one‐electron radical (NiI/II) reactivity and achieving selective cross‐coupling. The ZrO₂‐ester catalyst was also able to achieve greater product yields compared to the molecular homogeneous catalyst because the hybrid nature of the ZrO₂‐ester catalyst prevented catalyst deactivation via molecular catalyst dimerization.

The hybrid catalyst on a γ‐alumina support (Al₂O₃‐ester, PZC = 9) also exhibited Ni(I/III) reactivity. The Al₂O₃‐ester catalyst, however, produced less cross‐coupled yield compared to the ZrO₂‐ester catalyst (25% vs. 60%), but with a much higher ratio of cross‐coupled to homocoupled product compared to the SiO₂‐ester catalyst (Figure S30). This result can be explained by the higher PZC value γ‐Al₂O₃ support acting as an electron donor to the nickel catalyst. Thus, reduction of the Ni(II) species of the Al₂O₃‐ester catalyst should be more thermodynamically demanding compared to reduction of the Ni(II) species of the SiO₂‐ or ZrO₂‐ester catalyst. This leads to the reduction of the Ni(II) complex, step 1 of the mechanism, to become the rate‐limiting step for the Al₂O₃‐ester catalyst. This rate‐limiting reduction of the Ni(II) precatalyst leads to less product formation over the same time frame compared to the ZrO₂‐ester catalyst, which can undergo facile one‐electron reduction of Ni(II) under the standard reaction conditions.

The cross‐coupling involving aryl chlorides typically requires electron‐rich nickel catalysts due to the bond dissociation energy trends [61, 62]. The more electron‐rich Al₂O₃‐ester catalyst, therefore, should be able to activate stronger carbon‐halogen bonds in comparison to the SiO₂‐ester and ZrO₂‐ester catalysts. The reactivity summary in Figure 4 compares the reactivity of the MOx‐ester catalysts for the coupling of various aryl halides. The electron‐deficient SiO₂‐ester catalyst showed no reactivity towards the aryl bromide or chloride substrates. Conversely, the ZrO₂‐ester catalyst shows a minimal amount of reactivity towards the aryl bromide. The Al₂O₃‐ester catalyst showed increased reactivity towards the aryl bromide substrate compared to the ZrO₂‐ester catalyst and exhibited cross‐coupling reactivity with the aryl chloride substrate. These reactivity trends support the hypothesis that as the PZC of the support increases, the anchored nickel catalyst exhibits reactivity more akin to electron‐rich molecular nickel catalysts.

The observed reactivity trends occurred by only changing the support while keeping the anchored catalyst and reaction conditions constant. It is worth noting, as shown in Table S2, none of the oxide supports alone exhibit any cross‐coupling reactivity. Mixtures of NiCl₂ salts and metal oxide supports in the absence of ligands also do not exhibit catalytic cross‐coupling activity. Previous reports have also shown that neither heterogeneous Ni nanoparticles on SiO₂ nor Rainey Ni can effectively perform Suzuki cross‐couplings [2]. In addition, the 4,4′‐bis(methylester)‐2,2′‐bipyridine ligand was anchored onto the SiO₂ support in the absence of nickel. Spectroscopic characterization of the ligand anchored to the support is shown in Figures S24 and S25. The ¹³C NMR spectrum is consistent with the previously reported spectrum of 4,4′‐bis(trimethylsilylester)‐2,2′‐bipyridine ligand [63], and the IR spectrum is consistent with the spectrum of nickel‐containing SiO₂‐ester catalyst. The C:N ratio determined by elemental analysis of 5.6:1 most closely matches a ligand singly bound to the oxide support, as illustrated in Figure 1, but cannot rule out a doubly bound species. Furthermore, elemental analysis was used to determine that 1.6 ligand molecules per nm² of support of anchoring were achieved. As reported in Table S2 entry 19, the metal‐less ligand anchored to the support exhibited no cross‐coupling reactivity. Taking the set of control reactions together shows that the hybrid catalyst as‐synthesized (i.e., SiO₂‐ester) is required for the observed catalytic activity.

To further examine the correlation between support PZC values and molecular catalyst reactivity, we also synthesized and examined MOx‐ester catalysts on TiO₂, ITO, and MgO supports. Performing the standard cross‐coupling reaction with the TiO₂‐ester catalyst led to a radical distribution of products similar to the product distribution obtained using the SiO₂‐ester catalyst (Figure S28). TiO₂ has a PZC value near 4 and thus can also be considered an electron‐withdrawing support similar to SiO₂ [41]. Conversely, performing the standard cross‐coupling reaction with the ITO‐ester catalyst led to more selective cross‐coupling compared to the SiO₂‐ester catalyst (Figure S33). The PZC value for ITO is near 6, thus there is an observed dependence between the PZC of the support and the product ratios between cross‐coupled and homocoupled products. A high PZC support, MgO (PZC = 13), was also examined. For this MgO‐ester catalyst, minimal product yield or conversion of any of the substrates was observed (Figure S34). The high PZC support becomes electron‐donating enough to stabilize the Ni(II) precatalyst and prevent appreciable reduction to the Ni(I) species and entrance into the catalytic cycle, and thus, minimal reactivity is observed. In all, utilizing metal oxide supports which spanned 10 orders of magnitude in PZC values led to a volcano plot type of reactivity as shown in Figure 5. Low PZC supports exhibited rate‐limiting oxidative addition reactions and limited the molecular nickel catalysts to valencies of either Ni(I) or Ni(II), which promoted radical reactivity. Supports with mid‐range PZC values showed facile one‐ and two‐electron redox events optimal for cross‐coupling. While high PZC values ≥ 9 exhibited reactivity trends are indicative of electron rich catalysts with rate‐determining reduction of the Ni(II) pre‐catalyst. These reactivity trends for the hybrid MOx‐ester catalysts suggest that the covalent ester linkages between the molecular catalysts and the metal oxide supports allow the metal oxide to act as an additional electron‐withdrawing or donating group on the molecular catalyst through electronic induction effects between the support and the molecular catalyst. Metal oxide supports with low PZC values withdraw electron density from the surface‐bound molecular catalysts, while supports with higher PZC values can act as electron‐donating moieties on the molecular catalyst. This additional aspect of control over hybrid catalyst reactivity is new and advantageous, especially considering the established advantages of hybrid catalysts, such as exhibiting the practicality of heterogeneous catalysts and preventing unwanted catalyst dimerization [64]. It is also worth noting that other properties of oxide supports, such as defect sites and crystallinity, can have an effect on metal catalysts when the metals are bound directly to the oxide support [65, 66]. However, for this motif, where the metal is spatially separated from the oxide support, and instead the ligand is bound to the support, the results show that the bulk PZC property of the support can be used to describe and help rationally design catalyst properties.

FIGURE 5.

Volcano plot relating cross‐coupled product selectivity to the PZC of the support. The product ratio is the moles of cross‐coupled product divided by the moles of the homocoupled iodotoluene product.

Post‐reaction analysis of the catalysts was performed using XPS to confirm molecular integrity of the catalysts. The XPS spectra of the SiO₂‐ester, ZrO₂‐ester, and Al₂O₃‐ester catalysts after performing cross‐coupling reactions are shown in Figures 1E, S12, and S14, respectively. The post‐reaction spectra maintain the same splitting patterns and relative peak intensities compared to the pre‐reaction samples, indicating the catalysts maintained molecular structure. A hot filtration experiment was also performed on the ZrO₂‐ester catalyst to test for possible leaching of catalytically active species into solution [9]. The solution filtrate did not exhibit catalytic activity for continued product formation when returned to reaction conditions after filtering off the solid hybrid catalyst (Figure S56). Whereas the solid ZrO₂‐ester catalyst filtered from the initial reaction was placed in a new reaction solution and continued to show catalytic activity (Figure S58).

Catalyst reusability experiments were also performed for the ZrO₂‐ester catalyst. After each reaction (standard reaction conditions described in Figure 4), the solid catalyst was filtered from the reaction solution, rinsed thoroughly with hexane and diethylether, dried, and then placed in a fresh reaction. For the cross‐coupling of 4‐iodotoluene and 4‐methoxyphenyl boronic acid, the product yields were 60%, 58%, and 53% for three consecutive reactions. These obtained yields show the catalyst exhibits stable reactivity over three reaction cycles, which is consistent with the post‐reaction structural characterization via XPS analysis. The slight drop in product yields from reaction one to reaction three can be attributed to possible incomplete recovery of the solid catalyst during the filtrations. In addition, the reaction results reported above for the ZrO₂‐ester catalyst were all performed at 2 mol% of nickel catalysts with respect to the aryl halide coupling partner, with the catalyst loaded at the maximum achieved catalyst loadings for ZrO₂ (0.24 wt% Ni, or one catalyst molecule per nm² of support, Table S1). Further reactivity studies were also performed using ZrO₂‐ester to explore how catalyst loading on the support and overall catalyst concentration versus the substrate affected the product yields. A reaction was performed with the ZrO₂‐ester catalyst with 0.1 wt% Ni loading at 2 mol% Ni, which resulted in a 54% yield of desired cross‐coupled product, consistent with the product yields obtained with the maximum loaded catalyst. Next, a reaction was performed with the maximum 0.24 wt% Ni loaded catalyst; however, at a 0.1 mol% with respect to the aryl halide. Under standard reaction conditions, this equated to just 12.5 mg of hybrid catalyst added to the reaction solution. Under these low catalyst loadings, a product yield of only 22% was achieved over the standard 24 h reaction time frame. This lower loading can be attributed to diffusional mass transport limitation under the batch reaction conditions of a stirred round‐bottom flask. This limitation may be overcome in a flow reactor design, but the optimization of cross‐coupled product yield is outside the scope of this study.

The comparison between these new hybrid MOx‐ester catalysts and the molecular methyl ester catalyst highlights that hybrid catalysts can be greater than the sum of their parts. The bonding of molecular catalysts to solid supports can prevent bimolecular catalyst interactions and thus prevent the dimerization of the catalysts, which can limit catalyst lifetimes and turnovers [2, 52]. Thus, hybrid catalysts can control the variance of nickel catalyst species during reactions by preventing dimerization of the catalysts and leading to longer catalyst lifetimes and product formation [17]. In addition, with the MOx supports acting as additional electron‐withdrawing or donating groups, this MOx‐ester catalyst motif helps further control the number of nickel species during reactions by controlling the thermodynamically available oxidation states under reaction conditions. The SiO₂‐ester catalyst limited the monomeric nickel catalyst to valencies of either Ni(I) or Ni(II) and promoted radical reactivity. While radical reactivity is not desirable for selective Suzuki cross‐coupling, this reactivity could be exploited for efficient catalysis involving radical generations. Conversely, the ZrO₂‐ester catalyst allowed for accessing monomeric nickel catalysts in Ni(I), Ni(II), and Ni(III) valencies while also promoting the two‐electron reactivity essential for selective aryl‐aryl cross‐coupling. Lastly, the Al₂O₃‐ester catalyst generated an electron‐rich nickel catalyst that could be optimized for activation of carbon‐heteroatom bonds.

3. Conclusions

In conclusion, we have reported the first systematic study on the anchoring of molecular catalysts to metal oxide supports through covalent ester bonds. This MOx‐ester hybrid catalyst motif exhibits many properties that are advantageous in comparison to previously reported hybrid catalyst motifs. The inductive effect the metal oxide induced on the molecular catalyst altered the reactivity, akin to recently observed inductive effects observed through applying electrochemical potentials. The synthesis procedure is able to achieve monolayer catalyst surface coverages with catalyst loadings as high as 2 catalyst molecules per nm² of support on a wide range of different metal oxide supports, which represents an order of magnitude greater surface coverage than some common hybrid catalyst motifs. Furthermore, covalently linking molecular catalysts through ester bonds to metal oxide support prevented catalyst dimerization. The influence of the metal oxide allowed for precise control of the nickel speciation and oxidation states and predictable switching between radical reactivity and controlled two‐electron oxidative addition reactions. While these reactivity trends were elucidated through catalytic Suzuki cross‐coupling, one can envision how the structure‐function properties determined in this report could be widely exploited for a range of catalytic reactions. Furthermore, one could imagine how this motif could lead to advances in devices that utilize bonding of molecular species to oxide supports, such as dye‐sensitized solar cells and photoelectrocatalytic systems.

Author Contributions

Joseph J. Kuchta: conceptualization, methodology, investigation, writing – original draft, writing – review and editing. Sarah M. Moody: investigation. Alexia M. Bradbury: investigation. D. M. S. C. Dissanayake: investigation. Laura C. Maybach: investigation. Santosh K. Balijepalli: methodology, investigation. John R. Regalbuto: methodology, supervision. Aaron K. Vannucci: conceptualization, methodology, funding acquisition, supervision, writing – original draft, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72776‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.