Bis- and Azabis(oxazoline)–Copper–Tungstophosphate Immobilized on Mesoporous Silica: Preparation and Use as Catalyst in Enantioselective Cyclopropanation

Abstract

Tungstophosphoric acid (TPA) has been supported on mesoporous silicas prepared using urea as the pore forming agent. The amount of urea (20, 30, or 40% w/w) influences the silica specific surface area (S_BET), total pore volume (Vp), and average pore diameter (Dp). The materials synthetized using 20% w/w (SiU20) display mainly mesoporous structures, with the highest Vp and Dp values being chosen to be used as TPA support. The SiU20-TPA solids with different TPA loadings (10, 20, or 30% w/w) have been used as supports for chiral copper catalysts with bis(oxazoline) or azabis(oxazoline) ligands. The catalytic efficiency of enantioselective cyclopropanation strongly depends on support morphology and TPA loading. SiU-TPA20 has been shown to be the optimal one. The stability of the complex is also a very important parameter, and the best results are obtained with an excess of chiral ligand to ensure the correct formation of the complex on the solid. In this way, with azabox-Cu/SiU20-TPA20 it is possible to obtain a highly selective (90% ee for the trans-cyclopropanes) and recoverable catalyst.

1. Introduction

The advantages of heterogeneous catalysis over homogeneous catalysis have driven the development of chiral catalysts that promote enantioselective reactions under mild conditions [1,2]. The covalent bonding of the chiral ligand to the support is the most used strategy for immobilization of chiral catalysts [3,4]. However, the use of non-covalent immobilization methods [5] presents several advantages, such as the simplicity of preparation and the utilization of the same ligands used in homogeneous catalysis, without requirement of additional synthetic steps. One of the non-covalent strategies is the immobilization of charged chiral catalysts with supports of opposite charge through electrostatic interactions. In the case of chiral catalysts acting as Lewis acids, it has been shown that the counterion plays an important role, on both activity and enantioselectivity [6]. This drives the search for solid counterions to be used as supports of chiral catalysts; typically, anions with poor coordinating capacity, derived from very acidic solids such as perfluorosulfonic polymers [7].

Keggin type heteropolyacids (HPAs) emerged as an alternative for that kind of solid acid. Keggin anions have the general formula of [XM₁₂O₄₀]ⁿ⁻, where X is the heteroatom (most common are P⁵⁺, Si⁴⁺ or B³⁺), M is the addenda atom (most common are Mo and W), and O represents oxygen. Our research group has vast experience in the synthesis and characterization of HPAs immobilized on different materials, such as silica, alumina, carbon, and titani, and their application as acid catalysts [8,9,10]. Moreover, they have been used as supports for chiral Rh-diphosphine [11], Ru-binap [12], and Rh-prolinamide [13] complexes.

Bis(oxazolines) (box) constitute a family of privileged chiral ligands, with a multitude of applications in homogeneous catalysis, such as intra and intermolecular cyclopropanation reactions, aziridination reactions, aldol reactions, Mannich reactions, etc. [14]. For that reason, their complexes have been immobilized on different kinds of supports through many different strategies [5,15,16]. For example, electrostatic methods have been used with clays [17,18], polymers [7], or Al-MCM-41 [19]. However, they have shown some stability problems due to decomplexation of metal in both the preparation and the recovery of the supported catalyst. Azabis(oxazolines) (azabox) [20] have emerged as an interesting alternative to bis(oxazolines), showing an improved stability against decomplexation [18], and hence better behavior in immobilization and recycling [21,22,23,24,25].

In a previous work, we described the electrostatic immobilization of azabox-Cu(II) complexes on tungstophosphoric acid (H₃PW₁₂O₄₀, TPA), supported on commercial silica obtaining azabox-Cu/TPA-SiO₂ catalysts that proved their efficiency in the reaction of cyclopropanation of styrene with ethyl diazoacetate [26].

In 1990, Yanagisawa et al. [27] described the preparation of mesoporous silicas with uniform pore size, and the next year, researchers at Mobil [28] reported the synthesis of aluminosilicates with unique pore size in the mesoporous range. Since then, mesoporous materials have attracted much attention in several fields related to catalyst preparation because of their high specific areas and narrow pore size distributions. They have been synthesized via sol–gel reactions using ionic and neutral surfactants as templates. However, the use of low-cost non-surfactant organic compounds as pore-forming agents, such as hydroxy acids and urea was reported [29,30]. Mesoporous silicas have been used as supports for different types of catalysts, including chiral catalysts [31], but there is no information on their use as supports for HPA-box/azabox-Cu complexes.

In this work, the goals were focused on the synthesis of the same type of chiral heterogeneous catalysts, but in this case, the chiral Cu-ligand complex was immobilized on mesoporous silica prepared using urea as the pore-forming agent. We also compared the performance of the two types of chiral ligands, box and azabox (Figure 1).

2. Materials and Methods

2.1. Preparation of Mesoporous SiO₂

Tetraethyl orthosilicate (10.7 mL) was stirred in absolute ethanol (88.6 mL) under an argon atmosphere at room temperature until a homogeneous solution was obtained (10 min) Then aqueous NH₄OH (25%, 1.55 mL) was added dropwise for 1 h and the final mixture was stirred for 3 h (Figure S3).

After that, an appropriate amount of urea–ethanol–water solution (1:5:1 weight ratio) was added under vigorous stirring. The amount of added solution (2.9, 4.3 or 8.8 mL) was fixed to obtain a urea concentration of 20, 30, or 40% (w/w) in the final material. The samples were named SiUX (X = 0, 20, 30 or 40).

The gel was kept in a beaker at room temperature for 1 week until dry. The solid was ground into a powder, washed with distilled water for 3 periods of 24 h to remove urea, and then dried at room temperature until a constant weight was achieved.

2.2. Immobilization of TPA on Mesoporous Silicas

Silica-supported TPA samples were prepared using the wet equilibrium impregnation technique. Impregnating solutions were prepared from the required amount of TPA and ethanol (10 mL) as the solvent. The amount of TPA was fixed to obtain a content of 10, 20, and 30% (w/w) in the final material. The solution and the support were allowed to stand in contact at room temperature until dry. The solids were calcined in air flow at 400 °C for 4 h. The solids were named SiUX-TPAY (X = 0, 20, 30, or 40% of urea; Y = 10, 20, or 30% of TPA). The TPA content in the SiUX-TPAY materials was determined by atomic absorption spectrometry, as described in the Supplementary Material. The silica-supported TPA was neutralized with triethylamine. For example, the immobilized heteropolyacid (SiU20-TPA30, 1 g) was neutralized by stirring in a solution of triethylamine (0.34 mmol) in dichloromethane (10 mL). The triethylamine amount added was chosen to achieve a triethylamine to TPA molar ratio of 3:3. Then, the solid was separated by filtration, thoroughly washed with dichloromethane, and dried under vacuum.

2.3. Preparation of Box- and Azabox-Cu/SiUX-TPAY

The immobilization of the chiral complexes was carried out through a previously described method [26]. The chiral ligands were synthesized following the method described in the literature [20]. The copper complex was prepared with Cu(OTf)₂ (0.02 mmol) and the corresponding chiral ligand (0.022 mmol) in the minimum amount of anhydrous dichloromethane under argon atmosphere. The solution was stirred for 15 min and filtered through a PTFE microfilter; thus, the solvent was removed under an argon flow. The complex was redissolved in nitroethane (5 mL, pretreated with Na₂CO₃ and distilled to eliminate all the acid traces). The silica-supported TPA was neutralized with triethylamine. Then, the neutralized support, in an amount necessary to achieve a 1.5 Cu:TPA molar ratio, was added to a solution of the chiral complex in nitroethane. The suspension was stirred under argon atmosphere for 24 h at rt. The solid was separated by filtration, thoroughly washed with nitroethane and dichloromethane, and dried under vacuum at rt.

2.4. Characterization of Catalysts

The method employed to determine the TPA content is described in the Supplementary Material.

The acidity of the supports was measured by potentiometric titration with a solution of n-butylamine in acetonitrile, using a MetrohmTitrino Basic 794 titrating device and a double junction electrode (see Supplementary Material for more details). Specific surface area (S_BET), total pore volume (Vp), and average pore diameter (Dp) of solid samples were determined by nitrogen adsorption/desorption techniques using Micromeritics ASAP 2020 equipment. X-ray diffractograms (XRD) were obtained using a Bruker D8 Advance A25 powder diffractometer using Cu radiation (Kα, λ = 1.5418 Å, intensity = 30 mA, and voltage = 40 kV). The diffractograms were collected in the range 2θ = 5–60°. Elemental analysis of final catalysts was performed in a Perkin–Elmer 2400 elemental analyzer. Copper and tungsten analysis was carried out by ICP in a Perkin–Elmer Plasma 40 emission spectrometer. The morphology of the samples was characterized by scanning electron microscopy (SEM) with a Philips 505 microscope. The energy dispersive X-ray analysis (EDX) of the samples was obtained using an EDAX 9100 analyzer at a working potential of 15 kV. FT-IR spectra of KBr wafers treated at 200 °C under vacuum (10⁻⁴ torr) were recorded in a Nicolet Avatar 360 FTIR spectrophotometer. Detailed operation conditions can be seen in the Supplementary Material.

2.5. Cyclopropanation Reaction

Ethyl diazoacetate (40 mg, 0.3 mmol) was slowly added (2 h) with a syringe pump over a stirred suspension formed by styrene used in excess as solvent, n-decane (50 mg, internal standard), and the catalyst 1 mmol% in dichloromethane (4 mL) under argon atmosphere at rt. The reaction was stirred for 24 h and the results were determined by GC as described elsewhere [23]. The solid was recovered by centrifugation, washed with hexane (3 × 3 mL), and reused without drying under the same conditions.

3. Results and Discussion

3.1. Preparation and Characterization of the Catalysts

The textural properties of the mesoporous silicas were determined from the N₂ adsorption–desorption isotherms. In all cases, the isotherms obtained (Figure 2) were of type IV (hysteresis H2) according to the IUPAC classification, which is characteristic of mesoporous materials [32].

The micropore specific surface area (S_MIC), the S_BET obtained by the t-plot method, and the Vp and the D_p obtained from the Barrett–Joyner–Halenda pore size distribution (Figure S4) are shown in

Table 1. Mesoporous SiUX samples’ textural properties.

Entry	Sample	SBET (m2/g)	SMIC (m2/g)	SMIC/SBET (%)	Vp (cm3/g)	Dp (nm)
1	SiU0	380	52	13.7	0.42	3.2
2	SiU20	464	51	11.0	0.74	5.6
3	SiU30	523	70	13.4	0.55	4.7
4	SiU40	440	42	9.5	0.53	4.9

.

In all cases, the S_BET is higher than that of silica prepared in the absence of urea, SiU0 (S_BET = 380 m²/g), and the highest S_BET was obtained using 30% of template (523 m²/g). A similar trend is observed with respect to pore volume and mean pore diameter, which are always higher than the values obtained for SiU0, demonstrating the role of urea as a pore template. It should be noted that these prepared silicas have higher S_BET than commercial ones used in related works (Ralt-Chemiesilice = 253 m²/g, Grace-Davison silica = 311 m²/g), but lower pore volume and pore diameter [10,23].

The contribution of microporous to the total surface is always lower than 15% (S_MIC/S_BET), but this contribution is similar in SiU30 and SiU0, whereas SiU20 and SiU40 materials present the highest percentage of mesoporous in relation to the total area (lower ratio S_MIC/S_BET). As S_BET of SiU20 is higher than SiU40, the first silica was chosen as support for TPA impregnation.

Steps for immobilization of TPA on mesoporous silicas are shown in Figure 3.

Based on the W amount determined by AAS that remained in the beaker after removing dried solids it was concluded that more than 90% of TPA was incorporated in SiUX-TPAY. The final amount of TPA for the prepared materials is shown in

Table 2. Chemical and textural analysis of silica-TPA solids.

Entry	Sample	% TPA (w/w)	SBET (m2/g)	SMIC (m2/g)	Vp (cm3/g)
1	SiU20-TPA10	9.6	259	67	0.74
2	SiU20-TPA20	18.1	233	58	0.69
3	SiU20-TPA30	29.3	201	54	0.60

. The presence of the bulky TPA anions (size ~1.2 nm) produces a significant decrease in surface area (Table 2), directly related to the TPA amount incorporated, mainly blocking the mesopores, while S_MIC (the surface related to micropores) remains almost invariant.

For all SiU20-TPA samples, a broad band in the 2θ range of 15–30° is observed, which is characteristic of amorphous silica (Figure S5). No peaks corresponding to H₃PW₁₂O₄₀ (JCPDS N° 75-2125) or its most common hydrates (H₃PW₁₂O₄₀·23H₂O and H₃PW₁₂O₄₀·6H₂O, JCPDS Nº 38-0178, and JCPDS N° 50-0304, respectively) are visible; this might be because TPA is well dispersed in the support as a non-crystalline phase.

The EDX measurements revealed that the ratio between Si Kα (at 1.74 keV) and W Lα (8.39 keV) signal area (ASık/AWL) in SiU20-TPA samples decreases with the increment of TPA content. EDX mapping images of Si and W elements (Figure 4) show that TPA is homogeneously distributed in the samples. In Figure 5, SiU20-TPA20 is shown as an example.

TPA shows four characteristic bands (1080, 982, 893, and 812 cm⁻¹) in the FT-IR spectrum [33,34,35], (Figure S6). The FT-IR spectra of SiU20-TPAY solids shows that some of them (1080 and 812 cm⁻¹) are overlapped with those of SiO₂ (Figure 5). However, the bands at 982 and 896 cm⁻¹ (stretching of W-O_terminal and W-O-W) are clearly observed. During the SiU-TPA synthesis, a partial transformation of the [PW₁₂O₄₀]³⁻ anions into the [PW₁₁O₃₉]⁷⁻ lacunar species could be produced. However, by comparing the samples FT-IR spectra with the corresponding to the lacunar anion, no characteristic bands (at 1100, 1046, 958, 904, 812, and 742 cm⁻¹) were detected. So, the transformation did not take place. The obtained results let us establish that the primary Keggin structure remains intact.

The titration curves of SiU20-TPA10, SiU20-TPA20, and SiU20-TPA30 samples are shown in Figure 6. The initial electrode potential (311, 359, and 366 mV, respectively) indicates the maximum acid strength of the sites. In all cases, the solids presented very strong (Ei > 100 mV) and strong (0 < Ei < 100 mV) acid sites (see Supplementary Material for more details about the classification). The area under the curve accounts for the total number of acid sites (N_AS) present in the titrated solid [36]. Ns values increased in the following order SiU20-TPA10 < SiU20-TPA20 < SiU20-TPA30 (0.09, 0.18, and 0.26 meg n-butylamine/g, respectively) in parallel with the TPA content increment.

Figure 7 shows the steps for preparation of azabox-Cu/SiUX-TPAY. The same method was carried out for box-Cu/SiUX-TPAY.

Due to the sensitivity of bis(oxazolines) to acids, the supported TPA was neutralized with an excess of triethylamine [23]. Figure 6 shows that in the neutralized materials, only weak acid sites remain. Then, the azabox-Cu and box-Cu complexes were exchanged in nitroethane as solvent, to prevent the leaching of TPA in more polar protic solvents, such as methanol.

Table 3. ICP analysis of azabox-Cu and box-Cu catalysts.

		mmol/g
Entry	Catalyst	Cu	W	P	Cu/TPA
1	azabox-Cu/SiU20-TPA10	0.032	0.422	0.032	1.00
2	azabox-Cu/SiU20-TPA20	0.054	0.586	0.046	1.17
3	azabox-Cu/SiU-TPA30	0.057	0.740	0.062	0.93
4	box-Cu/SiU20-TPA10	0.057	0.431	0.034	1.67
5	box-Cu/SiU20-TPA20	0.055	0.553	0.044	1.26
6	box-Cu/SiU20-TPA30	0.072	0.797	0.063	1.14

shows the ICP analysis of the different prepared chiral catalysts. It can be seen that the limit for exchange seems to be around 1 copper complex per TPA unit for azabox-Cu complex, whereas this ratio can be higher (up to 1.67 Cu per TPA) with box-Cu, perhaps due to the easier accommodation of the phenyl groups in comparison with the tert-butyl ones. Taking into account the three negative charges of the Keggin anion, the maximum amount of exchanged Cu(II) complex should be 1.5 in case of total exchange of the original triflate anions, or 3 in case of exchange of only one of the starting triflates [23]. The values under 1.5 obtained might be due to steric effects, regarding both the cationic complex with a bulky chiral ligand and the Keggin anion interacting with the bulky silica surface. The analysis also reveals a certain degree of TPA leaching during the neutralization and cation exchange process, which is more important with the increasing TPA loading, from almost no leaching for SiU20-TPA10 up to around 40% in SiU20-TPA30. It is likely that the leaching is produced by a weak interaction of part of the TPA units with the surface, due to a partial saturation of the silica surface.

The presence of the copper complex was also confirmed by FT-IR, where all bands in the skeletal range of the spectrum, mainly the imine bands (aprox. 1600 cm⁻¹) of the ligand are observed, indicating that the structure is intact (Figure S7).

3.2. Catalytic Results

These catalysts were tested in the cyclopropanation reaction of styrene and ethyl diazoacetate (Figure 8), using styrene as solvent, given the better results obtained under those conditions [23]. Styrene is the most widely used substrate for this kind of reaction, and allows a direct comparison with the already published results.

The reaction mechanism of cyclopropanation starts with the formation of the true catalyst by reduction of Cu(II) species to Cu(I) with diazoacetate. Then a second molecule of diazoacetate reacts with the immobilized Cu(I) complex to form an intermediate copper(I)-carbene, which is the active species for concerted cycloaddition to the olefin.

In all cases, once the reaction was finished, the catalyst was filtered and conditioned to be reused. Moreover, after the addition of an excess of diazoacetate to the resulting solution, no significant increase in the yield was observed, so it can be concluded that there is no significant leaching of active species, and therefore it reaffirms the heterogeneous nature of the reaction.

The results are shown in

Table 4. Results of the cyclopropanation reaction using chiral Cu/SiU20-TPAY catalysts ^a.

Entry	Catalyst	Yield (%) (TON) c	trans/cis	% ee trans	% ee cis
1	azaboxCu/SiU20-TPA10	10	53/47	42	36
	cycle 2	8	57/43	18	15
2	azaboxCu/SiU20-TPA20	40	70/30	65	49
3	box-Cu/SiU20-TPA20	22	65/35	60	46
	cycle 2	20	61/39	56	40
	cycle 3	18	67/33	48	32
4	azaboxCu/SiU20-TPA30	7	54/45	35	34
	cycle 2	3	57/43	24	19
5	azabox2Cu/SiU20-TPA20 b	42	71/29	90	71
	cycle 2	35	70/30	88	70
	cycle 3	37	71/29	87	71
	cycle 4	35	70/30	80	72
6	box2Cu/SiU20-TPA20 b	25	70/30	64	52
	cycle 2	26	69/31	60	47
	cycle 3	11	69/31	61	50
7	box2Cu/SiU20-TPA10 b	11	50/49	41	32
8	box2Cu/SiU20-TPA30 b	11	53/47	34	31

^a Reaction conditions: ethyl diazoacetate (0.3 mmol, slow addition); all reactions reach complete conversion of ethyl diazoacetate. ^b Catalysts prepared with a ligand/Cu ratio of 2. ^c TON ≈ yield due to 1 mol% catalyst.

. When comparing azabox-Cu catalysts on the three different supports, it can be seen that the behavior of SiU20-TPA20 (entry 2) is much better than that of SiU20-TPA10 (entry 1) and SiU20-TPA30 (entry 4), with higher yields (40% vs. 10 or 7%) and selectivities. The trans/cis diastereoselectivity is the typical value for this kind of catalyst in solution (70/30), whereas it is much lower in the case of the other two supports. The enantioselectivities are always below the values obtained in the homogeneous phase (around 90% ee), with a moderate result for SiU20-TPA20 (65% ee for the trans- cyclopropanes), but still higher than the enantioselectivities observed with the other two supports. This support was also tested with the box ligand (entry 3), leading to moderate activity (only 22% yield) and selectivities (60% ee for the trans cyclopropanes), similar to those obtained in solution with the same ligand.

The low yields are the consequence of the chemoselectivity between the desired cyclopropanation and the competitive dimerization (and further polymerization) of the diazocompound. In our experience this is greatly affected by the nature of the anion, with highly coordinating anions being detrimental for catalytic performance. In spite of its large size and possible charge delocalization, the Keggin anion seems to be behaving as a coordinating anion whose behavior might be modulated by changing its composition and/or structure [37].

Regarding recycling results, the solids leading to worse results in the first cycle are also poorly recyclable, with decrease in both activity and selectivities. The low enantioselectivity in the first reaction and the even lower enantioselectivity after recycling is a symptom of low stability of the ligand-Cu complex under the immobilization conditions, which is also dependent on the support. In this case, the optimal distribution of TPA on silica for box-Cu immobilization is 20%.

One possible solution for this problem is the use of an excess of ligand to favor the formation of the complex on the solid. The solid azabox₂Cu/SiU20-TPA20 was prepared in this way, and the results (entry 5) showed a significant improvement of the enantioselectivity, reaching values close to those of the reaction in solution (90% ee for the trans- cyclopropanes). Additionally, the catalyst was stable, with only a minor drop in both activity and selectivity up to the fourth cycle (35% yield, 70:30 trans/cis, 80% ee trans, and 72% ee cis). In the case of box ligand (entry 6), this strategy produced values of enantioselectivity close to the value obtained in solution, but improved the stability against recycling, with similar enantioselectivity during three cycles (61% ee trans and 50% ee cis). However, the activity in the third cycle decreased, probably due to the poisoning by coordination of by-products, such as diethyl maleate and fumarate produced by dimerization of ethyl diazoacetate. However, this method had no positive effect on the other supports (entries 7 and 8), demonstrating the important role of the support in the catalytic performance of the immobilized box-Cu complexes.

The supported Keggin polyoxometallate is acting as a solid counter-anion for the box/azabox-Cu complex after exchange of the starting triflates of the homogeneous one. Ideally both triflates should be exchanged to ensure the stability of the copper complex on the solid once it is reduced to Cu(I). In case of only one exchange, the reduced complex might be box-Cu^I-OTf, that would be soluble and then released to the solution. With both triflates exchanged, the only possible reduced species would be box-Cu^I-Keggin which would remain immobilized on the solid.

Whereas this immobilization method ensures the stability of the cation on the solid, the stability of the chiral complex depends on the equilibrium constant with the free ligand and metal, and the presence of active metal without ligand will produce the cyclopropanation reaction in a non-enantioselective manner. This equilibrium will be shifted to the complexed form by adding some additional ligand to the solution, and hence a larger proportion of the reaction will take place through the enantioselective pathway.

In principle, the presence of additional ligand should not affect the surface structure beyond the coordination sphere of the copper cation. In case of adding even more ligand, this might force double coordination to copper, making both the copper reduction and then the formation of the copper–carbene intermediate more difficult, with a detrimental effect on reaction rate and final yield.

It should be noted that the solid azabox₂Cu/SiU20-TPA20 (entry 5) leads to better results in comparison with the catalyst immobilized on laponite (46% yield, 71:29 trans/cis, 83% ee trans, and 76% ee cis) and nafion-silica (30% yield, 68:32 trans/cis, 88% ee trans and, 81% ee cis) [21].

Although with azabox₂Cu/SiU20-TPA20 a lower yield was obtained than with azabox₂Cu/Ralt-Chemie silica used in previous work (60%, trans/cis 70:30, 96% ee trans and 84% ee cis) [23], the latter drastically loses its activity after two uses (10% yield, trans/cis 66:34, 20% ee trans and 21% ee cis for the third use).

4. Conclusions

We have shown that the main textural properties of silicas (S_BET, Vp, and Dp), the TPA loading, as well as the kind and amount of the chiral ligand are important factors in the catalytic performance of azabox(box)-Cu/SiUX-TPAY solids for the enantioselective cyclopropanation reaction of styrene with ethyl diazoacetate. In fact, the best results of activity (42% yield), trans/cis diastereoselectivity (71/29), enantioselectivity (90% ee) and recyclability (at least four cycles) are obtained with azabox₂-Cu/SiU20-TPA20, demonstrating the importance of the fine tuning of all the parameters in the preparation of this kind of enantioselective catalyst.