Introduction

Dirhodium(II) complexes such as tetrakis[(S)-(-)-N-(p-dodecylphenylsulfonyl)prolinato]dirhodium(II) (Rh2(S-DOSP)4) are well established as catalysts in many processes, making them indispensable in producing pharmaceuticals and fine chemicals (Chen et al. 2017; Boruta et al. 2012; DeAngelis et al. 2009; El-Deftar et al. 2012; Gadakh et al. 2015; Gutmann et al. 2015; Li and Davies 2010; Liu et al. 2015; Rackl et al. 2017). The versatility of dirhodium(II) catalysts includes C–H, C–O, C–N, and C–C bond as well as ylide formations, X–H insertions, carbene-type reactions like cyclopropanations and many more (Adly et al. 2016; Davies and Denton 2009; Davies and Manning 2008; Davies and Morton 2011; Doyle et al. 2010; Huang et al. 2019; Miyazawa et al. 2016; Reddy et al. 2006; Roizen et al. 2012; Zhu et al. 2016). A protruding aspect in dirhodium catalyst research is enantioselective synthesis (Davies and Walji 2003; El-Deftar et al. 2012; Li and Davies 2010; Natori et al. 2015; Panne et al. 2008; Vos et al. 2002; Doyle et al. 2010; Miyazawa et al. 2016). The importance of this topic arises mainly from the biological activity of enantiomerically pure substances, especially in drug production (Williams and Lee 1985; Takeda et al. 2010; Li et al. 2020). To highlight one of the many uses of enantioselective dirhodium catalysis, metal catalyzed asymmetric carbenoid cyclopropanation is an example of great scientific interest (Sun et al. 2002, Singh and DattaGupta 1997, Maas 2004, Li et al. 2021a, Li et al. 2020, Doyle et al. 1990, Davies and Pelphrey 2011, Davies and Morton 2011, Davies and Antoulinakis 2001, Bianchini and Lee 2000, Awata and Arai 2012, Anciaux et al. 1980, Adly et al. 2016). The products of many asymmetric cyclopropanation reactions, namely polysubstituted cyclopropane derivatives, are widely used as drugs, such as the antidepressant Milnacipran (Doyle and Hu 2001; Bonnaud et al. 1987). In most of these reactions, homogeneous catalysts are used due to their high activity and enantioselectivity. These catalysts, however, pose disadvantages, especially concerning catalyst separation and recovery (Corma and Garcia 2006; Liu et al. 2017; Zhao et al. 2006). The separation is essential in the production of pharmaceuticals since remaining toxic heavy metals like rhodium may contaminate the product. Next, the high cost of rhodium is a compelling argument towards the simplification of the recovery procedure.

A possible solution to overcome these issues is the immobilization of homogeneous dirhodium catalysts on insoluble support materials, which is the focus of many research groups (Adly and Ghanem 2016; Davies and Walji 2005; Davies et al. 2004; Dikarev et al. 2010; Doyle et al. 2002, 2003; Hultman et al. 2003; Liu et al. 2018; Oohara et al. 2012; Sun et al. 2002; Takeda et al. 2010, 2011; Trindade et al. 2012). They have presented approaches for immobilizing dirhodium catalysts on mesoporous silica (Li et al. 2021a; Liu et al. 2017; Gutmann et al. 2015) or on crystalline nanocellulose (Liu et al. 2015). Here, the solid support replaces costly separation steps by simple filtration or centrifugation after the reaction. Recently, dirhodium complexes have also been heterogenized as metal–organic coordination-polymers as shown by Kaskel and co-workers (Nickerl et al. 2014) and some of the authors who prepared achiral as well as chiral coordination-polymers (Li et al. 2020, 2021b; Liu et al. 2016, 2018). Although, these results have demonstrated the feasibility to immobilize achiral and chiral dirhodium catalysts, there is still potential for improvement. The immobilization on inorganic supports like SBA-15, while showing good enantioselectivities, poses the problem of recovery of the precious metal after deactivation of the catalyst, since the silica material cannot easily be removed from the catalyst, i.e., harsh conditions like strong acids and microwave irradiation or hydrofluoric acid have to be applied to recover the precious metals. The immobilization as a metal–organic coordination-polymer necessitates the use of high amounts of rhodium, making this approach rather expensive. A solution to overcome both of these drawbacks is the immobilization of homogeneous, enantioselective catalysts on organic supports like polystyrene (Davies and Walji 2003; Takeda et al. 2010). Organic support materials can easily be burned after deactivation of the catalyst and the synthesis of the immobilized catalysts requires lower amounts of the active metal compared to the synthesis of metal–organic coordination polymers. A further environmentally friendly improvement to the use of supports such as polystyrene or other synthetic polymers is the use of naturally occurring biopolymers such as cellulose, chitin or chitosan. For example, cellulose can be easily obtained, is stable as a solid in a wide variety of solvents and reaction conditions and carries accessible functional groups on the surface, making it an excellent material as catalyst support (Chauhan and Yan 2015; Eberhardt et al. 2001; Keshipour et al. 2013; Li et al. 2018; Seyednejhad et al. 2019; Yamada et al. 2020; Yasukawa et al. 2015; Liu et al. 2015). The synthesis of modified biopolymers by a grafting-type modification compared to monomer modification with subsequent polymerization offers the advantage of simple preparation pathways due to the easy separation of the heterogeneous reaction products from reactants. In these works, exclusively achiral catalysts are used. To the best of our knowledge, for rhodium only one example has been published so far, where a cellulose supported catalyst was employed in enantioselective synthesis (Yasukawa et al. 2020). Their proposed approach however used a cellulose-based rhodium nanoparticle catalyst and chiral, homogeneous auxiliaries to obtain selectivity. While this is an advance compared to the use of neat homogeneous catalysts, the synthesis and especially the separation and recycling of the homogeneous auxiliaries still necessitate additional reprocessing after each catalytic cycle. Thus, there is a great demand to synthesize highly enantioselective, immobilized catalysts on cellulose materials as well as to monitor the immobilization steps using a combination of established and advanced analytical techniques.

In this work, we present a synthesis strategy for novel biopolymer supported enantioselective catalysts by immobilization of chiral dirhodium catalysts on cellulose. The approach is based on the attachment of a long-chain, pyridine-based linker molecule (TTDS-Nic) to microcrystalline cellulose. The functionalized biopolymer is then used to immobilize the homogeneous Rh2(S-DOSP)4 catalyst. The synthesis steps are monitored by means of thermogravimetric analysis (TGA), scanning electron microscopy (SEM), and liquid nuclear magnetic resonance spectroscopy (NMR) as well as liquid chromatography (LC). Furthermore, dynamic nuclear polarization enhanced solid-state NMR (solid-state DNP) is used as a powerful technique to investigate structural changes at the nanoscale during the functionalization of the cellulose materials. This technique uses the three-order of magnitude higher polarization of electron spins, typically from an external source such as from stable binitroxyl radicals, and transfer it into nuclear spin polarization (see recent reviews and references therein (Rankin et al. 2019; Lilly Thankamony et al. 2017; Gutmann et al. 2019)). This allows to drastically enhance sensitivity in solid state NMR experiments and enables the detection of small amounts of surface functionalization and insensitive nuclei in natural abundance on cellulose-based materials as it has been impressively demonstrated by some recent works (Berruyer et al. 2021; Höfler et al. 2021; Groszewicz et al. 2020; Gutmann et al. 2017; Zhao et al. 2014; Perras et al. 2017; Takahashi et al. 2012). This cel-cat is then applied in catalytic activity tests to monitor the enantioselectivity employing the intermolecular cyclopropanation reaction between styrene and methyl (3E)-2-diazo-4-phenyl-3-butenoate (4) as model reaction. Finally, the reusability and leaching properties of the cel-cat are inspected to validate the stability and sustainability of the novel catalyst system.

Experimental

Materials

All purchased materials were used without further purification unless specified otherwise. Pentane, ammonium chloride, THF, toluene, piperidine, ethanol, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC∙HCl), DMF, 4-dimethyl-aminopyridine (DMAP), magnesium sulfate, hydrochloric acid, sodium bicarbonate and N,N-diisopropylethylamine (DIPEA) were purchased from Carl Roth GmbH + Co. KG. Styrene, diethyl ether, 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), 4-acetamidobenzenesulfonyl azide (p-ABSA), methyl (3E)-4-phenyl-3-butenoate, 3-fluoro-4-pyridinecarboxylic acid, isonicotinic acid, methylene chloride, microcrystalline cellulose, ethyl acetate and 4,7,10-trioxa-1,13-tridecanediamine (TTD) as well as deuterated solvents for NMR analyses were purchased from Sigma-Aldrich. Rh2(S-DOSP)4 was purchased from Strem Chemicals, Inc., succinic anhydride from Acros Organics, acetonitrile (MeCN) and water for HPLC analyses from VWR and N-(9-fluorenylmethoxycarbonyloxy)succinimide (Fmoc-OSu) from abcr GmbH & Co KG.

Characterization

1H- and 13C-NMR spectra in liquid phase were recorded at room temperature by the analytical department of the Technical University of Darmstadt using a Bruker DRX Avance spectrometer. The 1H spectra were recorded at a frequency of 500.13 MHz, 13C spectra at a frequency of 125.77 MHz, respectively, employing single pulse excitation. The chemical shift is given in ppm using the signals of the solvent as secondary reference to TMS in CDCl3 (δ = 0 ppm).

1H → 13C CP MAS DNP measurements were performed on a Bruker Avance III HD 600 MHz DNP spectrometer at 14.1 T leading to a frequency of 600.00 MHz for 1H and 150.87 MHz for 13C. The spectrometer is equipped with a triple resonance 1H/X/Y low temperature MAS probe operating at 100 K using the maximum cooling power. This spectrometer is connected to a high-power, high frequency gyrotron system producing continuous microwaves with a frequency of 395 GHz. The samples were prepared by incipient wetness impregnation according to ref. (Sauvée et al. 2013) using a 15 mM AMUPol solution in DMSO-d6/D2O/H2O (60/30/10). Thereby, 15 mg of the sample were impregnated with 15 μL of this solution. The spectra were measured using 1024 scans and a recycle delay of 1.3∙TB where TB is the build-up time obtainable from a 1H saturation recovery experiment of the sample, measured with microwave irradiation. The contact time was set to 2.5 ms and a linear ramp from 100 to 50 was used to efficiently transfer 1H spin polarization to 13C under MAS conditions. The acquisition time was set to 10 ms and spinal64 (Fung et al. 2000) proton decoupling was applied during data acquisition. The spectra were recorded at 10.5 kHz spinning frequency and referenced to a silicon plug inside the rotor (0 ppm).

19F MAS NMR measurements were performed on a Bruker Avance II + 400 MHz spectrometer at 9.4 T corresponding to a frequency of 376.48 MHz for 19F. The spectrometer is equipped with a 3.2 mm HF/X/Y MAS NMR resonance probe. The spectra were measured using a one pulse sequence with 908 scans and a recycle delay of 300 s to ensure to be quantitative. The acquisition time was set to 82 ms and a π/9 pulse with a length of 0.46 μs was used. Details of the quantification using BaF2 are given in the ESI.

The High-Performance Liquid Chromatograms were recorded using a Waters Alliance e2695 HPLC System equipped with a Waters 2998 Photodiode Array Detector and a Reprosil Chiral OM-R 5µ 250 × 4.6 mm column or a ISAspher 100–3 C18 50 × 4.6 mm column, depending on the analyte. As eluents, mixtures of water and MeCN with 0.1% of trifluoracetic acid as additive were used in different gradients.

The mass spectrometric analyses were performed on an Impact II mass spectrometer from Bruker Daltonic, with a qTOF analyzer using either ESI or APCI as methods for ionization.

Thermogravimetric analyses were performed on a Netzsch TG 209 F3 Tarsus®. After drying the sample under nitrogen flow (40 mL/min) at 120 °C for 1 h to remove solvents, the sample was heated with a rate of 15 K/min to a temperature of 950 °C holding that temperature for 30 min. The measured weight differentials were corrected with a blank measurement.

SEM images were recorded on a XL30-FEG electron microscope from Philips. The images were recorded with an acceleration voltage of 15 kV and a spot size of 3. The magnification of all images was set to 12 000 to achieve comparable results. The samples were all coated with a gold coating using a sputter coater in an argon atmosphere.

UV/Vis spectra were acquired on a TIDAS S 500 MCS UV/NIR spectrometer of J&M Analytik AG with quartz cuvettes of 10 mm width. All samples were measured against a dark scan and a reference scan employing the solvent without analyte. The absorbance was measured in single scan mode with an integration time of 30 ms and 200 accumulations.

Determination of the degree of functionalization (DF) of mod-cel-1

The degree of functionalization (DF) of Fmoc-TTDS-Cellulose (mod-cel-1) was determined by UV-spectroscopic analysis of the concentration of the dibenzolfulvene-piperidine adduct formed in the deprotection step of mod-cel-1. 20 mg of Fmoc-TTDS-Cellulose were suspended in 5 mL of a 20% (v/v) solution of piperidine in DMF. After ultrasonication for 5 min, the suspension was filtered and topped up through the used filter to 20 mL with the piperidine solution to rinse out remaining dibenzolfulvene-piperidine adduct. The absorption of the resulting solution was then measured at 301 nm and the concentration calculated using an external standard (see supporting information Figure S8).

Synthesis of Fmoc-TTDS (3)

N-Fmoc-N′-succinyl-4,7,10-trioxa-1,13-tridecanediamine (Fig. 1) was synthesized based on the procedure in the literature (Zhao et al. 1999). 5.00 g (1.00 eq.) of succinic anhydride dissolved in 200 mL MeCN were added dropwise to a stirred solution of 11.02 g (1.00 eq.) of TTD in 200 mL MeCN at 0 °C over the course of 1 h. The solution was then decanted from the precipitating intermediate N-succinyl-4,7,10-trioxa-1,13-tridecanediamine, which was then dissolved in 500 mL of a mixture of Water and MeCN (50/50, v/v). To this solution, 21.93 g (1.30 eq.) of FmocOSu in 250 mL of MeCN were added dropwise over the course of 45 min at 0 °C, before adjusting the pH to a value of 8 with DIPEA. The mixture was allowed to warm to room temperature and stirred overnight. After removing the solvent from the mixture under reduced pressure, the resulting crude product was mixed with 500 mL of an aqueous sodium bicarbonate solution. The mixture was then extracted three times with portions of 250 mL ethyl acetate. The aqueous phase was adjusted to a pH of 1 using concentrated hydrochloric acid before extracting it again three times with portions of 250 mL ethyl acetate. After drying of the organic phase over magnesium sulphate and removing the solvent under reduced pressure, 17.3 g of the product (3) as a colorless oil were obtained.

Fig. 1
figure 1

Molecular structure of Fmoc-TTDS (3)

Full size image

Yield: 64%; 1H NMR (500 MHz, CD2Cl2): δ 7.78 (d, J = 7.6 Hz, 2H), 7.62 (d, J = 7.6 Hz, 2H), 7.40 (t, J = 7.5 Hz, 2H), 7.32 (t, J = 7.4 Hz, 3H), 4.38 (d, J = 7.0 Hz, 2H), 4.25 (dt, J = 23.6, 6.7 Hz, 1H), 3.62–3.47 (m, 6H), 3.31 (q, J = 6.1 Hz, 2H), 3.25 (q, J = 6.5 Hz, 2H), 2.62 (dd, J = 8.8, 4.6 Hz, 2H), 2.48 (t, J = 6.6 Hz, 2H), 1.74 (m, 4H); 13C NMR (126 MHz, CD2Cl2): δ 175.38, 173.14, 157.05, 144.65, 141.75, 128.11, 127.52, 125.57, 120.39, 70.79, 70.48, 69.61, 66.77, 47.78, 39.33, 38.52, 31.19, 30.47, 30.01, 29.27; HPLC: Rt = 13.6 min, > 96% @ 214 nm.

Immobilization of Fmoc-TTDS (3) on cellulose

The synthesis of mod-cel-1 (Fig. 2) was inspired by a procedure in the literature (Cateto and Ragauskas 2011). 1 g Cellulose (1 eq. primary OH-groups) and 1.13 g (1.5 eq.) DMAP were mixed in 100 mL DMF and homogenized in an ultrasonic bath. After the addition of 2.37 g of EDC∙HCl (2 eq.) and a second homogenization step, 7 g Fmoc-TTDS (3) (2 eq.) dissolved in a small amount of DMF were added to the mixture. The mixture was then stirred for 24 h at room temperature before quenching with 10 mL of 2 N hydrochloric acid. The solid was filtered off and washed with 100 mL portions of DMF, ethyl acetate, water, and ethanol before drying it in vacuum yielding 1.06 g of the product (mod-cel-1).

Fig. 2
figure 2

Modified monosaccharide fragment of Fmoc-TTDS-Cellulose (mod-cel-1)

Full size image

Yield (6.8% DF): 87%.

Deprotection of Fmoc-TTDS-cellulose

0.94 g Fmoc-TTDS-Cellulose were shaken briefly in 10 mL of a 20% (v/v) solution of piperidine in DMF. The product was filtered off and washed with 50 mL portions of DMF, methylene chloride, water, ethanol, and ethyl acetate. After drying in vacuum, 0.86 g of the product TTDS-Cellulose (Fig. 3, mod-cel-2) were isolated.

Fig. 3
figure 3

Modified monosaccharide fragment of TTDS-Cellulose (mod-cel-2)

Full size image

Yield (6.8% DF): 94%.

Coupling of (fluorinated) isonicotinic acid to TTDS-cellulose

Analogous to the Steglich-type esterification in the immobilization of Fmoc-TTDS on cellulose the formation of an amide bond between the amino group of the TTDS linker and the nicotinic acid derivative is carried out using EDC∙HCl and DMAP. To 0.5 g TTDS-Cellulose (1 eq. amino groups) suspended in 100 mL DMF, 0.57 g DMAP (1.5 eq.) were added. The mixture was homogenized by means of ultrasonication. After the addition of 1.18 g EDC∙HCl (2 eq.) and another ultrasonication step, 0.76 g isonicotinic acid (2 eq.) were introduced to the reaction mixture. The reaction was carried out over the course of 24 h at room temperature. The reaction was then quenched using 10 mL of 2 N hydrochloric acid before removing the solvents by filtration. The filter cake was washed with 50 mL portions of DMF, ethyl acetate, water, and ethanol before drying under vacuum. 0.45 g of the product (Fig. 4, mod-cel-3) was isolated.

Fig. 4
figure 4

Modified monosaccharide fragment of TTDS-Nic(F)-Cellulose (mod-cel-3 and mod-cel-3a)

Full size image

Yield (6.8% DF): 86.7%.

The synthesis of the fluorinated analogue (TTDS-NicF-Cellulose) was carried out in the same way using 0.87 g (2 eq.) of 3-fluoro-4-pyridinecarboxylic acid instead of isonicotinic acid. 0.435 g of the product (Fig. 4, mod-cel-3a) was isolated.

Yield (6.8% DF): 83.8%.

Immobilization of Rh 2 (S-DOSP) 4 on mod-cel-3

The immobilization of Rh2(S-DOSP)4 by axial coordination was carried out in 15 mL of degassed toluene under an argon atmosphere. To a suspension of 0.1 g (1 eq.) mod-cel-3 in toluene, 0.18 g Rh2(S-DOSP)4 were added. The mixture was then gradually heated to 120 °C over the course of 2 h. After 6 h of reflux, the solvent was removed by filtration and the product was washed two times each with 30 mL portions of toluene, THF, and ethanol, resulting in 0.13 g of the product (Fig. 5, cel-cat) after drying under vacuum.

Fig. 5
figure 5

Modified monosaccharide fragment of TTDS-Nic-Rh2(S-DOSP)4-Cellulose (cel-cat)

Full size image

Yield (6.8% DF): 77.84%; ICP-OES: 4.35 wt.-% Rh.

Synthesis of methyl (3E)-2-diazo-4-phenyl-3-butenoate (4)

Methyl (3E)-2-diazo-4-phenyl-3-butenoate (Fig. 6) was synthesized according to the procedure described in the literature (Ibbeson et al. 2014). To 2.00 g (1 eq.) methyl (3E)-4-phenyl-3-butenoate in 100 mL of dry, degassed MeCN at 0 °C, 4.11 g (1.1 eq.) of 4-acetamidobenzenesulfonyl azide were added under argon atmosphere. After the addition of 3.4 mL (1.1 eq.) of DBU over the course of 15 min, the solution was stirred for additional 3 h at 0 °C under argon atmosphere. The reaction was then quenched adding 50 mL of a concentrated ammonium chloride solution. The resulting mixture was then extracted three times using 50 mL diethyl ether each. The combined organic extracts were then dried over magnesium sulphate before removing the solvents under reduced pressure in an ice bath. The resulting red oil was then purified by column chromatography yielding 2.83 g of the product (4) as a red oil.

Fig. 6
figure 6

Molecular structure of methyl (3E)-2-diazo-4-phenyl-3-butenoate (4)

Full size image

Yield: 60%; 1H NMR (500 MHz, DMSO): δ 7.42 (d, J = 7.2 Hz, 2H), 7.31 (t, J = 7.7 Hz, 2H), 7.19 (t, J = 7.1 Hz, 1H), 6.52 (d, J = 16.3 Hz, 1H), 6.48 (d, J = 16.4 Hz, 1H), 3.78 (s, 3H); 13C NMR (126 MHz, DMSO): δ 164.82, 136.77, 128.65, 126.93, 111.20, 63.48; HPLC: Rt = 17.17 min, > 92% @ 214 nm.

Catalytic asymmetric cyclopropanation

The catalytic cyclopropanation reactions (Scheme 1) were carried out at -78 °C, 0 °C, 10 °C and 35 °C respectively. To a degassed solution of 1.2 M (5 eq.) styrene in 1.10 mL of pentane, 54 mg (1 eq.) methyl (3E)-2-diazo-4-phenyl-3-butenoate (4) in 2.2 mL pentane were added. 8 mg (0.01 eq.) of the immobilized catalyst (cel-cat) were then suspended in the solution and the mixture was stirred for 24 h at the given temperatures above. The reaction mixture was then filtered and analyzed by chiral HPLC using a gradient from 30 to 80% (v/v) of MeCN in water over the course of 25 min with a flow of 1 mL/min employing a Reprosil Chiral OM-R 5µ 250 × 4.6 mm column (see ESI Fig. S9 for an exemplary chromatogram).

Scheme 1
scheme 1

Catalytic cyclopropanation of methyl (3E)-2-diazo-4-phenyl-3-butenoate (4) and styrene

Full size image

To identify the retention times of the enantiomers, the reaction given above was carried out as well with enantiopure homogeneous Rh2(S-DOSP)4 and Rh2(R-DOSP)4 catalysts at − 78 °C.

Recycling of the catalyst

To test the recyclability of the heterogenized catalyst, it was used in the cyclopropanation reaction for 4 consecutive times. For this purpose, 32 mg of the immobilized catalyst were suspended in 5 mL of degassed pentane under an argon atmosphere. To the suspension, 700 µL (5 eq.) styrene were added, and the mixture was heated to 35 °C. To the reaction mixture, a solution of 213 mg of methyl (3E)-2-diazo-4-phenyl-3-butenoate (4) in 10 mL degassed pentane were added dropwise.

Samples were taken for HPLC analysis directly after the addition of the substrate as well as after each hour in the first 12 h, and then every 12 h to a total reaction time of 48 h.

Results and discussion

Synthesis of the materials

In Scheme 2, the synthesis route for the Fmoc protected linker molecule (3) is shown.

Scheme 2
scheme 2

Synthesis strategy for the linker molecule Fmoc-TTDS (3)

Full size image

This synthesis is based on a Schotten-Baumann type acylation with succinic anhydride on the linear diamine TTD (1) without the addition of a base to achieve the single-sided amide TTDS (2). After Fmoc-protection of 2, the isolation of the desired product Fmoc-TTDS (3) is readily achieved using the pH-dependent solubility of the linker molecule in comparison with the side products. The protection of the linker molecule is necessary to prevent intramolecular amide formation or dimerization, respectively, during the coupling of the linker to the cellulose. Despite the exclusive stoichiometric control of the reaction to form the mono- instead of the disubstituted diamine, the linker is synthesized with a yield of 64% and a high purity of > 96% (HPLC) without further purification.

As shown in Scheme 3, Fmoc-TTDS-Cellulose (mod-cel-1) is then obtained by a Steglich esterification of the linker molecule 3 with the microcrystalline cellulose (cel). This step is followed by the deprotection of the amine group to obtain the modified cellulose TTDS-Cellulose (mod-cel-2). Finally, the isonicotinic acid derived head group is attached to mod-cel-2 via an amide bond formation to enable the coordination of the chiral Rh2(S-DOSP)4 catalyst.

Scheme 3
scheme 3

Immobilization steps of Rh2(S-DOSP)4 on Cellulose. The functionalized derivatives of cellulose are depicted as a modified monosaccharide fragment of the respective polymer

Full size image

For analytical reasons, the isonicotinic acid derived head group is used in two different derivatives: One with a fluorine marker (mod-cel-3a) to determine the degree of functionalization via 19F NMR and one without (mod-cel-3) for the attachment of Rh2(S-DOSP)4. The attachment of the Rh2(S-DOSP)4 to mod-cel-3 is then achieved by axial coordination to the nitrogen of the head group to obtain the immobilized catalyst Rh2(S-DOSP)4-TTDS-Nic-Cellulose (cel-cat). A closer depiction of the proposed coordination of the paddlewheel type Rh2(S-DOSP)4 catalyst to the head group of mod-cel-3 is shown in Fig. 7.

Fig. 7
figure 7

Spatial representation of the coordination of Rh2(S-DOSP)4 in cel-cat

Full size image

Characterization of the materials

Employing scanning electron microscopy, the microscopic structure of the unmodified cellulose (cel (A)), the modified celluloses (mod-cel-2 (B) / mod-cel-3 (C)) and the final immobilized catalyst (cel-cat (D)) are analyzed and compared. Figure 8 shows that there is almost no change in the morphology between the samples. This result is underlined by analyses of the DNP enhanced 1H → 13C CP MAS NMR data of mod-cel-1, mod-cel-2 and mod-cel-3a which suggest no significant changes of structural order during the modification of the cellulose (spectra are shown in the supporting information Fig. S5). Furthermore, the separation characteristics of the cellulose by filtration or centrifugation did not change during the modification process. This allows the separation of the immobilized catalyst from reaction mixtures in a similar way compared to cellulose.

Fig. 8
figure 8

Comparison of the morphology of cel (a), mod-cel-2 (b), mod-cel-3 (c) and cel-cat (d) at 12 000-fold magnification

Full size image

After immobilization of Fmoc-TTDS (3) on the cellulose material, leading to mod-cel-1, the degree of functionalization (DF) is quantitatively analyzed by UV/Vis spectrometry. Here, the amount of the dibenzofulvene-piperidine-adduct which is formed during cleavage of the protecting group from the amino group of the linker on the cellulose using piperidine in DMF, is quantified. With complete deprotection of the amino groups, which can be assumed under the present reaction conditions (Greene and Wuts 2007), the DF of the primary hydroxy groups of the cellulose is determined to be 6.8%. This DF is expected to remain constant, since the reaction conditions in the next step, the attachment of the linker head, are the same as in the attachment of the linker itself.

Since the surface of the material and therefore the obtained degree of functionalization is low compared to the bulk of the cellulose material, the identification of functional groups is a challenging task that requires specific analytical techniques. Here, DNP enhanced 1H → 13C CP MAS NMR is employed to identify modifications of the cellulose at the molecular level. Figure 9 shows the comparison of the DNP enhanced 1H → 13C CP MAS NMR of cellulose modified with the TTDS linker containing different chemical modifications as they occur during the synthesis steps. For the Fmoc protected TTDS linker grafted on cellulose (mod-cel-1, Fig. 9a), characteristic signals at 120–130 ppm as well as at 135–145 ppm referring to aromatic carbons of the Fmoc protection group and a signal at 155 ppm referring to the carbamate carbon of the protecting group are obtained. These signals vanish after the deprotection step (Fig. 9b), indicating the successful Fmoc protection and deprotection of the TTDS linker. On the other hand, the signals at 165–175 ppm which can be assigned to carbon 1 in the ester group and carbon 4 in the amide group do not vanish after deprotection. This observation confirms qualitatively the stable binding of the TTDS linker on the cellulose as well as the stability of the linker itself under the conditions of the deprotection.

Fig. 9
figure 9

Comparison of the DNP enhanced 1H → 13C CP MAS NMR spectra measured at 10.5 kHz spinning rate of mod-cel-1 (a) (grey/solid line), mod-cel-2 (b) (red/dashed line) and mod-cel-3a (c) (blue/dotted line). The asterisks indicate the positions of spinning sidebands from the cellulose and radical matrix signals verified by MAS experiments using different spinning rates (see ESI for details)

Full size image

After the binding of the nicotinic acid derived linker head (mod-cel-3a, Fig. 9c), several signals in the range between 115 and 160 ppm appear that are clearly attributed to aromatic carbons originating from the nicotinic acid derived head group. More importantly a signal at 156 ppm appears which is clearly assigned to the carbon in amide groups, indicating the successful binding of the head group to the linker. Again, after this modification step the signals at 170–175 ppm are preserved which proves the intactness of the linker and its binding to the cellulose after the modification step.

After coupling of the nicotinic acid derived linker head, the degree of functionalization is determined on mod-cel-3a as model system for mod-cel-3 by quantitative 19F NMR according to the procedures described by some of the authors earlier (Li et al. 2020; Liu et al. 2017). Comparing the integrals of the 19F signal of the fluorinated linker derivative at -127 ppm and the signal of the barium fluoride reference at -14 ppm (see ESI for detailed calculation) leads to a loading of 0.5 mmol/g of the fluorine labeled linker head with respect to the weight of the material or a degree of functionalization with respect to the primary hydroxy groups of the cellulose of 9.7% (see ESI for detailed calculation). This value is higher than the DF of primary hydroxy groups obtained after immobilization of Fmoc-TTDS and deprotection (6.8%, see above). It can therefore be assumed, that the linker head not only attaches to the linker via amide bond formation, but also to the primary hydroxy groups of the cellulose itself forming ester-bonds. In the final step, the Rh2(S-DOSP)4 is coordinated to the mod-cel-3 material yielding the heterogeneous cel-cat catalyst. The amount of rhodium in this sample is determined using TGA analysis. The burning of the sample in an oxygen rich atmosphere at a maximum of 950 °C leaves a residue of 6.6 wt% of rhodium(III)oxide denoting a rhodium content of 5.4 wt%. with respect to the weight of the heterogeneous catalyst. Additionally, ICP-OES analysis has been applied to the sample to determine the rhodium content. Here, a value of 4.4 wt.-% rhodium has been determined as the rhodium content of the immobilized catalyst which is close to the value obtained by TGA taking the error of the TGA analysis into account. With a loading of 0.5 mmol/g of the coordinating linker head in the material, the degree of complexation can be denoted to 72% (see ESI for detailed calculation).

Catalytic activity tests

The catalytic activity and selectivity of the heterogeneous cel-cat are tested in the cyclopropanation reaction (Scheme 4) of styrene with methyl(E)-2-diazo-4-phenylbut-3-enoate (4) in pentane as solvent. The identity of the products of the cyclopropanation is confirmed by mass spectroscopy. To evaluate the quality of the catalytic activity and selectivity of the cel-cat catalyst, it is compared to the analogous homogeneous Rh2(S-DOSP)4 and Rh2(R-DOSP)4 catalysts for which catalytic tests were performed at − 78 °C to obtain good enantioselectivity. Since the activity of the heterogeneous cel-cat is assumed to be lower compared to the homogenous ones, the catalytic tests on cel-cat are performed at different temperatures up to 35 °C where still no significant loss of enantioselectivity for cel-cat is expected.

Scheme 4
scheme 4

Cyclopropanation reaction of styrene with methyl(E)-2-diazo-4-phenylbut-3-enoate (4) to test the catalytic activity of cel-cat

Full size image

The identification of the products is performed by comparing the HPLC chromatograms of the known products of the cyclopropanation using the homogeneous Rh2(S-DOSP)4 and Rh2(R-DOSP)4 with the chromatograms of the products using the heterogeneous cel-cat catalyst. Since the enantiomeric excess of d1 produced by the homogeneous Rh2(S-DOSP)4 is known, and additionally, the ratio of d1 to ent-d1 is reciprocal in the reaction of Rh2(R-DOSP)4, the peaks of d1 and ent-d1 can be clearly assigned in the HPLC chromatogram (see ESI Fig. S9 for a sample chromatogram, using a chiral HPLC column). Analogous to the reaction using the homogeneous Rh2(S-DOSP)4 catalyst, the predominant enantiomer that is formed using the heterogeneous cel-cat is the (S,S)-product d1. The enantiomeric excess will be referred to this (S,S)-product d1. The diastereomers d2 and ent-d2 are not found in any of the product mixtures.

As depicted in Fig. 10, an almost constant enantiomeric excess of 85% at 35 °C during the course of the reaction is achieved for the heterogeneous cel-cat compared to 98% at − 78 °C reported for the homogeneous Rh2(S-DOSP)4 catalyst (Davies et al. 1996). This clearly shows the excellent catalytic performance at higher temperatures up to 35 °C of cel-cat with respect to enantioselectivity.

Fig. 10
figure 10

Evolution of the enantiomeric excess (red/circles) in relation to the turnover of the substrate (grey/squares) in the model cyclopropanation at 35 °C using the heterogeneous cel-cat catalyst

Full size image

Even larger enantiomeric excesses can be achieved for cel-cat at lower temperatures but with a remarkable decrease in catalytic activity. In Fig. 11, the enantiomeric excess at 24 h of reaction time (red squares) is depicted with the associated turnover at the same time. At a temperature of − 78 °C, the catalytic activity of cel-cat is too low to produce enough product in a reasonable time to calculate a reliable enantiomeric excess. In the temperature range between 0 and 35 °C the turnover after 24 h increases from 18 to 92% while the enantiomeric excess decreases by less than 10% from 92% at 0 °C to 85% at 35 °C. These results clearly demonstrate the excellent enantioselectivity in a broad temperature range where an optimum of catalytic activity and selectivity is expected to be reached at a temperature between 10 and 35 °C.

Fig. 11
figure 11

Enantiomeric excess (red squares) and turnover (blue columns) at different temperatures after a reaction time of 24 h using cel-cat in the cyclopropanation reaction of styrene and diazo compound 4

Full size image

Recyclability and leaching

Finally, the recyclability of the cel-cat catalyst as well as its leaching properties are inspected. For the recyclability test, the catalyst is used 4 consecutive times at a temperature of 35 °C, employing recovery, washing, and drying of the catalyst after each cycle.

As shown in Fig. 12, the activity of the catalyst decreases after the first two cycles, and then reaches an almost constant activity in cycle 3 and 4. The selectivity of the catalyst on the other hand, while remaining almost constant during the cycles, rises from cycle 1 to 4. This observation partially coincides with the observation made earlier in the examination of the selectivity and activity at different temperatures. A lower activity of the catalyst also leads to a higher selectivity in these experiments. However, the difference in activity between cycle 3 and 4 is not as prominent as the difference in selectivity, which let us assume that there are more factors involved which influence the behavior of the catalyst. One possible factor is the catalysts mass concentration in the reaction mixture. Due to losses during the recovery process, the mass concentration of the catalyst decreases from the first to the last cycle. This obviously influences the activity but may also have an influence on the selectivity. Further studies to optimize the reaction conditions would be necessary to provide more information on this topic, which is beyond the scope of this work. To analyze the leaching properties, after the first cycle, the supernatant reaction mixture has been analyzed via ICP-OES to reveal the amount of rhodium in the solution used for the catalytic reaction. The ICP-OES (see supporting information Table S2) shows only negligible amounts of rhodium in the reaction mixtures, suggesting a solid bond between the catalyst and the support. These results together with the results from the recyclability test clearly demonstrate the preservation of the heterogeneous catalysts in the cyclopropanation reaction. However, further detailed characterization of the catalyst after each individual cycle in the recycling test has to be performed to investigate changes in the catalyst structure which is beyond the scope of the present work.

Fig. 12
figure 12

Comparison of the turnover and ee in the recovery cycles. It has to be noted that the catalyst used in this investigation was obtained from an earlier, non-optimized batch leading to the lower selectivity values

Full size image

Conclusion

A novel cellulose supported heterogeneous catalyst based on the homogeneous Rh2(S-DOSP)4 catalyst has been successfully synthesized and applied in the cyclopropanation reaction of styrene with methyl(E)-2-diazo-4-phenylbut-3-enoate (4). The synthesis of the enantioselective catalyst has been monitored by DNP enhanced 1H → 13C CP MAS NMR, TGA and ICP-OES showing stable coupling of the linker, the coordinating ligand, and the catalyst to cellulose. With UV/Vis, quantitative 19F MAS NMR as well as TGA and ICP-OES, the degree of functionalization and the amount of rhodium have been determined.

The enantiomeric excess produced in the model cyclopropanation reaction has been proven to be excellent for the heterogeneous catalyst with 85% of the (S,S)-product d1 at 35 °C. At lower temperatures, even higher enantiomeric excesses of up to 92% have been obtained, however also a significant loss of activity at 0 °C. The recycling experiments showed only minor loss in activity during 4 cycles. Leaching of rhodium species in the catalytic reaction is negligible as stated by ICP-OES results, demonstrating the stability of the heterogeneous catalyst.

The newly formed cel-cat catalyst with its versatile linker, opens an easy and sustainable pathway to immobilize different types of dirhodium catalysts to conduct various enantioselective reactions such as cyclopropanations. The length of the linker as well as the coordinating linker head can be easily varied in the synthesis, leaving space for research on chiral catalyst immobilization on biopolymers.