Dennis
Kühbeck‡
a,
Judith
Mayr‡
a,
Marleen
Häring‡
a,
Martin
Hofmann
a,
Françoise
Quignard
b and
David
Díaz Díaz
*ac
aInstitut für Organische Chemie, Universität Regensburg, Universitätsstr. 31, 93053 Regensburg, Germany. E-mail: david.diaz@chemie.uni-regensburg.de
bInstitut Charles Gerhardt Montpellier-UMR 5253 CNRS-UMII-ENSCM-UMI, Matériaux Avancés pour la Catalyse et la Santé, 8 rue de l'École Normale, 34296 Montpellier Cedex 5, France
cIQAC-CSIC, Jordi Girona 18-26, 08034 Barcelona, Spain
First published on 15th January 2015
Understanding the ability of biopolymers to promote selective C–C bond formation could provide important insights into the molecular mechanisms underlying evolution, and could help in the design of safer and “greener” catalysts in the future. This work describes the preparation of metal ion-crosslinked alginate materials and the evaluation of their potential catalytic activity towards the Henry (nitroaldol) reaction between a variety of aldehydes and nitroalkanes. The results demonstrated broad substrate specificity and a remarkable influence of the solvent nature, metal ion, and physical state of the alginate-based catalysts. In this sense, Ca2+–alginate hydrogel beads in DMSO used as a heterogeneous and recyclable catalyst were found to be the most efficient system, although the reaction could also proceed in an aqueous medium. Moreover, both small and large-scale model reactions between 4-nitrobenzaldehyde and nitromethane were performed successfully at room temperature. The possibility of coupling the reaction with an enzymatic resolution of the so-obtained β-nitroalcohol products was also assessed. All materials were characterized using different techniques including, among others, FT-IR, TGA, ICP-OES, and FE-SEM.
Among the most abundant polysaccharides, alginates are interesting anionic polymers isolated from brown algae that have been widely applied in drug delivery,8 entrapment of bioactive molecules,8,9 and as supports in biocatalysis.10–14 These biopolymers are linear, unbranched block copolymers of (1 → 4)-linked β-D-mannuronate (M) and α-L-guluronate (G) monomers that are organized in blocks (i.e., containing (M)m, (G)n, and (M,G)x sequences) along the polymer chain (Fig. 1). One of the biggest advantages of alginates is that their solutions (monovalent metal ions form soluble salts) can readily form strong hydrogels with water content >95% upon metal exchange with many divalent or multivalent cations, especially Ca2+ (a borderline Lewis acid).15 This property is at the core of their natural function as water-storage agents in living organisms, as well as their main applications as immobilization supports and food thickening agents.16 The 3D network hydrophilic structure of these gels has been described by the so-called “egg-box” model where each calcium atom is coordinated to the carboxylates and hydroxyl groups of four G monomers from two adjacent chains of the polymers (Fig. 1).17,18 The properties of alginate gels are highly influenced by the ratio and sequencing of uronic monomers,19 concentration of metal ions, and maturation time.20 Importantly, the mechanical strength of the hydrogels can be tuned using monomer ratios, allowing easy handling in different shapes (e.g., beads, tubes, membranes).
The overall potential of metal–alginate complexes in the broad field of non-enzymatic heterogeneous catalysis has only been scarcely explored during the last decade.21 In 2005, Quignard's lab reported the effectiveness of the Pd(TPPTS)3 complex immobilized on alginate aerogel microspheres as a supported catalyst for the allylic substitution of methyl–allylcarbonate with morpholine.22 A few years later, Reddy and his group described the immobilization of Cu2+ ions on alginate gel beads, and their activity towards both the regioselective 1,3-cycloaddition of alkynes and azides, and the oxidative coupling of 2-naphthols and phenols in aqueous medium.23 In 2011, Guibal and co-workers described the use of Pd supported on alginate/ionic liquid porous monoliths for the hydrogenation of 4-nitroaniline in the presence of either sodium formate or formic acid as a hydrogen donor.24 At the same time, Fe3+–alginate gel beads were successfully used for the oxidative degradation of azo dyes under visible light irradiation.25 In 2012, Oudeyer, Dez and co-workers published the catalytic activity of metal or ammonium alginates in the 1,2-addition of silyl nucleophiles to carbonyl compounds.26 Quignard's group also reported the synthesis of bimetallic alginate aerogels, obtained upon partial exchange of metal cations with Pd2+, followed by their reduction into Pd-nanoparticles, and their excellent catalytic activity towards the Suzuki–Miyaura cross-coupling reaction.27 Shi and co-workers also reported the catalytic activity of binary Cu–Pd–alginate dry beads for hydroxylation of phenol.28 Finally, Ag-nanoparticles immobilized on alginate hydrogels have shown good performance in 4-nitrophenol reduction.29–31
The objective of the present study was to perform a critical assessment of the potential ability of crosslinked monometal–alginates complexes to mediate the Henry (nitroaldol) reaction, one of the most important methods for the formation of C–C bonds.32
The FT-IR spectra of the AHG beads showed the expected absorption bands at 1597 cm−1 and 1411 cm−1 (asymmetric and symmetric stretching bands of the carboxylate functional groups, respectively), as well as the typical bands for polysaccharides between 1176 and 1028 cm−1 (C–O, C–C, and C–O–H stretching absorptions). In general, the presence of the divalent/trivalent metal ions in the AHG beads had no major impact on the main vibration modes of the starting sodium alginate. The broad bands observed between 3379 and 3076 cm−1 correspond to the stretching vibrations of the –OH functional groups (Fig. S4–S6, ESI†).33
TGA thermograms of the alginate-based xerogel materials showed a typical gradual weight loss below 100 °C which was attributed to the vaporization of the adsorbed moisture and lactonization.34 A major and rapid weight loss starting around 200 °C can be associated with polymer decomposition via decarboxylation, whereas further weight loss above 280 °C is usually ascribed to the degradation of the alginate backbone and cleavage of numerous hydroxyl groups (Fig. S7, ESI†).35 In agreement with previous studies, a trend of stability in the interval between 200 °C and 300 °C could be observed as a function of the metal ion (i.e., stability increases in the order Fe3+ (lowest stability) < Cu2+ < Na+ < Ni2+ ≈ Co2+ ≈ Zn2+ ≈ Ca2+).36
As part of the characterization of materials, SEM images of the xerogels showed porous surfaces with rough-veined structures of high aspect ratios, in contrast to the amorphous structure of commercially powdered sodium alginate (Fig. S10, ESI†). Although the method used to prepare the xerogels (i.e., air-drying, freeze-drying) had a clear impact on the observed morphologies,21 the changes were consistent in all samples (Fig. 3). This heterogeneous surface structure of the beads could favor the adsorption of small molecules present in the medium via electrostatic, hydrogen bonding, and/or van der Waals interactions. In addition, the pore size distribution of these materials are affected as a function of the metal ion.36 However, the catalytic activity of these materials with different metal ions is likely related to the exact structure of the coordination complexes, which is out of the scope of this manuscript and will constitute the focus of future studies.
Taking everything into consideration, as well as the fact that calcium is one of the cheapest commercially available metals, the fifth most-abundant element in earth's crust, and non-toxic,46,47 we decided to investigate the ability of Ca2+-containing alginate biohydrogels to promote the Henry reaction. In preliminary experiments using the model reaction between 4-nitrobenzaldehyde (1a) and nitromethane (2a), we were delighted to observe the gradual formation of the desired β-nitroalcohol 3a in the presence of Ca2+–AHG beads under mild conditions (DMSO, room temperature). Initial optimization studies revealed that 10 equivalents of 2a and one Ca2+–AHG bead were necessary to drive the reaction to completion within 24 h (Table S1, ESI†). It is worth mentioning that the use of 6 mol% of CaCl2 instead of Ca2+–AHG did not promote the Henry reaction (i.e., 0% yield of the desired product was observed after 24 h). Although the Henry reaction is simple to carry out, the elucidation of the underlying mechanisms under different conditions is always interesting. In the case of Ca2+–AHG a plausible mechanism would involve the carboxylate groups as moderately basic charged ligands and the metallic centers as Lewis acids that activate the nitroalkane towards deprotonation, similarly to the mode of action of copper acetate–bis(oxazolines).48,49 Carboxylic acid moieties and/or nitromethane molecules (used in large excess with respect to the aldehyde) could participate in the protonation step of the alcoholate intermediate (Fig. S9, ESI†). In such dual-activation mechanisms, the embedment of the metal center in the biopolymer matrix could cause a shielding effect preventing the assembly of a highly ordered transition state,49 which could explain the observed lack of enantioselectivity.
Table 1 outlines the results of solvent screening for the Ca2+–AHG-promoted Henry reaction. The best results were achieved in DMSO (>99% conversion, 88% yield) (entry 1), where the control experiment in the absence of Ca2+–AHG showed no conversion. On the other hand, only 16% yield could be achieved in H2O (entry 3), although the addition of a phase transfer additive such as TBAB increased the yield by almost three times (entry 4). Nevertheless, in sharp contrast to DMSO, the background reaction (control in the absence of the catalyst) in the H2O–TBAB system was ca. 6%. Low yields were observed in EtOH (entry 5), whereas only traces or no product could be detected when the reaction was carried out in CH3CN, THF, CH2Cl2 or toluene (entries 6 and 7). Interestingly, the presence of one of these solvents in the reaction was tolerated well if combined with DMSO in a specific ratio (i.e., DMSO/toluene 3:1 (v/v), Table S7, ESI†) (entry 2). Unfortunately, unambiguous correlations between solvent properties and reaction kinetics could not be drawn at this point (Table S10, ESI†), suggesting the existence of a complex relationship between these parameters.
Entry | Solvent | Yield 3ab (%) |
---|---|---|
a Reaction conditions: 4-nitrobenzaldehyde (1a, 0.1 mmol), nitromethane (2a, 1.0 mmol), Ca2+–AHG bead (1 bead, ca. 5 mol% of metal with respect to 1a), solvent (0.5 mL), room temperature, 24 h. b Determined using 1H NMR analysis of the crude product (9.2 μL of DMA was used as internal standard). The reported values correspond to the average values of at least two independent experiments (STDV = ±2). c Control experiment made in the absence of Ca2+–AHG bead. d DMSO/toluene ratio = 3:1 (v/v). e Tetra-n-butylammonium bromide (0.04 mmol). | ||
1 | DMSO | 88 (0c) |
2 | DMSO–toluened | 83 |
3 | H2O | 16 (<2c) |
4 | H2O–TBABe | 44 (6c) |
5 | EtOH | 9 |
6 | CH3CN, THF | <2 |
7 | CH2Cl2, toluene | 0 |
In terms of kinetics, the rate of the Ca2+–AHG bead-catalyzed Henry reaction (Fig. 4) was found to be slower than standard base catalysts such as Et3N, but was comparable to other metal-catalyzed reactions (Table S9, ESI†). At this point, we hypothesized that low-molecular-weight calcium alkyl carboxylates may also mediate the formation of β-nitroalcohols in good yields under the studied conditions. Indeed, when the model reaction between 1a and 2a in DMSO was carried out in the presence of soluble calcium 2-ethylhexanoate (ca. 2.1 mol%), a well-established catalyst for homogeneous industrial polymerizations, the desired β-nitroalcohol 3a was formed in 90% yield within 3 h (Table S3, ESI†). Despite the higher catalytic activity of calcium 2-ethylhexanoate in comparison to Ca2+–AHG beads, heterogeneity and recyclability are sacrificed in the former case. In addition, and in agreement with the corresponding MSDS, both calcium 2-ethylhexanoate and its dissociation product (i.e., 2-ethylhexanoic acid) are known to be irritants to the eyes, respiratory system, and skin.
Fig. 4 Kinetics analysis of the Ca2+–AHG-promoted Henry reaction between 4-nitrobenzaldehyde (1a) and nitromethane (2a) as described in Table 1. Each data point represents the average of at least two independent measurements. C∞ = final concentration at infinite time; Ct = concentration at given time t; C0 = initial concentration at t = zero time. Apparent rate constant, k = 0.113 ± 0.04 h−1. Inset: characteristic signal of the benzylic proton in the 1H NMR spectrum of the β-nitroalcohol product. |
Equally importantly from a practical perspective, the scale-up of the target model reaction from 0.1 mmol to 7.5 mmol (1.13 g) of 1a also lead satisfactorily to the formation of 3a with an isolated yield of 82%. With these results in hand, we chose the model reaction between 1a and 2a in DMSO for further investigations.
Considering the differences in the Lewis acid character of each metal, as well as the fact that both morphology and porosity of the hydrogel networks are different depending on the metal center,50 we expected to observe a major effect of the metal nature on the kinetics of the Mn+–AHG bead-catalyzed Henry reaction.2,3Table 2 summarizes the results of this investigation using ca. 4 ± 1 mol% of the metal catalyst with respect to the aldehyde (ESI†). The best result was achieved with Ca2+–AHG beads (88% yield, Table 2, entry 1), although modest yields between 50 and 60% could also be obtained using Co2+–AHG (entry 3) and Ni2+–AHG beads (entry 4). Zn2+–AHG beads afforded the corresponding β-nitroalcohol 3a, albeit in lower yields (entry 5), whereas the use of Cu2+–AHG and Fe3+–AHG provided the desired product in very poor yields (entries 2 and 6 respectively). It is noteworthy that all hydrogel beads, except Fe3+–AHG, remained physically stable under the reaction conditions without visual deterioration. It should be noted, however, that further detailed studies using electron microscopy are necessary in order to precisely evaluate any morphological change of the hydrogels caused by their immersion into different solvents. Although we have not carried out experiments in order to improve the lower yields, it is expected that extension of the reaction time and/or slight increase of the temperature could increase them.7
Entry | Catalyst | Yield 3ab (%) |
---|---|---|
a Reaction conditions: 4-nitrobenzaldehyde (1a, 0.1 mmol), nitromethane (2a, 1.0 mmol), Mn+–AHG (1 bead, 4 ± 1 mol% of metal with respect to 1a), DSMO (0.5 mL), room temperature, 24 h. b Determined using 1H NMR analysis of the crude product (9.2 μL of DMA was used as the internal standard). Reported values correspond to the average values of at least two independent experiments (STDV = ±2). c Fe3+–AHG bead was not stable and underwent severe fragmentation during the reaction. | ||
1 | Ca2+–AHG | 88 |
2 | Cu2+–AHG | 4 |
3 | Co2+–AHG | 50 |
4 | Ni2+–AHG | 59 |
5 | Zn2+–AHG | 36 |
6 | Fe3+–AHG | 5c |
On the other hand, negligible enantioselectivities (ee ≤ 5%) were detected in all the experiments (ESI†), indicating that the reaction is not sensitive under the reported conditions to the primary chirality of the polysaccharide backbone or to a possible secondary chirality due to the formation of a helical conformation upon metal coordination. In general, the lack of enantioselectivity has been also reported for other catalysts and biocatalysts bearing intrinsic chirality.4,5,51–53 Moreover, we found that the use of DMSO as a solvent could also frustrate, at least to a certain extent, the formation of highly ordered transition states necessary to induce stereoselectivity in the Henry reaction. We corroborated this statement in our run of the model reaction in the presence of 2,2-bis((4S)-(−)-4-isopropyloxazoline)propane as the chiral ligand in combination with Cu(OAc)2·H2O (the ligand was synthesized as previously reported and displayed the expected spectroscopic data48). In agreement with previous observations, the bisoxazoline complex showed a significant enantioselectivity (ca. 65% ee) in MeOH,49 whereas only the racemic product could be isolated in DMSO.
During our previous research regarding the intrinsic catalytic activity of biopolymers and proteins, we have established that the morphology and/or physical state of the biocatalyst plays an important role in the kinetics of aldol-like reactions under heterogeneous or semi-heterogeneous conditions.2,3 Therefore, we decided to carry out here a comparative evaluation of the model reaction between 1a and 2a using different forms of the catalyst in DMSO at room temperature (Table 3). In sharp contrast to the hydrogel form of the catalyst (entry 1), the xerogels obtained either by freeze-drying (entry 2) or air-drying (entry 3) the corresponding Ca2+–AHG beads afforded the desired product in very poor yields (<5%) (Tables S5 and S6, ESI†). Considering the measured water content of the hydrogel beads (i.e., 434 μL per bead), we carried out another control experiment using the freeze-dried form of the catalyst in combination with the amount of water held in the hydrogel form (entry 4). However, only a slight improvement could be achieved (i.e., 14% yield of the desired product). This slight positive effect could be also observed upon using commercial available sodium alginate powder (Na+-AP) as the catalyst. Specifically, when the model reaction was run using 0.4 mg of Na+-AP (ca. 2.1 mol%)54 in DMSO, the desired product was formed in only 6% yield. However, the addition of 434 μL of H2O to the reaction containing Na+-AP increased the product yield up to 13%. Herein, the control reaction performed in DMSO + H2O (434 μL) in the absence of Na+-AP provided null conversion. Considering the amorphous microstructure of Na+-AP (Fig. S10, ESI†), the modest positive effect caused by the addition of H2O could also be attributed to its potential role as a plastifier of polysaccharides, increasing the accessibility to amorphous domains.22 Moreover, the direct use of commercial amorphous calcium alginate powder resulted in very poor yields (entry 5).
Entry | Catalyst | Yield 3ab (%) |
---|---|---|
a Reaction conditions: 4-nitrobenzaldehyde (1a, 0.1 mmol), nitromethane (2a, 1.0 mmol), catalyst (1 bead, ca. 5 mol% of metal with respect to 1a), DMSO (0.5 mL), room temperature, 24 h. b Determined using 1H NMR analysis of the crude product (9.2 μL of DMA was used as the internal standard). Reported values correspond to the average values of at least two independent experiments (STDV = ±2). c Freeze-dried form of the alginate hydrogel bead. d Air-dried form of the alginate hydrogel bead. The average volume of the obtained xerogel beads was ca. 3% of the average volume of the hydrogel beads. e Freeze-dried form of the alginate hydrogel bead and 434 μL of H2O. f Commercial powder calcium alginate (1 mg). g Alginate aerogel bead prepared from the corresponding hydrogel bead as described in the Experimental section. | ||
1 | Ca2+–AHG | 88 |
2 | Ca2+–AHFDc | 4 |
3 | Ca2+–AHADd | 1 |
4 | Ca2+–AHFGe | 14 |
5 | Ca2+–APf | 4 |
6 | Ca2+–AAGg | 89 |
In general, the afore-described results suggest that the characteristics of the hydrogel also play an important role in the Henry reaction under the described conditions. In general, catalysts with sponge-like and rough-veined microstructures showed higher activities than those with closed and agglomerated surfaces where the molecular adsorption is disfavored (Fig. 3 and Fig. S9, ESI†). Last but not least, the use of the mesoporous aerogel form of the catalyst (Table 3, entry 6), prepared by supercritical CO2 drying (see Experimental section), provided only slightly superior yields and kinetics (Fig. S3 and Table S8, ESI†) in comparison to the Ca2+–AHG beads of lower surface area, albeit without significant enantioselectivity (ee ≈ 6%). Overall, these results suggest the existence of a complex reaction mechanism involving both diffusion-controlled processes and surface reactivity. Evidence of surface reactivity instead of nanoreactor55–58 activity was also obtained from further experiments where the reactive contact surface area of the hydrogel beads was modified (Fig. S8, ESI†).
At this point we explored the substrate scope of the Ca2+–AHG-promoted Henry reaction in DMSO at room temperature (Table 4). In general, aromatic aldehydes with moderate or strong electron-withdrawing groups were converted selectively into the corresponding β-nitroalcohols in very good yields (entries 1–6), and without a dramatic influence of the position of the substituents in 1 (entries 1–3). Aromatic aldehydes with weak electron-withdrawing groups could be converted into the desired products, albeit with lower yields (entries 8–10). However, both benzaldehyde and 4-methylbenzaldehyde were poorly converted (entries 7 and 11). Remarkably, heteroaromatic systems such as 2-pyridinecarboxaldehyde also led to 50% yield (entry 12). In contrast, aliphatic aldehydes could not be converted into the desired product. On the other hand, similar results were obtained when using either nitromethane (2, pKa = 10.2)59 or nitroethane (4, pKa = 8.6)59 as nucleophiles (entries 2, 6, 10). Therefore, neither the size nor the pKa of the carbanion plays a major role in determining the outcome of the reaction. Although the diastereoselectivity (anti/syn ratio) observed in the Ca2+–AHG-catalyzed reactions with 4 was minor, an interesting trend towards the syn isomer was noticed, which was in contrast to the outcome of the control experiment where the anti isomer was favored.
Entry | R1CHO | R2 | Yield 3b (%) | drc |
---|---|---|---|---|
a Reaction conditions: aldehyde (1, 0.1 mmol), nitroalkane (2, 1.0 mmol), DMSO (0.5 mL), Ca2+–AHG (1 bead, ca. 5 mol% of metal with respect to aldehyde), room temperature, 24 h. b Determined using 1H NMR analysis of the crude product. The reported values correspond to the average values of at least two independent experiments (STDV = ±3). c Diastereomeric ratio (anti/syn) determined by 1H NMR analysis. NA = Not applicable. d Control experiment performed in the absence of Ca2+–AHG. | ||||
1 | H | 88 (0d) | NA | |
2 | CH3 | 91 (6d) | 1:1.3 (1.4:1d) | |
3 | H | 86 | NA | |
4 | H | 83 | NA | |
5 | H | 91 | NA | |
6 | CH3 | 91 | 1:1.3 | |
7 | H | 7 | NA | |
8 | H | 32 | NA | |
9 | H | 18 | NA | |
10 | CH3 | 40 | 1:1.2 | |
11 | H | 3 | NA | |
12 | H | 50 | NA |
Regarding recyclability, Ca2+–AHG beads could be reused at least three times without significant loss of activity. However, a gradual deactivation was detected after the 3rd cycle (Fig. 5A). In agreement with previous reports, blocking of the catalyst surface by aldehyde molecules,26 poisoning of the polysaccharide backbone,27 or textural instability22 under the reaction conditions could be responsible for the observed activity loss.
Additional control experiments highlighted the heterogeneous nature of the process catalyzed by the Ca2+–biohydrogels. For example, one Ca2+–AHG bead was matured for 24 h in DMSO without substrates, filtered off, and reused in a new mixture including the substrates of the model reaction. In this case the yields were almost identical to those obtained using freshly prepared Ca2+–AHG beads. A negligible catalytic effect caused by possible leaching of metal ions was demonstrated by nearly null catalytic activity (≤2% yield) of (1) the maturing solution obtained from the previous experiment or after a catalytic cycle, and (2) a stock solution containing the maximum theoretically possible leaching for a given bead (this solution was used as a potential catalyst instead of the corresponding Mn+–AHG bead – i.e., 7.14 mM CoCl2·6H2O solution for M = Co – Table S3, ESI†). As expected, stagnation of the reaction after removal of the catalyst was observed (Fig. 5B).
In order to access optical pure β-nitroalcohols a combination of the Ca2+–alginate catalysis with a lipase PS-catalyzed acyl kinetic resolution (KR) could be carried out.60 Thus, the synthesized racemic β-nitroalcohols could be subjected to a KR in the presence of vinyl acetate (VA) as an acyl donor and immobilized Amano lipase PS in toluene at room temperature (Table 5). In reasonable agreement with previous studies,61 selected β-nitroalcohols were isolated in good yields (ca. 30–40%) and ee values (>95%) after the KR. A rigorous “one-pot” process was not feasible because (1) the solvents required for achieving good yields in each step were incompatible (e.g., the lipase-catalyzed KR was not effective in DMSO, whereas the Ca2+–alginate catalysis did not proceed in toluene, moreover, attempts to use DMSO–toluene mixtures led to unsatisfactory results), and (2) a control experiment carried out by incubation of alginate beads, nitromethane and VA showed the consumption of the acyl donor. Nevertheless, we found that VA and lipase PS-IM could be added to the reaction flask after the completion of the first step. Further shaking (not stirring!) at room temperature for 18 h allowed for the KR step to take place in the same flask (e.g., (S)-2-nitro-1-(4-nitrophenyl)ethan-1-ol could be obtained in 40% using this procedure).
Entry | Ar | Yield 5b (%) |
---|---|---|
a Reaction conditions: β-nitroalcohol (3, 0.1 mmol), vinyl acetate (46 μL, 0.5 mmol), lipase PS-IM (50 mg), toluene (0.5 mL), room temperature, 18 h. b Determined using NMR analysis. HPLC optical purity > 95%. | ||
1 | 4-(NO2)–C6H4 | 31 |
2 | 4-(Cl)–C6H4 | 31 |
3 | C6H5 | 36 |
4 | 4-(CH3)–C6H4 | 38 |
5 | 4-(OCH3)–C6H4 | 41 |
Footnotes |
† Electronic supplementary information (ESI) available: Preliminary optimization experiments, metal loading, leaching study, kinetics, FT-IR spectra, TGA curves, 1H NMR spectra, HPLC chromatograms, and additional experiments. See DOI: 10.1039/c4nj02178a |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2015 |