Zhangjun
Huang
ab,
Jorge G.
Uranga
a,
Shiliu
Zhou
b,
Haiyan
Jia
ab,
Zhaofu
Fei
*a,
Yefeng
Wang
a,
Felix D.
Bobbink
a,
Qinghua
Lu
*b and
Paul J.
Dyson
*a
aInstitut des Sciences et Ingénierie Chimiques, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. E-mail: zhaofu.fei@epfl.ch; paul.dyson@epfl.ch
bSchool of Chemistry and Chemical Engineering, The State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai, P. R. China. E-mail: qhlu@sjtu.edu.cn
First published on 17th October 2018
We show that ionic liquids (ILs) interact with electron-rich, porous polyphosphazene (PPZ), to form hybrid PPZ-IL nanoreactors able to simultaneously capture and transform CO2 into carbonates. The PPZ nanospheres swell in organic solvents and effectively absorb IL cations by virtue of the electron-rich sites, while leaving the anions exposed and increasing their nucleophilicity. This leads to considerably higher catalytic activity compared to the IL alone in the cycloaddition reaction of CO2 to epoxides. The cation shielding effect is dependent on the structure of the IL cation and, hence, the catalytic activity can be tuned by varying the structure of the cation in the IL and DFT calculations were used to rationalize the experimentally observed differences in catalytic activity. These studies indicate that PPZ nanospheres could find widespread uses in catalysis, acting as active nanosupports for homogeneous catalysts, not only for the transformations of CO2, but also for other substrates.
Porous polymers have recently been evaluated in CO2 separation and conversion processes. For example, a flexible porous polymer exhibits gate opening-type abrupt adsorption for C2H2, but not for CO2, leading to an appreciable separation for CO2 from CO2/C2H2 mixtures at near ambient temperature (273 K).11 A porous polymer containing an embedded ionic-polymer was used to catalyze the cycloaddition reaction of CO2 to epoxides under ambient temperature,12 although Cu(OAc)2 was required as a co-catalyst to achieve high efficiency under ambient conditions. Cross-linked ionic polymers based on poly(styrene) and encompassing imidazolium cations catalyze this reaction in the absence of a co-catalyst and under mild conditions.13
Porous polyphosphazenes (PPZs) are prepared from the polycondensation of compounds containing o-dihydroxybenzene/o-phenylenediamine groups with hexachlorocyclophosphazene (HCCP).14 PPZs possess unique frameworks with each aromatic plane linked in a perpendicular fashion to the plane of the cyclophosphazene ring, leading to large gaps between the layers of the polymer and preventing the formation of π–π stacking or Lewis acid–base interactions, which leads to flexible amorphous structures that self-assemble to form spheroids.
Herein, we describe the preparation of a new electron-rich PPZ material that forms nanospheres which exhibit solvent-dependent size and porosity. The PPZ nanospheres readily absorb a range of ionic liquid (IL) cations, leaving the anions largely exposed. The resulting hybrid nanospheres catalyze the reaction of CO2 and epoxides to form cyclic carbonates considerably more effectively than the pure IL. This reaction has been extensively investigated and is even conducted on an industrial scale.15
The 13C solid state NMR spectrum of the PPZ nanospheres (Fig. 1a) shows signals centered at 9, 20, 40, 112, 140, 219 ppm, all of which are consistent with the presence of the ATC building block, providing further evidence of the incorporation of ATC into the PPZ nanospheres. The 31P solid state NMR spectrum shows one broad singlet at −1 ppm characteristic of P(V) in a single environment (Fig. 1b).14b
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) shows that the PPZ material comprises nanospheres (Fig. 1c and d) that are free-standing and well dispersed. Statistical analysis indicates that the PPZ nanospheres have a mean diameter of 121 nm and a reasonably narrow size distribution (Fig. S4†). The expected lack of long-range order is apparent from the powder X-ray diffraction (PXRD) analysis (Fig. S5†), which shows three amorphous broad peaks around 16, 39 and 41°. The amorphous character of PPZ material is caused by the interspersal of domains with eclipsed ordering and domains with staggered ordering,17 and the peak around 16° reveals some ordering in the range 4.9 to 6.5 Å, presumably corresponding to shoulder-to-shoulder packing (4.9 Å, staggered ordering, Fig. S6a†) and head-to-head packing (6.5 Å, eclipsed ordering, Fig. S6b†) of the layers. Considering the PXRD signal is relatively weak, most of the layers should pack obliquely, which leads to the macroscopic spherical shape.
The porosity of the PPZ nanospheres was established using Brunauer–Emmett–Teller (BET) surface area analysis. The N2 adsorption–desorption isotherms of the PPZ nanospheres (Fig. 1e) provide a surface area value of 128.8 m2 g−1 at 77 K, with a hierarchical pore size distribution of 7, 15 and 25 Å for half pore widths (inset of Fig. 1e). Smaller pores, i.e. those with a half pore width < 10 Å, are expected to correspond to the holes created by the stacking of the frames,18 and should open if the PPZ nanospheres swell. The pores with half pore widths of 15 and 25 Å presumably correspond to the intrinsic size of the frameworks and the gap between the random packed frameworks, respectively. CO2 adsorption–desorption isotherms of the PPZ nanospheres conducted at 273 K (Fig. S7a†) reveal that CO2 uptake reaches a value of 1.5 mmol g−1 at 1 atm. A broad adsorption/desorption hysteresis loop indicates the presence of intra-pore CO2–PPZ interactions, presumably due to polarization of CO2 by the electron-rich material leading to dipole–dipole interactions that inhibits the release of CO2 at low pressures. Notably, such dipole–dipole interactions are unfavorable with CH4 which is much less polarizable (Fig. S7b†).19
The ability of the PPZ nanospheres to capture CO2, observed previously with a related material,14a should be advantageous in reactions employing CO2 as a substrate. Since the PPZ nanospheres are electron-rich, they can potentially interact strongly with the cations of ILs, although the dense nature of the PPZ nanospheres and relatively low porosity could reduce their accessibility. However, when the PPZ nanospheres are dispersed in solvents, the polymeric layers separate and the nanospheres swell. AFM images of the PPZ nanospheres show that the overall size of the nanospheres swells from ca. 120 nm (Fig. 2a and c) in dry state to ca. 236 nm (Fig. 2b and d) in styrene oxide (SO). Further swelling of the PPZ nanospheres is accompanied by an increase in porosity, allowing interactions with solvent molecules and solvates. Thus, the PPZ nanospheres and ILs combine in situ to afford PPZ-IL nanoreactors.
![]() | ||
Fig. 2 AFM images of the PPZ nanospheres (a) in a dry state and (b) after swelling in SO. (c) and (d) are single particle images corresponding to (a) and (b), respectively. The size of the nanospheres was determined at 25 °C at a concentration of 0.1 mg mL−1. (e) Data for the average size and standard derivation of the PPZ nanospheres when dry (determined from SEM) and when swelled in solvents (determined from DLS, Fig. S8†). |
The cycloaddition of CO2 to epoxides was selected as an ideal test reaction20 as the PPZ nanospheres interact with CO2, and appear to polarize it, and therefore the absorbed CO2 could be somewhat activated. Moreover, the cycloaddition of CO2 to epoxides is of industrial importance,21 and ILs are good catalysts for this reaction,13a with highest activities obtained for catalysts with highly nucleophilic dynamic light scattering (DLS) shows that the extent of swelling of the PPZ nanospheres is solvent dependent (Fig. 2e). Swelling of the PPZ nanospheres in styrene oxide (SO) doubled their size, which was accompanied by an increase in porosity and catalytic activity. Due to the electron rich nature of the PPZ nanospheres, IL cations should interact strongly with them, increasing the nucleophilicity of the anions, and potentially enhancing catalytic activity. In this respect, the rate-determining step of the reaction catalyzed by imidazolium salts involves ring-opening of the epoxide by the anion.22 In this respect, the rate-determining step of the reaction catalyzed by allowing interactions with solvent molecules and solvates. Thus, the PPZ nanospheres and ILs selected as solvent and substrate for cycloaddition with CO2. Although many catalytic systems have been established for this highly atom-economic reaction,15 certain amorphous polymers have advantages including excellent stability, high efficiency under mild conditions, and they are readily recyclable and reusable.23 A series of ILs (Fig. 3a) with different structures were combined with the PPZ nanospheres and a schematic of a PPZ-IL interaction is shown in Fig. 3b. The PPZ nanospheres do not catalyze the cycloaddition of CO2 to SO in the absence of IL (Table 1, entry 1). In contrast, ILs that dissolve in SO under the given conditions, i.e. 57 °C and 1 atm CO2, catalyze the reaction to afford styrene carbonate (SC) in moderate yields, with the activity being comparable to structurally related ILs.24 The PPZ-IL nanoreactors, which form in situ, catalyze the reaction considerably more efficiently, which is as expected due to the enhanced nucleophilicity of the IL anion. Notably, the magnitude of the enhancement in activity of the PPZ-IL nanoreactors relative to the pure IL strongly depends on the structure of the cation, with the catalytic efficiency of less sterically encumbered ILs increasing by over 100% (Table 1, entries 2/3, 5/6 and 11/12), and over 80% (Table 1, entries 7/8, 9/10 and 13/14). In contrast, those with bulky substituents, e.g. PhIm, increase by only ca. <10% (Table 1, entries 15/16) and a decrease in activity is observed for the extremely bulky cation in 4PhIm (Table 1, entries 19/20). These differences in the magnitude of the reaction enhancement may be attributed to the strength of the interactions between the PPZ nanospheres and the IL, i.e. the least bulky IL cations interact strongly with the PPZ leaving the IL anion more exposed and reactive. The largest increase in activity was observed for the non-hindered bis-imidazolium salt, Bu2Im, with a –(CH2)4– linker connecting the rings, i.e. PPZ-Bu2Im led to near quantitative yields (Table 1, entry 22) whereas Bu2Im alone resulted in a yield of 14% (Table 1, entry 21). This represents an increase in activity of 7-fold. It has previously been shown that hydroxyl groups can enhance the catalytic efficiency of the cycloaddition reaction of CO2 to epoxides,25 however, in the PPZ nanoparticles the majority of hydroxyl groups in the ATC starting material are consumed, as demonstrated by the IR spectrum of the product (Fig. S3†), Moreover, the ATC starting material combined with Bu2Im is not a particularly active catalyst for the reaction (Table 1, entry 24), implying that the cation shielding effect provided by the electron-rich PPZ material is responsible for the enhancements in catalytic activity. While the ATC:
HCCP substrate ratios of PPZ was 3
:
1 (PPZ3) and 1
:
1 (PPZ1), the conversion efficiency of corresponding nanoreactors decreased (entries 26/27), due to the interfere of defects to the interaction between Bu2Im and PPZ.
![]() | ||
Fig. 3 (a) Structures of the ILs used to prepare PPZ-IL nanoreactors. (b) The proposed representation of interactions of the IL in the PPZ-Bu2Im system. |
Entry | Catalyst | Yield (%) | Entry | Catalyst | Yield (%) |
---|---|---|---|---|---|
a Reaction conditions: SO (480 mg, 4.00 mmol), CO2 (1 atm, using a balloon), 57 °C and 20 h. ILs (2.5 mol% halide), PPZ (19.2 mg, containing 2.5 mol% N). Yield determined by 1H NMR spectroscopy. The selectivity of the reaction is all >99% (determined by GC-MS). Entry 22 is shown as an example in Fig. S9 ESI.
b Using PPZ-TBAB in acetone.
c Using PPZ-Bu2Im in ethanol.
d PPZ3-Bu2Im and PPZ1-Bu2Im.
e Were synthesized from with ATC![]() ![]() ![]() ![]() ![]() ![]() |
|||||
1 | PPZ | 0 | 15 | PhIm | 58 |
2 | TBAB | 44 | 16 | PPZ-PhIm | 63 |
3 | PPZ-TBAB | 89 | 17 | StFImB | 13 |
4 | PPZ-TBABb | 89 | 18 | PPZ-StFImB | 34 |
5 | TBAI | 37 | 19 | 4PhIm | 48 |
6 | PPZ-TBAI | 88 | 20 | PPZ-4PhIm | 45 |
7 | EMImC | 66 | 21 | Bu2Im | 14 |
8 | PPZ-EMImC | 81 | 22 | PPZ-Bu2Im | 99 |
9 | EMIm | 73 | 23 | PPZ-Bu2Imc | 98 |
10 | PPZ-EMIm | 84 | 24 | ATC + Bu2Im | 32 |
11 | EMMIm | 41 | 25 | HCCP + Bu2Im | 14 |
12 | PPZ-EMMIm | 89 | 26 | PPZ3-Bu2Imd | 63 |
13 | C18Im | 67 | 27 | PPZ1-Bu2Ime | 73 |
14 | PPZ-C18Im | 88 |
In order to better understand the differences in catalytic activity, DFT calculations were performed on three ILs as representative examples, including the systems with the best and worst enhancements in activity when combined in PPZ-IL nanoreactors, i.e. PPZ-Bu2Im, PPZ-EMIm and PPZ-4PhIm.
First, the three ILs, Bu2Im, EMIm and 4PhIm, were independently optimized, and the most stable conformation showed hydrogen bonds (HBs) as the main interaction. Only in the case of EMIm the conformation with “on top” electrostatic I−–imidazolium+ interaction was also important, but for comparative purpose our analysis only considered the HB conformation. For Bu2Im and EMIm, the iodide interacts with the C2–H bond from the imidazolium ring, whereas, 4PhIm exhibits a HB network between the chloride anion and the four aromatic C–H bonds at the ortho position with an average HB distance of 2.44 Å. Optimized geometries of the ILs are presented in Fig. 4a, with HBs shown in yellow and the optimized structures of the adducts, i.e. PPZ-Bu2Im, PPZ-EMIm and PPZ-4PhIm are illustrated in Fig. 4c, with HBs shown in green and yellow. The PPZ moiety is represented by means of electrostatic potential map (see Fig. 4b) and, as expected, the phosphazene and dihydroxyphenyl rings are more electron rich than other regions. Hence, the phosphazene ring interacts preferentially with the IL cations.
Once the PPZ-Bu2Im adduct is formed, the stronger HB interactions are between the imidazolium cation and the phosphazene ring. Moreover, the orientation of imidazolium ring is not parallel to any of the π systems in the PPZ structure, i.e. π–π stacking interactions are not observed (see adducts A and B, Fig. 4c). Instead, HBs (marked in green in Fig. 4c) appear to be responsible of adduct formation. In contrast, the smaller IL (EMIm) interacts forms π–π stacking interactions between the imidazolium and PZ ring (see adduct D) or phenyl ring in the polymer (see adduct E) with an average distance between the rings of 3.50 Å for adduct D, and 3.55 Å for adduct E. In the case of the more sterically hindered 4PhIm system, the main interactions between PPZ and IL cation comprise HBs with the aromatic and benzylic C–H bonds (see green lines in adduct C). This leads to a more compact adduct in which the 4PhIm occupies most of the free space inside the PPZ framework. Note that due to the high volume of 4PhIm only one stable conformation was found.
A binding energy (BE) of 34.7 kcal mol−1 was calculated for the most stable conformer of PPZ-Bu2Im in the gas phase (adduct A in Table 2). Adduct B, is slightly less stable than adduct A (binding energy = 33.7 kcal mol−1). Lower binding energies were computed for PPZ-EMIm, with values of 21.1 kcal mol−1 for adduct D and 25.1 kcal mol−1 for adduct E. Intermediate stability is predicted for PPZ-4PhIm.
Structure | Binding energya | rCsp2H⋯Xb/Δr rCsp2H⋯Xc | νCsp2–He/ΔνCsp2–Hf |
---|---|---|---|
a Calculated binding energies in kcal mol−1. b H⋯Y distances in Å. c Computed changes in CH⋯X distance in Å. d Average (Cortho)H⋯X distance in Å. e Calculated C2–H or Cortho–H stretching frequencies in cm−1. f Computed changes in above mentioned frequencies. g Asymmetric Cortho–H stretching. h Symmetric Cortho–H stretching. (ν1) vibrational mode for the imidazolium ring 1 and (ν2) vibrational mode for ring 2. | |||
Isolated Bu2Im | (I1![]() |
3036 | |
PPZ-Bu2Im (A) | 34.7 | (I1) 2.73; (I2) 2.71, Δr1 = 0.22; Δr2 = 0.20 | (ν1) 3222; (ν2) 3249, Δν1 = 186; Δν2 = 213 |
PPZ-Bu2Im (B) | 33.7 | (I1) 2.64; (I2) 2.73, Δr1 = 0.13; Δr2 = 0.28 | (ν1) 3213; (ν2) 3195, Δν1 = 178; Δν2 = 159 |
Isolated 4PhIm | 2.44d | 3164g, 3173h | |
PPZ-4PhIm (C) | 25.8 | 2.46d, Δr = 0.02 | 3163g, 3169h, Δνg = −1; Δνh = −4 |
Isolated EMIm | 2.46 | 2987 | |
PPZ-EMIm (D) | 21.1 | 2.58, Δr = 0.12 | 3150, Δν = 163 |
PPZ-EMIm (E) | 25.1 | 2.56, ΔI = 0.10 | 3110, Δν = 123 |
HBs in ILs are evidenced by changes in the Csp2–H stretching frequencies in their IR spectra.26 Notably, the Csp2–H stretching frequency moves to lower energy as the strength of the HB increases and the distance between the CH group and the halide shortens.27 Taking these properties into account, our analysis was carried out considering the geometric and vibrational changes on the IL structure before and after adduct formation, as a descriptor of HB strength between the IL fragments. The frequency shift was calculated according to the formula Δν = νCHPPZ-IL − νCHIL, where νCHPPZ-IL represents the stretching frequency of the Csp2–H bonds in the adduct, and νCHIL the corresponding stretching frequency in the isolated ionic liquid. Accordingly, the analyzed frequency shifts were derived from the calculated νCHIL for the isolated ILs as reference (see Table 2), i.e. 3036 cm−1 for Bu2Im, 2987 cm−1 for EMIm, and 3166 and 3173 cm−1 for the asymmetric and symmetric stretching modes in 4PhIm. In addition, the changes in CH⋯X distances were estimated from the initial values of isolated ILs, i.e. 2.51 Å for PPZ-Bu2Im, 2.46 Å for PPZ-EMIm and 2.44 Å for 4PhIm (see Table 2).
In the case of adduct A, which involves a double imidazolium iodide structure, there are two different C(2)–H⋯I1 and C(2)–H⋯I2 moieties in which a reduction of the C–H stretching frequencies from the initial value was calculated, 186 cm−1 for C(2)–H⋯I1 and 213 cm−1 for C(2)–H⋯I2 (Table 2, adduct A). In parallel the C(2)–H⋯I1 HB is elongated by 0.22 Å and the C(2)–H⋯I2 HB is elongated by 0.20 Å relative to the initial value of 2.51 Å (Table 2, Bu2Im). In adduct B, frequency shifts relative to those in the isolated IL of 159 cm−1 and 177 cm−1 for C(2)–H⋯I1 and C(2)–H⋯I2 are calculated, respectively, with corresponding H⋯I elongations of 0.13 and 0.18 Å. In the case of adduct D, frequency shift of 163 cm−1 and H⋯I distance increase of 0.12 Å were obtained for the corresponding C(2)–H⋯I moiety. Similarly with adduct E Δν and Δr were 123 cm−1 and 0.10 Å respectively, upon adduct formation. These changes indicate that formation of the PPZ-EMIm adduct has a weaker effect on the ion–pair association compared to the PPZ-Bu2Im system (adducts A and B). Thus, HB strength calculations may be used to rationalize the changes in reactivity of systems before and after adduct formation with the PPZ material, in that the anion interacts less strongly with the cation in the PPZ-Bu2Im than in the PPZ-EMIm systems, and is therefore more nucleophilic and hence more reactive.
For PPZ-4PhIm, no important geometric and vibrational changes in IL structure were observed before and after adduct formation. For example, the average C–H⋯Cl distances (H-bonding network) in the adduct PPZ-4PhIm are comparable with those observed in isolated 4PhIm. The same was calculated for the Cortho–H bonds for which stretching frequency were reduced by only 1 and 4 cm−1. This suggests that the interaction between IL ion pairs is still significant and the chloride anion is less nucleophilic. Indeed, the chloride anion shown to be encapsulated within the HB network of the cation and in part by the surrounding PPZ material. As a result, it is less accessible and thus less reactive.
According to the computational results, the structure of the cation in ILs is strongly influenced by the interaction with the PPZ support and, consequently, determines the degree of activation of the anion which could be estimated from the strength of the H-bonds.
The kinetic profiles for the PPZ-Bu2Im nanoreactors and the Bu2Im IL are compared in Fig. S10,† showing that the former is considerably more active. These differences between the PPZ-Bu2Im nanoreactors and the other PPZ-IL nanoreactors may be attributed to the superior interaction of the Bu2Im cation with the PPZ nanospheres (see computational results above and Fig. 4c). The PPZ-IL nanoreactors form in various solvents such as acetone or ethanol and exhibit similar catalytic activities to those formed in situ in SO (Table 1, cf. entries 4 and 23 with entries 3 and 22). The PPZ-Bu2Im nanoreactors were recycled and reused 5 times with the yield of SC remaining above 95% (Fig. S11†). The slight decrease in activity may be attributed to ca. 2% loss of the Bu2Im IL after each reaction. No changes to the catalyst were detected by solid-state 31P NMR spectroscopy (Fig. S12†). The SEM image of the PPZ-Bu2Im nanoreactors after catalysis (Fig. 5a) shows them to be coated by a smooth 14 nm thick layer (Fig. 5b), corresponding to the precipitated Bu2Im IL in ethyl acetate (poor solvent). After removing the excess Bu2Im coating by washing with anhydrous ethanol and ethyl acetate, the PPZ nanomaterial presented a bigger diameter of ca. 150 nm (Fig. 5c) than the that of the freshly prepared PPZ (ca. 121 nm), due to the residual Bu2Im inside (elemental mapping of EDS), Fig. 5d, e and S13,† which has a detection depth of c.a. 100 nm.28 The amount of the residual IL was estimated to be 24% from thermogravimetric analysis (Fig. S14†).
The scope of the PPZ-Bu2Im nanoreactors in the cycloaddition of CO2 to various epoxides was studied at atmospheric pressure (Table 3), with the corresponding carbonates obtained in excellent yields.
Entry | Epoxide | Product | Time (h) | Yield (%) |
---|---|---|---|---|
a Reaction conditions: epoxide (4.00 mmol), CO2 (entry 1, 10 bar CO2 in an autoclave; others, 1 atm using a CO2-filled balloon), 57 °C. Bu2Im (23.7 mg, 1.25 mol%), PPZ (19.2 mg, containing 2.5 mol% N). Yield was determined by 1H NMR spectroscopy. The selectivity of the products is all above 99% by GC-MS. | ||||
1a |
![]() |
![]() |
3 | 99 |
2 |
![]() |
![]() |
3 | 99 |
3 |
![]() |
![]() |
4 | 99 |
4 |
![]() |
![]() |
20 | 99 |
5 |
![]() |
![]() |
4 | 99 |
Typical procedure for reactions under high CO2 pressures: to a 100 mL stainless steel autoclave equipped with a glass vial and a magnetic stirrer, propylene oxide (232 mg, 4 mmol), Bu2Im (23.7 mg, 1.25 mol%) and PPZ (19.2 mg, containing 2.5 mol% N) were added. The autoclave was sealed and purged three times with CO2 and then set to 10 bar. The autoclave was heated in a 57 °C oil bath. After reaction, the autoclave was cooled in an ice bath and the yield of the product was determined by 1H NMR spectroscopy.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ta08856j |
This journal is © The Royal Society of Chemistry 2018 |