Iman Noshadi‡
abc,
Baishali Kanjilal‡d,
Tahereh Jafarid,
Ehsan Moharrerid,
Nasser Khakpashe,
Ting Jianga and
Steven L. Suib*adef
aDepartment of Chemicals, Materials & Biolmolecular Engineering, University of Connecticut, 91 North Eagleville Road, Storrs, Connecticut 06269-3222, USA. E-mail: steven.suib@uconn.edu
bHarvard-MIT Health Science and Technology, Massachusetts Institute of Technology, 77 Massachussets Avenue, Cambridge, 02139, USA
cWyss Institute for Biologically Inspired Engineering, Harvard University, MA 02115, USA
dInstitute of Materials Science, University of Connecticut, 91 North Eagleville Road, Storrs, Connecticut 06269-3222, USA
eDepartment of Materials Science and Engineering, University of Connecticut, 97 North Eagleville Road, Storrs, Connecticut 06269-5233, USA
fDepartment of Chemistry, University of Connecticut, 55 North Eagleville Road, Storrs, Connecticut 06269-3060, USA
First published on 9th August 2016
This paper presents a new class of octamethylcyclotetrasiloxane (D4 siloxane) adsorbent based on the copolymer of divinylbenzene and a novel methacrylate monomer. The novel cyclic amine based methacrylate monomer was synthesized employing click chemistry and was polymerized to form a mesoporous adsorbent under solvothermal conditions and tested for siloxane removal. The D4 adsorption capacity of the novel adsorbent is 2220 mg g−1, which is greater than the adsorption capacities of mesoporous poly(divinylbenzene) and commercial activated charcoal. The adsorbent retains 47% regeneration capacity after 10 usage cycles. The high specific adsorption is due to a combination of physisorption, caused by the mesoporosity and pore volume and chemisorption, as evidenced by spectroscopic results. The incorporation of functional groups into a mesoporous structure with significantly enhanced specific adsorption offers future opportunities towards tailored polymer properties for efficient industrial applications in siloxane removal.
Several techniques have been studied for siloxane removal. They include absorption, bio-filtration, adsorption, peroxidation, cryogenic separation, membrane separations and catalytic processing.7–16 The limiting factors for siloxane capture by adsorption include adsorption capacity, regeneration, and reuse. Inorganic adsorbents such as silica, alumina, aluminosilicate and activated charcoal (AC)11,13,15,16 have been studied, in addition to polymeric adsorbents such as Tenax and Amberlite.13,14 However, they all exhibit low siloxane adsorption capacities (100–400 mg g−1) and poor regeneration. While activated charcoals have shown promise in terms of high siloxane adsorption capacities of (800–1700 mg g−1) and high surface areas, their reusability is challenging due to strong binding between adsorbates and charcoal adsorbent.1,9 A comparison of some recently reported adsorbents with their maximum capacities has been provided in our earlier work.3 In addition to the strong interaction between siloxanes and activated charcoal, polymerization of siloxanes on the adsorbent surface and its co-adsorption with water further compound complications in regeneration and reduce performance.17
Porous polymers have been successfully applied for adsorption of organics such as dyes and oils.18 The main advantage that a polymeric structure offers is the ability to tailor absorptive properties and hydrophilic–lyophilic balance.19–21 This edge needs to be exploited to develop highly adsorptive material in order to solve current problems in biogas purification.
In our previous study3 the selective capture of siloxanes, with an emphasis on octamethylcyclotetrasiloxane removal from biogas, was presented. The adsorbents used were mesoporous co-polymeric structures based on divinylbenzene (DVB) and 1-vinylimidazole, synthesized under solvothermal conditions. The adsorbent, while being hydrophobic, exhibited abundant mesopores, good air-stability, super wettability for siloxane removal, and hence very high adsorption capacities.
In this study, we further develop the performance of the mesoporous hydrophobic structure of the adsorbent by the use of a novel methacrylate monomer and its copolymer with DVB, synthesized under solvothermal conditions. This paper describes the synthetic route to obtaining the novel monomer based on allylcyclohexylamine, its characterization, copolymerization with DVB, its application in selective siloxane removal and the elucidation of a siloxane retention mechanism based on complementary physicochemical characterization. As reported in our previous study,3 CO2 and moisture were seen to influence the adsorption activity toward siloxanes. Hence future studies shall be conducted to evaluate the effect of the presence of impurities. The adsorbent reported in this work has the largest reported capacity per unit surface area to date for siloxane removal.3 The high adsorption capacity coupled with easy tailor-ability of properties opens up novel vistas for high performance materials for designing industrially scalable processes.
The vaporization temperatures of the thiolene click adduct and the subsequent ACAM monomer were also estimated using thermogravimetry analysis (TGA) as shown in Fig. 2(c). Both the thiolene click adduct and the ACAM monomer were viscous amber colored fluids with high boiling points. In an alcohol (thiolene adduct) some molecular association can be attributed to the presence of hydrogen bonding. Corresponding esters are usually lower boiling compared to alcohols. In this case, the loss of the hydrogen bonding due to conversion of the primary hydroxyl in the click adduct was offset by an increase of the molecular weight of the moiety leading to an increase in the boiling point by nearly 35 °C.
The nuclear magnetic resonance (NMR) spectra for monomer synthesis are shown in the Fig. S1.† The allyl protons labeled 1, 1′ and 2 in the ACA NMR appear at roughly δ of 5 and 6 ppm as seen in Fig. S1(a).† In the NMR of the thiolene click adduct there is a significant reduction of the allyl proton signal due to the reaction between ACA and mercaptoethanol (Fig. S1(b)†). The reason the residual allyl proton signal is seen in the thiolene click adduct is due to the slight stoichiometric excess of ACA used. The hydroxyl proton is labeled ‘α’ in the thiolene click adduct NMR (Fig. S1(b)†) and appears at δ ∼ 3.7 ppm. The thiolene click adduct was esterified to the ACAM monomer and its 1H NMR is shown in Fig. S1(c).† A reduction of the hydroxyl proton signal at δ ∼ 3.7 ppm is observed. The NMR of the ACAM monomer also shows the protons of the methacrylate double bond, labeled β,β′ with signals at δ ∼ 5.5 to 6.5 ppm. The OC–O–CH2– protons of the methacrylate monomer are labeled as γ and they appear at δ ∼ 4.2 ppm.
The thermal stability of the P(DVB-ACAM) copolymer was also evaluated by thermogravimetry analysis which is shown vis-à-vis the thiolene click intermediate and the purified ACAM monomer (Fig. 2(c)). Thermal degradation of the P(DVB-ACAM) copolymer commences at temperatures greater than 400 °C underscoring the thermal stability of the copolymer in its native form. High thermal stability is a requirement for the intended application as an adsorbent which needs to see temperatures up to 100–120 °C for recovery purposes.
The P(DVB-ACAM) copolymer was evaluated for its wetting characteristics and compared to a pure PDVB film and the results are shown in Fig. 2(d). While a water contact angle of 156° was observed with PDVB, a slight reduction in hydrophobicity with the incorporation of the comonomer in P(DVB-ACAM) was observed with the water contact angle of 150°. However, a high water contact angle still underscores the significantly hydrophobic characteristics of P(DVB-ACAM). The scanning electron microscopy (SEM) image of P(DVB-ACAM) confirms the porous structure of the polymer (Fig. 2(e) and (f)).
X-ray photoelectron spectroscopy (XPS) was used to identify chemical composition of P(DVB-ACAM) material. Fig. 3(a)–(d) shows the XPS spectra of P(DVB-ACAM) with signals associated with carbon, nitrogen, sulfur, and oxygen indicating that methacrylate species have been successfully introduced into the polymer surface. N 1s is detected at 400 eV, which is in the expected range for amine nitrogen in aromatic polymers.26–30 The S 2p peak is detected at 165 eV, which is about 2 eV above the expected sulfide peak for organic structures.31,32 The O 1s peak is detected at 533 eV where the carboxylic oxygen is expected to be.33,34 The shift towards higher binding energy suggests interactions between sulfide and carboxylic groups on methacrylate species. The carbon peak shift from 284 eV to 286 eV implies electrophilic interactions between methacrylate species and PDVB carbon.
Fig. 3 XPS analysis for P(DVB-ACAM) (a) carbon 1s (b) nitrogen 1s (c) sulfur 2p and (d) survey spectra overlay of P(DVB-ACAM) and P(DVB). |
Fig. 4(a) and the inset show the N2 adsorption–desorption isotherms and pore size distribution of P(DVB-ACAM), respectively, prior to siloxane adsorption (blue dotted line). The isotherms are observed to be type IV curves with a sharp capillary condensation step at a relative pressure of P/P0 = 0.8–1.0 which suggests abundant presence of mesopores in the polymer.35 P(DVB-ACAM) polymer synthesized showed BET surface areas of 271 m2 g−1 and bi-modal pore size distribution centered at 6 and 15 nm. This is the first such report of a mesoporous copolymer of DVB with ACAM co-monomer.
Fig. 4(a) and its insert capture the BET isotherms and pore size distribution for the adsorbent both before and after D4 capture. While the blue dotted line represents experimental data of the fresh adsorbent, the red line is representative of observations after siloxane adsorption. The surface area, pre-adsorption, computed as 271 m2 g−1 is reduced to a surface area of 7 m2 g−1 after saturation. This demonstrates filling up of all pores of the adsorbent with siloxane leading to blocking of the entire surface of the polymer. The average pore volume, after adsorption is reduced to 0.02 cc g−1 after exposure to the high concentration of siloxane for several hours. This happens, plausibly due to the presence of multiple adsorption and entrapment sites inside each pore, which fill up gradually. This is indicated by the gradual decrease in the average pore size. This continues until the pores can accommodate no more siloxane moieties and becomes unavailable for further adsorption.
Fig. 4(b) shows a typical TGA trace of D4 saturated P(DVB-ACAM). While the fresh P(DVB-ACAM) adsorbent is stable up to temperatures of 400 °C, Fig. 4(b) shows rapid weight loss (67%) of P(DVB-ACAM) after exhaustion with D4 (red line), in the temperature range up to 150 °C. The adsorbent is stable up to 150 °C and no weight loss was observed. A comparison of fresh adsorbent (blue dotted line) versus used adsorbent features that a significant amount of D4 was captured by P(DVB-ACAM) after exposure to siloxane.
Fig. 4(c) shows the FTIR overlay of fresh P(DVB-ACAM) (blue dotted line) versus that after siloxane adsorption (red line). The FTIR characterization used the reflectance mode. The IR bands at 1068 and 1265 cm−1 can be an indication of the presence of D4 and they are delineated in Fig. 4(c). The asymmetric stretching of the Si–O–Si siloxane bridge bonds of siloxane D4 appears around 1068 cm−1, while The Si–CH3 bond can be seen clearly at 1265 cm−1.
In order to obtain the adsorptive activity of P(DVB-ACAM), 5 sccm of nitrogen flow was bubbled in D4, which passed through the adsorbent bed for 2 h. The absorptive capacity of P(DVB-ACAM) is shown in Fig. 5(a) at room temperature. No exhaustion occurred during the adsorption process for P(DVB-ACAM) and commercial activated charcoal.
Fig. 5 (a) Adsorption capacity of P(DVB-ACAM) at room temperature (b) regeneration and reusability of P(DVB-ACAM) verified up to 10 cycles (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). |
Table 1 shows textural and adsorptive comparison of P(DVB-ACAM), PDVB and commercial activated charcoal of which the two latter materials were reported previously.3 The available surface area for adsorption as well as the pore volume of P(DVB-ACAM) is lower than that for PDVB and activated charcoal. The absolute siloxane adsorption of P(DVB-ACAM) is 13.8% and 159% more than PDVB and activated charcoal, where P(DVB-ACAM) has a specific siloxane uptake which is nearly 7 times higher than that adsorbed by activated charcoal and nearly 4 times higher than that adsorbed by PDVB. This suggests thermodynamic and affinity factors at play in its active adsorption of siloxane. This is the largest reported capacity to date for siloxane removal per unit surface area.3
Materials | Surface area (m2 g−1) | Pore volume (cc g−1) | Capacity (GC) (mg g−1) | Normalized capacity (mg m−2) | Capacity (TGA) (mg g−1) |
---|---|---|---|---|---|
Commercial AC | 695 | 0.52 | 856 ± 54 | 1.23 | NA |
PDVB | 831 | 1.66 | 1951 ± 74 | 2.35 | 1980 ± 25 |
P(DVB-ACAM) | 271 | 0.97 | 2220 ± 53 | 8.19 | 2125 ± 32 |
To obtain the capacity (mg D4/g adsorbent), the solid adsorbent was saturated by siloxane under continuous gas flow for several hours. The exhausted adsorbent was washed off by hexane and then the solvent was diluted and tested by GC/mass spectrometry (MS) to evaluate the amount of captured D4. Fig. S2† (along with its inset) shows the GC pattern of washed P(DVB-ACAM) and commercial activated charcoal. Fig. S2† also displays the GC pattern of D4 standard solution of 10 ppm wherein the elution time for D4 was observed at 5.78 minutes. Additionally, the capacity was also calculated based on TGA data in which all weight loss below 150 °C was attributed to D4 in the material.
Table 1 also compares the calculated capacity of P(DVB-ACAM) based on GC and TGA data. The calculated capacity based on TGA is a little lower than that calculated from the hexane extraction experiment. This might point to a better mass transfer regime in the liquid phase than heat transfer induced recovery in siloxane adsorbed on the mesoporous polymer system.
In addition to high specific adsorption properties, an important property to ascertain the viability of its usage is recyclability. The reusability of P(DVB-ACAM) was verified up to 10 cycles and the results are represented in Fig. 5(b). After each exhaustion step and extracting the adsorbed siloxane by hexane, the adsorbent was heated to 100 °C in air overnight for regeneration for further usage. The results on regeneration indicated that P(DVB-ACAM) is reusable for several cycles with the average capacity loss of 7% per cycle leading to 53% loss after 10 cycles. The polymerized D4 (as indicated in Fig. S2†) in the polymer pores leads to blockage of accessible surface area to some extent (Fig. S5† shows the N2 adsorption–desorption isotherms before and after 10 cycles regeneration). D4 possibly remains partly chemically unbound to the P(DVB-ACAM) matrix. Fig. S6† shows the characterization of P(DVB-ACAM) after 10 cycles of regeneration indicating no change in functional groups of polymer matrix. The surface area after 10 cycles of recovery drops slightly from 271 to 254 m2 g−1. The considerable loss of capacity is in contrast to the slight changes of textural and functional properties which could be due to inefficient removal of lingering adsorbate in between regeneration steps. However, the degassing stage (150 °C) prior to N2 adsorption–desorption is capable of inducing complete desorption of remained siloxanes. Besides, nitrogen is more flexible in finding accessible surface area even when partly blocked by remaining siloxanes. The D4 is adsorbed via a combination of physisorption and chemisorption into the adsorbent matrix. The specific adsorption may be on one hand due to polar, yet hydrophobic interactions of the D4 moieties with the co-monomer sites in the P(DVB-ACAM) matrix which is superior to that faced by D4 with the PDVB control. On the other hand, there could be possible chemical bond formation which augments the adsorption capacities. This indicates a great advantage of this material over commercially activated charcoals which require high recovery temperatures in the range of 250–300 °C.
With PDVB control, the structure of the polymer rules out the formation of specific functional interactions. Instead, the adsorption phenomenon can be attributed primarily to physical adsorption. This also underscores the relative ease of desorption. With the P(DVB-ACAM) structure, even with lower surface area, the adsorption properties are significantly enhanced. Thus this phenomenon may be thought of as a combination of physisorption and chemisorption entailing some irreversible chemical bond formation.
The mechanism of chemisorption in AC adsorbents is well studied.1 The spontaneity of adsorption is incumbent on the free energy for adsorption which is brought about by non-electrostatic and electrostatic contributions. Coulombic interactions predominate only where adsorbate is electrolytic and highly polar in nature. However, given the nature of the Si–O–Si bond in siloxanes, wherein the difference in electronegativity between Si and O is not high enough to warrant a large charge separation between the atoms, non-electrostatic interactions with the adsorbents in this work are considered.
For ACs, the surface chemistry depends primarily on the surface oxygen content, which determines the surface charge and hydrophobicity.36 For a high surface area carbonaceous material, in contact with an aqueous solution, surface group dissociation causes the development of surface charges, such as negative charges from acidic group based complexes, such as carboxyl and phenolic groups.36 This eventually leads to a decrease in hydrophobicity.37,38
In the mesoporous PDVB control, there is an absence of such specific functional groups. Non-electrostatic attractive, yet weak interactions such as van der Waals forces allow for the physisorption of siloxanes into the surface pores. However, given the low strength of such attractions, the extent of specific adsorption is lower than that achieved due to the presence of functional groups.
In the functionalized P(DVB-ACAM) mesoporous polymer, an increased specific adsorption is observed. As in surface oxygen functionalized ACAs, the secondary amine functionality in P(DVB-ACAM) is also seen to reduce the hydrophobic character as seen from the results of the contact angle experiments. The secondary amine from ACA functionalization is capable of hydrogen bonding. Polarization of the N–H bond causes the H to become electrophilic which then attracts the lone pair of electrons on oxygen and forms a hydrogen bond, underscored by the loss of hydrophobicity. This phenomenon is also applicable for oxygen in the Si–O–Si bond of the siloxane adsorbate. A dative covalent bond may form between the electron lone pair on oxygen and the electron deficient H on the secondary amine of the adsorbent.
Functionalized siloxane plates have been used for the separation of amines.39–41 In addition, chemisorption and physisorption of ammonia on silica surface have been studied. Griffiths et al. reported that the chemisorption of ammonia on silica surfaces via formation of NH2–O–Si bonds.39 Strained siloxanes are more reactive and facilitate the chemisorption of NH3 with resultant Si–NH2 groups on silica surface.40 Morrow et al. showed that for unsymmetrical siloxane bridges where one Si bond is electron deficient, an initial fast reaction is followed by a slow reaction involving highly strained sites allowing for low temperature ammonia adsorption.41 In addition, the existence of water molecules in siloxanes here should also be taken into account since they may result in rupture of less-reactive siloxane bridges and lead to formation of SiOH groups, which are capable of hydrogen bonding with the secondary amine of the adsorbent.
However, in this work evidence of the formation of Si–N bonds is seen in the Fourier transform infrared spectroscopy (FTIR)42,43 (Fig. 4(c) and ESI Fig. S3,† inset b) after adsorption at a wavenumber of 872 cm−1. This indicates a probable reaction of the siloxane adsorbate with the N–H bond of the ACA as per the proposed schematic represented in Fig. 6.
Fig. 6 Reaction schematic of the D4 siloxane adsorbate with the secondary amine bond of the ACA as a part of the mesoporous adsorbent structure. |
There exists a small percentage of N–H bonds in the adsorbent matrix for the reaction. A single D4 siloxane molecule may be thought to consume a single N–H bond via reaction. The rest of its molecular structure retains the Si–O–Si bonds dominating the FTIR signature. Additionally, the typical N–H stretch band between 3000 and 3500 cm−1 (ref. 42 and 43) in the adsorbent, prior to adsorption, is seen to decrease after adsorption as seen in the inset (ESI Fig. S3,† inset d). This supports the depletion in the concentration of the N–H bonds in the adsorbent as they are being used up by the siloxanes to form the Si–N bonds. There is no signature of the Si–H bond in the 2000–2500 cm−1 range given that the adsorbates are dimethyl cyclic siloxane structures.42,43 The Si–CH3 group is also visible in the adsorbent FTIR after adsorption as a strong sharp band at around 1260 cm−1 (ref. 42 and 43) (ESI Fig. S3† inset c). Siloxanes also show strong IR signals in the range of 1000–1100 cm−1 for Si–O–Si.42,43 This comes up as a single band for disiloxanes or smaller rings. With branching or with larger rings, the band becomes broader and this is shown in the expanded inset (c) of ESI Fig. S3.†
A small signature of D5 and D6 was seen in the gas chromatograph of washed and diluted commercial activated charcoal and P(DVB-ACAM) after they had adsorbed D4 siloxane (ESI Fig. S2†). The peak of D4 siloxane is predominant (at 5.78 minutes). This signifies very little ring opening polymerization which may have contributed additionally to the siloxane retention mechanism in the pores. Polymerization, however, is a known phenomenon for mesoporous adsorbents such as silica gel or activated charcoal (AC) which contain reactive groups such as OH, the equivalent of which is the –NH group in P(DVB-ACAM). These groups, in the presence of moisture, form a hydrogen bond with the Si–O–Si skeletal framework of the siloxane. In the presence of moisture and a slightly acidic pH, the cleavage of Si–O–Si bond may be fostered followed by the condensation of Si–OH bonds which leads to polymerized structure formation.44 The formation of D5 and D6, albeit in small amounts, for both AC and P(DVB-ACAM) may be explained by multipoint fragmentation of Si–O–Si bonds in the siloxane, followed by condensation of the formed Si–OH ends. This is schematically represented in Fig. 7.44 However, given the small intensity of peaks pertinent to higher homologues, compared to the dominant D4 peak, the phenomenon of polymerization may be trivial in comparison to chemisorption and bond formation.
Fig. 7 Schematic of a possible route of D4 siloxane bond cleavage followed by condensation via Si–OH bonds forming higher homologues. |
Additionally, the XPS data on the adsorbent after adsorption (ESI Fig. S4†) show a small Si peak. Since XPS occurs in high vacuum, it should lead to desorption of most of the D4, that is physisorbed or bonded via hydrogen bonding or polar interactions. At the same time, the observation that some Si signal is retained supports the formation of some amount of strong chemical bonds.
The primary constituents of biogas are methane, carbon dioxide, water, and different organic volatile compounds. The influence of humidity on the adsorption capacity of siloxanes on various porous adsorbent structures such as zeolites, granular activated carbon, and activated carbon cloth has been studied in detail.45 In such studies, the effect of humidity was found to be greater with granular activated carbon than other carbon structures. The difference was attributed to more open microporosity in granular activated carbon structure.
In the current study, the disparity in results under high humidity and low humidity is expected to be low. The adsorbent in this study has been shown to be hydrophobic, as seen in the water contact angle experiments (Fig. 2(d)), despite the incorporation of the secondary amine based functionality. The micropores and mesopores are expected to repel water from their inner surface hence leading to preferential adsorption of siloxanes rather than water vapor. This phenomenon is also observed with zeolites in earlier studies, where the hydrophobic nature of zeolites mitigated the siloxane adsorption capacity loss.
In addition to water vapor, interference can come from the primary components of biogas – methane (CH4) and carbon dioxide (CO2). These are the major components and their presence in large amounts can drive preferential adsorption due to high concentrations as compared to siloxanes. In an earlier study1 with zeolites and carbon structures, reductions in adsorption capacity of up to 20 mg g−1 have been reported, and these have been attributed to competitive adsorption between methane and siloxanes.
In this study, the adsorption characteristics are expected to be different primarily because of the mechanisms elucidated for adsorption in earlier sections. The D4 may be thought of being adsorbed via a combination of physisorption and chemisorption. As mentioned earlier, a part of the specific adsorption is believed to be due to polar, yet hydrophobic interactions. In methane (CH4), a symmetrical molecule, the overall null separation of charges means that there is a very low probability of polar interactions. The only interactions that may be thought of with the P(DVB-ACAM) are hydrophobic, nonpolar interactions, which are weaker than electrostatic interactions between polar moieties. Thus methane poses a lower competitive force compared to siloxane for physical interaction mediated adsorption. Similarly, carbon dioxide is also electrostatically neutral due to its symmetry and would behave in ways similar to methane and is expected to have inferior physical interaction mediated adsorption as compared to siloxanes.
On the other hand, there are possibilities of a chemical bond formation between the siloxane units and the functional group on the adsorbent augmenting the adsorption capacities, as delineated earlier. For CO2 there is a possibility of acid–base interactions with the basic secondary amine. Carbon dioxide is rendered acidic by absorption into water that converts to carbonic acid. In such a situation, the acid–base interactive forces are mitigated by the presence of water as a hydration sphere around the carbonate and hydrogen carbonate anions, which is expected to be repelled by the inherently hydrophobic mesopores.
D4 siloxane | Octamethylcyclotetrasiloxane |
AC | Commercial activated charcoal |
DVB | Divinylbenzene |
ACA | Allylcyclohexylamine |
ACAM | 3-((3-(Cyclohexylamino)propyl)thio)propyl methacrylate |
P(DVB-ACAM) | Copolymer of ACAM and divinylbenzene |
PDVB | Poly(divinylbenzene) |
TGA | Thermogravimetry |
XPS | X-ray photoelectron spectroscopy |
NMR | Nuclear magnetic resonance |
SEM | Scanning electron microscopy |
GC | Gas chromatography |
FTIR | Fourier transform infra red spectroscopy |
MS | Mass spectrometry |
EToAC | Ethyl acetate |
THF | Tetrahydrofuran |
TEA | Trimethylamine |
AIBN | 2,2′-Azobis(2-methylpropionitrile) |
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11382f |
‡ These authors contributed equally. |
This journal is © The Royal Society of Chemistry 2016 |