Reactive Epoxy Nanofiltration Membranes with Disulfide Bonds for the Separation of Multicomponent Chemical Mixtures.

This article reports the fabrication of organic solvent nanofiltration membranes containing a labile disulfide bond, which is broken by reaction with a chemical stimulus. These membranes are a new generation of smart membranes that have tailored selectivities and flux that can be altered by reacting with a chemical stimulus. The selectivity and flux of chemicals through the membranes was controlled by varying the concentration of disulfide bonds in the membrane. When the disulfide bonds were cleaved, the pores in the membrane became larger and yielded different separation properties. The membrane selectivity was changed by up to 70% and flux was increased up to 5×. The rapid change in selectivity of the membrane allowed for the separation of three-component mixtures. A three-component mixture of 33.3% m-dinitrobenzene, 33.3% triphenylmethane, and 33.3% 1,3,5-tris(diphenylamino)benzene (TDAB) was separated into three different fractions that were significantly enriched in one of the three molecules. The first fraction contained m-dinitrobenzene at 82% purity and 84% yield, the second fraction contained triphenylmethane at 67% purity and 49% yield, and the third fraction contained TDAB at 71% purity and 88% yield.

Introduction

The chemical industry is critically important worldwide with chemical sales in the United States alone reaching $800 billion in 2015.[1] In order to produce high-quality chemicals and products, chemicals are purified using many different methods including distillation, chromatography, crystallization, liquid–liquid extraction, sublimation, and much more.[2] Chemical separations account for as much as 15% of the total global energy consumption.[3−5] In part due to these costs, energy efficiency is one of the 12 Principles of Green Chemistry that are widely recognized as important principles to guide the future of chemistry in academic and industrial settings.[6−11] Membrane-based separations are attractive alternatives to distillations and other energy intensive separation methods and can be applied to chemicals that do not form crystals, do not sublime, or are challenging to purify with other methods. One of the emerging fields in membrane science is organic solvent nanofiltration (OSN),[12] which is concerned with the separation of chemicals with molecular weights ranging from 100 to 1000 g mol–1 in organic solvents. Many important chemicals have molecular weights that fall in this range such as catalysts, ligands, solvents, monomers used to make plastics, and pharmaceutical drugs. The purification of these chemicals is important for companies to produce high-quality products, as impurities can significantly affect how materials function or how drugs react in the body. OSN is a promising alternative for chemical separations that can save energy and reduce greenhouse gas emissions when compared to a more conventional technique such as distillation.[13,14]

Membranes separate mixtures into a retentate and permeate; if two chemicals are present then each chemical may be isolated in high purity. If the initial mixture has three or more chemicals, then typically only one chemical will be isolated in high purity unless multiple membranes with different separation properties are used in a membrane cascade.[15] Despite the need for membranes that can separate more than two chemicals, this area of research is understudied, with only one example in the recent literature involving OSN.[16] This study used a molecularly imprinted membrane (MIM) to separate a three-component mixture of an active pharmaceutical ingredient (API), a catalyst, and the API building block. However, when the selective sites in the MIM reached saturation, the membrane lost selectivity.

In this article, we report a new method of controlling flux and selectivity in OSN membranes by fabricating new membranes which incorporate disulfide bonds into cross-linked epoxy nanofiltration membranes. Cleavage of the disulfide bond changes the separation properties of the membranes, and incorporating different concentrations of disulfide bonds in the membrane allows for control over the changes in selectivity and flux of chemicals. We also report the application of these membranes to isolate three chemicals in significantly enhanced purity from a three-component mixture.

Research on membranes that can change pore sizes has received increasing attention in recent years.[17−20] Membranes can respond to various stimuli such as pH,[21] temperature,[22] light,[23] electricity,[24] and magnetism.[25] These membranes are highly desired because they give the user control over flux and selectivity, which are two of the most important characteristics of a membrane. Both properties are controlled by membrane pore size, and in a stimulus-sensitive membrane the pore size can either decrease or increase in response to specific stimuli.

To the best of our knowledge, there has not been a report of a membrane with cleavable disulfide bonds, despite the fact that disulfides have been incorporated into other smart materials such as self-healing polymers[26,27] and reprocessable thermosets.[28] Disulfide bonds are easily cleaved through chemical or thermal processes, which makes them excellent candidates for incorporation into a membrane.

In this paper, we report the fabrication of epoxy nanofiltration membranes with cleavable disulfide bonds using a one-pot step polymerization with a diepoxide and a mixture of two dianilines (Figure a). The reaction between amines and epoxides creates a highly cross-linked polymer membrane rich in disulfide bonds, and the space between the physical and chemical cross-links forms the pores within the membrane. Chemicals diffuse into the membrane and through the space between the cross-links to permeate the membrane. A chemical stimulus is used to cleave the disulfide bonds and open up the pores in the membrane (Figure b,c). The change in flux and selectivity of chemicals through the membrane is dependent on how many disulfide bonds are cleaved during exposure to a chemical stimulus.

Figure 1

(a) Chemical reaction between two monomers to synthesize a cross-linked polymer membrane is shown. The disulfide bonds are intact when the membrane is fabricated. (b) Photograph of an epoxy membrane with cleavable disulfide bonds showing the color change after treatment with a chemical stimulus. Only the center part of the membrane was exposed to the stimulus and the yellow polymer was completely etched away leaving behind the white, inert solid support. (c) Chemical structure of the polymer membrane after it has been treated with a chemical stimulus. The disulfide bonds have been cleaved to thiols.

We also report how these membranes were used to produce three purity-enriched streams of chemicals from a three-component mixture using one membrane. After collecting the first chemical in high purity, the disulfide bonds in the membrane were cleaved to increase the pore size and the second, larger chemical permeated the membrane in high purity. The third chemical was too large to permeate the membrane, so it was rejected by the membrane and retained in high purity.

Results and Discussion

Selection of Monomers for Epoxy Nanofiltration Membranes with Disulfide Bonds and Characterization of the Membranes

Various diamines and diepoxides were investigated for their ability to fabricate epoxy membranes that were robust but permeable for selected chemicals (Figure ). All membranes were fabricated with a 1:1 mole ratio of NH bonds to epoxides, yielding a highly cross-linked polymer network. Chemicals diffuse into the membrane and through the physical and chemical cross-links which make up the pores of the membrane.

Figure 2

(a) Amines and (b) epoxides that were screened in the fabrication of epoxy nanofiltration membranes.

Attempted synthesis of diamine A, which bears a disulfide, began with transesterification of 3,3′-dithiopropionic acid to the diester, which proceeded smoothly in 91% yield. Treatment of the diester with ethylenediamine to form diamine A proved unsuccessful, as the primary amines of ethylene diamine reacted with the aliphatic disulfide. Because the aliphatic disulfide was unstable in the presence of primary amines, an aromatic disulfide monomer was investigated. Commercially available dianiline B (4-aminophenyldisulfide) was stable in the presence of a primary amine for over a month (Figure S1). Fabrication of membranes using dianiline B and diepoxide 1 yielded membranes that were brittle and showed little permeation of chemicals (<1%) through the membrane after 10 d (Table S1). However, use of epoxide 2 in the polymerization allowed chemicals to permeate the membrane (Table S2).

Incorporating a chemically responsive bond, such as a disulfide bond, in the membrane allows modification of the cross-links and pores in the membrane. If a membrane is fabricated with 100% disulfide monomer in the polymerization, the membrane will completely dissolve upon cleavage of the disulfide bond. If a membrane is fabricated with 50% of disulfide monomer and 50% of a comonomer that does not bear a disulfide, only 50% of the cross-links will break, leaving behind a cross-linked membrane with much larger pores. After cleavage of the cross-links in a membrane, the separation properties will change, and the degree of this change is controlled by varying the molar equivalents of disulfide comonomer used in the fabrication of the membranes.

To make testable variations on this membrane, it was necessary to select a diamine that had epoxide reactivity similar to that of monomer B. Dianilines D and E were mostly unreactive with epoxide 2 and were unable to form a solid membrane after 2 weeks at 60 °C. These results are likely due to the fact that both monomers possess strong deactivating groups between the aromatic rings. By contrast, dianiline C has an activating ethylene group between the two aromatic groups, and polymerization with epoxide 2 was successful, allowing chemicals to permeate the membrane (Table S3).

Diepoxide monomer 2 and diamine monomers B and C were then used in the polymerizations or copolymerizations to fabricate epoxy nanofiltration membranes. Membranes are henceforth referred to by a simplified nomenclature. For example, a membrane fabricated with diamine B and diepoxide 2 is referred to as membrane B-2. If a mixture of the diamine comonomers B and C is used, the membrane is referred to as BC-2, where the superscripts denote the molar equivalents of the comonomers. For example, membrane B1C3-2 would contain 25% of monomer B, and 75% monomer C.

The thickness of the membranes was characterized via scanning electron microscopy (SEM). A membrane was fractured to obtain a cross-sectional image that showed the epoxy membrane fabricated on top of the solid support (Figure a). SEM micrographs showed that the thicknesses of the epoxy membranes were approximately 1 μm.

Figure 3

(a) SEM micrograph of epoxy membrane B1C3-2 fabricated on top of a porous solid support. (b) FT-IR spectra of the epoxide peaks at different times during the fabrication of membrane C-2.

The polymerization reaction was monitored by Fourier-transform infrared (FT-IR) spectroscopy by observing the decrease in % transmittance of the epoxide peaks at 910 and 860 cm–1 (Figure b).[29,30] The reactions were deemed complete when there was no further change in the spectra for the peaks at 910 and 860 cm–1. The fabrication of membrane C-2 was completed within 5 d because the FT-IR spectra collected at 8 d overlapped with the 5 d spectra. The reaction to form membrane B-2 was completed within 7 d (Figure S2), and the fabrication of membranes B3C1-2, B1C1-2, and B1C3-2 was also monitored by FT-IR and they reached completion within 7 d (Figure S3–S5).

Selection of Chemical Stimulus To Break the Disulfide Bonds

It was important to select a chemical stimulus that would cleave the disulfide bonds, while retaining membrane integrity. Cleavage of the disulfide bonds would result in a large increase in flux because of the large increase in the size of the membrane pores. Cysteamine was investigated for its ability to cleave the disulfide bonds and break the cross-links in the membrane. Cysteamine was chosen because it is a small chemical that was expected to have a rapid flux through the membranes, and it is well known for its ability to cleave disulfide bonds in biology. Initial experiments with membrane B-2 and cysteamine were promising based on the color change of the membranes. The membrane was off-yellow, but after the center part of the membrane was exposed to cysteamine in dichloromethane (DCM) the yellow color disappeared and the white solid support was visible (see Figure b). This is due to cysteamine cleaving all of the disulfide bonds, which breaks the cross-links and dissolves the membrane.

The use of cysteamine to break the disulfide bonds in the membranes was investigated by several methods. First, the flux of pNB was investigated before and after treatment with cysteamine for membranes B-2 and C-2 as control experiments (Figure a,e). If all of the disulfide cross-links in membrane B-2 were cleaved, the membrane would wash away as soluble organic chemicals, and membrane C-2 should remain unchanged because it contains no disulfide bonds. The flux of pNB through membrane B-2 was 4.7× faster after treatment with cysteamine, but the flux was unchanged through membrane C-2 after treatment with cysteamine. Flux was determined following a previous procedure.[31]

Figure 4

Flux before treatment with cysteamine and after treatment with cysteamine for membranes (a) C-2, (b) B1C3-2, (c) B1C1-2, (d) B3C1-2, and (e) B-2. The chemicals were pNB, triethylamine (Et3N), tripropylamine (Pr3N), and TPM.

Next, the flux of pNB, triethylamine, tripropylamine, and triphenylmethane (TPM) through intact membranes B-2 and C-2 was investigated. The relative flux of these chemicals was measured for separations through membranes B-2 and C-2, and the results in Figure demonstrate that their rates of permeation decreased in progression from pNB, to triethylamine, to tripropylamine, and to TPM. This result is understood based on the different sizes of these chemicals as described elsewhere.[32] When the membranes were exposed to cysteamine for 18 h and the separation of the chemicals were investigated, the relative flux of the chemicals was unchanged for membrane C-2 but changed dramatically for membrane B-2. The relative flux of the chemicals through membrane B-2 after treatment with cysteamine was very similar to the relative flux of chemicals through the solid support in the absence of any epoxy membrane.

Raman spectroscopy was used to characterize the disulfide bond in the membranes before and after exposure to cysteamine by investigation of two peaks that were unique to the disulfide monomer (Figure a,b). The peak at 480 cm–1 was unique to the disulfide bond and the peak at 1600 cm–1 was only present in the spectrum for membranes fabricated with the disulfide monomer. Neither was present in the spectrum for membrane C-2 that did not contain any disulfide bonds (Figure S6).[33] The peak at 1600 cm–1 was a shoulder peak to the peak at 1620 cm–1; both peaks were due to the aromatic ring vibrations, and the peak at 1620 cm–1 was seen in the spectra for both membranes B-2 and C-2. The disappearance of the peaks at 480 and 1600 cm–1 demonstrated that the disulfide bonds were no longer present in the membrane. Figure a shows the Raman spectra for membrane B1C1-2 before treating with cysteamine, with both peaks at 480 and 1600 cm–1 showing strong intensities. Following treatment with cysteamine for 3 h, another Raman spectrum was collected and there was a significant decrease in both peaks related to the disulfide bond. The disulfide peak at 480 cm–1 has decreased by 68%, and the aromatic peak from the S–S had been reduced to a small shoulder to the left of the larger peak at 1620 cm–1. Longer reaction times did not reduce the peak at 480 cm–1. Raman spectra were collected for all membranes used in this study before and after treatment with cysteamine (Figures S6–13). After characterizing the cleavage of the disulfide bonds, the membranes were investigated for their ability to separate chemicals before and after cleavage of the disulfide bonds.

Figure 5

(a) Raman spectra of membrane B1C1-2 before treatment with cysteamine. (b) Raman spectra of membrane B1C1-2 after treatment with cysteamine. Note diminution of disulfide peaks at 480 and 1600 cm–1.

Selectivity in Epoxy Nanofiltration Membranes before and after Cleaving the Disulfide Bonds

To investigate how the cleavage of disulfide bonds affected the selectivities of membranes, they were fabricated with 0, 25, 50, 75, and 100 mol % of diamine monomer B. By varying the amount of the disulfide in the membrane, we controlled the number of cleavable cross-links in the membrane and thus controlled the change in selectivity and flux. It was hypothesized that membranes containing more disulfide bonds would show a larger increase in flux and a loss of selectivity after the disulfide bond was cleaved.

In nanofiltration, the two models most commonly used to describe the flux of chemicals through a membrane are the solution-diffusion model[34] and the pore-flow model.[35] In this report, we will use the solution-diffusion model to describe the permeation of chemicals through the membrane, both for its simplicity and because the chemicals were separated in a diffusion-based separation apparatus (Figure S14). Chemicals were dissolved in DCM and added to the retentate side of the diffusion apparatus.

The concentration of chemicals in the permeate stream was monitored using 1H NMR spectroscopy and used to determine the relative and absolute flux of the chemicals through the membrane before treatment with cysteamine (Figures and S5–S10). For the membrane series of B1C3-2, B1C1-2, and B3C1-2, the selectivities for pNB to TPM increased from 12.4:1, to 14:1, and up to 15.9:1. As the selectivity of this membrane series increased, the flux of pNB through the membrane decreased (Figure a). One possible reason for this decrease was that membranes containing more of monomer C had lower conversions of the epoxide functional group, as determined by FT-IR spectroscopy (Table S4). The density of cross-links decreased as the conversion of the reaction between epoxides and amines decreased.

Figure 6

(a) Flux of pNB through the membranes before and after treatment with cysteamine is shown. (b) Graph showing the relative increase in flux of pNB through the membrane, and the % decrease in selectivity for pNB over TPM after the disulfide bonds were cleaved. The red data points correspond to the right axis and represent the decrease in selectivity for pNB to TPM, and the green data points correspond to the left axis and represent the increase in flux for pNB through the membrane.

This decrease in density of cross-links would yield larger pores consistent with a faster flux of pNB through the membrane and a lower selectivity for pNB over TPM. After characterizing the flux and selectivities of chemicals through the membranes before exposure to a chemical stimulus, the membranes were treated with cysteamine. Using the initial data as a baseline, a clear difference in flux in pNB was observed before and after the disulfide bonds were broken in each membrane. This was consistent with the existence of larger pores after reaction with cysteamine (Figures a and S5–10). As more disulfide bonds were broken, the membranes showed a greater increase in flux for pNB (Figure a). For the membrane series B1C3-2, B1C1-2, and B3C1-2, the flux of pNB was 1.1×, 2.2×, and 5.3× faster when compared to the membranes that had the cross-links still intact (Figure b). This result demonstrates that the flux and selectivities of chemicals through the membranes were altered by adding a chemical stimulus (Table ).

Table 1

Flux of p-Nitrobenzaldehyde (pNB) through Membranes B-2 and C-2 before and after Treatment with Cysteamine

	membrane B-2		membrane C-2
	flux before treatment with cysteamine (mol/cm2 h)a	flux after treatment with cysteamine (mol/cm2 h)a	flux before treatment with cysteamine (mol/cm2 h)a	flux after treatment with cysteamine (mol/cm2 h)a
pNB	8.42 × 10–7	3.96 × 10–6	1.37 × 10–6	1.41 × 10–6

All flux values carry a ±4.24% relative error.

The selectivity of the membranes also decreased after reaction with cysteamine, which was consistent with the membranes possessing larger pores (Figure b). Figure b and Figures show that as the molar equivalents of disulfide in the membrane increased, the selectivity between pNB and TPM decreased because more disulfide bonds were broken. Membrane C-2 showed minimal changes because cysteamine does not react with this membrane, and membrane B-2 showed significant loss of selectivity (86%), and the flux of pNB was close to that of the solid support in the absence of an epoxy membrane (Table S10). This result was expected because membrane B-2 dissolved after treatment with cysteamine. For the membrane series of B1C3-2, B1C1-2, and B3C1-2, the membranes lost 29, 58, and 75% respectively of their initial selectivity (Figure b), which correlated with the molar equivalents of disulfide bonds cleaved in each membrane. Repeated experiments to reform the disulfide bond after reaction with cysteamine did not succeed, and selectivities of the original membranes could not be completely recovered after reaction with cysteamine.[36−39]

Separation of Chemicals in a Three-Component Mixture

Cleavage of the disulfide bonds yielded membranes with different selectivities, which would allow for a simple separation of a three-component mixture of chemicals A, B, and C (Figure a). The first separation would remove the smallest chemical, A, from two larger chemicals, B and C, that were retained by the membrane (Figure b). Next, the cross-links in the membrane could be cleaved to increase the pore size and change the separation properties (Figure c). This change in separation properties of the membrane would allow B to permeate the membrane while retaining C (Figure d). The result would be the separation of three chemicals using a single membrane.

Figure 7

Depiction of a three-component separation. The yellow membrane indicates a membrane that has not been treated with a chemical stimulus, and a white membrane indicates that the membrane has been treated with a chemical stimulus. (a) Initially, at t = 0, a mixture of three chemicals on the retention side of the membrane is shown. (b) Over time, the red molecule has permeated the membrane and is extracted from the permeate side of the membrane. (c) Membrane has been treated with a chemical stimulus to change the membrane selectivity and fresh solvent is added to the permeate side. (d) After some amount of time, the green molecule permeates the membrane, and the blue molecule is retained by the membrane.

A three-component mixture of equal molar equivalents of m-dinitrobenzene, TPM, and 1,3,5-tris(diphenylamino)benzene (TDAB) was added to a separation apparatus using membrane B1C1-2. m-Dinitrobenzene is the smallest chemical with a molecular weight of 168.11 g mol–1, and TPM and TDAB have molecular weights of 244.34 and 579.79 g mol–1, respectively. The first fraction of permeate was collected before treating the membrane with cysteamine, and it was found to be highly enriched with m-dinitrobenzene (Table ). The permeate contained 84% of the initial starting mass of m-dinitrobenzene, and the purity was increased from 33.3 to 82%.

Table 2

Separation of a Three-Component Mixture Using One Epoxy Membrane

	m-dinitrobenzene		TPM		TDAB
fractions	massa (mg)	mol %	massa (mg)	mol %	massa (mg)	mol %
initial	22.8	33.3	33.0	33.3	80.5	33.3
first permeate (before breaking cross-links)	19.2	81.6	4.8	14.0	3.5	4.3
second permeate (after breaking cross-links)	3.6	21.7	16.2	67.1	6.4	11.2
retentate			12.0	28.7	70.6	71.3

All isolated masses have a standard deviation of ±0.2 mg.

After collection of the first permeate, the remaining mixture was removed from the separation apparatus and the membrane was treated with cysteamine to cleave the cross-links. After the cross-links were cleaved, the retentate was reintroduced to the separation apparatus, solvent was added to the permeate side of the membrane, and another permeate fraction was collected. The second permeate fraction was enriched in TPM, containing 49% of the initial starting material at 67% purity. The retentate was collected after isolation of the second permeate, and it was significantly enriched in TDAB. The final retentate contained 88% of the starting TDAB chemical at 71% purity. The epoxy membrane was able to separate the three chemicals into three different fractions, all significantly enriched in purity compared with purity levels in the starting mixture.

Conclusions

We report the first examples of epoxy nanofiltration membranes that have cleavable disulfide bonds within the polymer matrix. These disulfide bonds reacted with the chemical stimulus cysteamine, modifying the flux and selectivity of chemicals through the membrane. Cleaving the disulfide bonds increased the pore sizes, increased the flux, and decreased the selectivities of the membranes. Furthermore, incorporating the disulfide bond in a membrane gave a single epoxy membrane with two different separation properties. These membranes were successfully applied to the separation of a three-component mixture. Though membrane modification is currently not reversible, in further research we hope to address this issue.

Not only is this new methodology useful in tackling chemical separations, but we believe that the introduction of labile disulfide bonds within a membrane may be useful in extending the lifetime of OSN membranes, which are typically 1–3 years.[40]

Over time, membranes age and undergo compaction, and a significant decrease in flux of chemicals and solvents through the membrane is often observed. If the membrane begins to suffer from compaction, cleaving a disulfide bond in the membrane could restore the flux and allow for a full recovery of membrane properties. When the membrane has aged further, cleaving more disulfide bonds could allow for the recovery of flux again, further extending the life of the membrane. Such an innovation may improve membrane performance and efficiency and spur further enthusiasm for this attractive green separation technology.

Experimental Section

Materials

All chemicals were purchased from Sigma-Aldrich, Acros, or VWR at their highest purity and used as received. Flat-sheet membrane PZ polyacrylonitrile solid supports (MWCO 30 000) were purchased from Synder Filtration and used as received. 1,4-Butanediol diglycidyl ether was purchased from Sigma-Aldrich and purified as described in a prior publication.[31]

Characterization

1H nuclear magnetic resonance (NMR) spectra were collected using a Bruker DPX-500 at 500 MHz or Bruker DRX-400 at 400 MHz at room temperature. NMR samples were referenced to trimethylsilane. Fourier transform-infrared spectra were collected at room temperature using an Avatar 370 FT-IR with a HP-DTGS-KBr detector. A Hitachi S-4800N scanning electron microscopy (SEM) system was used to collect SEM micrographs. Raman spectra were collected using a custom Raman instrument. The components include a diode-pumped CW laser (Excelsior, Spectra physics) producing 300 mW of 532 nm radiation, a Horiba iHR320 spectrometer, and an Andor Newton BV-CCD detector.

Fabrication of Membrane B-2

Diamine monomer B (4-aminophenyldisulfide, 0.25 g, 1.0 mmol) was dissolved in dimethylformamide (0.32 mL) in a scintillation vial. Purified diepoxide monomer 2 (1,4-butanediol diglycidyl ether, 0.37 mL, 2 mmol) was added to the reaction mixture, and the vial was placed in an oil bath at 60 °C for 12 h. The vial was then removed, and 0.05 mL of the reaction mixture was cast onto a polyacrylonitrile solid support. The membrane was placed in an oven at 60 °C and allowed to cure for 7 d.

Fabrication of membranes C-2, B1C3-2, B1C1-2, and B3C1-2 followed a similar procedure, and experimental details are described in the Supporting Information.

Flux of pNB, Triethylamine, and Tripropylamine

A membrane was clamped between two O-rings and two glass vessels. pNB (0.20 g, 1.3 mmol), triethylamine (0.18 mL, 1.3 mmol), tripropylamine (0.25 mL, 1.3 mmol), and DCM (25 mL) were added to one side of the membrane (retentate). DCM (25 mL) was added to the other side of the membrane (permeate). Approximately 7.07 cm2 of membrane was in contact with solvent on both sides. Solvent on both sides of the membrane was continuously stirred at room temperature. Samples (1 mL) were removed from the permeate and retentate sides of the membrane at various time points. p-Toluenesulfonic acid monohydrate (1 mL of a 0.53 M solution) dissolved in methanol was added to each sample that was removed from the separation apparatus to form salts with the amines. This increased their boiling points so the volatile amines would not evaporate. Tetraethylene glycol dissolved in DCM (0.4 mL of 0.023 M solution) was added to each sample, and solvent was removed by evaporation. The samples were analyzed by 1H NMR spectroscopy to find concentrations of the chemicals on both sides of the membrane.

This experiment was repeated using pNB and TPM.

Flux of pNB, Triethylamine, and Tripropylamine after Treating the Membrane with Cysteamine

A membrane was clamped between two O-rings and two glass vessels. DCM (8 mL), MeOH (7 mL), and cysteamine (1.0 g, 13.0 mmol) were added to the side of the apparatus facing the membrane (retentate). DCM (8 mL) and MeOH (7 mL) were added to the other side of the membrane (permeate). After 18 h, both solutions were removed and both sides of the apparatus were washed with 50 mL of MeOH and 50 mL of DCM. After washing, chemicals were added to the retentate side, and the experiment followed the same procedure as described previously.

Three-Component Separation of m-Dinitrobenzene, TPM, and TDAB through Membrane B1C1-2

Membrane B1C1-2 was clamped between two O-rings in a glass separation apparatus. m-Dinitrobenzene (22.8 mg, 0.14 mmol), TPM (33.0 mg, 0.14 mmol), and TDAB (80.5 mg, 0.14 mmol) were dissolved in DCM (10 mL) and added to one side of the membrane (retentate). DCM (70 mL) was added to the other side of the membrane (permeate). Extractions were performed every 24 h by decanting the permeate and removing the solvent in vacuo. Fresh DCM (70 mL) was added to the permeate side of the membrane. Permeate samples were weighed and analyzed via 1H NMR spectroscopy. After 7 d, the retentate was removed, analyzed by 1H NMR spectroscopy to determine its composition, and saved. The membrane was then treated with cysteamine as described above. After treatment with cysteamine, the saved retention fraction was redissolved in DCM (10 mL) and returned to the retentate side of the membrane. DCM (70 mL) was added to the downstream. Extractions were performed again every 24 h. After 13 d, the retentate was decanted and analyzed by 1H NMR spectroscopy to determine composition of the retention.

Monitoring the Polymerization via FT-IR Spectroscopy

Monomers used to fabricate an epoxy membrane were mixed together as described above. A couple of drops were removed from the mixture and placed between two polished NaCl salt plates. The first FT-IR spectrum was collected immediately. After collecting the first spectrum, the salt plates were placed in a 60 °C oven. Subsequent spectra were collected at different time points by removing the salt plates from the oven, collecting a spectrum, and placing the salt plates back in the 60 °C oven. The reaction was monitored until there were no further changes in the spectra, indicating that the reaction had reached completion and the membranes were ready for use.

Collecting Raman Spectra of the Membranes

Epoxy membranes were fabricated as described above and in the Supporting Information. Before acquiring Raman spectra, the spectrometer and camera were calibrated using known emission lines from a neon lamp. The laser radiation incident on the samples was focused to spot size of ca. 100 μm2, and the beam energy was attenuated to achieve irradiance on the samples between 7 and 0.5 mW to avoid sample damage. The scattered radiation from the sample was collected via a Nikon f/1.2 camera lens with a focal length of 50 mm and passed through two holographic notch filters (Kaiser optical systems) to eliminate 532 nm radiation. The collected radiation was focused onto a 0.05 mm slit before entering the spectrometer. Spectra are composed of 50 accumulations with exposure times of 2 s per accumulation (100 s total acquisition time). Spectra were collected for membranes C-2, B1C3-2, B1C1-2, B3C1-2, and B-2 before exposure to cysteamine and after 3 h of exposure to cysteamine.