Synthesis of Alkaline Anion Exchange Membranes with Chemically

Apr 13, 2018 - Department of Chemistry and Chemical Biology, Baker Laboratory, Cornell University, Ithaca , New York 14853-1301 , United States...
3 downloads 0 Views 2MB Size
Article Cite This: Macromolecules XXXX, XXX, XXX−XXX

Synthesis of Alkaline Anion Exchange Membranes with Chemically Stable Imidazolium Cations: Unexpected Cross-Linked Macrocycles from Ring-Fused ROMP Monomers Wei You, Kristina M. Hugar, and Geoffrey W. Coates* Department of Chemistry and Chemical Biology, Baker Laboratory, Cornell University, Ithaca, New York 14853-1301, United States S Supporting Information *

ABSTRACT: In order to prepare base-stable, mechanically strong, and synthetically feasible alkaline anion exchange membranes (AAEMs) for applications in alkaline fuel cells, an imidazolium-fused cyclooctene monomer was prepared and subjected to ring-opening metathesis polymerization (ROMP) conditions. Surprisingly, macrocyclic oligomers were obtained instead of high molecular weight polymers. High-performance AAEMs were synthesized by using a bifunctional monomer to cross-link the macrocycles. The resultant AAEMs showed high ionic conductivities (σOH− = 59 mS/cm at 50 °C), robust mechanical properties, and excellent alkaline stabilities.



INTRODUCTION Alkaline fuel cells (AFCs) that are assembled with alkaline anion exchange membranes (AAEMs) have several significant advantages in comparison to state-of-the-art proton exchange membrane fuel cells (PEMFCs).1−6 (1) Increased pH in AFCs accelerates the rate of the oxygen reduction reaction, which lowers fuel cell cost if non-platinum electrode catalysts are used.7−9 (2) Oxidation of direct alcohol fuels (e.g., methanol and ethanol) is also significantly faster in AFCs.10,11 (3) Perfluorinated polymers (i.e., Nafion) for PEMFCs not only are expensive but also hamper the recycling of Pt.12−14 Hundreds of AAEMs have been prepared over the past decades for the development of AFCs1−6 as well as other applications, including high purity H2 production from water electrolysis,15,16 redox flow batteries,17,18 and gas separation.19 However, widespread applications of AAEMs have not been achieved yet because most AAEMs degrade rapidly under the standard operating conditions (e.g., high pH and high temperature). An ideal AAEM would have high hydroxide conductivity, good chemical durability, strong mechanical properties, and a simple chemical synthesis. AAEMs usually consist of polymeric structures with appended cations, and both the polymer backbones and the cations play an important role in the alkaline stabilities.1−6 The cations contribute to the ionic conductivities, while the polymer backbone influences the mechanical properties, preventing the membranes from excessive swelling. Cation degradation under alkaline conditions reduces ionic conductivity, and degradation of polymer backbones reduces the mechanical properties. 20,21 The membranes’ micromorphology may also influence the conductivity and stability.22 Tetraalkylammonium cations are widely used in AAEMs, but numerous studies have reported that they undergo degradation © XXXX American Chemical Society

under alkaline conditions, especially benzyltrimethylammonium (BTMA) derivatives.20,23−29 Cyclic tetraalkylammonium cations (e.g., pyrrolidinium or piperidinium) have been recently reported to exhibit better alkaline stabilities.30−36 Other cationic groups have also been extensively studied as candidates in AAEMs, including but not limited to imidazolium,37−45 benzimidazolium,46−51 and phosphonium52−56 groups. In 2012, our group described an alkaline-stable AAEM functionalized with a tetrakis(dialkylamino)phosphonium cation.56 Despite its good ionic conductivity and outstanding alkaline stability, the synthesis of the phosphonium monomer required nine steps, and it was difficult to introduce structural variations.55 In 2015, we reported multisubstituted imidazolium cations that are readily synthesized and have excellent alkaline stabilities (Scheme 1).29 Imidazolium alkaline stability is significantly improved with aromatic substituent at the C2 position, alkyl/ Scheme 1. Alkaline-Stable Imidazolium Monomers for AAEM Synthesis

Received: January 29, 2018 Revised: March 8, 2018

A

DOI: 10.1021/acs.macromol.8b00209 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules Scheme 2. Synthesis of Imidazolium Monomers 1−4

Figure 1. Alkaline stabilities of monomers 3 and 4.

using cyclooctene (COE)-substituted monomers (Scheme 1).56 By using a bifunctional cross-linker, novel AAEMs with high ionic conductivities, robust mechanical properties, and excellent alkaline stabilities were obtained.

aryl substituents at C4 and C5, and alkyl substituents (other than methyl or benzyl) on the two nitrogen atoms.29 No cation decomposition was detected when the multisubstituted imidazolium was heated with 5 M KOH in methanol at 80 °C for 30 days.29 Because of the high base stability and straightforward synthesis, we investigated incorporation of imidazolium cations into polymeric membranes. Polyethylene-based backbones have been known as ideal AAEM backbones due to their excellent alkaline stabilities and mechanical properties.56−60 Mecking and co-workers reported synthesis of imidazolium-functionalized polyethylene through direct Pd-catalyzed copolymerization of ethylene and allylimidazolium, but the imidazolium incorporation was low (800:1), we got low molecular weight products rather than high molecular weight polymers, which was confirmed by gel permeation chromatography (GPC) analysis (Mn = 570 Da

the imidazolium ring, were then designed to reduce the number of hydrophobic components in the monomer. Monomers 3 and 4 can both be easily prepared in five steps without column chromatography (Scheme 2b).66 One of the key intermediates, cyclooct-5-ene-1,2-dione, was achieved by modified Swern oxidation.67,68 The unoptimized overall yield was 38% and 32% for 3 and 4, respectively. Both monomers also showed excellent stabilities under strong alkaline conditions (Figure 1). Only 1% degradation was observed for monomer 3 when it was treated with 5 M KOH in CD3OH at 80 °C for 30 days. Although 4 degraded by 4% after 30 days in 5 M KOH/CD3OH treatment, it is noteworthy that the C

DOI: 10.1021/acs.macromol.8b00209 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules Scheme 3. Proposed Reaction Pathways To Form Macrocyclic Products

Table 1. Preparation of Cross-Linked AAEMsa

AAEMs

4:3:COEb

IECc (mmol/g Cl−)

water uptaked (%)

dimensional changee (%)

σf (OH−, 22 °C, mS/cm)

σf (OH−, 50 °C, mS/cm)

AAEM-1 AAEM-2

160:0:800 160:100:800

1.32 ± 0.01 1.37 ± 0.01

70 ± 3 94 ± 3

17 ± 1 17 ± 1

29 ± 2 37 ± 2

43 ± 3 59 ± 2

a

See the Supporting Information for more details. Grubbs II catalyst = (Cl2Ru(IMes)(PCy3)CHPh); Crabtree’s catalyst = [(COD)Ir(py)(PCy3)]PF6. bFeeding equivalents ratio. cIon exchange capacity (IEC) determined by back-titration, average of three trials. dGravimetric water uptakes of the fully hydrated membranes, average of three trials. eDimensional length change of the fully hydrated membranes, average of three trials. fHydroxide conductivities of the AAEMs fully immersed in degassed water at 22 and 50 °C, average of three trials.

and Đ = 4.54; see Figure S36). Additionally, there was consistently 20% unreacted monomer 3 after the copolymerization. To study the detailed polymerization process, the reaction was monitored by 1H NMR spectroscopy (Figure 2).66 It was observed that COE was converted approximately 10 times faster than sterically unhindered monomer 3: 96% COE monomer was consumed in 10 min, while monomer 3 slowly approached 80% conversion after about 6 h (Figures 2a and 2b, respectively). There were almost no further monomer conversions from 6 to 24 h. To test whether catalyst decomposition had occurred, additional COE was added. Rapid polymerization was observed, demonstrating that the catalyst was still active. Since COE was fully polymerized within 15 min, we hypothesized that 3 may undergo slow secondary metathesis reactions with polycyclooctene (polyCOE). To test this hypothesis, a well-defined polyCOE sample (Mn = 54.4 kDa and Đ = 1.91) was mixed with 3 and Grubbs II catalyst in CDCl3; a similar conversion rate of 3 was observed (Figure 2c). When 3 alone was treated with Grubbs II catalyst, the conversion was slower, and a maximum conversion of ca. 60% was observed (Figure 2d).

Electrospray ionization−high resolution mass spectrometry (ESI-HRMS) was used to analyze the reaction products.69−71 The products were easily observed by ESI-HRMS (positive mode), and the numbers of charges were identified by analysis of isotope peaks (Figure 3).66 The ESI-HRMS spectrum of 3/ COE copolymer was almost identical to the product of 3 with polyCOE.66 To our surprise, instead of linear polymers, most of the products were macrocyclic oligomers with different numbers of ring-opened COE and 3 units. The HRMS data confirmed that the structures were macrocycles due to the lack of end groups (e.g., for [AB2]nn+, calculated: 501.4203; found: 501.4185, −3.6 ppm). The [A]nn+ peak in Figure 3 corresponds to the monomer 3 or its homooligomers. According to 1H NMR analysis, 20% unreacted 3 remained, so the peak intensities in ESI-HRMS suggested that the macrocyclic products were the majority, especially the cyclic oligomers in the forms of [AB2]nn+, [AB]nn+, and [A2B]n2n+. The m/z region of 501.4−503.4, which is assigned to [AB2]nn+, is enlarged in Figure 3 to show the isotope peaks. These low molecular weight features are consistent with the above-mentioned GPC results.66 Additionally, through similar ESI-HRMS analysis, the products of the homopolymerization reaction of 3 (Figure 2d) D

DOI: 10.1021/acs.macromol.8b00209 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules were identified to be oligomer macrocycles including dimers, trimers and tetramers (Figure S31).66 This type of ring-opening metathesis copolymerization to form macrocyclic compounds has never been reported before.72,73 Two reasons are proposed to explain the macrocycle formation (Scheme 3). The first is the moderate ring strain of the bicyclic fused COE. According to the preliminary density functional theory (DFT) calculations, the ring-opening energies of 3 and COE are −7.5 and −11.5 kcal/mol, respectively (Table S5).66 The moderate ring strain resulted in slow ROMP propagation and more likely for secondary metathesis (Scheme 3, paths b and c).74−76 The second reason is the planar nature of the fused imidazolium ring. The ringopened 3 is near planar due to the 4,5-substitution on the imidazolium ring, and it can act as a “U-turn” in the propagating chain. In contrast to the ROMP of simple cyclic alkenes, the intramolecular secondary metathesis to produce macrocycles is much more facile. During the reaction of 3 with polyCOE, cross-metathesis is likely occurring to incorporate 3 while triggering the depolymerization of polyCOE. Consistent with these results, Grubbs reported that the diene ring-closing metathesis (RCM) to form aromatic- or trans-[8,6]-fused COE showed much higher yield than more strained and nonplanar cis-[8,6]-fused COE.77 ROMP of trans- and cis-diacetyl-fused COE was also reported.78 It was found that the less-strained trans-monomer can only be polymerized under very concentrated conditions (>2 M). Nuckolls and co-workers also reported that cis,cis-dibenzo[a,e]cyclooctatetraene is unreactive under ROMP conditions.79 Copolymerization of these monomers with COE has not been studied. The results in this work suggested that these planar-fused low-ring-strain monomers may be candidates to depolymerize normal ROMP polymers into oligomeric macrocycles. Since the macrocyclic materials had poor mechanical properties, cross-linker 4 was added to the polymerization (Table 1).57 The membranes were produced by the reactive casting method, in which the solvent slowly evaporated during the reaction process. Crabtree’s catalyst was added into the reaction mixture during the casting process, and the crosslinked membranes were directly hydrogenated to yield the saturated polymer. It was noticed that the membranes’ longterm mechanical stabilities were significantly improved by hydrogenation. The hydrogenated cross-linked membranes showed very good ionic conductivities, and the optimal composition was [Ru]:[COE]:[3]:[4] = 1:800:100:160 (Table 1, AAEM-2). High hydroxide conductivity (37 mS/cm at 22 °C and 59 mS/cm at 50 °C) was observed, and less than 20% dimensional change was detected under fully hydrated conditions. The tensile strength at break (hydrated iodide form) was determined to be 4.7 MPa with more than 140% strain, suggesting that the membrane was not brittle or stiff. It is noteworthy that the optimal loading of Grubbs II catalyst is much lower than our group’s previous cross-linked membrane ([Ru]:[alkenes] = 1:90), likely due to the unexpected reaction pathways.57 The membranes’ alkaline stabilities were evaluated in 1 M aqueous KOH solution at 80 °C for 30 days (Figure 4). An initial conductivity drop to 29 mS/cm was observed after the first 3 days, which may be caused by dissolution of the unlinked macrocycles. The hydroxide conductivity was then stabilized at around 30 mS/cm, and a minor conductivity drop (