Article pubs.acs.org/Macromolecules
Anhydrous Proton Transport in Polymerized Ionic Liquid Block Copolymers: Roles of Block Length, Ionic Content, and Confinement Christopher M. Evans,†,‡ Gabriel E. Sanoja,∥ Bhooshan C. Popere,†,‡ and Rachel A. Segalman*,†,‡,§ †
Materials Research Laboratory, ‡Department of Chemical Engineering, and §Materials Department, University of California, Santa Barbara, Santa Barbara, California 93106, United States ∥ Department of Chemical and Biomolecular Engineering, University of California, Berkeley, Berkeley, California 94705, United States S Supporting Information *
ABSTRACT: Anhydrous proton transport has been investigated in a series of proton conducting polystyrene-blockpolymerized ionic liquid (PS-b-PIL) copolymers spanning a range of molecular weights and compositions. The PIL is a macromolecular analogue of imidazolium bis(trifluoromethane sulfonimide) (ImTFSI), a well-known proton conducting ionic liquid, and consists of imidazole linked to a polymer backbone via the 5-carbon. In contrast to prior work on nitrogen-linked imidazolium PILs, carbon-linked imidazolium has two nitrogens which can both function as proton donor/acceptors and participate in Grotthus mechanism conduction. The conductivity of the PIL block is shown to be dramatically impacted upon confinement by a PS block and can exceed the conductivity of the homopolymer in the range of 30−130 °C for PIL-rich block copolymer composition. At high temperature the conductivities track with ionic content while at room temperature the conductivities are nonmonotonic. X-ray scattering reveals a suppression of the peak associated with ionic aggregation in all block copolymers relative to the homopolymer consistent with the higher conductivities observed at room temperature. The dependence of ionic conductivity on temperature, as quantified by the VFT strength parameter D, decreases with decreasing PIL block length corresponding to a change in the packing efficiency of the conductive block. These changes in packing are hypothesized to lead to the different temperature dependences of conductivity which cause the nonmonotonic block copolymer conductivities observed at room temperature. Finally, we demonstrate that the fraction of PIL in the block copolymer is the main factor governing the high temperature ionic conductivity of these materials while confinement effects become important at room temperature.
■
INTRODUCTION Ion conducting polymers are key components of energy storage and conversion devices such as batteries,1−5 fuel cells,6−9 supercapacitors,10,11 and solar fuel cells12 where flexible, solid state devices are desired. Polymerized ionic liquids (PIL), macromolecules based on the covalent attachment of familiar ionic liquid (IL) chemistries to a polymer backbone, have recently become an attractive route to solid polymer electrolytes due to their ability to exhibit high ionic conductivities particularly under anhydrous conditions.13−41 While water is required for ion transport in conventional polyelectrolytes such as Nafion or polystyrenesulfonate, PILs have the potential to be used at temperatures where these materials dry out (>100 °C). In contrast to studies that swell polymer domains with ILs to form ion gels, PILs do not suffer from ion leakage because one of the ions is tethered to the polymer backbone. Additionally, the current is essentially carried entirely by the untethered ion (transference number approaching unity), and the mechanical integrity is greatly improved compared to ILs. As research in PILs has grown over the past decade, there are still many fundamental questions remaining concerning the relationship © XXXX American Chemical Society
between structure and conductivity as well as a precise understanding of the mechanism of charge transport compared to conventional ionomers. Recent work on PILs has primarily focused on understanding ion transport in homopolymers, 13−25 random copolymers,23,29,31 and block copolymers23,26−29 based on tethered imidazolium cations and various hydrophobic, fluorinated counterions. In most of this prior work, the imidazolium was attached to the polymer backbone via one of the two nitrogens followed by quaternization of the second nitrogen. These PILs can conduct various anions such as BF4, PF6, and bistrifluoromethyl sulfonimide (TFSI) with a transference numbers of ∼1 and have revealed a surprisingly large dependence of conductivity, viscosity, and dielectric constant on the anion species as well as the length and polarity of the spacer connecting the imidazolium to the polymer backbone.15,16 In these types of materials, the ion pairs are present within a low Received: October 6, 2015 Revised: November 20, 2015
A
DOI: 10.1021/acs.macromol.5b02202 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules Scheme 1. Synthesis of Protic and Nonprotic Polymerized Ionic Liquids
increasing the molecular weight;20 however, processing becomes more difficult, and the material may still not be sufficient to make a robust membrane. With copolymers, the molecular weights and composition of each block can be varied to design a material with an optimal balance of mechanical properties and conductivity. Additionally, the role of block copolymer confinement on the PIL conductivity is an emerging area of interest in PILs as recent works have indicated that the conductivities of ionic liquids may be increased by 3 orders of magnitude when confined into nanoporous frameworks.48 Block copolymers with a sulfonated PS block swollen by IL have also demonstrated confinement enhanced conductivity.49 By controlling the PIL block length, the domain size and thus the degree of confinement can be systematically tuned to investigate its effect on conductivity. We have investigated a series of proton conducting PS-blockpoly(5-aminoethylimidazolium acrylamide) TFSI (P(5AEImH+)TFSI) copolymers spanning a range of molecular weights and composition in order to disentangle the roles of block length and overall ionic content. The conductivity of the P(5AEImH+)TFSI block is shown to be dramatically impacted upon confinement by a PS block and can exceed the conductivity of the homopolymer in the range 30−130 °C for PIL-rich block copolymer composition. At high temperature, the conductivities appear to merge, and the extrapolated high temperature conductivity of the block copolymers, determined from a Vogel−Fulcher−Tammann (VFT) fit, never exceeds that of the homopolymer. For shorter P(5AEImH+)TFSI blocks, the conductivity exceeds that of the homopolymer at room temperature but is lower at higher temperatures, indicating the importance of measurement temperature in discussing block copolymer confinement effects. This is due to the manner in which confinement impacts the
dielectric constant environment presented by the polymer backbone. The pairs tend to cluster to minimize their interaction with the medium and form aggregates containing multiple ion pairs. Aggregate formation has been studied via simulations where a variety of structures ranging from spherical to string-like aggregates have been observed.42−44 Experimentally, the main technique for probing aggregation is X-ray scattering which reveals multiple correlation peaks on the length scale of ∼3−7 Å.44 The structure of the aggregates, and in particular how they are impacted upon confinement within a block copolymer, is expected to have a profound influence on the conductivity, viscosity, and dielectric constant. Recent work28 has demonstrated that imidazole can be attached to the polymer via one of the carbon atoms instead of a nitrogen. The two free nitrogens can function as proton donor/acceptor sites and engage in proton transport. In the neutral state, a single acidic proton alternates between the two nitrogens. Upon protonation with a strong acid like HTFSI, both nitrogens become protonated and the imidazolium ring bears a positive charge. These materials are the polymer equivalent of imidazolium TFSI (ImTFSI), a well-known protic IL which operates via the Grotthus mechanism of conduction.45−47 If these protic PILs can be made to exhibit reasonably high proton conductivities in the absence of water, they provide a clear advantage over existing proton exchange membranes. Block copolymers provide a versatile route to independently tune conductivity and mechanical properties by attaching a PIL block to a mechanical component such as polystyrene (PS). While PIL homopolymers are more mechanically robust than ILs, they generally have subambient glass transition temperatures and are typically not robust enough to form freely standing membranes. The viscosity of a PIL can be increased by B
DOI: 10.1021/acs.macromol.5b02202 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
three times and then polymerized at 90 °C for 24 h followed by precipitation in methanol. Molecular weight determination was performed on a THF GPC using styragel columns and PS standards (Figure S2). The ASI block was added by dissolving the PS macroinitiator in DMF (0.5 g/mL) with AIBN at a 10:1 molar ratio of RAFT chain ends to AIBN. The block copolymer was precipitated in methanol, and the PASI block was converted to a PIL as previously described. Block copolymer composition was verified by 1H NMR in d-DMF. Materials Characterization. 1 H NMR spectroscopy was performed on a 500 MHz Bruker spectrometer in anhydrous dDMSO (99.9 atom %, Aldrich) to confirm functionalization of PASI into the imidazolium acrylamide polymers. For PS-block-PASI copolymers composition determination, d-DMF was used because it is a solvent for both blocks. The PS blocks were characterized by GPC with THF mobile phase at 30 °C, Styragel columns, and PS standards for calibration. Thermal analysis was performed using a PerkinElmer DSC 8000 to measure the Tg of the homo- and block copolymers on second heating at 20 °C/min via the half delta Cp method. Conductivity Measurements. Impedance measurements were performed on a Biologic VSP-300 using circular stainless steel blocking electrodes in a through plane configuration. The PIL was placed on an electrode covered with Kapton tape as an ∼50 μm spacer. A hole of precise diameter was punched in the tape to control the electroactive area of the sample. A second electrode was placed on top and then consolidated with a Carver press (140 °C, 1 ton, 5 min). The film thicknesses were measured using a digital micrometer (±1 μm precision) on the assembly with and without the PIL. Impedance measurements were performed from 3 MHz down to 30 mHz with an amplitude of 100 mV and the complex impedance (Z* = Z′ − iωZ″) was converted to complex conductivity according to50
temperature dependence of ionic conductivity as quantified by the VFT strength parameter D. Finally, the fraction of PIL in the block copolymer is demonstrated to be the main factor governing the high temperature ionic conductivity of these materials while confinement effects become important at room temperature.
■
EXPERIMENTAL SECTION
Materials. Azobis(isobutyronitrile) (AIBN, Aldrich), N-acryloxysuccinimide (TCI Chemicals), cyanomethyldodecyl trithiocarbonate (CMDDT, Aldrich), 3-(aminopropyl)imidazole (Aldrich), histamine dihydrochloride (Aldrich), methyl iodide (Spectrum Chemical), lithium bis(trifluoromethanesulfonimide) (LiTFSI, Aldrich), HTFSI (Aldrich), and triethylamine (TEA, Aldrich) were used as received. Styrene monomer (Aldrich, 99%) was passed through basic alumina prior to polymerization. Anhydrous dimethylformamide (DMF, Aldrich) and dimethyl sulfoxide (DMSO, Aldrich) were used as received for polymerizations. Synthesis. Protic and nonprotic PIL homopolymers were synthesized as depicted in Scheme 1 starting with synthesis of the poly(n-acryloxysuccinimide) (PASI). The succinimide is a good leaving group for subsequent amidations affording a versatile platform for PIL synthesis. N-Acryloxysuccinimide was dissolved in anhydrous DMF (0.5 g/mL) followed by addition of the RAFT agent CMDDT and AIBN in a 10:1 molar ratio to a 25 mL Schlenk flask. The solution underwent a minimum of three freeze−pump−thaw cycles to remove dissolved oxygen from the system. Polymerization was performed at 70 °C for 18 h and then precipitated in methanol. 1H NMR end-group analysis of the alkyl tail on the RAFT agent was used to determine the Mn of PASI in deuterated DMSO. The polymer was not soluble in THF, chloroform, or toluene for GPC characterization. To synthesize a protic PIL, the PASI homopolymer was dissolved in anhydrous DMSO along with 1.2 equiv of histamine dihydrochloride and 3 equiv of TEA to neutralize the HCl and the displaced hydroxysuccinimide groups. The reaction mixture was stirred overnight at 60 °C. The amidation was quantitative as confirmed by 1H NMR via the disappearance of the succinimidyl protons and appearance of two methylene units between the imidazole and acrylamide (Figure S1). Poly(5-aminoethylimidazolylacrylamide) (P(5AEIm)) was dialyzed for 48 h in DI water and 24 h in methanol before drying in a vacuum oven at 120 °C overnight. Protonation with HTFSI in methanol yielded the final poly(5-aminoethylimidazolylacrylamide) TFSI (P(5AEImH+)TFSI). For the nonprotic PIL poly(3-methyl-1-aminopropylimidazolylacrylamide) TFSI ((P3MAPIm+)TFSI), PASI was dissolved in anhydrous DMSO with 1.2 equiv of 3-(aminopropyl)imidazole and 1 equiv of TEA and reacted overnight at 60 °C. This polymer was then dissolved in methanol with 1.2 equiv of methyl iodide to quaternize the free nitrogen. Salt metathesis was performed using a 10× mol equiv of LiTFSI in methanol at 50 °C to exchange TFSI for Li, and the polymer was dialyzed for 48 h in DI water and 24 h in methanol followed by drying at 120 °C overnight under vacuum. Details of the homopolymers made from the same PASI precursor are summarized in Table 1. Block copolymers were synthesized by first polymerizing a polystyrene block using RAFT. Styrene monomer was passed through basic alumina and then dissolved in anhydrous DMF (0.2 g/mL) with a RAFT:AIBN ratio of 10:1. The solution was freeze−pump−thawed
σ ′(ω) =
The frequency independent plateau of conductivity was taken as the “dc conductivity” following literature protocol.50 Prior to measurement the samples were dried in a vacuum oven overnight at 140 °C followed by hot pressing and another 2 h under vacuum. Conductivity measurements were performed on an INSTEC hot stage with a continuous dry nitrogen purge provided by a liquid nitrogen dewar. The samples were equilibrated for ∼1 h at the highest temperature, and the conductivity was monitored until it stabilized. Measurements were then taken upon cooling at a rate of ∼3 °C/min. X-ray Scattering. Samples for small- and wide-angle X-ray scattering were prepared by removing the polymer from the conductivity cells and placing the material in a washer with Kapton windows attached via silicone epoxy to provide a moisture barrier. The assembly was formed in a nitrogen glovebox, and samples were stored over Drierite until they were placed in the beamline. Experiments were performed at both beamline 7.3.3 at the Advanced Light Source at Lawrence Berkeley National Lab and beamline 1-5 at the Stanford Synchrotron Radiation Lightsource. Data analysis was performed using the Nika package for Igor pro, and sample-to-detector distances were calibrated with AgB. Sample intensities were background corrected by subtracting the signal of an empty sample holder, with both sample and blank intensities normalized by their total transmission.
■
RESULTS Two homopolymer PILs were synthesized as described in the Experimental Section to investigate the fundamental difference between a nonprotic and protic PIL. When imidazolium is attached via the 1-nitrogen, the 3-nitrogen is subsequently quaternized with a methyl group and then salt exchanged to yield a TFSI− counterion (Scheme 1, P(3MAPIm+)TFSI). The attachment of imidazolium via the 5-carbon leaves the ring available for protonation and both H+ and TFSI− are able to conduct (Scheme 1, P(5AEImH+)TFSI). As mentioned previously, the small molecule analogue ImTFSI is known to
Table 1. Molecular Weight and Glass Transition Temperatures of Homopolymer PILs polymer
Mn PSa (kg/mol)
Tg (°C)
P(5AEImH+)TFSI P(3MAPIm+)TFSI
39 39
51 16
Z′(ω) k[Z′(ω)2 + Z″(ω)2 ]
a
Determined by 1H NMR end-group analysis of RAFT chain transfer agent and succinimide protons of precursor. C
DOI: 10.1021/acs.macromol.5b02202 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
Figure 1. (a) Real part of the complex conductivity measured from 30 to 130 °C as a function of frequency for the protic polymerized ionic liquid P(5AEImH+)TFSI. The plateau is taken as the “dc conductivity” which increases with increasing temperature. Conductivity of P(5AEImH+)TFSI (blue diamonds) compared to the conductivity of a nitrogen-linked P(3MAPIm+)TFSI (red triangles) is shown as a function of (b) temperature and (c) temperature normalized to the calorimetric Tg illustrating the contribution of proton transport to PHA conductivity. (d) A VFT fit to the P(5AEImH+)TFSI data illustrating the non-Arrhenius temperature dependence of conductivity.
conduct protons.45−47 Figure 1a shows the real part of the complex conductivity (σ′) determined from the measured complex impedance Z* where the plateau is taken as the “dc conductivity” which increases with increasing temperature, as expected for ionic materials.50 On a raw temperature scale, the conductivity of the nonprotic system is higher than the protic case (Figure 1b). However, when normalized to the calorimetric glass transition Tg, the conductivity is substantially higher in the protic P(5AEImH+)TFSI (Figure 1c). This is consistent with a strong contribution from proton diffusion in addition to the TFSI anion, whereas the P(3MAPIm+)TFSI only has mobile TFSI.51 In the impedance measurement, for each half cycle the cations will drift to the negative electrode, and the anions will drift toward the positive electrode. Both species create a concentration gradient with a concomitant potential that opposes the direction of the applied field. Thus, both ions add constructively to the impedance response of the materials, and the ionic motions of H+ and TFSI− do not cancel. Conductivity for the neutral precursor is shown in Figure S3, indicating a vanishingly small conductivity due to proton hopping of neutral imidazole. This data also verifies that there is not an appreciable amount of ionic impurities in the polymer prior to protonation. In both materials, a strongly non-Arrhenius behavior was observed as expected for polymers where one ion is covalently attached to the polymer backbone (Figure 1d). Such behavior is ubiquitous for polymers with one tethered ionic group and is
attributed to the coupling of ion transport with polymer segmental dynamics.13,15,16,19,32,34 In this case, the conductivity is described by the VFT equation: ⎛ DT0 ⎞ σ = σHighT exp⎜ − ⎟ ⎝ T − T0 ⎠
(1)
where σHighT is the limiting high temperature conductivity, D is the so-called strength parameter (inversely related to fragility), and T0 is the Vogel temperature where the conductivity hypothetically diverges to zero. Physically, the prefactor σHighT corresponds to the theoretical highest conductivity that could be achieved by heating the material if decomposition did not occur. It represents a regime where the material is far above Tg such that the polymer dynamics and ion dissociation are no longer a significant barrier to ion transport. Two different molecular weights of PS (7.2 or 21.2 kg/mol) were investigated with a range of P(5AEImH+)TFSI block lengths to deconvolute the roles of block domain size and overall ionic content on conductivity. Table 2 summarizes the block copolymers used in the present study. The morphology of all of the block copolymers in the present work is lamellar as determined by SAXS (Figure 2). The circles above the data correspond to integer multiples of the primary peak wavevector q*. The intense scattering is due to the large mismatch of electron density between PS and P(5AEImH+)TFSI as well as the strong segregation due to the polarity difference of the blocks. In order to achieve these well-ordered patterns, samples D
DOI: 10.1021/acs.macromol.5b02202 Macromolecules XXXX, XXX, XXX−XXX
Article
Macromolecules
conductivity, but a stronger temperature dependence leads to a crossing with the 26 mol % block copolymer. Finally, for the highest P(5AEImH+)TFSI content (23 kg/mol, 43 mol % PIL), the conductivity exceeds that of the homopolymer over the entire temperature range investigated (30−130 °C). It is interesting to note that the PS7-b-P(5AEImH+)TFSI23 and homopolymer conductivities appear to be merging at higher temperature. For the different PIL block lengths, the varying temperature dependence of conductivity is attributed to the manner in which confinement alters the packing. The shortest conducting blocks are most confined, and we hypothesize that they are being forced to pack more effectively due to chain stretching in a strongly segregated block copolymer and due to confinement by the polystyrene. This tighter packing would be consistent with a structure that does not evolve as much with temperature and thus a conductivity which also has weaker temperature dependence than in the less confined block copolymers. The manner in which confinement alters the packing and thus the temperature evolution of conductivity leads to room temperature conductivities which show a nonmonotonic dependence on ionic content. Block copolymers with a 21 kg/mol PS block show nearly identical trends (Figure 3b) as the lower molecular weight system. At high P(5AEImH+)TFSI content the conductivity exceeds that of the homopolymer over the entire temperature range but appears to merge with the homopolymer line at higher temperature. At room temperature all of the block copolymers are more conductive than the homopolymer as with the 7 kg/mol PS diblocks. To gain insight into the nature of these block copolymer confinement effects, WAXS experiments were performed to investigate the ionic aggregate peak that has been described in previous PIL studies (Figure 4).15,44 The protonated version of the P(5AEImH+)TFSI shows an additional peak, not present in the neutral precursor, at q = 0.88 Å−1 corresponding to ionic aggregation (Figure 4a). The ion pairs form clusters due to the low dielectric constant of the polymer backbone, and they give rise to scattering on the ∼5−7 Å length scale. Intuitively, one expects a larger extent of aggregation to correspond to a lower conductivity as the number of free ions will be diminished. A representative block copolymer WAXS pattern reveals that the strength of the ion aggregate peak is suppressed relative to the P(5AEImH+)TFSI homopolymer at room temperature (Figure
Table 2. Summary of Block Copolymer Molecular Weights and Compositions polymer name
a
PS7-b-P(5AEImH+)TFSI11 PS7-b-P(5AEImH+)TFSI16 PS7-b-P(5AEImH+)TFSI23 PS21-b-P(5AEImH+)TFSI21 PS21-b-P(5AEImH+)TFSI38 PS21-b-P(5AEImH+)TFSI62
Mn PS, kg/mol (dispersity)b
mole fraction PIL, f ILc
Mn PIL,c kg/mol
NPIL
7.2 (1.09)
0.26 0.35 0.43 0.16 0.26 0.37
10.8 15.8 23.1 20.7 37.8 62.1
24 35 51 46 84 138
21.2 (1.12)
a
Subscripts denote molecular weight of the block in units of kg/mol. Determined via THF GPC at 30 °C against PS standards. c Determined via 1H NMR in d-DMF. Error in composition determinations is taken to be
