Advanced Materials

Poly(alkylene biguanides) as Proton Conductors for High-Temperature PEMFCs

Authors


Abstract

Poly(alkylene biguanides) are novel high-temperature proton conductors. This long-known class of polymers is presented as surprisingly stable high-temperature proton-conducting materials in the form of water-free HCl conjugates. Proton conductivity is dominated by free volume relaxation. Application in the context of fuel-cell membranes is discussed.

original image

A solid polyelectrolyte membrane (PEM) is key to the function and performance of any low-temperature fuel cell (PEMFC).1–3 The membrane should be permeable for solvated protons that are produced from the fuel at the surface-anchored catalyst but it also needs to exhibit barrier properties for the fuel, notably hydrogen or methanol (in case of a direct-methanol fuel cell)4 or other fuels like hydrocarbons. For reasons inherent to the nature and rate of the catalytic process,1 the fuel cell is optimally driven at temperatures well above the boiling point of water, that is, temperatures around 150 °C and higher.5, 6 This constitutes a severe drawback for membrane materials that function because they are swollen with water as the essential medium for the proton solvent, in which case the charge carriers are mobile hydronium ions. The best-known case is NAFION,7 a polymer composed of a poly(fluorocarbon) backbone and fluorocarbon side chains with sulfonic acid end groups, but there are other similar polymers. These polymers fall dry at temperatures above 80 °C rather quickly due to evaporation of water from the matrix. Thus, complex “water management” under pressurized conditions is necessary to run fuel cells efficiently if they contain such polymers as the functional component. These and other related considerations have given birth to widespread research efforts to create alternative materials as high-temperature proton conductors.6, 8–14

An obvious suggestion was to replace water by other materials that have the ability to solvate protons. Two such attempts are noteworthy: replacement of water by a) imidazole 5, 8 and b) phosphoric acid.15–18 While both are good solvents for protons forming onium-type complexes, they are also subject to electro-osmotic drag (as is water) and a mechanism needs to be implemented in order to maintain a constant local concentration throughout the whole membrane. This has led to the concept of a “polymer-bound solvent.”5 If the membrane-forming polymer carries the solvating entity as an integral part of its structure, the protons are expected to migrate by a hopping mechanism in the electrochemical gradient across the fuel cell. Polymers bearing imidazole groups in the side chain5, 8 as well as polymers having pendant phosphonic acid groups have been examined recently.6, 19–21 In particular, poly(vinyl phosphonic acid) has been evaluated in great detail as a potential membrane-forming material. However, the phosphonic acid groups undergo water elimination to form anhydrides at the temperatures at which fuel cells are operated, making water management necessary once again. Polymers having imidazolium groups involved in proton transport do not give high enough mobilities to be really useful in the context of fuel cells.

In our search for more efficient PEM materials we found poly(alkylene biguanides) as potential proton-conducting materials. Following the seminal work of Rose et al.,22–24 these polymers can be readily obtained in their hydrochloride form by polycondensation of sodium dicyanamide with α,ω-alkylenediamine dihydrochlorides according to Scheme 1. The biguanide residue in the repeat unit of the polymer is a strong base with a pKa of ca. 13, while its conjugate with protons acts as a strong acid with a pKa of ca. 3.25 The protonated form is actually a mixture of tautomeric forms as shown in Scheme 2. This is not only the reason for the high acidity of the protonated form but also hints for high proton mobility in the polymers. The expectation is that the barrier for hopping of protons among various sites either intra- or intermolecularly will be low as a consequence of the high density of sites available for residence at a shallow potential-energy distribution among these sites. Moreover, the biguanide groups are expected to form a dense, 3D network of hydrogen bonds,26 which has been found to be favorable to create a matrix for proton diffusion.6, 19

Scheme 1.

Synthesis and molecular structure of poly(alkylene biguanides) HCl conjugates.

Scheme 2.

Possible tautomers of the protonated form of the biguanide moiety.

Poly(alkylene biguanides) and to some extent also poly(arylene biguanides) have been originally investigated for their antibacterial activity and have found widespread application in medical and veterinary practice.27 More specifically poly(hexamethylene biguanide) (PHMB, 1) (Scheme 1) is currently used in a wide variety of disinfection and preservation applications. Synthesis and characterization of PHMB have been extensively described by East et al.,28 while the more recent work of O'Malley et al. has demonstrated that PHMB has the expected end groups, namely a combination of amine, guanidine, and cyanoguanidine residues.29 PHMB has been described as a water-soluble, fiber-forming material, and its potential application as an ion-exchange material or complexant for heavy metal ions has been mentioned.30 Its stability against hydrolysis even at elevated temperatures in aqueous media was also noted.28 However, to the best of our knowledge the potential of this or related polymers as a proton-conducting material has not been investigated so far. As a consequence, we will describe in the following the synthesis and characterization of the poly(alkylene biguanides) 14 as proton-conducting materials (2 =  poly(tetramethylene biguanide) (PTMB), 3 = poly(ethylene biguanide) (PEB), and 4 = poly(methylene biguanide) (PMB)).

The poly(alkylene biguanides) 14 were synthesized essentially following the procedure described by Rose and Swain.22 They were obtained and purified by reprecipitation in hydrochloride form (Scheme 1). They were characterized by IR, 1H-, and 13C-NMR spectroscopies. The analytical data support the structure and agree with previous literature. All polymers were soluble in water or methanol to give viscous solutions. Our polymers did not show IR absorptions indicative of the presence of substituted melamine or amino-s-triazine structures that have been noted in earlier research with 1, albeit when it was subject to excessive thermal treatment in the course of synthesis by melt polycondensation.28 We have no indication of side reactions in the course of the synthesis and believe that the polymers have a linear chain structure. The absorption of the [BOND]C[TRIPLE BOND]N end groups was clearly visible at ca. 2170 cm−1 in all polymers. Polymers 14 were carefully dried and turned out to be amorphous.

Attempts to define conditions for determination of molecular weight (Mw) by size exclusion chromatography (SEC) or related techniques such as high-performance liquid chromatography (HPLC) were unsuccessful in our hands. The behavior of these polymers as polyelectrolytes in aqueous solution and the very strong dependence of solution properties on the presence of a neutral salt have been noted previously as the key factors for precise Mw-determination difficult.28 The presence of oligomers having the expected end groups (all in agreement with the results of ealier work29) were revealed by matrix-assisted laser-desorption ionization time-of-flight (MALDI-TOF) spectrometry, but a full analysis of the Mw-distribution remains a task for later work.

Differential scanning calorimetry (DSC) showed a glass transition temperature (Tg). The value of Tg was dependent on the length of the alkylene spacer m between the biguanide groups and was highest for 1 at 356 K and lowest for 4 at 303 K, 2 (343 K) and 3 (306 K) being intermediate.

Thermal gravimetric analysis (TGA) of 14, each dried in vacuo at 100 °C for 1 week prior to measurement, revealed appreciable thermal stability of the materials in agreement with the literature.28 The polymers started to decompose under nitrogen at a given heating rate at a temperature Td of 445 K for 4, 459 K for 3, 495 K for 1, and 501 K for 2 (Figure 1). The dependence of Td on m and the increasing thermal stability as this length increases is an indication that the biguanide residue is thermally rather stable but that the degradation is related to processes linked to the structure of the spacer. The stability of 1 against hydrolysis in boiling water has been noted previously.28

Figure 1.

TGA traces of the poly(alkylene biguanides) 14 (heating rate 10 K min−1 under N2 atmosphere).

The ion conductivity of the poly(alkylene biguanide) hydrochlorides was characterized by impedance spectroscopy following established procedures of the evaluation of the temperature and frequency dependence of the dielectric loss.31 The measurements were performed on thin films between stainless steel electrodes. The films were cast from solution in methanol and were investigated after removal of solvent and water by annealing in vacuo at 100 °C for typically 1 week. Direct-current (DC) conductivity, σ(ω,T), as a function of frequency, ω, and temperature, T, reaches very high values for polymers 3 and 4 at temperatures between 100 and 190 °C, that is, at temperatures well below the onset of thermal decomposition but in the range desired for application in PEMFCs. A plot of the experimental data in terms of (log σ) versus reciprocal temperature is shown in Figure 2. It reveals clearly that σ depends on temperature as described by the Williams–Landel–Ferry (WLF) formalism. In this formalism the temperature-dependence of a typical transport property, here σ, is described with regard to a reference temperature. Tg is conveniently used as the reference temperature:

equation image((1))
Figure 2.

WLF plot of the temperature-dependence of conductivity for polymers 14 in the dry state. The solid line through the points is the best fit to WLF formalism with parameters C1, C2, and Tg tabulated in Table 1. The Tg of each polymer is indicated by an asterisk. The inset shows the master curve for all four polymers.

The parameters C1 and C2 have to be evaluated by a best fit to the experimental data. The full lines through the experimental points in Figure 2 represent the best fits in terms of the WLF formalism; the respective Tgs of the polymers are indicated by asterisks. These were obtained from DSC data as mentioned above. The values of C1 and C2 for each of the four polymers are tabulated in Table 1.

Table 1. Characteristic thermal data of the synthesized proton conductors as well as WLF-fitting parameters for the temperature-dependent ionic conductivities of the poly(alkylene biguanides) 1, 2, 3, 4, and 5.
sampleTg [K] [a]Td [K] [b]log σ(Tg) [S cm−1]C1 [K−1]C2 [K]T0 [K] [c]
  • [a]

    Via DSC-determined Tg (invariant).

  • [b]

    Via TGA-determined Td (invariant).

  • [c]

    T0 = TgC2 (invariant).

  • [d]

    With respect to the melting point at 308 K.

1356495−9.259.7993.24263
2343501−9.3210.5199.75243
3306459−9.3010.7876.49230
4303445−9.8011.4779.53224
5308 [d]444−6.01 [d]5.7470.72237 [d]

The validity of the WLF formalism allows the creation of a master curve. It describes the behavior of all four polymers simultaneously. For that purpose the conductivity is rescaled with regard to the conductivity at the reference temperature Tg. Log(σ(T)/σ(Tg)) is then plotted against the temperature difference between the actual temperature and a reference temperature. The master curve thus obtained is shown in the inset in Figure 2. We note that the best fit to a master curve was obtained when the rescaled conductivity was plotted against (T–T0), with T0 = TgC2 (Vogel temperature).32, 33 The finding of a common master curve hints that the same transport mechanism is effective in the four different polymers, while the WLF-type dependence on temperature indicates that the ion transport is controlled by segmental relaxation processes. These are at the origin of free volume diffusion in the polymer matrix.

The function of the biguanide moiety as the origin of ion conductivity can also be demonstrated by a low Mw model. 1,1'-Hexamethylene-bis-5-(4-chlorphenyl)biguanide (HMCB) is a commercially available substance that finds application as a biocide and antibacterial agent.34, 35 It has a melting point of 135 °C that is lowered to 35 °C when converted to its trifluoroacetic acid (TFA) conjugate: HMCB × 2 TFA (5). The pKa of the latter is ca. 3 while the pure HMCB shows a pKa of ca. 13 in aqueous solution. The pure HMCB model compound exhibits a conductivity of 5.47 × 10−5 S cm−1 at 170 °C whereas the TFA conjugate reaches 3.59 × 10−3 S cm−1 at this temperature. The structure of this doubly charged model compound and its 1H-NMR spectrum in DMSO-d6 are displayed in Figure 3. Only one of the possible tautomers is depicted while the inspection of the spectrum clearly reveals the presence of the other tautomers from the splitting of the resonance signals in the N–H region. Very similar NMR spectra of polymers 14 in the bulk state have been obtained, however, the temperature evolution of the spectra that reveals the temperature- (and solvent-) dependence of the population of the different tautomers goes beyond this Communication. Here we concentrate on the description of the ionic conductivity at temperatures well above 100 °C.

Figure 3.

1H-NMR spectrum of model compound 5 in DMSO-d6. Only one of the possible tautomers is shown in this Figure.

In this context it was worthwhile to investigate the effect of water on the conductivity of the polymer acid conjugates in the form of hydrochlorides. As a typical example, the hydrochloride of 3 reached a magnitude of conductivity of 4.7 × 10−3 S cm−1 at 190 °C and showed σ = 2.3 × 10−3 S cm−1 at 153 °C under dry conditions. When in contact with water vapor at a partial pressure of 1 atm the conductivity quickly increased and stabilized at 9.8 × 10−3 S cm−1 at 153 °C. The same experiment at 121 °C gave 1.1 × 10−2 S cm−1 for the sample in contact with water vapor while the dry polymer showed 4.4 × 10−4 S cm−1. When the water was pumped off, the original conductivities of the dry material were reestablished. These results require further experimental studies to determine the temperature- and pressure-dependent degree of water sorption of the polymers as was done for other polymers containing phosphonic acid residues.19, 36 This together with pulsed-gradient NMR studies would allow the two major effects of water onto the magnitude of conductivity to be separated: i) water acts as a plasticizer, reduces Tg, and enhances free volume relaxation processes; ii) water acts as a source of further mobile ionic species, that is, hydronium ions. The latter would contribute to the conduction processes in their own way. At this point in our investigation we cannot resolve these factors properly. However, we believe that the superior behavior of the poly(alkylene biguanides) as expressed by Figure 2 deserves more work to bring out the various contributions to ionic conductivity in the context of potential applications in PEMFCs. A further comment concerns the nature of the ionic species that contributes to the observed conductivities. Impedance spectroscopy probes all mobile ions and since it works with blocking electrodes we cannot differentiate between positive- and negative-charge carriers. However, with reference to our earlier work on anion conductivity in various ionenes37 we hypothesize that the contribution of negatively charged ions is only a small fraction of that of the positively charged ones, namely protons and their conjugates. This point deserves further study as well.

Poly(alkylene biguanides) are excellent ion conductors at temperatures well above 100 °C and up to 200 °C in their conjugate acid forms (as with HCl). They also show a remarkable thermal stability. Although such polymers are well-known and currently find application as antibacterial agents and biocides in medical and veterinary practice, their potential as a novel class of high-temperature ion conductors used for the construction of membranes in fuel cells is demonstrated here for the first time. Ion conductivity as probed by impedance spectroscopy follows WLF behavior with regard to temperature dependence. This reveals that free volume relaxation processes dominate and control the motion of the protons, which are assumed to be the main source of ionic conductivity. The biguanide residues in the repeat unit exist as tautomers in their protonated form and thus provide a large density of sites separated in a shallow energy landscape. Hopping of protons occurs among these sites. This and their ability to form a dense 3D network of hydrogen bonds creates a path for proton transport as is depicted in Scheme 3. Here, the interaction of three individual biguanide moieties is shown in a 2D projection, where a section out of the suggested hydrogen-bond network is depicted with protons placed as constituents of the network. The arrows indicate the type of segmental motion that may be associated with proton transport in between these segments of the network. It is this segmental motion that gives rise to the WLF temperature-dependence of conductivity that is experimentally observed. The very large values of up to 5 × 10−3 S cm−1 at 190 °C under dry conditions can be enhanced by at least one order of magnitude if in contact with water vapor (under pressure). This and the interesting structure–property relationships in the poly(alkylene biguanides) make further research in this class of materials worthwhile.

Scheme 3.

Hydrogen-bond network and proton hopping in bulk poly(alkylene biguanides). Segmental motions necessary to reorganize the structure after each hopping step are indicated by the arrows. A possible hopping trajectory is indicated by the shadowed path linking the hydrogen-bonded segments.

Experimental

Materials

Chemicals were received from Acros, BASF, Aldrich, Merck, and Arch Chemicals in p.a. quality and used without further purification. 1-Butanol was dried and distilled under argon over sodium. Methanol, diethyl ether, and (low-boiling) petrol ether were dried with activated molecular sieves (4 Å).

Measurements

The 1H-NMR and 13C-NMR spectra were recorded on Bruker AC-300 and Bruker Spectrospin 250 spectrometers at room temperature. For FTIR measurements a Perkin Elmer Paragon 1000 at a resolution of 4 cm−1 was used. TGA was performed under nitrogen at a heating rate of 10 K min−1 using a TGA/SDTA-851 (Mettler-Toledo). DSC was carried out on a Mettler–Toledo DSC-30 under nitrogen at a heating rate of 10 K min−1. Impedance spectroscopy (IS) was recorded as a function of temperature using a Solartron SI 1260 impedance/gain phase analyzer with a high-resolution dielectric converter (Alpha High Resolution Dielectric Analyzer; Novocontrol) in the range from 10−2–107 Hz. The measurements were performed using stainless-steel electrodes with a temperature-controlled cryostat under nitrogen (Novocontrol).

Preparation of the Films

Solid homopolymers were dissolved in dry methanol or ethanol. These solutions were dropped onto coin-like stainless-steel electrodes with a diameter ranging from 10–20 mm to form thin films about 1-mm thick after complete evaporation of the organic solvent at 100 °C in vacuo. Finally the counter electrodes were placed upon the thin films. Highly viscous homopolymers and the HMCB × 2 TFA conjugate were placed between stainless-steel electrodes that were kept at a fixed distance of about 0.1 mm by means of a Teflon spacer.

Synthesis of the Poly(alkylene biguanides)

Methylene- and tetramethylenediamine dihydrochloride are commercially available and were used as obtained. Ethylene- and hexamethylenediamine dihydrochloride were precipitated from previously freshly distilled ethylene- and hexamethylenediamine by concentrated HCl in a molar ratio of 1:2 under cooling. For the polycondensation 5 g of the respective alkylenediamine dihydrochloride and equimolar amounts of sodium dicyanamide were mixed and suspended together in 1-butanol. The suspension was heated at ca. 134 °C under reflux for 3 days. Thereafter sodium chloride and the remaining educt were removed from the reaction mixture by stepwise extraction, each time with water (5 mL), such that the major part of the product remained in the organic phase. The concentrated organic phase was then dissolved in water (10 mL) and re-extracted with diethyl ether (5 mL) in order to remove residual educt and butanol. Finally the product was treated with cold petrol ether (50 mL) whereupon a colorless solid formed. After filtration and drying at 100 °C in vacuo, 3.44 g (13 mmol; 51% yield) 1, 3.21 g (17 mmol; 54% yield) 2, 2.39 g (15 mmol; 39% yield) 3, and 2.26 g (15 mmol; 36% yield) 4, all of them colorless polymers, were obtained.

1: 1H NMR (250 MHz, DMSO-d6, 25 °C, TMS, δ): 1.28 (sb, 2(2H)n; γ-CH2), 1.43 (sb, 2(2H)n; β-CH2), 3.08 (sb, 2(2H)n; α-CH2N), 6.90–7.80 (mb, 6(1H)n; NH); 13C NMR (250 MHz, DMSO-d6, 25 °C, TMS, δ): 25.3 (γ-CH2), 28.5 (β-CH2), 118.0 (α-CH2N), 155.4−160.9 (C[DOUBLE BOND]N); IR (KBr): ν = 3311 (s), 3249 (m; ν([BOND]N[BOND]H)), 2929 (m; ν(C[BOND]H)), 2173 (s; ν(C[TRIPLE BOND]N)), 1638 (s), 1542 (m; δ([BOND]N[BOND]H)), 1459 (m), 1438 (m; δ(CH2)), 1351 (s), 1144 (w), 1029 (m; ν(C[BOND]N)), 723 cm−1 (m; δ(CH2).

3: 1H NMR (250 MHz, DMSO-d6, 25 °C, TMS, δ): 3.17–3.52 (mb, 2(2H)n; α-CH2N), 6.77–8.46 (mb, 6(1H)n; NH); 13C NMR (250 MHz, DMSO-d6, 25 °C, TMS, δ): (42.0, 43.6), 117.7 (α-CH2N(H+)), 157.0–161.2 (C[DOUBLE BOND]N); IR (KBr): ν = 3300 (s), 3204 (m; ν([BOND]N[BOND]H)), 2921 (w; ν(C[BOND]H)), 2177 (s; ν(C[TRIPLE BOND]N)), 1630 (s), 1599 (m), 1560 (s; δ([BOND]N[BOND]H)), 1441 (m), 1420 (w; δ(CH2)), 1384 (s), 1297 (m), 1150 (w), 1076 (m; ν(C[BOND]N)), 779 cm−1 (m; δ(CH2).

Acknowledgements

The authors gratefully acknowledge the gift of PHMB by Arch Chemicals GmbH, Ratingen, Germany. Assistance in synthetic work by A. Manhart is gratefully acknowledged as well as help in gaining impedance data by Dr. A. Kaltbeitzel and C. Sieber.