Permethyl Cobaltocenium (Cp^*₂Co⁺) as an Ultra-Stable Cation for Polymer Hydroxide-Exchange Membranes

Abstract

Hydroxide (OH⁻)-exchange membranes (HEMs) are important polymer electrolytes enabling the use of affordable and earth-abundant electrocatalysts for electrochemical energy-conversion devices such as HEM fuel cells, HEM electrolyzers, and HEM solar hydrogen generators. Many HEM cations exist, featuring desirable properties, but new cations are still needed to increase chemical stability at elevated temperatures. Here we introduce the permethyl cobaltocenium [(C₅Me₅)₂Co(III)⁺ or Cp^*₂Co⁺] as an ultra-stable organic cation for polymer HEMs. Compared with the parent cobaltocenium [(C₅H₅)₂Co(III)⁺ or Cp₂Co⁺], Cp^*₂Co⁺ has substantially higher stability and basicity. With polysulfone as an example, we demonstrated the feasibility of covalently linking Cp^*₂Co⁺ cation to polymer backbone and prepared Cp^*₂Co⁺-functionalized membranes as well. The new cation may be useful in designing more durable HEM electrochemical devices.

Polymer hydroxide (OH⁻)-exchange membranes (HEMs) are attractive electrolytes for electrochemical energy conversion devices such as fuel cells¹,²,³,⁴,⁵,⁶,⁷,⁸,⁹,¹⁰,¹¹,¹²,¹³,¹⁴,¹⁵,¹⁶,¹⁷,¹⁸, electrolyzers¹⁹,²⁰, and solar hydrogen generators²¹, largely due to their intrinsic compatibility with non-precious-metal catalysts²,²² and superior CO₂ tolerance²³ compared to liquid alkaline electrolytes²⁴. The active hydroxide-conducting component of an HEM is a cation that is covalently linked to a polymer backbone. Organic cations based on nitrogen [ammonium²⁵,²⁶, pyridinium²⁷, guanidinium²⁸,²⁹, imidazolium⁶,³⁰], phosphorus [phosphonium³,⁸], sulfur [sulfonium³¹], and ruthenium [bis(terpyridine)ruthenium⁷] have been introduced, featuring specific HEM properties including improved solubility³, enhanced thermal stability³¹, and increased basicity³². New cations with chemical stability at elevated temperatures (>80 °C³³,³⁴) are still desired to reduce CO₂ poisoning, increase catalyst activity, and improve heat management in HEM electrochemical devices.

Alkali metal cations (e.g., Li⁺, Na⁺, and K⁺) provide the highest OH⁻ conductivity and show excellent stability, but at present cannot be covalently tethered to a polymer backbone for HEM applications³⁵. Organic bis(cyclopentadienyl) metallocenium cations based on VIIIB family metals [(C₅H₅)₂M(III)⁺ or Cp₂M(III)⁺, M = Co, Rh, Ir, or Mt] satisfy the 18-valence electron stability rule and resemble alkali metal cations: They bear one unit of positive charge, may be precipitated by the addition of excess anions, and with hydroxide as counter-ion absorb CO₂ and water from ambient air³⁶. In particular, cobalt is the smallest atom in the VIIIB family and forms the strongest metal-ring bonds, resulting in the most stable metallocenium cation, Cp₂Co⁺ (cobaltocenium)³⁷. While Cp₂Co⁺ has been used in strongly basic anion-exchange resins³⁸,³⁹ and water-soluble redox-active oligomers/polymers⁴⁰,⁴¹, none of these compounds appears suitable for HEM applications. In addition, the stability of Cp₂Co⁺ may be still limited due to its unsubstituted Cp rings.

Here we introduce the concept of using permethyl cobaltocenium [(C₅Me₅)₂Co(III)⁺ or Cp^*₂Co⁺, Fig. 1a,b] as a highly stable cation for polymer HEMs that are required in designing able and durable electrochemical devices.

Figure 1. Structure of Cp^*₂Co⁺ cation and Cp^*₂Co⁺-functionalized polysulfone (Cp^*₂Co⁺-PSf).

(a) Chemical structure of Cp^*₂Co⁺. (b) Molecular structure of Cp^*₂Co⁺. (c) Chemical structure of Cp^*₂Co⁺-PSf. (d) Molecular structure of Cp^*₂Co⁺-PSf (one repeat unit of polysulfone shown, predicted by the software MOPAC and drawn in Jmol, version 13.0).

Results

Cation stability: Cp^*₂Co⁺ vs. Cp₂Co⁺

Complete methylation of the Cp ring in Cp^*₂Co⁺ enhances electron donation to the metal center, strengthening the metal-ring bond and delocalizing positive charge away from the metal center⁴². Calculations confirm that Cp^*₂Co⁺ has a lower heat of formation (ΔH_f) than Cp₂Co⁺, suggesting stronger bonding: 499 vs. 775 kJ mol⁻¹ (a 36% reduction), as predicted by the semi-empirical quantum chemistry software MOPAC2012 (Table 1). The same calculation confirms that the charge on cobalt (δ_Co) is lower in Cp^*₂Co⁺ than in Cp₂Co⁺: +0.988 vs. +1.058 e (a 6.6% reduction). (Note that the overall system charge remains exactly +1 for both cations, regardless of delocalization.) This reduced charge is consistent with the substantial negative shift (ca. 600 mV) in redox potential (φ) observed for Cp^*₂Co(III)⁺/Cp^*₂Co(II) vs. Cp₂Co(III)⁺/Cp₂Co(II) (−1.24 V vs. −0.63 V, referring to the standard hydrogen electrode, SHE, with CH₂Cl₂ as solvent⁴³).

Table 1. Comparison between Cp ^* ₂ Co ⁺ and Cp ₂ Co ⁺ .

Cation	ΔHfa (kJ mol−1, cation)	δCob (e, or 1.602 × 10−19 C)	ϕc (V vs. SHE)	θd (degree)	pKbe (for cation hydroxide)
Cp*2Co+	499	+0.988	−1.2443	40.3–49.6	2.7
Cp2Co+	775f	+1.058	−0.6343	72.4	5.4

^aΔH_f: heat of formation, predicted by the software MOPAC2012 (Stewart Computational Chemistry).

^bδ_Co: partial charge at the cobalt atom, predicated by the same software.

^cϕ: formal reductive potential of cation in CH₂Cl₂.

^dθ: accessible angle formed by cobalt and the edges of the circumcircles of hydrogen atoms in cation (Fig. S1).

^eK_b: base dissociation constant for cation hydroxide (Fig. S6).

^fThis calculated value is close to the measured value reported in literature (823 kJ mol^{−1 68}).

The reduced positive charge on the metal center makes Cp^*₂Co⁺ less susceptible to nucleophilic attack by hydroxide, which is, in general, the first step in HEM degradation. In addition to the electronic effect, steric hindrance⁴² from the methyl groups in the Cp^* ligands may also further protect Cp^*₂Co⁺ from hydroxide attack. The difference in steric hindrance between Cp^*₂Co⁺ and Cp₂Co⁺ can be quantified⁶ by comparing the accessible angle (θ) that is formed by the cobalt center and the edges of circumcircle of hydrogen atoms (Fig. S1). Cp₂Co⁺ has an accessible angle of 72.4° while Cp^*₂Co⁺ has much smaller accessible angles of 40.3–49.6° (Table 1). The smaller accessible angle reduces the chance for the cobalt atom to be attacked.

Indeed, a high-temperature alkaline stability test (at 140 °C in 1 M NaOD/D₂O) showed that Cp^*₂Co⁺ is significantly more stable than Cp₂Co⁺. After six weeks (1,000 hours) only 8.5% of the initial Cp^*₂Co⁺ had been found to degrade (¹³C NMR, Fig. S2) whereas Cp₂Co⁺ degraded completely after only one week (Fig. S3). Further, Cp^*₂Co⁺ showed no change in UV-vis absorption after the stability test (Fig. S4). For comparison, a typical ammonium cation (trimethyl benzylammonium) degraded by 18% in 24 hours (Fig. S5). Cp^*₂Co⁺ demonstrates a level of stability that has not been achieved by any known HEM cations (Fig. 2).

Figure 2. Alkaline stability of Cp^*₂Co⁺ and other reported cations.

Test conditions: 1 M KOD or NaOD in D₂O, 20% degradation threshold on ¹H NMR basis (unless otherwise noted). Cobaltocenium: Cp^*₂Co⁺ (¹³C NMR spectroscopy, this work). Ammonium: benzyl-trimethylammonium (btmAm) [80 °C in 1 M NaOD/(D₂O+CD₃OD)⁸ and 100 °C in 1 M NaOD/D₂O, this work]. Imidazolium: benzyl-1-methyl-limidazolium (bmIm)⁶⁷, 1,3-dimethyl-2-phenyl-benzimidazolium (dmpBIm) (0.3 M KOH)⁶, 1,3-dimethyl-2-(2,4,6-trimethylphenyl)-benzimidazolium (dmtmpBIm) (1.3 M KOH)⁶. Guanidinium: benzyl-pentamethylguanidium (bpmGu) (this work). Pyridinium: benzylpyridinium (bPy) (this work). Phosphonium: tetrakis(dialkylamino)phosphonium (tkdaaPh) [1 M NaOD/(D₂O+CD₃OD)]⁸. Sulfonium: (4-methoxyphenyl)-diphenylsulfonium (mopdpSu)³¹. Ruthenium: bis(terpyridinine)ruthenium (ttpRu) (UV-vis spectroscopy). The chemical structures of those cations are shown in Table S3. ¹H NMR spectra of btmAm, bpmGu, and bPy are not shown.

The reduced charge on cobalt also improves the basicity of the cation hydroxide, since weakened cation-anion interaction favors dissociation. Indeed, the base dissociation constant (K_b) is much larger for Cp^*₂Co⁺OH⁻ than for Cp₂Co⁺OH⁻: pK_b = 2.7 vs. 5.4 (a 500-fold increase in K_b, Table 1), as measured by correlating OH⁻ concentration with total organic base concentration in aqueous solution (Table S1 and Fig. S6). Improved basicity is expected to increase the hydroxide conductivity of the corresponding HEMs.

Superior stability and basicity suggest that Cp^*₂Co⁺ is a much more desirable cation than Cp₂Co⁺ for HEM applications.

Synthetic strategy for Cp^* ₂Co⁺-PSf polymer

Cp^*₂Co⁺-PSf was synthesized by a diamine-bridge strategy (Fig. S7). The feasibility of this approach was established by a series of small-molecule model reactions (bromination, Fig. S8; anion exchange, Fig. S9; amination, Fig. S10; and cation incorporation, Fig. S11), in which benzyl chloride functioned as a surrogate for the halomethylated polymer. The Cp^*₂Co⁺ cation was first brominated to introduce a single halomethyl group. In addition to ¹H NMR spectroscopy (Fig. S8), the bromination step was further confirmed by ¹³C NMR (Fig. S12) and mass spectroscopy (Fig. S13, no sign of multiple bromination). In parallel, one amine group from hexamethylenediamine (HMDA) was linked to benzyl chloride by amination (leaving the other amine group intact). Then, the remaining unreacted amine group from HMDA was used to link the brominated Cp^*₂Co⁺. Reaction conditions for all steps were optimized (yield >95%, ¹H NMR spectroscopy).

In a similar manner, Cp^*₂Co⁺-PSf was synthesized (Figs. S14 and S15) and then membranes were prepared. Unlike the small-molecule model reaction, proper reaction conditions are critical for amination of polysulfone. Firstly, a low reaction temperature (e.g., 20 °C) avoids the ammonolysis that leads to polysulfone depolymerization (observed at 40 °C and higher). Secondly, a large excess of amine (e.g., 20 equiv.) avoids cross-linking in which one amine molecule reacts with two chloromethyl groups. Once cross-linked, the polymer becomes insoluble, and is no longer suitable for further reactions.

Note that the hydrophobic⁴⁴,⁴⁵ PF₆⁻ anion in Br-Cp^*₂Co⁺PF₆⁻ must be exchanged for a hydrophilic anion (e.g., Cl⁻) prior to linking the cation to the backbone, due to the difficulty of exchanging PF₆⁻ for the ultimately desired OH⁻ in polyelectrolytes⁷. The counter-ion was later exchanged from Cl⁻ to OH⁻ with 1 M KOH. The Cp^*₂Co⁺-PSf membranes were found be to flexible, uniform in thickness, and transparent (Fig. 3a).

Figure 3. Characterization of Cp^*₂Co⁺-PSf membrane.

(a) Photograph (2” x 2”, 100 μm thick). (b) Dynamic mechanical analysis (DMA) test curve (ambient humidity and temperature, 10 mm min⁻¹ cross-head speed). (c) Thermal gravimetric analysis (TGA) and derivative thermal gravimetric (DTG) curves (10 °C min⁻¹, nitrogen atmosphere).

Microstructure and properties of Cp^* ₂Co⁺-PSf membranes

Due to the presence of cobalt, high-contrast transmission electron microscopy (TEM) images were obtained without the need to employ a dye anion (e.g., PdCl₄^{2− 46} or WO₄^{2− 16}), directly revealing detailed features of the membrane microstructure (Fig. S16a). Overall, the hydrophilic (dark) and hydrophobic (light) domains are homogenously interspersed, which is essential to simultaneously provide ionic conduction and mechanical robustness¹⁶. The hydrophilic domains show an average size of 15 nm, similar to those in typical ammonium-based membranes (10–30 nm) with the same polysulfone backbone and similar ion exchange capacity (IEC, 1.14 mmol/g)⁴⁶. The presence of cobalt was also confirmed by energy dispersive X-ray (EDX) spectroscopy during imaging (Fig. S16b). High tensile strength (40 MPa with 10% elongation at break, Fig. 3b) shows that the membrane is strong and robust. Under similar test conditions, the commercial proton exchange membrane (PEM) Nafion^® 212 is half as strong (around 20 MPa⁴⁷).

Two membranes of different degrees of functionalization (DFs, 100% and 123%) were characterized (Table 2). Both hydroxide conductivity and water uptake increase with ion exchange capacity (IEC), as with HEMs based on other cations. Note that the measured IECs are very close to the theoretical ones: 1.04 vs. 1.09 mmol g⁻¹ for Cp^*₂Co⁺-PSf100 membrane; 1.16 vs. 1.20 mmol g⁻¹ for Cp^*₂Co⁺-PSf123 membrane. Such an agreement indicates the Cp^*₂Co⁺ cations in Cp^*₂Co⁺-PSf membranes are fully functional. For comparison, HEMs with similar theoretical IEC and similar backbone (when available) but different cations are summarized in Table 2. The Cp^*₂Co⁺-PSf membranes are similar to other membranes in terms of hydroxide conductivity (10–22 vs. 0.19–45 mS cm⁻¹) and water uptake (41%–68% vs. 8.2%–240%) in deionized water at room temperature, as well as IEC-normalized hydroxide conductivity (HC_IEC, 9.2–18 vs. 0.15–38 mS∙g cm⁻¹ mmol⁻¹), suggesting similar cation basicity and hydroxide conduction efficiency. As expected, hydroxide conductivity of Cp^*₂Co⁺-PSf membranes increases with temperature. E.g., the hydroxide conductivity of Cp^*₂Co⁺-PSf123 membranes reached 49 and 64 mS cm⁻¹ at 60 and 80 °C, respectively.

Table 2. Membrane properties of different cation-based HEMs (polysulfone backbone when available, chemical structures of cations shown in Table S3).

Cation	HEMa	IECb (mmol/g)	HCc (mS/cm)	HCIECd (mS∙g/cm∙mmol)	WUe (%)	TODf (°C)	Ref.
Cobaltocenium	Cp*2Co+-PSf123	1.20	22	18	68	305	This work
	Cp*2Co+-PSff100	1.09	10	9.2	41		This work
Ammonium	btmAm-PSf58, 64g	1.08, 1.18	11, 19	10, 19	150, 240	150 (air)	2,46
	bteAm-PEKh	N/A	14	—	17	200	69
	ptmAm-PSf70, 100i	1.01, 1.32	15, 26	15, 20	8.7, 12	N/A	70
	badmAm-PPO20, 30j	1.08, 1.48	7, 14	6.5, 9.5	8.2, 14	N/A	71
	dbdmAm-PSf11	1.21	11	8.8	N/A	175	72
	bdabcoAm-PSfl	N/A	22	—	N/A	N/A	73
	bdmpAm-PSf67m	1.23	11 (50 °C)	7.3	50	N/A	74
	pdmAm-PDDA39n	1.24	0.19	0.15	21	285 (Cl−)	75
	mamAm-PBI50, 84o	1.15, 1.68	0.6, 2.9	0.52, 1.7	28, 50	160	76
	btamAm-PStp	N/A	21	—	21	150	49
Imidazolium	bmIm-PSf66, 80q	1.28, 1.39	9, 16	7.0, 11	25, 8.5	140, 258	77,78
	amIm-PStAE20, 30r	0.80, 1.20	25, 31	31, 26	87, 116	210	79
	bdmIm-PETFEs	1.70	17 (HCO3−)	10	32 (Cl−)	N/A	80
	admIm-PFS200t	1.04	24	23	17	220	56
	aamIm-PStAE60, 80u	1.12, 1.50	5, 12	4.5, 8	41, 65	220	57
	btmtmopIm-PPO39v	1.36	34	25	125	N/A	81
	dmpBIm-PBIw	N/A	N/A	—	N/A	230	82
	dmtmpBIm-PBI64, 80x	1.00, 1.50	13.2, 10.1	13.2, 6.7	82, 119	N/A	6
Guanidinium	bpmGu-bpPSf40, 60y	0.86, 1.21	5, 12	5.8, 9.9	12, 17	165	28
	apmGu-PPS100z	1.39	22	16	32	200	55
	ppmGu-fPSf85aa	1.03	21	20	10	N/A	29
Pyridinium	aPy-PVPSt50ab	1.76	0.6	0.76	30	230 (Br−)	27
	bPy-PSt34ac	1.29	14	11	N/A	130 (DSC)	83
Phosphonium	bttmopPh-PSf124, 152ad	1.09, 1.17	27, 45	25, 38	70, 137	187	3,32,48
	tkdaaPh-PCoe17ae	0.93	22	24	52	N/A	8
Sulfonium	mopdpSu-PSf46af	0.69	15	22	27	242	31
Ruthenium	ttpRu-PN9, 17ag	1.00, 1.40	14, 29	14, 21	30, 126	N/A	7

^aThe number at the end of the name indicates degree of functionalization.

^bIon exchange capacity based on theoretical calculation.

^cHydroxide conductivity in deionized water at room temperature.

^dIEC-normalized hydroxide conductivity.

^eWater uptake.

^fOnset decomposition temperature from TGA test (N₂ atmosphere, 10 °C min⁻¹).

^gBenzyl-trimethylammonium (btmAm).

^hBenzyl-triethylammonium (bteAm), poly(ether ketone) (PEK).

ⁱPhenyl-trimethylammonium (ptmAm).

^jBenzyl-alkyl-dimethylammonium (badmAm), poly(phenylene oxide) (PPO).

^kDibenzyl-dimethylammonium (dbdmAm).

^lBenzyl-1,4-diazabicyclo-[2.2.2]-octane-ammonium (bdabcoAm).

^mBenzyl-(1,4-dimethyl)piperazine-ammonium (bdmpAm).

ⁿPyrrolidine-dimethylammonium (pdmAm), poly(diallyldimethylammonium) (PDDA).

^oMorpholine-alkyl-methylammonium (mamAm), polybenzimidazole (PBI).

^pBenzyl-1,3,5-triazine-methylammonium (btamAm), polystyrene (PSt).

^qBenzyl-1-methyl-imidazolium (bmIm).

^rAlkyl-1-methyl-imidazolium (amIm), crosslinked poly(styrene acrylonitrile ethylene) (PStAE).

^sBenzyl-1,2-dimethyl-imidazolium (bdmIm), grafted poly(ethylene tetrafluoroethylene) (PETFE) backbone.

^tAlkyl-1,2-dimethyl-imidazolium (admIm), poly(fluorene sulfone) (PFS).

^uAlkyl-1-alkyl-2-methyl-imidazolium (aamIm).

^vBenzyl-1,4,5-trimethyl-2-(2,4,6-trimethoxyphenyl)-limidazolium (btmtmopIm).

^w1,3-Dimethyl-2-phenyl-benzimidazolium (dmpBIm).

^x1,3-Dimethyl-2-(2,4,6-Trimethylphenyl)-benzimidazolium (dmtmpBIm).

^yBenzyl-pentamethylguanidinium (bpmGu), biphenylene-type polysulfone (bpPSf).

^zAlkyl-pentamethylguanidinium (apmGu), poly(phenolphthalein sulfone) (PPS).

^aaPhenyl-pentamethylguanidinium (ppmGu), fluorinated polysulfone (fPSf).

^abAlkylpyridinium (aPy), poly(vinylpyridine-styrene) (PVPSt).

^acBenzylpyridinium (bPy), crosslinked backbone.

^adBenzyl-tris(2,4,6-trimethoxyphenyl)-phosphonium (bttmopPh).

^aeTetrakis(dialkylamino)phosphonium (tkdaaPh), polycyclooctene (PCoe).

^af(4-Methoxyphenyl)-diphenylsulfonium (mopdpSu).

^agBis(terpyridinine)ruthenium (btpRu), polynorbornene (PN).

Thermal and alkaline stability of Cp^* ₂Co⁺-PSf membranes

The membranes showed very high thermal stability, consistent with the aforementioned stability of the Cp^*₂Co⁺ cation. The onset of decomposition temperature of 305 °C (Fig. 3c) is the highest among all reported HEMs: e.g., 60 °C higher than a sulfonium-functionalized PSf³¹, 120 °C higher than a phosphonium-functionalized PSf⁴⁸, or 150 °C higher than ammonium-functionalized PSfs⁴⁶,⁴⁹. Note that the thermal gravimetric analysis (TGA) data only reflect short-term thermal stability and are only useful for comparison under similar test conditions.

The Cp^*₂Co⁺-PSf membranes also showed improved alkaline stability: At 80 °C in 1 M KOH, the IEC loss of Cp^*₂Co⁺-PSf membranes was 18% and 27% of the initial value after 1,000 and 2,000 hours, respectively (Fig. S17). After the 1,000-hour stability test, solid-state ¹³C NMR spectroscopy confirmed that Cp^*₂Co⁺ functional groups remained almost unchanged (Fig. S18), but there were clear signs of PSf backbone scission. This result is consistent with that of a recent study⁵⁰, and it further highlights the need for more stable backbones in developing next-generation highly durable HEMs.

Under a more aggressive test at 100 °C for 2,000 hours, the IEC lost was about 50% for Cp^*₂Co⁺-PSf membranes. Considering that higher test temperature leads to shorter membrane lifetime, the membrane lifetime is plotted against the test temperature (Fig. S19). With 20% loss of initial IEC as the failure criterion, Cp^*₂Co⁺-PSf membranes are more stable than all other cation-based HEMs reported to date.

Discussion

Polysulfone was used as the polymer backbone in this work primarily to demonstrate linking Cp^*₂Co⁺ to polymer and to compare Cp^*₂Co⁺ to other cations reported. However, polysulfone backbone is sensitive for HEM degradation⁵⁰,⁵¹, and more stable polymer backbones are needed to better match the stable Cp^*₂Co⁺ cation. Very recently, alternative polymer backbones, such as polystyrene⁵² and poly(phenylene)⁵³, have been shown to have better stability than polysulfone in alkaline media. In principle, the synthesis reported here may be modified to employ those more stable polymer backbones. The incorporation of Cp^*₂Co⁺ cation to more stable polymer backbones and the synthesis and stability of their resulting membranes warrant important future research.

In the amination step, the chain length of the diamine linker between the polymer backbone and the Cp^*₂Co⁺ cation was found to affect membrane flexibility, likely resulting from the incompatibility between the rigid hydrophobic polymer backbone and the bulky hydrophilic cation. Increasing the chain length (e.g., ethylene⁵⁴, propylene⁵⁴,⁵⁵, and hexamethylene⁵⁶,⁵⁷,⁵⁸) between cation and backbone helps alleviate this incompatibility. Indeed, HMDA, with six carbon atoms and two nitrogen atoms, was found to be a good choice for preparing flexible, robust membranes.

In the cation incorporation step, although Br-Cp^*₂Co⁺PF₆⁻ could react at two possible amine sites — either close to the polymer backbone (the basal secondary amine) and or at the end of side chain (the terminal primary amine) (Fig. S7) — a reaction was only observed at the terminal site, even for the less-hindered model molecule (Fig. S10). Such good selectivity results from strong steric hindrance in the bulky Br-Cp^*₂Co⁺PF₆⁻ molecule and higher nucleophilicity of the terminal amine group (pK_a of amines’ conjugate acids: 10.21 and 10.21 for the basal amine vs. 9.26 and 9.93 for the terminal amine, of HMDA-aminated chloromethylated polysulfone polymer and HMDA-aminated benzyl chloride small molecule, respectively; calculated by the MarvinSketch software, Table S2).

During the model cation durability test, ¹³C instead of ¹H NMR spectroscopy was used to monitor stability because both Cp^*₂Co⁺ and Cp₂Co⁺ undergo rapid H-D isotopic exchange upon exposure to alkaline media, rendering ¹H spectroscopy inappropriate for accessing stability⁵². Figs. S20 and S21 show the degree of H-D isotopic exchange is 78% and 29% for Cp^*₂Co⁺ and Cp₂Co⁺, respectively, under the same test conditions (60 °C, 40% KOH/D₂O in methanol, 30 min). This difference in degree of exchange is consistent with the observation from the literature that methyl protons are slightly more acidic than ring protons in cobaltocenium [pKa in methanol: 22.7 for methyl protons of 1,1’-dimethyl-cobaltocenium, (1-Me-Cp)₂Co⁺, vs. 23.9 for ring protons of Cp₂Co⁺]⁵⁹,⁶⁰.

Table 2 summarizes the HEM samples with similar backbone and IECs for the purpose of comparing different cations. It is noted that HEMs with very high hydroxide conductivity have been reported. In general, high conductivity can be achieved by providing either high IEC or high efficiency of hydroxide conduction. With high IECs (> 2.0 mmol g⁻¹), HEMs exhibited high hydroxide conductivities (at ~20 °C unless otherwise noted) such as 43, 50, 65, 67, and 68.7 mS cm⁻¹ for a badmAm-based PPO (2.75 mmol g⁻¹)⁹, a btmAm-based fluorenyl-containing poly(ether sulfone ketone) (2.54 mmol g⁻¹, 30 °C)⁶¹, a btmAm-based di-fluorinated bpPSf (2.62 mmol g⁻¹)⁶², a bpmGu-based bpPSf (2.15 mmol g⁻¹)²⁸, and a badmAm-based cross-linked polyalkylene (2.3 mmol g⁻¹)⁵, respectively. High conductivities can also be realized with relatively low IECs via methods that likely improve the efficiency of hydroxide conduction. Examples include the use of architecture of aliphatic side chain (45 mS cm⁻¹, 30 °C, with 1.0 mmol g⁻¹ for a btmAm-based hexyl-modified PSf)¹³, saturated aliphatic backbone (48 mS cm⁻¹ with 1.5 mmol g⁻¹ for a trimethyl ammonium-based polyalkylene)⁶³, all-benzene-ring backbone (50 mS cm⁻¹ with 1.57 mmol g⁻¹ for a btmAm-based poly(phenylene)⁶⁴, hydrogen-bonding group (1,2,3-triazole) (62 mS cm⁻¹ with 1.8 mmol g⁻¹ for a trimethyl ammonium-based triazole-containing PPO)⁶⁵, and block copolymer backbone (80 mS cm⁻¹ with 1.93 mmol g⁻¹ for a poly(fluorenyl sulfone)-co-PSf)¹⁶.

In summary, Cp^*₂Co⁺ shows improved stability and basicity compared with Cp₂Co⁺, making Cp^*₂Co⁺ an ultra-stable cation for polymer HEMs. We have also incorporated Cp^*₂Co⁺ cations to a typical polymer backbone (polysulfone) and the resulting membranes exhibited enhanced (short-term) thermal stability and improved (long-term) alkaline stability. More stable polymer backbones are needed to better match the outstanding stability of Cp^*₂Co⁺ cation; and the availability of such a stable organic cation might help to develop the next-generation affordable and durable HEM electrochemical devices.

Methods

Computational methods for calculating ΔH _f, δ _Co, and pK _a

Heat of formation in gas phase (ΔH_f) of the cation and partial charge at the cobalt atom (δ_Co) were calculated by the software MOPAC2012 (Stewart Computational Chemistry) through the graphical user interface Vega ZZ (version 3.0.1.22, Drug Design Laboratory at the University of Milan). The keyword CHARGE =1 was used to specify the total charge of the cation, and the keyword PRECISE was used to tighten convergence criteria. Conjugate acid dissociation constants (K_a) of two amine groups (basal and terminal) in hexamethylenediamine connected to polysulfone polymer and a benzyl group were predicted using the pK_a module of the software MarvinSketch (version 5.12.3, ChemAxon Ltd.) through an empirically parameterized method based on partial charge distribution⁶⁶. One repeat unit of polysulfone with hydrogen atoms as end-groups was used in place of the whole polymer for the pK_a calculation.

Experimental method for measuring pK _b

Consider the dissociation equilibrium of a general base: MOH M⁺ + OH⁻. In the equilibrium state, K_b = ([M⁺]∙[OH⁻])/[MOH], where K_b is the dissociation constant of the base; [M⁺], [OH⁻], and [MOH] are concentrations of dissociated M⁺, dissociated OH⁻, and undissociated MOH, respectively. When [H⁺] is neglected (reasonable in base), [M⁺] = [OH⁻], so K_b = [OH⁻]/(C₀/[OH⁻]−1), where C₀ is the initial concentration of MOH. K_b can be obtained from fitting the slope of the line of [OH⁻] vs. (C₀/[OH⁻]−1). The pH of Cp^*₂Co⁺OH⁻ and Cp₂Co⁺OH⁻ solutions was measured at different initial concentrations (ranging from 0.25 to 50 mM) and the slope was fitted to give K_b. Pure Cp^*₂Co⁺OH⁻ and Cp₂Co⁺OH⁻ were prepared by ion exchange of Cp^*₂Co⁺PF₆⁻ and Cp₂Co⁺PF₆⁻, respectively, with a strongly-basic fast-kinetics anion-exchange resin (Amberjet^® 4200, Dow Chemical Co.), which had been pre-exchanged with hydroxide in 50 wt% water : 50 wt% methanol, followed by filtration to remove resin and evaporation to remove solvent.

Synthesis of Br-Cp^* ₂Co⁺Cl⁻compound

Permethyl-cobaltocenium hexafluorophosphate (Cp^*₂Co⁺PF₆⁻) was brominated by N-bromosuccinimide (NBS) with dibenzoyl peroxide (BPO) as radical initiator and 1,1,2,2-tetrachloroethane (TCE) as solvent. Specifically, 0.474 g (1 mmol) of Cp^*₂Co⁺PF₆⁻ was dissolved in 10 ml TCE, and then 0.178 g (1 mmol) of NBS and 0.0121 g (0.05 mmol) of BPO were added into the solution. After 24 hours of reaction with stirring at 100 °C, brominated permethyl-cobaltocenium hexafluorophosphate (Br-Cp^*₂Co⁺PF₆⁻) was precipitated by pouring the reacted solution into excess diethyl ether. Subsequently, the Br-Cp^*₂Co⁺PF₆⁻ precipitate was filtered, washed thoroughly with diethyl ether, and dried in vacuum at room temperature. The degree of bromination was found to be 95% by NMR spectroscopy (¹H, Fig. S8 and ¹³C, Fig. S12) and mass spectroscopy (Fig. S13). Second, the Br-Cp^*₂Co⁺PF₆⁻ was ion exchanged with a strongly basic fast-kinetics anion exchange resin balanced with Cl⁻ to produce brominated permethyl-cobaltocenium chloride (Br-Cp^*₂Co⁺Cl⁻). Specifically, Br-Cp^*₂Co⁺PF₆⁻ was dissolved in deionized water and then the excess (20 equiv.) anion exchange resin (Amberjet^® 4200 with chloride anion) was added into the solution. With stirring, the ion exchange was held at room temperature for 24 hours, and then the resin was removed by filtration. Br-Cp^*₂Co⁺Cl⁻ was obtained as a yellow powder by evaporating water from the filtrate. The anion exchange was confirmed by ¹⁹F NMR spectroscopy with 97% replacement of PF₆⁻ by Cl⁻ (Fig. S9).

Synthesis of BAHA-Cp^* ₂Co⁺Cl⁻ compound

First, benzyl chloride was aminated by excess (20 equiv.) hexamethylenediamine (HMDA). Specifically, 0.127 g (1 mmol) benzyl chloride was dissolved in 10 ml N-methyl-2-pyrrolidone (NMP), then 2.32 g (20 mmol) HMDA and 0.652 g (2 mmol) cesium carbonate were added into the solution. The amination reaction was carried out for 24 hours with stirring at room temperature, then the solvent was removed by adding excess diethyl ether, leaving behind the mixture of desired product (i.e., N-benzylhexane-1,6-diamine, BHDA) and residual HMDA. The HMDA was removed by addition of excess saturated K₂CO₃ aqueous solution. The leftover BHDA white power was thoroughly washed with deionized water, followed by drying at 40 °C for 48 hours under vacuum. The synthesis was confirmed by ¹H NMR spectroscopy with complete mono-amination (Fig. S10). Second, (6-(benzyl amino)hexylamino permethyl-cobaltocenium chloride (BAHA-Cp^*₂Co⁺Cl⁻) was synthesized by reacting BHDA with Br-Cp^*₂Co⁺Cl⁻. Specifically, 0.206 g (1 mmol) BHDA was dissolved in 10 ml NMP, and 0.444 g (1 mmol) Br-Cp^*₂Co⁺Cl⁻ was added into the solution. The reaction was carried out at 80 °C for 24 hours with stirring, then the solvent was removed by adding excess diethyl ether and the BAHA-Cp^*₂Co⁺Cl⁻ was obtained after washing by saturated K₂CO₃ aqueous solution and deionized water sequentially. The reaction was confirmed by ¹H NMR spectroscopy with complete conversion (Fig. S11) and complete selectivity for the terminal amine over the basal amine.

Synthesis of Cp^* ₂Co⁺-PSf polymer

First, chloromethylated polysulfone (CM-PSf) was synthesized with trimethylchlorosilane and paraformaldehyde as co-chloromethylating agent and stannic chloride as catalyst as reported in detail in our previous work³². Second, CM-PSf was aminated with excess HMDA similar to the case of benzyl chloride. Specifically, 0.502 g (1 mmol polysulfone repeat unit or 1.23 mmol chloromethyl group for 123% of DC) CM-PSf was dissolved in 10 ml NMP, then 2.85 g (24.6 mmol) HMDA and 0.802 g (2.46 mmol) cesium carbonate were added into the solution. The amination reaction was held at room temperature for 24 hours with stirring, then the HMDA-aminated polysulfone (HMDA-PSf) white powder was precipitated by pouring the reacted mixture into excess deionized water. After filtration and thorough washing with deionized water, the HMDA-PSf was dried at 40 °C for 48 hours under vacuum. The synthesized HMDA-PSf is completely soluble in organic solvents such as chloroform and NMP, and neither crosslinking nor depolymerization was found for the HMDA-PSf. The synthesis was confirmed by ¹H NMR spectroscopy with 96% conversion of the chloromethyl groups (Fig. S14).

Second, Cp^*₂Co⁺-PSf was synthesized by reacting HMDA-PSf with Br-Cp^*₂Co⁺Cl⁻. Specifically, 0.600 g (1 mmol polysulfone repeat unit or 1.23 mmol terminal amine group) HMDA-PSf was dissolved in 10 ml NMP, then 0.546 g (1.23 mmol) Br-Cp^*₂Co⁺Cl⁻ was added into the solution. The reaction was completed by heating the mixture at 80 °C for 24 hours with stirring. The reacted solution was cast onto a glass plate and the Cp^*₂Co⁺-PSf membranes were obtained after drying at 60 °C for 72 hours. The balancing anion of Cp^*₂Co⁺-PSf membranes was exchanged with hydroxide by treatment in 1 M KOH at room temperature for 48 hours. After thorough washing and immersion in deionized water for another 48 hours to completely remove residual KOH, the membranes were ready for use. The synthesis of Cp^*₂Co⁺-PSf was confirmed by solid-state ¹³C NMR spectroscopy (Fig. S15).

Additional Information

How to cite this article: Gu, S. et al. Permethyl Cobaltocenium (Cp*₂Co⁺) as an Ultra-Stable Cation for Polymer Hydroxide Exchange Membranes. Sci. Rep. 5, 11668; doi: 10.1038/srep11668 (2015).