Insight into degradation mechanisms of Anion Exchange Membranes.

MSc Chemistry

Molecular Sciences

Literature Study

Insight into degradation

mechanisms of

Anion Exchange Membranes.

by

Vibhav

Yadav

12197254

Period : 10 June 2020 to 10 August 2020

12 ECTS

Supervisor:

Examiners:

Dr. Ning Yan

Dr. Ning Yan

&

Abstract

Anion Exchange Membranes (AEMs) are electrolytes that conduct various ions like OH-_{,CO32- etc. via the aid of covalently bonded cationic charges that are essential for}

providing hopping sites. In comparison to Proton Exchange Membranes (PEMs) they work exactly in the opposite manner in high alkaline medium. There have been recent advances for the potential of AEMs to be used in fuel cells (APEFCs) due to different reaction chemistry which allows the use of non-expensive non-Platinum group metals (PGMs). Even though APEFCs provide a scope for cost reduction which is the number one criteria for commercialisation they suffer from serious degradation due to highly unstable cationic groups at higher pH values. 


Investigation of AEMs has been proposed as an alternative to proton exchange membranes for application in fuel cells due to various aspects like use of low cost electrode materials (Pt/C or non PGMs), use of “dirty” H2 feed instead of pure H2.

Investigating the structural and dynamical properties, charge transport, membrane degradation and electrochemical performance in AEMs using various experimental and simulation techniques provides a guideline in the design of new high-performance membrane fuel cells.

However, the two main problems perceived with AEMs are low stability in alkaline medium and low conductivity of the hydroxide ion. The lower conductivity value can be tackled by increasing the Ion exchange capacity (conductivity ∝ mobility*concentration). Few approaches have been reported with new fixed cation chemistries or modification of the polymer electrolyte membrane network (e.g., cross-linking, blockpolymers, etc.). In recent literature the methods were successful enough to facilitate the hydroxide conductivity to about 10-2 _S.cm-1 _{.Whereas, there are still studies ongoing on the stability of AEMs in}

alkaline media and at elevated temperatures. This study will focus on AEMs for fuel cell applications with peculiar attention towards its degradation routes at operating conditions.

The first few chapters will give a general overview on AEMs by giving a few examples on the state of the art AEMs in the recent literature. The mechanism of OH nucleophile inside the polymer and also drawing a parallel with the mechanism of proton in PEMs.

Then we will dwell upon the potential use of AEMs in H2/O2 fuel cell, DMFCs and some

unconventional systems like carbon based fuel cells eventually moving on to discuss a few pitfalls like CO2 poisoning of the electrodes and consecutive reduction in conductivity

due to formation of carbonates and bicarbonates.

The last part of the study will mainly focus on the reasons for the poor chemical stability of the AEMs in alkaline media and a few methods on how to counteract this problem. Due to the potent nature of the OH nucleophile, AEMs have received much less attention in the sector of commercial applications. It is well known that the most common polymer bound cation, the quaternary ammonium, undergoes degradation via different routes involving substitution (SN2), Hofmann elimination and other rearrangements via ylide

formation.

This study will make use of various characterisation techniques like NMR and electrochemical measurements in tandem with computational approach to gain a meaningful insight into the working behaviour of AEMs.

Acknowledgments

I would start by thanking Dr. Ning Yan my supervisor, who entrusted me with this project for his constant support that helped me to realize my eagerness for the scientific field. It has been a great experience to work under his supervision; the past two months have helped me a lot both as an individual and as a chemist. I cannot imagine completing the study without his guidance.

Next, I would like to thank my examiners at University of Amsterdam, Dr. Bernd Ensing for taking their time out to evaluate the literature study.

Furthermore, I would like to take the time to thank Charlotte Armengaud for the invaluable conversations we shared and for her exceptional editing skills. Finally, I would like to thank my family and friends, especially my sister, as their support motivated me to pursue my studies abroad.

Table of Contents

Abstract ii

Acknowledgments iii

Table of Contents iv

List of Figures v

List of Tables vii

Nomenclature and abbreviations viii

Chapter 1: Introduction 1 1.1 Fuel Cell 1

1.1.1 Design of Fuel Cells 1

1.2 Characterisation of membranes 7 1.2.1 Conductivity 7 1.2.1.1 Ion Migration 7 1.2.2 Stability 12 1.2.2.1 Backbone Stability 12 1.2.2.2 Cation Stability 12

Chapter 2: Degradation routes 16 2.1 Degradation characterisation 16

2.1.1 Reaction Kinetics 16

2.1.2 Nuclear Magnetic Resonance 18

2.2 Degradation 21 2.2.1 Quaternary Ammonium 21 2.2.2 Quaternary Phosphonium (QPs) 26 2.2.3 Imidazolium 28 2.2.4 Backbone 32 Chapter 3: Remedies 38 3.1 Metal Cation Anion Exchange Membranes 38

3.2 Phase segregation 38 3.3 Choice of fuel 39 3.3.1 Cathodic reactions 39 3.3.2 Anodic reactions 40 3.4 Concluding remarks 41 References 43

List of Figures

Fig.1.1 : Schematics of a Fuel Cell, alkaline medium (left), acidic medium (right) 2 Fig.1.2: Structure of electrolyte polybenzimidazole (PBI) 3 Fig.1.3. Structure of electrolytes, sPEEK (sulfonated poly(ether)-ether ketone; above) and

phosphotungstic acid (below) 4

Fig.1.4: Various cationic moieties; (a) benzyl trialkylammonium (benchmark), (b) alkyl side chain quaternary ammonium, (c) imidazolium, (d) pyridinium, (e) phosphonium, (f)

guanidinium, (g) ruthenium metal systems 5 Fig.1.5: Various vackbone chemistries; (a) polyvinylidene flouride (PVDF), (b) polyethylene,

(c) polysulfones (PSFs) 6

Fig.1.6: Solvation sphere of the hydronium ion; Eigen structure 8 Fig.1.7: Speculated structure of the H3O2- ionic complex. 9

Fig.1.8: Simulated structure of the hydroxide ion and the plausible mechanism. 11 Fig.1.9: Degradation of the PVDF backbone via hydroxide attack. 12 Fig.1.10: Degradation of the cationic moiety via nucleophilic attack of the hydroxide

ion. 13

Fig.1.11: Degradation of the cationic moiety via proton abstraction leading to elimination

of the trialkylamine. 14

Fig.1.12: Degradation of the cationic moiety by abstraction of proton from a bronsted base (OH-_{), and following rearrangements due to formation of a radical intermediate}

(above) and ylide intermediate (bottom). 14 Fig.2.1: Magnetic spin of the nucleus in/against the direction of the external magnetic field (above), Splitting of the energy levels of the nucleus (below). 18 Fig.2.2: Precession of the magnetic moment around z-axis with larmor frequency (above), T1 and T2 relaxation mechanisms (below). 19

Fig.2.3: 2-D COSY pulse sequence (top), Interaction of energy levels that give rise to

coupling (bottom). 20

Fig.2.4: TG-EGA via infrared spectroscopy of the ramped ther-mal decomposition of (CH3)4NOH·5H2O. Temperature ramp is 0.25°C min-1 [31]. 21

Fig.2.5: Degradation sequence of trimethylammonium pentahydrate via nucleophilic

Fig.2.6: Antiperiplanar orbital configuration (left) and molecular orbital energy diagram

(right). 24

Fig.2.7: Diaza-bicyclooctane (DABCO) (left), Newmann projection of DABCO (right). 25 Fig.2.8: Optimised transition state of SN2 (left) and ylide formation (right). 26

Fig.2.9: Tertiary phosphonium considered for the reaction kinetics study (where Y=NO2,

Cl, H, CH3 or OCH3). 26

Fig.2.10: Proposed mechanism of tertiary phosphonium degradation under alkaline

medium (top), speculated attack of the hydroxide in a synchronous fashion (bottom). 27 Fig.2.11: Tertiary phosphonium reported in recent literature with exceptional stability due to sterics and electron donating ligands (2,4,6-trimethoxyphenyl). 28 Fig.2.12: Structure of the ionic liquid

poly(1-[(2-methacryl-oyloxy)ethyl]-3-butylimidazolium hydroxide) (poly(MEBIm-X) {where X : Br, OH, CO3, HCO3}. 29

Fig.2.13: Conventional nucleophilic attack of chloride ion leading to formation of two degradation products (above), ring opening mechanism via hydroxide attack prevalent in high alkaline media (bottom). 30 Fig.2.14: A thermodynamically less stable product (-CH=CH-) formed during

rearrangement. 31

Fig.2.15: A series of substituted imidazolium salts considered for their stability. 32 Fig.2.16: A proposed backbone degradation via NMR studies leading to formation of quaternary alcohol (above) and phenol (below) due to the presence of electron

withdrawing sulfone and QA moiety. 33 Fig.2.17: Alkaline stability of PPO based AEMs (TMA–PPO, TMA–PPO&TTMPP–PPO) in nitrogen degassed and oxygen saturated 1 M potassiumhydroxide

(at 60ºC for 30 days). 34

Fig.2.18: Formation of superoxide radical and hydroxyl radical and regeneration of the

ylide species. 35

Fig.2.19: A stable radical adduct via addition of superoxide. 35 Fig.2.20: A stable radical adduct via addition of hydroxyl radical. 36

List of Table

Table 2.1: Effect of different length of alkyl-spacer chains on the exchange capacity

before and after exposure to alkaline solution[32]_{. 23}

Table 2.2: Effect of different length of alkoxy-spacer chains on the exchange capacity before and after exposure to alkaline solution[32]_{. 24}

Table 2.3: Calculated value of the activation barrier via DFT simulations[35]_{. 25}

Nomenclature and Abbreviations

AEMs Anion exchange membranes PEMs Proton exchange membranes

PEMFCs Proton exchange membrane fuel cells APEFCs Alkaline polymer electrolyte fuel cell RTCFCs Room temperature carbonate fuel cell NOx Nitrogen oxides

SOx Sulfur oxides

ATP Adinosine tri-phosphate ADP Adinosine di-phosphate pH logarithm of proton activity

AAEM Alkaline anion exchange membranes TPB Triple Phase Boundary


HOR Hydrogen oxidation reaction ORR Oxygen reduction reaction AFCs Alkaline Fuel Cells

PGM Platinum group metals

sPEEK sulfonated-poly(ether ether ketone) PBI polybenzimidazole

HPAs Heteropoly acids

QAs Quaternary Ammoniums BET Brunauer–Emmett–Teller IEC Ion exchange capacity PMR Proton magnetic resonance IR Infrared resonance

NMR Nuclear Magnetic Resonance
 PSF polysulfone


PVDF polyvinylidene diflouride E2 bimolecular elimination

SN1 unimolecular substitution nucleophilic

SN2 bimolecular substitution nucleophilic

1-D 1 dimensional 2-D 2 dimensional

RDS Rate determining step ν Reaction rate ln logarithm μ magnetic moment μ mobility γ gyromagnetic ratio ω larmor frequency

COSY Correlated Spectroscopy J Coupling constant

TG Thermogravimetric DME Dimethyl

TMA+ _{tetramethylammonium cation}

σ conductivity

σ sigma bonding orbital σ* sigma antibonding orbital DABCO diaza-bicyclooctane

DFT Density Functional THeory QPs Quaternary Phosphonium

TPQPOH tris(2,4,6-trimethoxyphenyl)polysulfone-methylene quaternary phosphonium hydroxide

π pi-bonding orbital

MEBIm [(2-methacryl-oyloxy)ethyl]-3-butylimidazolium EDMIm 3-ethyl-1,2-dimethyl imidazolium

EIMIm 3-ethyl-2-isopropyl-1-methylimidazolium EMPhIm 3-ethyl-1-methyl-2-phenyl- imidazolium PPO poly-propylene oxide

TTMPP tris(2,4,6-trimethoxyphenyl) phosphine

DIPPMPO 5-diisopropoxy-phosphoryl-5-methyl-1-pyrrroline-N-oxide DCPD dicyclopentadiene

GDL Gas diffusion layer

ETFE Ethylene tetraflouroethylene FEP Flourinated ethylenepropylene

Chapter 1: Introduction

The energy appetite for the masses has grown considerably since the 1760s, dependance on conventional fuels have surged. The two big industrial giants, United States of America and China, have had a cumulative emission of about 39% of the total of 51.8 gigatonnes of CO2 equivalent in 2019[1]. The current usage predicts depletion of Oil reserves by 2050

and of Coal by 2080. Even the sustainable usage of conventional fuel would result in continual flush of anthropogenic emissions of greenhouse gases like NOx and SOx,

leading to rise in global mean temperature, reduction in permafrost zone, acidification of soil, etc. Keeping in mind the availability, the environmental impact and the sustainability of various fuels the concern is to shift the trend of consumption towards more non-conventional sources of energy.

Even the shift of conventional fuel to renewables is met with various hurdles of intermittency, geographical limitations and high upfront costs. Therefore the only recourse is to jointly capture, manage and store energy at a large scale and low cost. One of the oldest ways to store energy is the physical method of interconversion of kinetic and potential energy, e.g. in hydropower plants. Recently liquefaction and compression of hydrogen has shown great promise, but high energy density of hydrogen also possesses safety issues. One of the emerging ways is to imitate the biological conversion of ATP to ADP which is accompanied by energy release. Few compounds like hydrazine borane and its derivatives are able to release multiple equivalents of hydrogen that can be used in-situ for fuel cell applications [2,3]_.

This study will focus on electrochemical storage devices that is capable of storing charges via redox reactions or charge separation. The electrical storage device comprises of two electrodes, a medium to separate two electrodes (electrolyte) and an external circuit. The electrodes are held at constant voltage to stimulate the diffusion of the conducting species through the medium.

An electrochemical device, fuel cell, converts the fuel (H2) using an oxidising agent(O2)

into electricity via a redox reaction. Design and components of the fuel cell will be discussed in detail in the next section.

1.1 Fuel Cell

Electrochemical cells, like Fuel cells, uses redox reaction between the fuel and the oxidising agent to generate electricity. The byproducts for a hydrogen-oxygen (considering pure feed) fuel cell is water and heat. Recent works have incorporated fuel cells in tandem with renewables (hydrogen generation via electrolysis) that could be later consumed as a feed in fuel cell.

1.3.1 Design of Fuel Cell

The fuel (H2/O2) undergoes two different redox reaction depending on the acidity of the

conducting electrolyte. The setup regardless of the fuel (H2, CH3OH etc.) has 3

components; anode, electrolyte and cathode. The electrodes are connected by an external circuit that transfers the electron to complete the circuit. The working of the cell and the reduction-oxidation reaction depending on the pH level is illustrated below.

Fig.1.1 : Schematics of a Fuel Cell, alkaline medium (left), acidic medium (right)

Therefore, depending on the working conditions, the Fuel cell can be categorised into two; Alkaline Polymer Exchange Fuel Cell (APEFC) and Proton Exchange Membrane Fuel Cell (PEMFC). As shown in the schematics above, in the PEMFC the Proton Exchange Membrane (PEM) conducts H+ _{from anode to cathode whereas in the APEFC the Alkaline}

Anion Exchange Membrane (AAEM) conducts OH- _{from cathode to anode, in the direction}

opposite to the H+ _conduction.

Electrode enables oxidation/reduction of the fuel, that creates a concentration gradient of oxidised species along the membrane, chemically selective towards one species, and along with the aid of the generated potential difference ,the dipoles in the membrane reorient in the direction of the applied stimulus to help in the transfer of ions from electrode to the other. The engineering of the interface of the electrolyte and the electrode where the fuel comes in contact, called the Triple Phase Boundary (TPB), to carry out the electrochemical reaction is essential. Different mechanisms (mass transfer, etc.) are responsible for bringing these phases in contact, therefore the kinetics of the reaction is one of the limiting factors.

E

= 1.23 V

2H

+ O

→ 2H

O

The overall reaction is similar in both the fuel cell, but due to difference in operating conditions the following differences are evident :

a.) Water is required for Oxygen Reduction Reaction (ORR) for APEFCs while it is a byproduct of the ORR in PEMFCs.

b.) Water is produced at cathode for PEMFC and at anode for APEFC.

In comparison with conventional aqueous alkaline fuel cells (AFCs), the use of polymer as a electrolyte avoids electrolyte seepage and can effectively separate the fuel (H2) and the

oxidant (O2) by a thin film barrier of approx 10microns in thickness. The main advantages

of APEFC over PEMFC is due to the alkaline media:

a.) Due to the low cathodic potential, we observe enhanced ORR catalysis. It enables the use of non-Pt catalyst or non-PGM metals which are inexpensive [4-6]_.

b.) Using methanol as a fuel has potential crossover issues in PEMFCs which is not evident in AEMs.

c.) Wider window in terms of fuel selection apart from a pure source of H2 (e.g. hydrazine

hydrate, dirty H2 etc.)

In the case of PEMs, the diffusion occurs from anode to cathode of H+ _{ions. Walter Grot,}

while working for DuPont in 1960s, discovered a polymer with poly(tetraflouroethylene) backbone with perflourinated vinyl ether side chains terminated by sulfonic moieties called Nafion™ [7]_{. It showed exceptional conductivity of around 10}-1 _S.cm-1 _{in humidity}

conditions of 95 %RH at temperature less than 358K. Even though it exhibits exceptional conductivity, but the operating conditions are limited due to humidity content as the temperature cannot be more than the boiling point of water. One of the major problem of the polymer electrolyte is the amorphous nature [8,9]_{, even though they show exceptional}

conductivity but little is known about the ion conduction mechanism. The amorphous polymers have no long range order and therefore fail to exhibit a X-ray diffraction pattern from which a plausible mechanism could be pointed out.

Apart from Nafion, recently there have been numerous literature on electrolytes with acidic moieties (sulfonic, phosphonic, etc.) or adsorbed cations (Mg+_{, Na}+_{) that are}

capable of migration on applied potential difference. Organic electrolytes like sPEEK[10]

(1.1x10-1 _S.cm-1 _{in DMAc solvent @ 298K) and PBI}[11] _(5x10-2 _S.cm-1_{,H3PO4 doped with}

600%PBI @ 413K), inorganic electrolytes like HPAs[12] _(2x10-2 _S.cm-1 _{for 80%-sPEEK}

composite), water assisted MOF like of Ti-carboxylate (MIL-177[13]_{, 2.6x10}-2 _{@ 298K and}

95%RH) are reported for their exception proton conductivity.

Fig.1.3. Structure of electrolytes, sPEEK (sulfonated poly(ether)-ether ketone; above) and phosphotungstic acid (below)

For AEMs, migration species is anion, e.g. OH- _{and Cl}- _{etc., in the direction opposite to}

PEMs from cathode to anode. The migration is aided by cationic groups covalently bounded to polymer backbone. Various cationic heads studied in the recent literature are summarised below :

a.) QAs like benzyltrialkylammonium, alkyl-bound QAs

b.) Heterocycles like imidazolium, benzimidazolium, pyridinium etc. c.) Guanidinium systems

d.) Phosphorus based systems

e.) Metal based systems known for their ability to have multiple charge carrier per metal site

These cationic heads are covalently bonded to polymer backbone often via -CH2 bridges. Few relevant backbones like polysulfones, poly(ether ketones), perflourinated, polybenzimidazole, polyethylene, etc. are studied extensively for their stability upon functionalisation with cationic moieties.

Although, considering the above points the main disadvantage of AEMs is the low conductivity of OH- _{ions when compared to PEMs for H}+ _{conduction and the stability of}

the membrane in highly alkaline medium. Consequently, it has been difficult to reach the conduction level of PEMs in terms of commercialisation.

We will discuss the pitfalls and the recent methodologies in place to tackle these problems in the next section.

Fig.1.4: Various cationic moieties; (a) benzyl trialkylammonium (benchmark), (b) alkyl side chain quaternary ammonium, (c) imidazolium, (d) pyridinium, (e) phosphonium, (f) guanidinium, (g)

ruthenium metal systems

(a) (b)

(c) (d)

(e)

Fig.1.5: Various vackbone chemistries; (a) polyvinylidene flouride (PVDF), (b) polyethylene, (c) polysulfones (PSFs)

(a)

(b)

1.2 Characterisation of membranes

1.2.1) Conductivity

Depending on the purity of the feed, O2 versus air, the conductivity of the AEMs can drop

due to production of less conductive CO32-, HCO3- forms. Therefore, the inhibition of the

CO2 from the oxidant feed is of utmost important, also the presence of CO2 is known to

cause poisoning of the Pt electrodes.

OH- _{+ CO2}⟹ HCO_3-

OH- _{+ HCO3-}⟹ CO_{32- + H2O}

One other issue with impure feed is the production of water which causes swelling and long term degradation issues. Excess quantity of CO2 causes blocking of the gas

diffusion layer

One way to avoid contamination by CO32- migration species is to ensure thorough Ion

Exchange. In practicality, the membranes are first synthesised in their chloride form and then immersed in a solution of the target ion. In order to attain maximum ion exchange, multiple immersion of the membrane are required with continuous fresh replacements of the solution containing the target ion in excess.

The available active sites for OH- _{adsorption is quantified via Ion Exchange Capacity}

(IEC), measured in number of equivalent per gram of the resin(polymer); eq.g-1_{. The IEC}

can be calculated via variety of methods like BET, Chromatography, back titration, etc. The scope of removing the ambiguity is narrow, therefore in order to compare literature across different groups it is essential to report the conductivity in both OH- _{(ion exchange}

via KOH) and CO32- (ion exchange via K2CO3) immersed membranes.

The conductivity of the ions can be calculated via the relation conductivity ∝ mobility*concentration. Therefore there are two ways to manipulate the conductivity, either by choosing a highly mobile ion or by increasing the concentration of the ion.  One way to increase the concentration of the ions is to ensure thorough ion exchange to reach the maximum theoretical value. Even though the IEC of AEMs (~1.1 meq.g-1_{) is}

more than of PEMs (0.98meq.g-1 _{for  Nafion-11x series)}[14]_{. The high level of target ion}

content results in high water uptake which causes swelling of membranes. The acid base equilibrium of tetramethylammonium hydroxide (RNMe3OH, unstable in anhydrous form)

is predominantly towards the right side (give equation) mainly due to the stability of the hydroxide ion by solvation effect.

Upon comparing the aqueous solution of various ions in water, the mobility follows the trend H+ _{> OH}- _{> Cl}- _{> K}+ _{> CO32- > Na}+ _{> HCO3-} [15-17]_{. The reason for this trend lies in the}

migration mechanism of each species, it is speculated that H+ _{migrates via forming a}

trigonal planar structure with water and diffuses via Grotthuss mechanism (Ea<0.4eV). The negative ions (F-_{, Cl}-_{) having strong polarisability forms a bulky 6 coordinated first}

solvation sphere further reducing the mobility[18]_{. In the next section we will dwell upon the}

migration mechanism of the hydroxide ion in the presence of water which is somewhat similar to the hydronium ion.

1.2.1.1) Ion migration

Before dwelling onto the migration mechanism of the hydroxide ion, we would look into the well established mechanism of proton in order to draw a comparison.

As discussed in the previous section, the reason for high proton conductivity when compared with different ions lies in the mechanism of migration between two oxygen centres in a hydrated medium.

A well established model was proposed be Eigen[19]_{, unlike water which is in a tetrahedral}

symmetry, the protonated water does not form H-bonds via the O nucleus and is solvated by 3 H2O molecules in a trigonal complex of H9O4+, called the Eigen cation. The first

solvation shell is tightly bound as suggested by the gas phase mass spectrometry, the bond dissociation enthalpy decreases as we move further away from the H3O+ suggesting

strong interactions within the first sphere of solvation[20]_.

Considering all the experimental data, Zundel and Fritsch postulated a proton transfer mechanims via structure fluctuation between the H9O4+ and H5O2+ complex[21,22], the

mechanism is as follows:

a.) The rate limiting step involves the cleavage of a hydrogen bond with an activation energy of 2.6 kcal.mol-1_.

b.) The 3 hydrogen bonds in the first solvation shell does not participate in complex activation.

c.) Due to high bond dissociation enthalpy in the first shell, the H-bond dissociation occurs in the second shell.

d.) The oxygen atom adjacent to the H3O+ makes a transition from tetrahedral geometry

to planar geometry.

e.) During the transition of the proton from the centre oxygen to oxygen in the first solvation shell, bonds and angles readjust to form a complex H2O … H … OH2, this

complex is stabilised by surrounding water molecules polarised towards the complex via dipole interaction.

f.) The oxygen atom which was protonated before upon loss of the proton makes a transition from trigonal planar geometry to tetrahedral symmetry by accepting a surrounding water molecule.

Having taken into consideration the migration mechanism of a proton, very less insight has been given into the hydroxide migration in the literature. It has been speculated, on a qualitative level, that the migration happens by structural diffusion, similar to the proton mechanism, continuous interconversion between the hydration sphere due to migration of a proton hole from one oxygen to another.

In this section we will discuss the solvation sphere of the hydroxide ion, which was earlier speculated to be OH-.(H2O)3, and the interconversion of the sphere to H3O2- species to

facilitate the proton transfer followed by nuclear rearrangements to form a new sphere. It has been long speculated via various experimental techniques like Raman spectroscopy and Proton Magnetic Resonance (PMR) that the hydroxide ion in an aqueous phase forms a H3O2- complex[23,24].

Earlier studies were done to verify the state of the OH- _{in the aqueous solution of the alkali}

metal hydroxides via PMR. Two models were hypothesised about the localisation of the hydroxide ion as follows :

a.) Na+ _{… |OH}- _{- H}+_|….OH- _{: The hydroxyl ion forms a complex with the metal through the}

polarised water of the ion.

b.) Na+ _….|OH-1/2 _{- H}+ _{- OH}-1/2_{| : Experiments also suggest that the hydroxide occurs in}

the form of H3O2- complex.

If the first hypothesis is true, then the nature of hydration sphere of the each cation would be different and the chemical shift of the proton would be effected by changing the type of the cation.

If the second hypothesis is true, then the nature of the cation would have very less impact on the chemical shift, rather it would be determined by the structure of the H3O2- complex

that would determining the J-coupling over space.

Considering the activity coefficient of alkali salts in the aqueous solution, it follows the order Li+ _{> Na}+ _{> K}+ _{> Cs}+ _{, therefore if the hydroxide ion occurs in the form of OH}-_{, the}

nature of the chemical shift should change if the cation changes. Zatsepina showed in his

Fig.1.7: Speculated structure of the H3O2- ionic complex.

work that the chemical shift in the PMR spectra is independent of the nature of the cation (Li, … ,Cs) in their aqueous solutions. Therefore, the large chemical shift of the proton is due to the structure of H3O2- complex, in which the proton is dangling in between the

oxygen atoms and is de-shielded by a greater extent.

They also used Raman spectroscopy to support the formation of H3O2- complex[25]. It has

more peaks. The spectra of pure water (vmax=3450cm-1) and KBr (vmax=3225cm-1) can be

decomposed into two bands, denoting symmetric and asymmetric nodes.

In the case of acidic solution, comprises of the bands similar to water with a third band, the shift of the peak towards lower frequency is associated with the elongation of the O-H distance to incorporate the change of geometry to form the H3O+ complex.

For the alkaline solution, two new bands are observed, one is analogous to the formation of a strong hydrogen bond between OH- _{and a water molecule, the other sharp peak is}

characteristic of the OH- _{stretching in the monomeric water molecules. The increase in}

the intensity of these peaks is evidenced with the increase in concentration of the alkali. The low frequency band of bending vibration of the OH group shifts to higher frequency as the alkali concentration increases. This can be supported if the postulation of formation of OH- _{…H—OH is correct.}

Having discussed the possibility of the formation of the H3O2- complex during the transfer

of a proton from one oxygen centre to the other. We will go into detail the plausible mechanism of migration of a proton hole.

Gas phase mass spectrometric measurements[26] _{suggest the coordination sphere of 3 for}

the hydroxide monomer, OH-.(H2O)3 (namely H7O4-). But Car-Parrinello simulations[27]

suggest a coordination number of around 5, with one water molecule weekly bonded in the first solvation sphere (H9O5-).

In order to address this problem, Parrinello studied the ab-initiio path integral of the hydroxide ion in the bulk water at room temperature. He simulated a periodic box of length 9.8 Å with 31 water molecules and a single hydroxide ion. After equilibration of the geometry, enormous number of possible configurations were generated. They sorted the configuration as follows:

A defected oxygen (O*) was located with having only 1 hydrogen in the close proximity. A variable ∂ was selected corresponding to ∂ = R(O*H)-R(ObH) {where; Ob is the nearest

oxygen to the defected oxygen, O*}. A low value of |∂| corresponds to a high propensity of proton transfer between two oxygen centres. Therefore, in order to differentiate between different solvation spheres, different configurations of large and small ∂ were selected. On integration of the RDF, the coordination number around the hydroxide ion for large ∂ was approximately 4.5 and the geometry as imported from the simulation suggested a planar configuration of OH-_.(H2O)4.

For small ∂, the integration value decreased to 4.1, with the number of accepted and donated H-bond around 3.2 and 0.9 respectively. This suggests a tetrahedral symmetry of OH-_{.(H2O)3 around the O* atom, meanwhile the donating H-bond takes part in proton}

transfer.

Therefore the mechanism can be summarised as follows :

a.) The hydroxide ion sits in an energy minimum by coordinating with 4 water molecules in a planar fashion.

b.) For activation of the proton transfer mechanism requires the transition from the planar to the tetrahedral geometry. This transfer of geometry is accompanied by a breaking of a H-bond which is an uphill reaction with a minimal activation energy of 1.16kcal/mol. c.) Further, after the change in geometry, the hydroxide ion in a tetrahedral geometry accepts a proton from a coordinated water. After the transfer of the proton, the original hydroxide ion becomes a tetrahedrally solvated molecule representing bulk water.

The graphical outline of the mechanism is depicted below.

1.2.2) Stability

The reason AEMs suffer from poor performance is due to two reasons, i.) poor OH

-mobility and ii.) low stability at high pH environment.

The degradation of the membrane can happen via two different routes, the most common researched is the low stability and susceptibility of the cationic heads in high alkaline media and the other being the degradation of the backbone via ether hydrolysis or hydroxide attack.

This study will go into depth into the mechanism of degradation of the membranes via different routes, but first a general overview of various modes of degradation will be provided.

One way to retard the degradation is to ensure sufficient amount of water level in the membrane so that the hydroxide ion remains hydrated. If the OH- _{stays tethered to the}

backbone then the OH- _{attack is imminent. But high hydration levels has its drawback, as}

it causes swelling of the membrane and could hamper with the mass transfer through the gas diffusion layer.

1.2.2.1) Backbone stability

One of the main way to characterise degradation is using proton NMR. It was long speculated that the backbone remains intact during high pH environment due to ambiguous signal characterisation. Due to highly electron withdrawing nature of the cation groups (QAs, phosphonium cations, etc.) the electron density in the backbone gets localised and the functionalities like ether or quaternary carbon that tether the phenyl moieties in the backbone are susceptible to hydroxide attack.

Fig.1.9: Degradation of the PVDF backbone via hydroxide attack.

Many membranes have enhanced stability against hydroxide ions like polysulfones (PSF)

[28] _{or flourocarbon type membranes. However, few membranes (e.g. polyvinylidene}

flouride;PVDF)[29] _{with halogen moieties which are good leaving groups in high alkaline}

medium undergo E2 elimination. 1.2.2.2) Cation head stability

This section will give a brief overview of various degradation mechanism possible for the cationic head. We will discuss in detail in the later sections.

The degradation scheme has been outlined in detail below using benzene trialkylammonium as a template due to possibility of Sommelet-Hauser and Stevens rearrangement.

The most common degradation pathway is the nucleophilic substitution, either SN1 or SN2

depending on the stability of the carbocation intermediate. Due to high electron withdrawing nature of the quaternary ammonium, the α-C either of the methyl or benzyl is susceptible to the nucleophilic attack resulting in methanol or benzyl alcohol respectively. The hydroxide ion could also behave as a bronsted base and abstract an acidic proton resulting in an ylide. This reaction progresses forward in two ways via Sommelet-Hauser or Stevens rearrangement, the difference being that Sommelet-Hauser progresses with deprotonation only at the methylene moiety of at the benzylic posiiton. The difference in mechanims is that the former proceeds via 2,3 sigmatropic shift and the latter proceeds via a disputed radical mechanism.

The hydroxide can also attack the β-H or α-H and result in a formation of an alkene in two possible ways. The β-H hydrogen elimination, also called Hoffmann elimination, requires anti-periplanar geometry and is a facile reaction at low temperature resulting in a formation of alkene and trialkylamine.

In the case of steric hinderance at the β-C position, the hydroxide abstracts a hydrogen from the α-C position followed by rearrangement and elimination of the amine group resulting in the same products as of Hoffmann elimination. The difference in mechanism can be pointed out by using isotopic labelling.

Fig.1.11: Degradation of the cationic moiety via proton abstraction leading to elimination of the trialkylamine.

Fig.1.12: Degradation of the cationic moiety by abstraction of proton from a bronsted base (OH-_{), and}

following rearrangements due to formation of a radical intermediate (above) and ylide intermediate (bottom).

Chapter 2: Degradation routes

Research undertaken on degradation of alkaline electrolytes before used to treat the stability of cationic groups separately from the backbone, as the stability of the membrane was long established. 


As new degradation routes and the mechanism are proposed a model was generated and the stipulated model was verified using chemical reaction kinetics. The products were tested for a change of signal in NMR methods so as to verify change in reactants. Due to the presence of different hydrogen in various chemical environments, 1-D NMR spectroscopy often had overlapping peaks and noises. Most often, methyl groups in polymer backbone of polysulfones or methyl groups after rearrangement via ylide formation had peaks at almost the same chemical shift. Using 2-D NMR spectroscopy it was easy to draw a comparison between the two via observing the coupling between two hydrogens. 


Hence, it was found that due to highly electron withdrawing nature of the cationic groups there is electron localisation in the backbone resulting in the attack of the backbone in a highly alkaline medium. 


In the following chapter we will discuss various ways of counteracting the problem of degradation at high alkaline medium. Most of the literature reports the conductivity of AEMs with various backbones and using QAs as cationic heads. One of the main reason is the stability of QAs upon comparison with other functionalities, NR4+ > PR4+ >SR3+ [30].

Recently, more complex and stable cationic heads discussed in the previous chapter have been reported consisting of phosphonium cation with bulky substituents or exotic metal complexes. One of the main reason for using AEMs is the use of non-expensive non Pt-group metals, and using cationic heads like bulky phosphonium groups or metal complexes would have a setback on cost as procuring precursors and synthesising QAs is relatively cheap.

2.1) Degradation characterisation

Having the knowledge of reaction kinetics is important to ascertain the mechanism of reaction, the slowest step of the reaction (RDS) and would give us insight into possible ways of rerouting the mechanism to retard the degradation.

One of the key steps in reaction kinetics is the concentration determination of the studied species as time evolves. Assuming a general reaction in which 1 mole of the target species reacts with b moles of hydroxide ions to yield products C and D, stated below :

(Eq.2.1)

Where, A and B are the cationic head and the hydroxide ion reactants respectively which react to give products whose chemical composition can be speculated using various techniques like mass spectroscopy, gas phase chromatography,etc. Their concentration at any specific time is denoted in parenthesis [.] as [A], [B] and so on.. with concentration at the start of the reaction denoted as [A]o ,[B]o and assuming the concentration of the

products at the start of the reaction as 0 units.

Sometimes one of the products is in a different phase than the reaction environment, i.e. it precipitates either as a solid or bubbles out as a gas. Therefore the activity of that product could be taken as unity. We can further simplify the reaction by taking one of the reactants in excess such that the concentration remains constant as the reaction progresses. Therefore the equation above reduces to (assuming D precipitates out as solid) :

(Eq.2.2)

Assuming the rate of the forward reaction is k1, the rate of disappearance of the B can be

written as :

(Eq.2.3)

Incorporating the constants together, we can write k1*b=K1 .We could progress the

reaction in various ways, the more traditional method is as follows :

a.) @ any time, t=to; assuming x moles of A has been degraded. Therefore, at t=to, the

activity of various species is [A] = Ai - x, [B] = Bi - bx and [C] = cx. Substituting these

values in the equation above, yields :

(Eq.2.4)

The degradation concentration can be mapped over time by carefully analysing the products or the reduction in moles of target species. The equation above in 3 variables (k1, a, b) is in a complex form and needs to be simplified by making some assumptions.

b.) Taking excess of one species, we can assume the activity of the species to remain constant with time and can monitor the order of the reacting species, 


taking excess of QA, we can incorporate the activity of QA in the rate constant as k1[A]=k2:

(Eq.2.5)

c.) Generally the cation degradation is either a first order reaction in hydroxide (for nucleophilic attack and formation of alcohols) or a second order (formation of ethers, formation of phosphonium oxide via hydroxide attack). Second order essentially means involvement of two hydroxide species either in two reaction steps or a synchronous attack. We will go into detail of these reaction in later sections.

(Eq.2.6) (Eq.2.7)

A + bB → cC

ν = − 1

_b

dB

_dt

= k

[A][B]

− dB

dt

= K

(A

− x)(B

− bx)

− dB

dt

= K

[B]

−∫

_[B]

1

dB = ∫

K

dt → lnB = lnB

− K

* t

−∫

_[B]

1

dB = ∫

K

dt → ( 1

B

− 1

B

) = K

* t

Therefore, after plotting the data of the concentration profile versus time, we can fit the above two equations in order to ascertain the time and calculate the reaction constant. If the graph of ln(B) vs. t is linear then it is first order in B with the slope of the graph giving the rate constant. If the graph of 1/B vs. t is linear then it is second order in B with the slope of the graph giving the rate constant.

In the next few sections, we will discuss how the plausible mechanism of degradation was proposed with the help of reaction kinetics.

2.1.2) Nuclear Magnetic Resonance

A charged spinning nucleus possesses a magnetic moment (μ) which can align on application of an external magnetic field (Bo). In the absence of Bo , the nucleus has the

highest population density in the two degenerate energy levels (n=0, l=0, ms = ± 1/2).

(Eq.2.8)

Fig.2.1: Magnetic spin of the nucleus in/against the direction of the external magnetic field (above), Splitting of the energy levels of the nucleus (below).

μ = γI

The magnetic moment (μ) is related to angular momentum of the nucleus via a proportionality constant unique to each nucleus called the gyromagnetic ratio (γ).

Upon application of Bo, these energy level split into two levels with the separation energy

defined by the strength of the magnetic field.

(Eq.2.9)

Each nucleus has it’s own larmor frequency (ω=γBo), with which it is precessing around

the z-axis. Data acquisition of the field can be done by placing a coil in the x-y plane. In order to rotate the magnetic field in the x-y plane a radio-frequency pulse of certain frequency (hard pulse, soft pulse : resonant to larmor frequency) has to be applied for short instance. 


Upon application of 90º and 180º pulses, the signal decays exponentially with different time constants:

a.) T2 (xyplane, spin-spin relaxation)

b.) T1 (z-axis, spin-lattice relaxation)

Fig.2.2: Precession of the magnetic moment around z-axis with larmor frequency (above), T1 and T2

relaxation mechanisms (below).

E = − γ h

_2π

B

m

A 180º rf pulse rotates the magnetisation vector in the -z direction, and the decay of the magnetisation to reach the equilibrium value decays exponentially with time constant T1:

(Eq.2.10)

A 90º rf pulse rotates the magnetisation vector in the xy plane, and due to inhomogeneities of the sample dephasing occurs in between the nucleus and the magnetisation decays to the equilibrium value with a time constant T2:

(Eq.2.11)

Using 2-D NMR techniques we can ascertain the coupling between hydrogen approximately over four bonds. Methods like COSY (Correlated Spectroscopy) are used to ascertain the coupling between hydrogen atoms. The 2-D spectroscopy is divided into 4 domains as preparation, evolution (t1), mixing and acquisition (t2) as depicted below.

Fig.2.3: 2-D COSY pulse sequence (top), Interaction of energy levels that give rise to coupling (bottom).

M

= M

(1 − 2e

)

M

= M

e

Since the I to S or S to I polarization transfers are the same, we’ll explain it for I to S and by symmetry we get the same for S to I. The main idea is to perturb I and analyze what happens to S. After the first π / 2, two I vectors are in the xy plane, one moving at ω+ J / 2 and the other at ω- J / 2. The effect of the second pulse is that it will put the components of the magnetization aligned with y on the -zaxis, which means partial inversion of the I populations.

For t= 0, we have complete inversion of the I spins (it is a π pulse and the signal intensity of S does not change). For all other times we will have a change on the S intensity that depends periodically on the resonance frequency of I. The variation of the population inversion for I depends on the cosine (or sine) of its resonance frequency.

Consider that we are on-resonance with one of the lines and if t = 1/4J: the magnitude of the vector along –z depends on J which is a measure of the polarization transfer. Therefore, with t1 = 1/4J the vectors would cancel each other and we wouldn’t record any

signal.

2.2) Degradation

2.2.1) Quaternary Ammonium

A comprehensive study was undertaken to elaborate the degradation temperature for QAs using an Evolved Gas Analysis in order to get an idea of the working temperature for the AEMs[31]_{. TG data showed the melting of tetramethylammonium pentahydrate}

[(CH3)4NOH.5H2O] at 68ºC, therefore vast literature records the stability tests of the AEMs

at around 60ºC. The low temperature transition are correlated to the loss of coordinated water (20 mol%) and a sharp peak with an onset temperature of 112ºC is associated with the degradation of the compound.

Fig.2.4: TG-EGA via infrared spectroscopy of the ramped ther-mal decomposition of (CH3)4NOH·5H2O.

Temperature ramp is 0.25°C min-1 [31]_.

The loss of coordinated water suggests that partial dehydration is required for hydroxide species to become reactive enough to start the degradation. At temperature above 110ºC, the thermal data shown above can be resolved into two different regions (region I -

105-120ºC, region II - 120-130ºC). Dimethyl ether (DME), trimethylamine and methanol are formed with large amount of water released, whereas in region two no methanol is detected using FTIR and mass spectroscopy.

From the data they proposed a mechanism as follows:

a.) loss of 2 equivalents of coordinated water (consistent with TG values). Reactive hydroxide can attack in two ways; as a nucleophile or as a bronsted base.

b.) Nucleophile:

1.) SN2 attack at methyl position resulting in the formation of methanol and

trimethylamine.

2.) Due to basic medium, hydroxide abstracts a proton from the generated methanol which in-turn acts as a nucleophile on TMA+ _{yielding DME.}

c.) Bronsted base:

1.) Abstracts a proton from a methyl generating a ylide intermediate. 2.) Insertion of the carbene into O-H bond of the water or already generated methanol yields methanol and DMA respectively.

Fig.2.5: Degradation sequence of trimethylammonium pentahydrate via nucleophilic attack and ylide formation.

The existence of the ylide intermediate can be confirmed via isotopic labelling by using [(CH3)4NOD.5D2O]. The carbanion generated would yield deuterated DME and methanol

and as well as via deprotonation/protonation of the TMA+ _{would give deuterated}

trimethylamines.

The main route for degradation of QAs is the β-H abstraction by hydroxide ion followed by elimination of the trialkylamine resulting in an alkene. In another route the hydroxide ion attacks the α-C resulting in a formation of an alcohol and trialkyl amine. The cationic head experiences serious degradation on temperatures increasing 60ºC.


In both the reactions, the proximity of hydroxide ion is key to the reaction, therefore few studies were undertaken by monitoring the reaction rate while increasing the length of the pendandt alkyl chain.

Tomoi et. al. studied the effect of spacer chains on the membrane performance[32]_{. They}

used copolymerisation of ω-bromoalkylstyrenes functionalised with 2-8mol% of divinylbenzene, which were quaternized with trialkylamine. The stability was monitored in deionised water at 100-140ºC for 30-90 days. They monitored the Exchange Capacity before and after testing different membranes in deionised water at elevated temperature for >30 days.

Table 2.1: Effect of different length of alkyl-spacer chains on the exchange capacity before and after exposure to alkaline solution[32]_.

Therefore, it is evident that increasing the alkyl chain length results in retention of the Exchange capacity. It could be due to the absence of reactive benzyl carbon that readily loses a proton. Although increasing the chain length might open a pathway for Hoffmann elimination, therefore substituting the carbon at β position (entry #3) should be more stable than a linear alkyl chain. 


But from the result ,comparable entry in column 4 for entry #3 and #2, it is evident that Hoffmann elimination doesn’t proceed in spacer modified cationic groups.

Spacer Chain DVB content

% Exchange Capacity mEq/L Ratio %

Initial Final CH2 1.42 1.12 79 (CH2)4 2 0.84 0.78 92 CH2CH2CH(CH3)CH2 2 0.77 0.71 92 4 1.05 0.91 86 (CH2)3CH(CH3)CH2CH2 2 1.01 0.97 96 (CH2)7 2 1.13 1.03 93 CH2CH2-(C6H10)-CH2 2 0.93 0.86 93

Table 2.2: Effect of different length of alkoxy-spacer chains on the exchange capacity before and after exposure to alkaline solution[32]_.

However, incorporation of ether functional groups in the spacer chain showed interesting results. The IR spectra showed two bands around 1640 and 990 cm-1 _{correlating to the}

CH=CH2 linkage. The spacer chain, propyleneoxymethylene showed drastic decrease in

exchange capacity which is speculated to enhanced Hoffmann elimination. The electronegativity of the oxygen atom stabilises the intermediate of the Hoffmann Elimination via inductive effects. Increasing the distance of the ether group from the β-C increases the stability as evident from the table above.

The most known strategy to subdue the rate of Hoffmann Elimination is to have no β-H, like in the case of benzyl-trialkylammonium. The correct geometry for a facile elimination of the trialkyl amine is to be antiperiplanar to the hydrogen as shown below. This enhances the overlap between the σC-H and the σ*C-X orbitals which ensures facile removal

of the leaving group without being an energy intensive reaction.

Fig.2.6: Antiperiplanar orbital configuration (left) and molecular orbital energy diagram (right).

Cationic heads like diaza-bicyclooctane (DABCO) have an ecliptic geometry that constraints the β-H out of the anti-periplanar geometry increasing the activation energy for the Hoffmann elimination[33,34]_{. But one of the main drawback of the constrained}

Spacer Chain DVB content

% Exchange Capacity mEq/L Ratio %

Initial Final CH2O(CH2)3 4 3.38 0.43 10 6 3.33 0.49 11 8 3.12 0.56 13 CH2O(CH2)4 4 3.42 3.20 94 6 3.21 2.87 89 8 3.32 2.93 88 CH2O(CH2)6 4 3.00 2.93 98 σ-σ* gap E2 gap σ* σ

geometry is the high ring-strain energy which increases the energy level of the QAs (LUMO) leading to degradation.

Fig.2.7: Diaza-bicyclooctane (DABCO) (left), Newmann projection of DABCO (right).

Recently, studies regarding computational analysis of the degradation pathways are used to ascertain the kinetics of different mechanism[35,36]_{. The knowledge of the reaction}

pathway is important to model these systems computationally. For a model reaction A + B → C, the structure A, B and C are optimised using density functional theory. Using the Self Consistent Field (SCF) energy, incorporates the electronic energy of the molecule, the energy of the reaction was calculated ΔE = ESCF(C) - { ESCF(A) + ESCF(B) }. Vibrational

frequency analysis is used to calculate the standard Gibbs Free energy of the species and of the reaction ΔGº = Gº(C) - { Gº(A) + Gº(B) }.

The transition state (1st _{order saddle point along the reaction coordinate) can be}

calculated by optimising the structure of A and B in close proximity and making a transition towards C via minimum energy path (methods like Growing string method). The state with highest energy along the minimum energy path and having a single imaginary vibrational frequency is the transition state. The result of the following calculations on various cations is summarised below :

Table 2.3: Calculated value of the activation barrier via DFT simulations[35]_.

QA Ylide

@ methyl SN2 Hoffmann Elimination

@

trans-methyl @ gauche-methyl @ non-methyl @ syn @ anti

TMA+ _{16.8 26.7 26.7 NA NA NA} EthylTMA+ _{18.2 24.7 25.1 27.0 17.5 21.4} n-propylTMA+ _{18.4 25.9 26.3 29.7 22.9 26.3} n-butylTMA+ _{- 25.3 - 31.1 23.9} -n-pentylTMA+ _{- 25.6 - 30.4 24.1} -n-hexylTMA+ _{- 25.6 - 29.7 23.7} -isobutylTMA+ _{- 25.7 - 28.3 21.7 24.8} neo-pentylTMA+ _{- 25.0 - 34.5 NA NA} phenylTMA+ _{18.4 22.6 24.3 NA NA NA} benzylTMA+ _{16.9 25.1 26.5 23.3 NA NA}

Fig.2.8: Optimised transition state of SN2 (left) and ylide formation (right).

It is clearly evident from the table above that for n-alkylTMA+ _{Hoffmann Elimination is the}

most feasible pathway for degradation. The barrier for elimination increases due to steric effects as the chain length increases. Moreover, we can alter the elimination pathway by increasing the barrier by substituting the β-C with different substituents.

2.2.2) Quaternary Phosphonium (QPs)

On of the aspects in commercialisation of membranes for fuel cell applications is the electrode assembly. Membranes show good conductivity values but when applied on electrode the performance drops drastically. Solvents like methanol, ethanol or n-propanol are used in electrode preparation because they are easy to evacuate post preparation, hence solubility of the membrane in these solvents is important. Therefore, the inability of QAs to form binders on the catalytic sites results in drop of performance due to inefficient mass transfer through the Triple Phase Boundary.

QPs have excellent solubility in methanol accompanied with strong basicity of tertiary cations resulting in efficient mass transfer across the electrodes and high conductivity of hydroxide via the membrane[37-40]_.

But even taking these factors into consideration, QPs also degrade under highly alkaline medium. McEwen et. al. in the late 50s proposed a mechanism of QPs (methylethylphenylbenzyl phosphonium, p-Y-benzyltribenzyl phosphonium; Y = NO2, Cl,

H, CH3 or OCH3) under alkaline medium[41,42].

Fig.2.9: Tertiary phosphonium considered for the reaction kinetics study (where Y=NO2, Cl, H, CH3 or

They made the following observations :

a.) concentration of the hydroxide was mapped by taking 20-fold excess of the cation, they found pseudo-second order concentration dependance of the hydroxide ion.

b.) ease of elimination of a substituent is proportional to the stability of the substituent as an anion

c.) ease of elimination of substitutent is directly correlated to the nature of non-departing groups.

They found that the rate of reaction decreased in the order NO2 >> Cl > H > CH3 > OCH3.

For the nitro substituted QPs the only degraded was p-NO2-Toluene with phosphine

oxide. Whereas for methoxy substituted QP the major degraded product was Toluene with phosphine oxide.

Apart from nitro substituted QP, all other compounds showed mixture of p-substituted toluene and toluene.

Therefore on the basis of this study they proposed the following mechanism :

Fig.2.10: Proposed mechanism of tertiary phosphonium degradation under alkaline medium (top), speculated attack of the hydroxide in a synchronous fashion (bottom).

The mode of attack of hydroxide is speculated, it can occur in various ways. Either in a consecutive manner with two different reaction constants, or a synchronous attack leading to a formation of phosphine oxide and water in one step, or coordination of two hydroxide to the phosphorus atom leading to a formation of tetragonal bipyramidal complex.

The substituents can be modified greatly in order to retard the degradation. Using substituents like 2,4,6-trimethoxyphenyl, a strong electron donor, stabilises the phosphonium cation towards degradation and protects the phosphonium cation and the α-C from high steric bulk. AEMs comprising of tris(2,4,6-trimethoxyphenyl)polysulfone-methylene quaternary phosphonium hydroxide (TPQPOH) have been synthesis recently showcasing good solubility in solvents like methanol with high hydroxide conductivity and good alkaline stability[37]_.

Fig.2.11: Tertiary phosphonium reported in recent literature with exceptional stability due to sterics and electron donating ligands (2,4,6-trimethoxyphenyl).

2.2.3) Imidazolium

One of the structure with multiple nitrogen atoms that has been extensively reported recently is the imidazolium heterocycles. They are remarkably known to suppress the SN2

and Hoffmann Elimination reaction due to the anion stability via π-conjuagted system. Even in the presence of a proton abstraction from C-2 position resulting in a formation of a ylide species, the resulting carbanion is stabilised in between two electronegative nitrogen groups suppressing further rearrangement reactions.

Synthesised and dried imidazolium cations with hydroxide as Ionic Liquids has been reported for their potential towards catalysis (stable under dry conditions) for Michael reaction[43,44] _{whereas QAs requires solvation for stability and are hence stored in a liquid}

form with low concentration[45]_.

Despite the promising aspects of the imidazolium cation they also suffer from degradation to some extent but which can be avoided largely. Aromatic heterocycles due to their planar nature are susceptible to an approaching hydroxide from both the sides, substitution of the nitrogen group and at C-2 position could increase the steric bulk and

retard the approach of the hydroxide ion. Initial studies on purine and guanosine derivatives suggest that a ring opening mechanism at the C-2 position could likely be the major degradation route for the imidazolium cations.

A comprehensive study on the degradation of the C-2 unsubstituted imidazolium tethered to a backbone was undertaken[46]_{. Polymerised ionic liquid (PIL)}

poly(1-[(2-methacryl-oyloxy)ethyl]-3-butylimidazolium hydroxide) (poly(MEBIm-X) {where X : Br, OH, CO3,

HCO3} was tested under broad range of humidity, temperature and alkaline concentration.

Thermal degradation (Td) of the membranes was monitored under nitrogen environment,

poly(MEBIm-Br) showed the highest degradation temperature with a nucleophilic attack at the α-C via SN2 reaction resulting in alkyl leaving group[47]_.

Fig.2.12: Structure of the ionic liquid poly(1-[(2-methacryl-oyloxy)ethyl]-3-butylimidazolium hydroxide) (poly(MEBIm-X) {where X : Br, OH, CO3, HCO3}.

The drop in the TG curve is consistent with the degradation product, ~39% weight loss due to 1-butylimidazole leaving group and the remaining 61% corresponding to poly(2-bromoethyl methacrylate). However, if we apply the same degradation route for the hydroxide, then the mass loss peak should correlate to 49% but the actual loss was 25% suggesting that a different mechanism is in play.

1_{H-NMR data and comparing the results with previous research focussed on degradation}

of purines[48,49]_{, it was speculated that the imidazolium degrades via ring opening at the}

C-2 position. The 1_{H-NMR data is consistent with the formation of two isomers as shown}

below, formation of two isomeric formyl groups result in peak splitting (@8.54 and 8.55ppm) and also with the isomeric forms of -CH=N- groups. The formation of the -CH=CH= group is confirmed @ 5.56 and 4.66 ppm although the it has been established that -CH=N- group is more thermodynamically stable.

Fig.2.13: Conventional nucleophilic attack of chloride ion leading to formation of two degradation products (above), ring opening mechanism via hydroxide attack prevalent in high alkaline media

Fig.2.14: A thermodynamically less stable product (-CH=CH-) formed during rearrangement.

The membranes were stable in oscillating humidity conditions between 90-10% RH reproducing the conductivity value after every cycle.

In a recent work a series of imidazolium salts, 3-ethyl-1,2-dimethyl imidazolium bromine ([EDMIm]-[Br]), ethyl-2-isopropyl-1-methylimidazolium bromine ([EIMIm][Br]),and 3-ethyl-1-methyl-2-phenyl- imidazolium bromine ([EMPhIm][Br]), were synthesised both with un-substituted and substituted C-2 carbons[50]_{. They tested the stability of the above}

mentioned salts in varying concentration of KOH alkaline solution.

Table 2.4: Extent of degradation in C-2 unsubstituted and substituted imidazolium[50]_.

Severe degradation was encountered for the unsubstituted imidazolium salt which was even more sever than the benchmark benzyltrimethylammonium. Substituted

salts KOH solution

(M) Time (h) Degradation (%) [EMIm][Br] 1 60 47.6 [EDMIm][Br] 1 168 0.0 [EIMIm][Br] 1 168 0.0 [EMPhIm][Br] 1 168 0.0 [EDMIm][Br] 2 168 0.0 [EIMIm][Br] 2 168 10.2 [EMPhIm][Br] 2 168 25.7 [EDMIm][Br] 6 10 5.4 [EIMIm][Br] 6 10 20.2 [EMPhIm][Br] 6 10 32.8 Benzyltrimethylammonium 1 M NaOH 96 20.0