Hydroxide AEM

Anion exchange membrane fuel cells

Increasing population and technological development over recent decades has caused a significant increase in energy consumption. Therefore, it has become vital to explore new renewable energy technologies. Fuel cell devices are among the most exciting such possibilities for renewable energy. Today, there are many types of fuel cell devices, mainly sharing the basic structure and mechanism, with different levels of efficiency and functionality.  Anion exchange membrane (AEM) fuel cells have received increased attention recently mainly due to  their ability to use low-cost non-precious electrocatalysts, the reduced risk of water flooding at the cathode, and fuel flexibility.  Presently, the most challenging technical hurdles for AEM fuel cells are the development of membrane materials that have high hydroxide ion conductivity and stability under high pH conditions. The research in our group is focusing on deriving a set of design rules for the creation of new anion exchange membranes with high hydroxide conductivity for use in emerging fuel cell technologies and other electrochemical device applications such as electrolyzers and water purifiers.

Working in conjunction with experimental efforts to synthesize and characterize new membranes (within the paradigm of the Materials Genome Initiative or MGI), we perform coarse-grained (dissipative-particle dynamics or DPD) simulations based on a assumption of local ion coordination structures and protocols for grouping individual atoms into coarse-grained “beads”.  These simulations are being done in collaboration with the group of Stephen Paddison at the University of Tennessee, and they permit us to obtain mesoscale morphologies of the proposed membranes, but they do not allow for structural diffusion of the hydroxide ions.  Thus, from the DPD simulations, we extract idealized local structures, and we employ fully atomistic ab initio molecular dynamics simulation to study hydroxide diffusion mechanism in model AEM architectures. In order to mimic the complicated AEM environment, our theoretical models are composed of fully hydrogenated carbon nanotubes and graphane sandwich structures to which various functional cationic groups are attached, thus allowing different cation chemistries to be investigated.  The space within a tube or between the graphene sheets is filled with water and hydroxide ions to achieve electrical neutrality. In addition to diffusion constants, local solvation patterns are extracted, and these are then fed back into the DPD coarse-grained model  in order to refine the definitions of the coarse-grained beads, and the feedback loop proceeds in this manner.  In the simulations, parameters that are varied include the chemical composition of the cation, the number of cations, the number of water molecules solvating the system, the cation spacing, and the available volume.  Parameter choices that promote the highest hydroxide diffusion constants are then suggested to the experimental groups who are able to turn these simulation results into actual membranes, and measured ion conductivities/diffusivities and morphologies allow the theoretical models to be refined, thus completing the theory/experiment feedback loop.

The figure above illustrates our multi-scale approach to AEM membrane design. On the left is a snapshot of a DPD simulation of the morphology of a polystyrene-b-poly(ethylene-co-butylene)-b-polystyrene (SEBS) block copolymer with tethered tetramethylammonium cations and a water/hydroxide ratio of 8.   The morphology is largely lamellar.  On the right is the atomistic model consisting of two graphane sheets with two tethered tetramethylammonium cations, water, and hydroxide ions.   Yellow spheres show the hydroxide oxygen. Hydrogen bonds in the system are represented with dotted lines.

Our previous work on hydronium and hydroxide ions in bulk water using ab initio molecular dynamics yielded a detailed mechanistic picture for a process known as structural diffusion governing how these ions diffuse through the hydrogen bond network.  This is an idea that is attributed to Christian Johann Theodor von Grotthuss in his 1805 paper entitled “Sur la décomposition de l’eau et des corps qu’elle tient en dissolution à l’aide de électricitè galvanique”.  The basic ideas expressed in this paper were more completely elucidated in the 20th century through extensive experimental and theoretical efforts.  The unifying concept that emerges is that of hydronium and hydroxide ions manifesting themselves as charged topological defects in the hydrogen bond network that propagate via a series of proton transfer reactions that transport this defect through the hydrogen bond network.  The basic steps of the structural diffusion mechanism for hydronium  and hydroxide ions are shown in the figures below (left — hydronium, right — hydroxide).  These snapshots are taken from our ab initio molecular dynamics calculations [see Berkelbach et al. Phys. Rev. Lett. 103, 238302 (2009) and Tuckerman et alAcc. Chem. Rev. 39, 151 (2006)].

The hydronium mechanism clearly shows how a three-coordinated hydronium ion is coordinated by water molecules each having a fourfold coordination shell. It is only when one of these coordinating waters loses an acceptor hydrogen bond that the proton can begin to transfer from the hydronium to this water molecule. The process is force to go to completion when a water molecule begins to form a hydrogen bond at the lone-pair site of a distorted hydronium ion, which, after the proton transfers, completes the coordination shell of the nascent water molecule. The “preparation” of the water molecule to receive the transferring proton via coordination shell equalization is a concept we refer to as the “pre-solvation” concept. In the hydroxide case, we see something unusual, namely, that the hydroxide forms a “hypercoordinated” coordination shell in which it accepts four hydrogen bonds from donating water molecules around the oxygen site. It is only when one of these hydrogen bonds breaks and the hydroxide hydrogen is able to donate a hydrogen bond to a nearby water molecule, thus completing a water-like coordination shell, that the proton is transferred from a neighboring water to the hydroxide oxygen site. This picture is known as the “dynamic hypercoordination” mechanism, and it is fully consistent with the pre-solvation idea. It also shows that the hydronium and hydroxide transport mechanisms, although governed by the same pre-solvation idea, are not merely mirror images of each other.

The hydroxide mechanism, in particular, has been intensely investigated experimentally and has withstood this experimental scrutiny. Importantly, however, the computed and experimentally measured diffusion constants of hydronium and hydroxide at room temperature show that hydroxide diffusion is roughly a factor of 2 slower than that of hydronium. Thus, a key goal of this project will be to find a set of membrane design parameters that not only promote this hydroxide diffusion mechanism but are able to enhance it. Clearly, then, understanding of the atomistic details of the diffusion mechanism is key to learning how to promote it.

Computational methods: Ab initio molecular dynamics is a technique in which Newton’s equations of motion are integrated numerically with forces generated “on the fly” from electronic structure calculations. Details can be found in the book Ab Initio Molecular Dynamics: Basic Theory and Advanced Methods by Marx and Hutter and in review articles such as that by M. E. Tuckerman J. Phys. Condens. Matt. 14, R1297 (2002). Very briefly, if the dynamics occurs on the ground-state Born-Oppenheimer surface, then the exact representation of ab initio molecular dynamics is expressed as

where MI are the nuclear masses, RI are the nuclear coordinates, Ψ0(R) is the exact ground-state electronic wave function Helec(R) is the electronic Hamiltonian, and U(R) is the nuclear-nuclear repulsion.  In all but the smallest of systems, it is not possible to compute the exact ground-state electronic wave function, hence, approximations are needed, and the most usual is to employ Kohn-Sham density functional theory.  The important feature of ab initio molecular dynamics pertinent to these calculations is its ability to describe chemical bond breaking and forming events and to capture electronic polarization effects.

Dissipative-particle dynamics is a coarse-graining approach in which specific groups of atoms become single beads that are then used in Langevin-like dynamics with very simple interactions and distance-dependent friction and random forces.  These are all illustrated in the figures below