Redox cycling stability of Fe₂NiO₄/YSZ composite storage materials for rechargeable oxide batteries

Highlights

• Fe₂NiO₄/YSZ composite storage material for rechargeable oxide batteries.

• Consistent long term cyclability of the composite with no visible capacity fading.

• Increasing reactive surface area with increasing cycle number.

Abstract

The usability of novel Fe₂NiO₄-spinel/YSZ composites as storage components for rechargeable oxide batteries (ROB) is assessed. Continuous thermogravimetric analysis while cycling between oxidizing and reducing conditions revealed a much better robustness of these composites as compared to iron oxide storage materials, which suffer from densification and exhibit severe charge capacity fading when redox-cycled. The scaffolded structure and the nickel containing redox-active spinel greatly reduce such densification and sustain fast redox reactions. Both oxidation and reduction rates of the composites are significantly enhanced by increasing their porosity, leading to better gas penetration of the redox shuttle gas used in ROBs to transfer oxygen between the storage component and the membrane electrode assembly.

Introduction

In 2015 the United Nations Framework Convention on Climate Change (UNFCCC) adopted the Paris climate accord [1] and by 2017 ​195 UNFCCC members had signed the agreement which aims to drastically reduce the global carbon footprint. With the current focus lying on renewable energy sources such as wind and solar energy, safe and efficient means of storage are required. Rechargeable oxide batteries (ROBs) are a relatively new concept in an attempt to deliver such a system [[2], [3], [4], [5], [6], [7], [8], [9], [10]], as they can provide a cheap and safe alternative due to its design and the materials utilized.

The conventional ROB design comprises an all solid state electrochemical cell with an oxide-ion conducting electrolyte, an air electrode, and a composite anode [[2], [3], [4], [5], [6], [7], [8], [9], [10]]. The cell is in contact with the storage component through a H2/H2O buffer gas which acts as an oxygen shuttle as depicted in Fig. 1. As the storage material is physically decoupled from the rest of the cell, the main constraints that apply are imposed by the available thermodynamic window of the composite anode of the cell [2,6]. These restrict the available capacity of the material based on the equilibrium reactions of the material that lie within the thermodynamic window of the anode (Fig. 2) [[11], [12], [13]]. Further constraints on the capacity develop over time through densification and agglomeration of the storage material. In the case of iron-based storage materials, this results in the formation of a dense, metallic iron layer that blocks the penetration of the shuttle gas into the material. As a consequence, the active surface area is now greatly diminished, as the inner surfaces of the porous material do not participate in the oxygen exchange during redox cycling. This leads to a reduced oxygen capacity of the material. The electrochemical degradation was shown by Leonide et al. where over 10000 charge/discharge cycles were completed using pure iron oxide [10]. A decline in round trip efficiency was reported in one case from 93% to 39%, and in a second test from 93% to 56%. This was attributed to chrome poisoning of the cell and densification of the iron. To combat the iron densification in iron-based systems, the addition of inert scaffolding such as yttrium stabilized zirconia (YSZ) was proposed by Menzler et al. [6] and Berger et al. [2,7]. They found that YSZ/Fe2O3 composites were effective in reducing the iron layer formation at the surface. At high YSZ volume fractions (70 vol-%) no layer formation occurred at all. Electrochemical investigations with a two cell stack using a storage material (30 vol-% YSZ/70 vol-% Fe₂O₃) was also reported showing 200 charge/discharge cycles. Performance reports are inconclusive as the cells were not operated fully reversible due to occasional corrections of the current density caused by leakages.

Further investigation into other oxides with potential utilization as scaffold material resulted in the calcium-iron oxide system [14], in which a scaffold of CaO is formed during the reduction process, which prevents the formation of larger iron agglomerates. It is also reported that during oxidation the calcium-iron oxide compound is reformed quickly preventing the formation of surface iron layers. However, the issue of capacity fading remains for this system, which still shows capacity drop of about 50% over the first 10 cycles.

Another approach is presented by Kim et al. [4] which utilized a catalyst material to improve the cyclability of iron oxide. They reported an increased conversion of metallic iron to Fe₃O₄ by about 40% of the total oxygen capacity (20% conversion without catalyst to 60% conversion with catalyst) over 8–10 ​h of oxidation with the addition of their catalyst (Ce_0.6Mn_0.3Fe_0.1O₂). This change remained relatively constant for 1%–20% of added catalyst material.

It is noted that all literature sources mention some form of degradation or densification. However, no quantification of the actual oxygen exchange with the surrounding is made. Furthermore, the electrochemical results mostly do not utilize the full capacity of the storage material, and are subject to degradation effects of the electrochemical set up, which makes it difficult to separate the effects that results directly from the storage material.

In this work we present an alternative storage material, namely a composite consisting of the nickel-iron spinel and YSZ. Thermogravimetric investigations of the oxygen exchange between material and surrounding show no signs of capacity fading, while retaining the ability to utilize the full capacity of the material comparably fast. Due to the thermogravimetric approach, a correlation between the sample morphology and the oxygen storage capacity of the material can be made.

In this work we look at a nickel-iron spinel (Fe₂NiO₄) as a possible redox mass in combination with YSZ scaffolding. The nickel-iron spinel operates in a similar thermodynamic window as pure iron oxide [15] making it a suitable storage material to be used combination with the commonly used Ni/YSZ anodes. The first step in determining the suitability of the composite is the general long term redox behavior of the material. Compared to pure iron oxide, the spinel offers the advantage of inhibiting cation mobility with increasing nickel concentration [16]. It is expected that this curbs the formation of dense surface layers, which are responsible for capacity fading.

The redox equilibria of the involved oxide phases are shown in Fig. 2. However, determining the exact redox equilibrium for Fe₂NiO₄ composite is difficult as Fe_2+xNi_1-xO₄ with varying nickel content may form during the redox processes, and the equilibrium shifts with varying nickel content [15]. The assumed reactions are shown in Eqs. (1), (2), (3)).

$$\text{Fe }+{\text{ H}}_2\text{O }\rightleftharpoons {\text{H}}_2\text{ }+\text{ FeO}$$

$$3\text{FeO }+{\text{ H}}_2\text{O }\rightleftharpoons {\text{H}}_2\text{ }+{\text{ Fe}}_3{\text{O}}_4$$

$$\text{ }(1-\text{x}){\text{Fe}}_3{\text{O}}_4+\text{xNi }+\text{ }\frac{4}{3}{\text{xH}}_2\text{O }\rightleftharpoons {\text{ Fe}}_{3-x}{\text{Ni}}_x{\text{O}}_4+\frac{4}{3}{\text{xH}}_2\text{ }$$

One of the main advantages of the ROB concept is the high energy density that comes with the metal-air storage system. The iron nickel spinel has a specific capacity of 915 ​mAh⋅g⁻¹ while pure hematite is slightly higher at 1007 ​mAh⋅g⁻¹. These materials still have high capacities even when applying large amounts of scaffolding (e.g. for an equimolar ratio of YSZ to Fe₂NiO₄ the specific capacity lies at 368 ​mAh⋅g⁻¹). If only the first two oxidation steps are considered (Fe → FeO → Fe₃O₄, with nickel remaining metallic), the specific capacity for the pure spinel is reduced by 33%–610 ​mAh⋅g⁻¹ (Fig. 3). The specific capacity of the composite would then further depend on the amount of necessary scaffolding. To ensure an interconnected scaffold, a minimum of 30 ​vol.-% is required. This translates to a composition of 75.8 ​mol.-% spinel to 24.2 ​mol.-% YSZ. Using these numbers, a theoretic specific capacity of 413 ​mAh⋅g⁻¹ can be determined.

The second consideration to be made is that the anode needs to remain electronically conductive, which for the case of an Ni/YSZ electrode would limit the highest pO₂ during the discharge process to the Ni/NiO equilibrium pO₂. This would limit the capacity to the formation of Fe₃O₄, but keep the voltage in the region between 0.9 and 1 ​V ​at 1073 ​K (Eq. (4)) as given by the Nernst equation.

$$\text{OCV}=-\frac{RT}{4F}\text{ln}\frac{p{\text{O}}_2^{\text{redox}}}{p{\text{O}}_2^{\text{air}}}$$

Here OCV represents the open circuit voltage, R the Boltzmann constant, F is Faradays constant, T is the temperature, and the ratio $p{\text{O}}_2^{\text{redox}}/p{\text{O}}_2^{\text{air}}$ is given by the oxygen partial pressures at the cathode (air electrode) and anode according to the schematic in Fig. 1.

The calculated values for the proposed material exhibit better specific and volumetric capacities despite the limitations imposed by the thermodynamic window and the scaffold than the current common intercalation cathodes used for lithium ion batteries. These are the LiCoO₂ (LCO) cathodes introduced by Goodenough [17] which provide a theoretical specific capacity of 274 ​mAh⋅g⁻¹, in addition to a multitude of other materials shown in a review by Nitta et al. [18]. Here the capacities for various lithium ion battery cathodes materials are shown for current technology that is commercially available as well as various materials which are being researched, as the cathode capacity is the limiting factor for lithium ion batteries [18,19]. The materials mentioned give theoretic capacities in a range of 140–490 ​mA ​h⋅g⁻¹ which has been visualized in Fig. 3. The materials included consist of various crystal structures (layered, spinel, olivine, and tavorite) and compositions ranging from LiNi_xCo_yMn_1-xO₂ known as NCM materials [19,20] to phosphates and sulfates such as LiMnPO₄ [21].

The achievable capacities for the lithium ion cathodes are all significantly below that of the proposed system in this work. Even when adding significant amounts of scaffolding to the proposed ROB redox masses, the theoretical values are at worst comparable with the theoretic values of the lithium battery cathodes. Furthermore, the theoretical values are mostly significantly higher than the actually achievable ones, so it will be of interest to see how the ROB material can compare. In theory the capacity up to the Fe₃O₄ phase should be fully usable, however possible side reactions with the stabilizing lattice could influence this.