Exsolved catalyst particles as a plaything of atmosphere and electrochemistry

Abstract

A new type of catalyst preparation yields its active sites not by infiltration but exsolution of reducible transition metals of its own host lattice. These exsolution catalysts offer a high dispersion of catalytically active particles, slow agglomeration, and the possibility of reactivation after poisoning due to redox cycling. The formation of exsolved particles by partial decomposition of the host lattice can be driven by applying a sufficiently reducing atmosphere, elevated temperatures but also by a cathodic bias voltage (provided the host perovskite is an electrode on an oxide ion conducting electrolyte). In addition, such an electrochemical polarisation can change the oxidation state and thus the catalytic activity of exsolved particles. In this work, we investigate the electrochemical switching between an active and an inactive state of iron particles exsolved from thin film mixed conducting model electrodes, namely La_0.6Sr_0.4FeO_3-δ (LSF) and Nd_0.6Ca_0.4FeO_3-δ (NCF), in humid hydrogen atmospheres. We show that the transition between two activity states exhibits a hysteresis-like behaviour in the electrochemical I-V characteristics. Ambient pressure XPS measurements proofed that this hysteresis is linked to the oxidation and reduction of iron particles. Furthermore, it is demonstrated that the surface kinetics of the host material itself has only a negligible impact on the particle exsolution, and that the main impact factors are the surrounding atmosphere as well as the applied electrochemical overpotential. In particular, we suggest a 'kinetic competition' between gas atmosphere and oxygen chemical potential in the mixed conducting electrode and discuss possible ways of how this process takes place.

Fig. 1. Sample sketch: mixed conductor (thin film) (1), buried Pt current collector (2) grid, yttria stabilised zirconia electrolyte (3), and the counter electrode consisting of GDC (spin-coated) (4), Pt-GDC (brushed) (5), Pt (brushed) (6).

Fig. 2. Sketch of the ex situ measurement setup: impedance analyser or DC source meter (1), multimeter measuring U_Lambda (2), inner tube (3), gas inlet (4), custom lambda probe consisting of a hollow YSZ-pipe with Pt-contacts at sample height (5), thermocouple (6), spring loaded middle tube (7), sealed outer tube (8), sample consisting of counter electrode, electrolyte and working electrode (9–11), silica spacer (12) and Pt contact sheets and mesh (13).

Fig. 3. Flowchart of the method used for exsolution onset determination where each measurement cycle consists of three steps: exsolution is triggered by application of a specific cathodic potential U_cath for a certain acquisition time t_hold. Then an I–V curve is recorded with varying step sizes. If no step change is observed the applied cathodic potential U_cath is decreased further for the next iteration.

Fig. 4. Exsolution onset determination with different (a) materials (LSF, NCF), (b) atmospheres (H₂ : H₂O = 10 : 1, 30 : 1) or (c) t_hold (1 min or 1 h). Each of the figures show the anodic branch of the I–V characteristics of a pristine sample (without any activity changes) and one after the first visible discontinuity in the anodic current (with the legend caption showing the necessary U_cath). (d and e) Summary of the switching and onset potentials of all varied parameters. The switching potential of a species for a given atmosphere remains unchanged. An increase in t_hold or a more reducing atmosphere result in a lower onset potential. The hatched sample had different manufacturing conditions (see text for further information).

Fig. 5. (a) In situ recorded I–V curve for LSF in H₂ : H₂O ≈ 16 : 1 with the arrows and numbers indicating measurement direction. (b) Zoom in on the part of the activity change (marked by the rectangle in (a)). (c) Fe 2p XPS spectra consisting of two species: One oxidised species at 710.5 eV and a metallic species at 707 eV (position indicated by the wedge). The large arrow indicates the recording direction. (d) Amount of Fe⁰ compared to the total amount of surface iron based on the applied overpotential.

Fig. 6. I–V Characteristics of LSF and NCF at very low oxygen partial pressures (specified as the ratio between H₂ and H₂O with increased H₂ content) showing the current step change due to oxidation or reduction of the exsolved iron particles in both directions (indicated by the dashed/dotdashed line). The arrows specify the measurement direction. The grey wedge represents the thermodynamically expected switching potential (see Section 4 for more information).

Fig. 7. I–V Characteristics of LSF and NCF at very low oxygen partial pressures (specified as the ratio between H₂ and H₂O with increased H₂O content) showing the current step change – if there is one – in both directions (indicated by the dashed/dotdashed line). The arrows specify the measurement direction. The grey wedge represents thermodynamically expected switching potential (see Section 4 for more information). The inset in (f) represents the current density over time for the highlighted region.

Fig. 8. Sketch to explain the difference in switching potential visibility based on measurement direction: (a) an increase in η results in a lower i which ends up being a real step change. (b) A further decrease in η only results in a steeper curve, but no real step change.

Fig. 9. (a) Sketch of all gradients in the oxygen chemical potential Δμ_O of the involved phases. (b–d) Three different possibilities for the switching behaviour where either Δμ^MIEC-gas_O ≈ Δμ^MIEC-part._O (b) or Δμ^MIEC-gas_O ≈ Δμ^part.-gas_O (c) applies as well as an intermediate one with both, MIEC and gas, affecting the particle (d). (e) Resulting Fe/FeO equilibria of all three cases shown in a semilogarithmic plot of the overpotential η at the working electrode versus the equivalent oxygen partial pressure in the atmosphere. The dashed lines represent the isobars of .

Fig. 10. Semilogarithmic presentation of the working electrode overpotential η versus the oxygen partial pressure of the gas analogous to Fig. 9e but including extracted step changes in the I–V plot for both mixed conducting electrode materials, LSF and NCF. Further distinctions are made with regards to the recording direction of the I–V curves (see legend) and the of the previous I–V recording, i.e. a higher (filled markers) or a lower (half-full markers) one.

Fig. 11. (a) Simulated I–V curve with a switching potential in the anodic branch showing a large hysteresis defining three significant points (schematic explanation in (b)) based on the measurement direction and the dominant reaction taking place.