Fluidized Electrocatalysis

Introduction

There has been extensive effort in developing new electrocatalytic nanomaterials, which are usually deposited and glued on the surface of an electrode with the help of binders and conductive additives, and then immersed in electrolyte to perform electrochemically driven reactions. A common problem in electrocatalysis is the rapid drop of catalyst performance (hereafter referred to as fatigue),^1–3 which can be attributed to a number of material degradation mechanisms summarized and illustrated in Figure 1a, including surface passivation or poisoning from reaction intermediates,⁴ dissolution, migration, agglomeration and even sintering of catalytic particles,^5–7 and detachment or pulverization of active components from their support or the electrode,^8,9 as well as mass transport limitation of electroactive molecules to the catalysts.^10,11 Fatigue of catalysts greatly reduces their efficiency, shortens their lifetime, and degrades long-term performance of the corresponding energy conversion and storage systems. There have been tremendous interests and efforts to make electrocatalysts more stable and fatigue resistant. For example, morphology of the catalytic particles can be tuned to expose their most active and robust crystallographic surfaces for the reactions.^12–16. The surface states and chemical composition of the catalysts can be adjusted such as by alloying, doping and mechanical strains to improve their stability while maintaining catalytic activity.^17–20 Support materials can also be customized to help prevent the catalyst nanoparticles from agglomeration and detachment.²¹ All of these strategies focus on improving the material stability of the catalysts in the electrochemical reaction, therefore, the corresponding solutions may be applicable only to a specific set of catalysts.

• Download figure
• Download PowerPoint

In electrocatalysis, electron transfer between the electrode and deposited catalyst is coupled with a number of slower processes such as adsorption of reactants, conversion to intermediates, formation and then desorption of products, as well as mass transport (i.e., molecular diffusion) between catalyst surface and solution, all of which are convoluted at or near the surface of the catalysts. Although electrical field is only needed momentarily for the electron transfer step, it is constantly applied through the electrode, exerting unnecessarily extra electrochemical stress on the catalysts. Here, the electrochemical stress refers to all the factors degrading the performance of catalysts that are triggered by the applied electrochemical potential. This is intrinsically a rather ineffective way to utilize the catalysts and the surface area of the electrode, which, inevitably, aggravates the fatigue problem regardless of any improvement over the catalytic materials.

Here we report a fluidized strategy to enhance fatigue resistance of electrocatalysis. As illustrated in Figure 1b, the catalytic materials, instead of being fixed on an electrode, are now fluidized in the electrolytes. Reaction is catalyzed by individual catalytic particles only when they interact with the electrodes, which collectively delivers a continuous, stable and scalable electrochemical current output. Since the particles now work in rotation, it drastically reduces the time scale of the electrochemical stress applied to the catalyst, thus suppressing many common fatigue mechanisms. Moreover, since electron transfer is now spatially and temporally decoupled from the other relatively slower steps (e.g., surface chemical reactions and mass transfer), fluidized catalyst experiences faster kinetics and can be much more efficiently used. As demonstrated below using three model reactions with different fatigue mechanisms, including oxygen evolution reaction (OER), methanol oxidation reaction (MOR) and hydrogen evolution reaction (HER), fluidized electrocatalysis can indeed deliver much more stable electrochemical performance over drastically extended reaction time, even with intentionally selected unstable catalysts.

It is worth noting that fluidized catalysis is conceptually different than previously studied fluidized bed electrodes,^22,23 in which the electrolyte and reactants are pumped through loosely packed, conductive beads that collectively maintain electrical contact to the electrode. Fluidized bed essentially acts as a dynamic but continuous, porous and higher surface area extension of the electrode, which is unrelated to fluidized electrocatalysis reported in this work.

Experimental Methods

Materials

Pt/C (i.e., Pt nanoparticles on carbon black powders) was chosen as the model catalyst in this work, because it is not an optimized, and often unstable (chemically or mechanically) catalyst in the 3 types of electrochemical reactions studied here. This makes them a good example to test the effect of the fluidized strategy. All Pt/C particles except those used Figure 2 were purchased from Alfa Aesar. Pt/C particles used in Figure 2 was purchased from Suzhou Technology Co., Led, which were used mainly for single particle level studies with the micro-working electrode. Both particles have a nominal Pt loading of 20% on carbon black (HiSPEC^TM 3000). H₂SO₄, methanol and NaOH were purchased from Sigma-Aldrich. All the electrolytes were prepared with deionized water.

• Download figure
• Download PowerPoint

Electrochemical measurements

All electrochemical measurements were conducted on an Autolab potentiostat (Metrohm-Autolab BV, Netherlands) with a three-electrode system at 298 K. Ag/AgCl electrode was used as the reference electrode (RE). Pt plate (20 mm × 20 mm, for MOR and OER) or carbon rod (for HER) was used as the counter electrode (CE). Glassy carbon microelectrode (d = 10 µm) was used as the working electrode (WE) for OER to reduce the number of particle-electrode collision events for resolving single particle level events. Higher current output was obtained using a large double-sided glassy carbon plate (25 mm × 25 mm × 1 mm) with an immersed area of 500 mm². All fixed electrode reactions and fluidized reactions were operated under magnetic stirring set at 1400 rpm in an electrochemical cell containing 50 mL of electrolyte. In general, the overall current output would increase with stirring, but it could eventually decrease when the particle-electrode collision is hindered at high shear.²⁴ In our experiments, all the stirring rates were set to be 1400 rpm, which were found to be nearly optimal for reactions using glassy plate electrode. Pt/C was first sonicated to disperse in water and drop casted on the glassy carbon plate surface or injected in a fluidized cell.

OER was carried out in a solution made of 0.1 M KOH under a fixed electrode potential of 1.6 V (vs. RHE) with magnetic stirring. For fluidized reaction, Pt/C loading in the electrolyte was kept at 1 mg/mL for microelectrode experiment, and 0.2 mg/mL for the large plate reactions. For fixed reaction, 2.5 mg Pt/C particles were modified on a glassy carbon plate. Chronoamperometric profile was recorded at high temporal resolution of 3 ms for microelectrode reactions, and 1 s when using large plate electrode (Figure 2). Control experiment was done without Pt/C particles present in solution, or with Pt/C but under an applied potential of 1.2 V (vs. RHE), which is well below the onset OER potential (See ). The effect of electrode area on the OER current (Figure 2e) was studied for fixed Pt/C at a glassy carbon plate with varying immersed electrode area. The effect of catalyst loading on the OER current (Figure 2f) was studied for both fixed and fluidized Pt/C at the glassy carbon plate with an immersed area of 500 mm². The currents are extracted from the corresponding chronoamperometric profiles recorded under 1.6 V (vs. RHE) at reaction time of 500 s.

MOR was carried out in 0.5 M H₂SO₄ and 1 M CH₃OH under a fixed electrode potential of 0.7 V (vs. RHE) at a glassy carbon plate. HER was carried out in 0.5 M H₂SO₄ at the potential of −0.15 V (vs. RHE). The voltage was chosen so that the HER show fast kinetics at Pt/C electrode but sluggish at a bare glassy carbon electrode (see ). For both OER and HER, 2.5 mg Pt/C particles modified on a glassy carbon plate were used for fixed catalyst reaction. 0.1 mg/mL and 0.16 mg/mL Pt/C loadings in the electrolyte were used for fluidized MOR and HER, respectively.

Electrical charge passed per spike from single particle experiments was obtained using “Peak Analyzer” function in Origin software. The duration of each spike was measured from where the current starts to rise from baseline to where it goes back to baseline.

TEM characterization of Pt/C particles before and after fixed and fluidized MOR and OER was done on a Hitachi H-8100 microscope. After reaction, fluidized particles were collected and drop cast on TEM grids. Particles fixed on plate electrodes were gently sonicated to disperse in water, and then drop cast on TEM grids. Control experiments showed that the sonication step does not alter the morphology of the Pt/C particles. Pt nanoparticle size was measured using DigitalMicrograph software.

Results and Discussion

Model reaction 1: OER and its fatigue mechanisms

OER was chosen as the first model reaction to demonstrate the proof-of-concept, because it is notoriously harsh for its catalysts due to its high operational potential. Studies have shown that the reaction tends to inflict significant structural damage to the catalytic nanoparticles, leading to severe dissolution and sintering of the nanoparticles.^7,25 Corrosion of the support materials (e.g., carbon) can also occur under such potential, which further escalates catalyst degradation. We chose the commercially available Pt/C material (i.e., Pt nanoparticles on carbon black powders) as our prototype catalyst due to its widespread use in electrocatalysis, and its poor stability under the OER conditions.^7,26 This helps to demonstrate how a rapidly degrading electrocatalytic reaction becomes fatigue resistant using the fluidized strategy.

Fluidized OER: Transient current and reaction timescale

In a fluidized electrocatalytic reaction, the suspended Pt/C particles collide on electrode surface, collectively generating a continuous current output. The overall current output should be larger when using higher particle concentration or a larger working electrode. In order to better resolve the transient currents generated by individual collision events, a glassy carbon microelectrode (10 µm in diameter) was used as the working electrode (WE) to reduce the frequency of particle collision events (Figure 2a). Working electrodes with larger area, such as a glassy carbon plate were used to collect higher overall current output (Figure 2b).

Figure 2c shows a segment of the high-temporal-resolution chronoamperometric profile of a fluidized OER reaction, recorded with the micro-carbon electrode, where single particle level current transients can be observed. Only those strong current transients with a signal-to-noise ratio greater than 3 are considered as valid signals. The overall chronoamperometric profile and the shape of current spikes (Figure 2c, inset) agree well with prior observations made in single particle electrochemistry,²⁷ which has already established that the duration of these faradic current transients corresponds to the time scale of particle catalyzed reaction during collision with the electrode. An analysis on the peak width of the current transients of over 100 spikes of the fluidized OER reaction at the microelectrode shows that over 90% of spikes are shorter than 40 milliseconds (see for additional chronoamperometric segments and for a histogram of spike duration), which is consistent with the time scale of particle-electrode collision observed in prior studies.²⁸ This drastically reduces the characteristic OER time scale and the electrochemical stress experienced by the Pt/C particles from continuum down to tens of milliseconds.

High-resolution chronoamperometric profiles like the one shown in Figure 2c and can be used to calculate the collision frequency of Pt/C particles (1 mg/mL) onto the microelectrode (10 µm in diameter) based on the number of signals observed in a given period of time, which is found to be around 0.4 s⁻¹. Control experiments were carried out by removing Pt/C particles (Figure 2d, black line) or holding the applied potential at 1.2 V (vs. RHE), which is well below the onset of reaction potential (see ). The corresponding chronoamperometric profiles show near zero current, and no transient current spikes (Figure 2d, inset), confirming that the electrochemical oxidation processes during the collision of Pt/C particles onto the electrode surface is the source of the current spikes. Note that the movement of Pt/C particles are driven by the flow of electrolytes rather than diffusion, and the size of the particles is much small than the size of the electrodes, thus the collision frequency is mainly influenced by the stirring speed but not the particle size. However, larger particle size does influence the height of current transients produced during collision events.

Collective OER current output: Effect of electrode area and particle concentration

Figure 2e displays plots of the current collected from a large plate electrode with a function of the immersed electrode area. Since larger electrode would experience more catalyst collision events, the current output of fluidized reaction scales nearly linearly with the electrode area. On the other hand, adding more particles to the fluidized reaction also increases the current output (Figure 2f, red line). This is in contrast to reactions using fixed catalysts, the current of which tends to saturate at very low mass loading levels of Pt/C (Figure 2f, black line) due to close packing of particles. Although only a small percentage of fluidized particles are contributing to the reaction at any given moment, however, as is discussed in the next section, the fluidized catalysts have much higher current efficiency than their counterparts fixed on electrode. This leads to better scalability of fluidized reaction (Figure 2f, black line). For example, the current output from 10 mg of fluidized Pt/C particles, although not saturated yet, has already reached about 40% of the maximal current that can be possibly delivered by fixed catalysts. Note that the current outputs from both fixed and fluidized OER are extracted from the corresponding chronoamperometric profiles recorded under 1.6 V (vs. RHE) at only 500 s after the reactions started. The comparison was made at this early stage of reaction before severe current decay of fixed OER. Although the initial current output from fluidized OER is smaller than the fixed one, overtime fluidized OER would deliver much higher overall output due to higher fatigue resistance (also see Figure 3 and discussion later).

• Download figure
• Download PowerPoint

Fluidized v.s. fixed OER: Particle-average current output and efficiency

As illustrated in Figure 1, fluidized catalyst particles should have much higher mass transport efficiency than those fixed on an electrode.^27,28 As electrochemical reaction proceeds, particles fixed on an electrode tend to develop an expanding depletion zone before reactants can be replenished near their surface. For well isolated particles (i.e., at low loading level or low surface coverage on the electrode surface), the diffusion mode of reactants near their surface is radial, which transits into a much less efficient linear mode as the particles are closely packed at higher loading levels. In earlier studies, it has been found that the single particle efficiency at very low loading level (e.g., 0.01–0.1% surface coverage of single particles) can be over two orders of magnitude higher than that from close packed particles.^29,30 Since mass transport efficiency is reflected by the Faraday current generated by single catalytic particles, the single particle efficiency (i.e., particle-average current output) of fluidized and fixed particles can be estimated based on their contribution to OER currents as follows.

For fixed catalyst Pt/C particles, contribution from individual particle can be estimated by dividing the total catalytic OER current by the total number of fixed particles. The density of Pt/C particles is estimated to be 2.3 g/cm³, based on the densities of carbon black (1.8 g/cm³) and Pt (21.45 g/cm³), and the Pt loading level (20 wt.%). The average diameter of dispersed Pt/C particles or clusters is taken to be around 600 nm based on dynamic light scattering measurement, and confirmed by TEM observations (see ). If we approximate the shape of Pt/C particles to be a sphere, the mass of one such particle is calculated to be 2.5 × 10⁻¹³ g. As discussed earlier and shown in Figure 2f, the OER current from fixed Pt/C electrode rapidly saturated as the particle loading increases. Therefore, here we calculate the particle-average current output of the fixed catalysts using the results obtained from near the lowest particle loading level (i.e., 0.25 mg), so that their single particle efficiency would not be underestimated. Since 0.25 mg of Pt/C contains 1 × 10⁹ particles and collectively delivered 0.5 mA, the particle-average contribution is calculated to be 5 × 10⁻¹³ A.

The particle-average current output for fluidized Pt/C is taken to be the average current produced during a collision event, which is calculated by dividing the total Faraday charges produced in a current spike (i.e., the area of the spike) over its duration. The particle-average current output from the fluidized OER is calculated to be 6 × 10⁻¹⁰ A based on the analysis of 100 transient spikes on high-resolution chronoamperometric profiles. This is about three orders of magnitude higher than the particle-average efficiency in the fixed OER, which highlights the potential of fluidized electrocatalysis. There should be plenty of room for improvement to scale up the collective current output of fluidized reactions, such as by optimizing flow profile of electrolyte and the geometry of the electrode and reaction vessels.

Fluidized v.s. fixed OER: Fatigue performance

Since particles in fluidized reaction work in rotation, they would not experience the buildup of electrochemical stress as in fixed catalyst reactions. Therefore, fluidizing the catalysts can significantly reduce the degree of degradation of the Pt nanoparticles (e.g., agglomeration or sintering), which helps to maintain a stable current output. Figure 3a shows the results of “stress test” (i.e., long term chronoamperometric measurement) of a fixed and a fluidized OER under the same operating voltage of 1.6 V (vs. RHE), using identical electrodes (500 mm² double-sided glassy carbon plate). The catalyst loading levels were optimized based on the results shown in Figure 2f, which are 2.5 mg for fixed reaction and 10 mg for fluidized reaction, so that both types of reactions started from their near-maximal currents. As expected, the current output from fixed Pt/C particles exhibited a sharp drop during the first 500 seconds, and diminishes to around 1% of the initial value after 13,000 seconds (Figure 3a, black line). Before reaction, the Pt nanoparticles are around 2–5 nm in diameter and evenly distributed on the carbon support (see TEM image in ). However, after just 500 seconds of reaction, the morphology of Pt/C particles has drastically changed. Transmission electron microscopy (TEM) observation revealed significant degree of sintering of Pt nanoparticles (Figure 3b, also see for additional TEM images), which is a clear sign of catalyst degradation. In contrast, the fluidized Pt/C particles yield a quite stable current (Figure 3a, red line) and show no significant change in morphology (Figure 3c, also see for additional TEM images) even after 60,000 seconds. The drastic different particle size distribution is evident in the histograms in Figures 3d and 3e, based on the analysis of around 250 particles after the corresponding reactions. Note that the particle-average contribution (i.e., electrochemical charges) after 60,000 seconds of fluidized OER has far exceeded that of 500 seconds of fixed OER, suggesting that Pt/C particles in the fluidized reaction indeed degrade much more slowly.

Model reaction 2: MOR

Pt/C catalyzed MOR is used as the second model system to test fluidized electrocatalysis, as its fatigue problem is notoriously complex and difficult.³¹ As shown in the chronoamperometric profile of a typical MOR (), its current output decays by half after just 200 seconds. The primary fatigue mechanism for MOR has been attributed to poisoning of Pt by reaction intermediates such as CO, formic acid and/or formaldehyde, among which CO is particularly hard to remove. Applying higher potential could help to remove CO by oxidation, but it escalates other fatigue mechanisms such as electrochemical sintering of Pt nanoparticles³² and the formation of inactive oxides on Pt surface.³³ Mass transport limitation of methanol also contributes to decaying MOR performance.¹⁰

Figure 4a shows the result of stress test of Pt/C particles under both fixed and fluidized conditions for MOR. While the current from the immobilized Pt/C decays by over 60% in just 7,000 seconds, the fluidized Pt/C delivers a stable current over a much longer period of 30,000 seconds. Although the initial current generated from fixed Pt/C is more than twice of that from fluidized particles, it is surpassed by the latter at around 9,000 seconds. TEM studies revealed that Pt particles on the fixed Pt/C have been extensively displaced, agglomerated and sintered after 7,000 seconds (Figure 4b and also see for additional TEM images), while no obvious change can be observed for the fluidized Pt/C after 30,000 seconds () or even after 150,000 seconds (Figure 4c, and also see for additional TEM images). The corresponding histograms showing the size distribution of Pt nanoparticles are shown in Figures 4d and 4e, based on the size analysis of over 150 particles after both types of reactions.

• Download figure
• Download PowerPoint

The fatigue resistance of fluidized MOR may also be partially attributed to the greatly shortened reaction time scale, which is known to generate different intermediates than continuous reactions. Earlier mechanistic studies have found that in the very early stage (e.g., within tens of milliseconds) of Pt/C catalyzed MOR, the dominating MOR intermediates are the more soluble species such as formic acid and/or formaldehyde, after which CO generation becomes dominating.³⁴ Therefore, if the characteristic electrochemical reaction time scale can be shortened to just tens of milliseconds, the CO generation pathway, which is responsible for the stubbornest Pt positioning mechanism, can be suppressed. In fluidized electrocatalysis, since the continuum of the reaction is broken into numerous transient electrocatalytic events, the time scale of individual reactions is now shortened to just tens of milliseconds (also indicated in Figure 2c).²⁷ Fluidizing Pt/C particles has additional benefits, such as expedited desorption of reaction intermediate³⁵ and mass transfer between the bulk solution and the surface of the particles to disrupt the buildup of a surface depletion zone. All of these features may have contributed to the high fatigue-resistance of fluidized MOR.

Model reaction 3: HER

Next, we tested another fatigue mechanism due to particle detachment during electrocatalysis. Pt/C catalyzed HER is chosen to study this problem, because Pt/C itself is quite a robust catalyst for HER, and thus is a good model system to highlight the effect of particle detachment in performance decay. In such gas evolution reactions, the nucleation and growth of gas bubbles occur in between the catalyst particles or on their surface, which eventually leave the particles. Electrolyte near the particles is repeatedly displaced and refilled during bubble evolution, applying cyclic local mechanical stress while flushing the catalytic particles, which are often held together and fixed on the electrode by binder materials. This weakens the connection between the particles, and their adhesion on the electrodes, especially when the reaction is carried out with large area electrodes or high catalytic loading. The problem is further aggravated at high gas evolution rate at high operating current. In such reactions (e.g., Pt/C catalyzed HER), even if the catalysts do not suffer significant materials degradation, pulverization or detachment from electrodes leads to rapid performance decay.⁸ Here, fluidizing the catalysts fundamentally avoids this problem. Figure 5a compares the performance of fixed and fluidized Pt/C for HER. A current decay of 65% was observed for fixed Pt/C after 10,000 s of reaction. Figure 5b displays zoom-in photos of the reaction cell after 10,000 s of fixed electrode reaction, showing significant amount of Pt/C sediment due to detachment from the electrode. In contrast, fluidized HER is free from such concern, and therefore maintains 80% of the initial current after 50,000 s.

• Download figure
• Download PowerPoint

Conclusions and outlook

In summary, fluidized electrocatalysis spatially and temporally de-convolutes electron transfer from other slower molecular processes in electrocatalysis, leading to higher catalyst efficiency, higher catalyst stability and better scalability. Fluidizing the catalysts avoids the buildup of excessive electrochemical stress, and suppresses or mitigates a number of degradation mechanisms of the active materials, thus drastically increasing their fatigue-resistance. The fluidized approach should be largely agnostics to catalytic materials and reactions, therefore, it could work in conjunction with the extensive effort in catalyst design^36,37 to improve the overall performance of electrocatalytic systems.

Since fluidized electrocatalysis would require the use of flowing catalytic particles, it could encounter volumetric constrains in volume-sensitive applications. However, it should be suitable for bulk volume applications, such as stationary power supplies or scaled up electrocatalytic syntheses. For the same catalyst-reaction combination, the fluidized version should deliver much more stable performance and overall higher output than the fixed one, and the catalyst can be more easily collected and recycled by simple separation techniques such as filtration. More work is needed to identify the most suitable electrochemical reactions that maximize the benefit of fluidized electrocatalyst. More quantitative understanding is also needed about how the particle-electrode collision frequency is affected by the electrolyte flow profile,²⁴ including the configuration and geometry of the electrodes and reaction vessel, and stirring speed (e.g., see ), so that collision frequency can be maximized at given particle concentrations to achieve optimal overall current output.

The fluidized approach could allow low cost, more abundant, but otherwise unstable electrocatalysts to be used with drastically extended time of operation. Balancing the volumetric effect and the catalyst cost, the fluidized system should be a promising alternative. The work-in-rotation strategy provides a new insight for high efficiency utilization of electrocatalysts, and for potentially designing new reaction mechanisms based on the drastically shortened reaction time scale and discontinued mode of operation.

Supporting Information

Supporting Information is available

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

All authors contributed to the design of the experiments and the analysis of the results. Y.K. performed all the TEM work. Y.Z. performed all the electrochemistry experiments at Northwestern and later at Hunan University, and drafted the paper with J.H., who conceptualized fluidized electrocatalysis.