Volume 8, Issue 2 p. 378-385
Article
Open Access

Indirect Formic Acid Fuel Cell Based on a Palladium or Palladium-Alloy Film Separating the Fuel Reaction and Electricity Generation

Dr. Elena Madrid

Dr. Elena Madrid

Department of Chemistry, University of Bath, Claverton Down, BA2 7AY Bath, UK

Search for more papers by this author
Catajina Harabajiu

Catajina Harabajiu

Department of Chemistry, University of Bath, Claverton Down, BA2 7AY Bath, UK

Search for more papers by this author
Robyn S. Hill

Robyn S. Hill

Department of Chemistry, University of Bath, Claverton Down, BA2 7AY Bath, UK

Search for more papers by this author
Dr. Kate Black

Dr. Kate Black

School of Engineering, University of Liverpool, Liverpool, L69 3BX UK

Search for more papers by this author
Dr. Laura Torrente-Murciano

Dr. Laura Torrente-Murciano

Department of Chemical Engineering and Biotechnology, University of Cambridge, Philippa Fawcett Drive, Cambridge, CB3 0AS UK

Search for more papers by this author
Dr. Angus J. Dickinson

Dr. Angus J. Dickinson

Johnson-Matthey Fuel Cells, Swindon, UK

Search for more papers by this author
Dr. Philip J. Fletcher

Dr. Philip J. Fletcher

Materials and Chemical Characterisation Facility (MC2), University of Bath, Claverton Down, BA2 7AY Bath, UK

Search for more papers by this author
Dr. Kenneth I. Ozoemena

Dr. Kenneth I. Ozoemena

School of Chemistry, Molecular Sciences Institute, University of the Witwatersrand, Private Bag 3, Wits PO, ZA-2050 Johannesburg, South Africa

Search for more papers by this author
Dr. Adewale K. Ipadeola

Dr. Adewale K. Ipadeola

School of Chemistry, Molecular Sciences Institute, University of the Witwatersrand, Private Bag 3, Wits PO, ZA-2050 Johannesburg, South Africa

Search for more papers by this author
Dr. Emeka Oguzie

Dr. Emeka Oguzie

Department of Chemistry, Electrochemistry & Materials Science Research Laboratory, Federal University of Technology Owerri, Owerri, Nigeria

Search for more papers by this author
Dr. Chris O. Akalezi

Dr. Chris O. Akalezi

Department of Chemistry, Electrochemistry & Materials Science Research Laboratory, Federal University of Technology Owerri, Owerri, Nigeria

Search for more papers by this author
Dr. Frank Marken

Corresponding Author

Dr. Frank Marken

Department of Chemistry, University of Bath, Claverton Down, BA2 7AY Bath, UK

Search for more papers by this author
First published: 21 December 2020
Citations: 8

Graphical Abstract

Divide and conquer: An indirect formic acid fuel cell allows catalytic hydrogen generation spatially separated from electricity generation with a thin palladium alloy film connecting the two compartments. Although demonstrated for formic acid, this concept could be applied to a wider range of fuels subject to catalyst performance and future development of improved thinner more robust hydrogen conducting membranes.

Abstract

An indirect fuel cell concept is presented herein, where a palladium-based membrane (either pure Pd with 25 μm thickness or Pd75Ag25 alloy with 10 μm thickness) is used to separate the electrochemical cell compartment from a catalysis compartment. In this system, hydrogen is generated from a hydrogen-rich molecule, such as formic acid, and selectively permeated through the membrane into the electrochemical compartment where it is then converted into electricity. In this way, hydrogen is generated and converted in situ, overcoming the issues associated with hydrogen storage and presenting chemical hydrogen storage as an attractive and feasible alternative with potential application in future micro- and macro-power devices for a wide range of applications and fuels.

1 Introduction

Micro-power supplies are employed in situations where low power consumption occurs over longer periods of time (e. g. in sensors or remote transmitters1). Often membrane-based fuel cell systems are employed with an air exposed cathode and a fuel exposed anode2 or bio-anode.3 The cross-contamination driven by diffusion of air/oxygen reaching the anode and fuel reaching the cathode, cause losses in energy (direct discharge) as well as long term contamination and catalyst degradation. Alternative micro-devices have been proposed for example, hydrogen-based permeation through siloxane membranes.4

Due to an excellent ability to bind hydrogen5 and permeability to hydrogen,6, 7 palladium and palladium alloys are very useful in hydrogen purification and separation.8, 9 Electrochemical cell designs centered around palladium membranes (as hydrogen conductor) have been known for many years.10 Recently, a novel electrosynthesis cell was proposed and developed by Sherbo, Delima, Berlinguette et al.11-13 based on the principle of separating the electrochemical reaction, from the catalytic hydrogenation process and adding electro-deposited palladium to enhance catalytic reactivity. Hydrogenation of both, ethene and ethyne derivatives, was demonstrated coupled to the oxidation of an alcohol in the electrochemical reaction compartment. Separated processes were employed in a “paired electro-synthesis” mode with optimum utilization of the electrical energy and without any waste products being generated. Here, a similar process is suggested for the conceptually opposite case of energy generation. A separation is introduced between a catalytic fuel compartment and an electrochemical power generation compartment. A palladium or palladium-alloy film membrane provides a short pathway for a locally generated hydrogen intermediate to diffuse from the catalysis to the electrochemistry reaction zones.

Although high temperature processes based on fuel dehydrogenation at palladium membranes14 for power generation have been proposed, there are scarce reports on room temperature processes. Vielstich and coworkers have investigated the room temperature behavior of chemical systems such as methanol, formaldehyde and formic acid, undergoing dehydrogenation at palladium film electrodes.15 The transport of hydrogen through the palladium membrane (and through palladium alloy membranes16) has been studied in detail by electrochemical methods.17 Only recently, new operando methods have been applied to reveal the effects of facets and vertices on palladium surfaces on the hydrogen uptake and release.18 The Devanathan-Starchurski electrochemical cell19 can be used to measure hydrogen transport quantitatively based on two potentiostats operating in conjunction. It has been shown that the diffusion coefficient (D) for hydrogen in palladium or in palladium alloys at room temperature is approximately 10−11 m2 s−1.20 Therefore, the transport time for hydrogen diffusing through a membrane of thickness L=25 μm or L=10 μm will be typically τL2/D=62 s or 10 s, respectively.

The chemical process by which the hydrogen permeates through solid metals such as palladium is termed solution-diffusion mechanism. Molecular hydrogen adsorbs onto the metal surface where dissociative chemisorption takes place, generating hydrogen atoms. Hydrogen atoms are absorbed into the bulk and diffuse to the opposite side through the palladium lattice. During diffusion, the hydrogen generates a distortion of the metal lattice.21, 22 Atomic hydrogen is accommodated in the octahedral interstices of the face-centered cubic palladium lattice. At higher concentrations, the presence of hydrogen generates a phase transition in the palladium (from α to β) associated with a change in volume. This phase change causes deformation and micro-strain associated with traps slowing the hydrogen permeation23 and associated with brittleness/pinholes. The use of palladium alloys such as PdxAgy has been shown to reduce the lattice distortions, thereby enhancing the membrane mechanical properties and the permeability to hydrogen.24

The spontaneous decomposition of formic acid on metals25 and on metal oxides has been studied in detail.26 Different mechanisms were suggested for the decomposition of the formic acid depending on the chosen catalyst. The decomposition is related to the intrinsic electronic properties of the metallic particles.27 It has been reported that palladium is the most active mono-metallic catalyst among the transition metal elements tested for formic acid decomposition in aqueous phase in direct formic acid fuel cell systems.28 Large terrace sites on palladium adsorb formic acid in bidentate form, which gives rise to the dehydrogenation pathway (hydrogen production), whereas surface-unsaturated sites boost the dehydration pathway (CO production).29 Herein, palladium is used as both, a hydrogen transporting membrane at ambient temperatures and a catalyst for the dehydrogenation reaction of formic acid.

Formic acid has been suggested as a potential hydrogen carrier in fuel cell energy sources.30 In the direct formic acid fuel cell, a Nafion membrane is employed between the anode and cathode to avoid direct discharge by cross-contamination of fuel and oxygen.31 We report a proof-of-concept device of an indirect formic acid fuel cell that integrates the functions of formic acid catalysis and hydrogen fuel cell by linking two separate compartments with a palladium membrane. Initially, a half-cell configuration (Figure 1A, three-electrode configuration) is employed to study the reactivity of formic acid in contact to pure palladium and Pd75Ag25 alloy.

Details are in the caption following the image

Schematic illustration of A) a half-cell (three-electrode configuration, reference, counter, and working electrode) and B) a full fuel cell (two-electrode configuration, cathode, and anode) with a palladium membrane separating a fuel compartment from an electrochemical compartment. C) Photograph of the laminated electrodes with exposed active region (diameter 2 mm) based on a porous carbon paper impregnated with catalyst (cathode) and a palladium film (anode).

Next, under optimized conditions, a combined air – formic acid “indirect fuel cell” is developed and tested (Figure 1B, two-electrode configuration). It is suggested that the indirect fuel cell concept is beneficial in stopping fuel crossing into the electrochemical compartment.

2 Results and Discussion

2.1 Hydrogen Production from Formic Acid at the Palladium Membrane

Initial experiments were performed to demonstrate hydrogen production from formic acid in aqueous solution at room temperature. This process is known to occur spontaneously at palladium nanoparticles32 and is employed here to generate molecular hydrogen locally at the palladium membrane surface. The hydrogen generation reaction can be expressed as in Equation 1.
urn:x-wiley:21960216:media:celc202001570:celc202001570-math-0001(1)

Chronopotentiometry data in Figure 2A (note that data traces are offset to enhance clarity; arrows indicate the time of fuel addition) show that upon addition of aqueous 0.1 M formic acid into the fuel compartment of the test cell (see Figure 1A) the zero current equilibrium potential reading in the electrochemical compartment changes from positive to negative. The initial positive equilibrium potential is likely to be dominated by ambient oxygen. The anticipated equilibrium potential for 1 bar hydrogen gas pressure (at 25 °C, in 10 mM H2SO4, pH approximately 1.8) would be approximately Eequilibrium=−0.241 – (0.059×1.8)=−0.347 V vs. SCE. The observed equilibrium potential approaches −0.2 V vs. SCE (see Figure 2A), but then drifts to more positive potentials possibly (at least in part) due to the presence of oxygen resulting in loss of hydrogen on both sides of the palladium membrane. The process is not at equilibrium and there could be further detrimental effects, for example from palladium surface deactivation (vide infra).

Details are in the caption following the image

A) Zero current chronopotentiometry in 10 mM H2SO4 in contact to a palladium membrane. Aqueous 0.1 M formic acid is injected (see arrows) into the fuel compartment to trigger hydrogen generation, hydrogen permeation through the palladium, and the potential response. Data in ambient air (i) and under argon (ii) are presented. B) Chronoamperometry data for the hydrogen oxidation response at E=0.1 V vs. SCE applied to the palladium membrane. Injection of aqueous 0.1 M formic acid (see arrows) into the fuel compartment triggers the anodic current. Data in ambient air (i) and under argon (ii) are presented.

Chronoamperometry data were obtained (Figure 2B) with the potential held at 0.1 V vs. SCE. After a few minutes of stabilization, the aqueous 0.1 M formic acid was injected into the fuel compartment and a current transient was observed. An anodic current peak was observed reaching a maximum of 3 μA for both ambient air conditions (i) and argon de-aerated conditions (ii). Following the current peak, a current decay occurred fast in ambient air. Under argon, the decay is considerably slower but still significant. The magnitude of the current and the current decay both seem inconsistent with rate limiting diffusion of 0.1 M formic acid towards the membrane and are, therefore, more likely to be associated with transport through the palladium membrane and/or the kinetics of hydrogen generation at the palladium membrane surface. This can be confirmed by modification of the palladium surface with electrodeposited palladium.

2.2 Effect of Palladium Electrodeposition on Hydrogen Production from Formic Acid

In order to increase the rate of catalytic hydrogen formation from formic acid, a modification of the Pd membrane surface was carried out. The roughness/ surface area of the palladium on the side of the catalytic compartment was increased by electrodeposition of PdCl42− from solution (see experimental). Scanning electron microscopy images (Figure 3) show the deposition process progressing for 15 s, 30 s and 45 s. Small grains of Pd metal formed when Pd was electrodeposited for 15 s, which increase in size to ca. 100 to 200 nm diameter as deposition time increased.

Details are in the caption following the image

SEM images of the palladium membrane surface with Pd electrodeposition for A) 0 s, B) 15 s, C) 30 s, and D) 45 s.

Data in Figure 4A show zero current potentiometry results for surface modified palladium membranes. When 0.1 M formic acid was injected (see arrows), hydrogen was generated in all cases. Compared to the case of bare palladium (see Figure 2A), and the equilibrium potential is more stable reaching Eequilibirum=−0.24 V vs. SCE for the 60 s palladium deposit. Additional cyclic voltammetry experiments starting at open circuit potential (OCP) and scanning into positive potential direction (see Figure 4B) show evidence for hydrogen permeation (see oxidation peak at Eox=0.2 V vs. SCE) at the palladium surface with 60 s Pd deposit.

Details are in the caption following the image

A) Zero current chronopotentiometry data (in 10 mM H2SO4) for palladium membranes modified on the catalysis side by palladium electrodeposition for 15 s (i), 30 s (ii), 45 s (iii), and 60 s (iv). Potential transients are observed after injection of aqueous 0.1 M formic acid (see arrows) into the catalytic compartment. B) Cyclic voltammetry data (three consecutive cycles, scan rate 10 mV s−1; start at OCP) obtained at the end of chronopotentiometry experiments in (A). C) Chronoamperometry data (in 10 mM H2SO4) for the hydrogen oxidation response at E=0.1 V vs. SCE applied to the palladium membrane with 60 s Pd electrodeposition. Injection of aqueous formic acid (FA) of 0.01 M (i), 0.1 M (ii), 1.0 M (iii), and 2.5 M (iv) concentration into the fuel compartment (see arrows) triggers the anodic current. D) Cyclic voltammetry data (scan rate 10 mV s−1; start at OCP) obtained at the end of chronopotentiometry experiments in C for formic acid concentrations of 2.5 M (i), 0.1 M (ii), and 0.01 M (iii).

Chronoamperometry (at 0.1 V vs. SCE) with a palladium membrane modified by 60 s Pd electrodeposition was performed by injection of aqueous formic acid with (i) 0.01 M, (ii) 0.1 M, (iii) 1.0 M, and (iv) 2.5 M (see Figure 4C; arrows indicate fuel injection). Compared to data obtained with the bare palladium film (Figure 2B), currents for hydrogen oxidation in the electrochemical compartment are substantially increased. When increasing the concentration of formic acid, a step-shaped transient (for lower concentrations) changes into a peak-shaped transient (for higher concentrations). A relatively constant current (approx. 10 μA) is obtained for 0.1 M formic acid. The decay of current with time for 1.0 M and 2.5 M formic acid could be associated with gas bubble formation and blocking or with secondary reactions such as CO formation and Pd catalyst surface poisoning.33 Cyclic voltammetry experiments (see Figure 4D) were performed starting at open circuit potential (OCP) to demonstrate that the amount of hydrogen undergoing oxidation in the peak at approximately 0.1 V vs. SCE is the highest for 2.5 M formic acid. A concentration of 0.1 M formic acid appeared to provide constant conditions and was therefore employed for further experiments with indirect fuel cells to maintain constant power production (vide infra).

2.3 Hydrogen Formation from Formic Acid at a Pd75Ag25 Alloy Membrane

Many materials offer hydrogen permeation capability, but palladium allows permeability as well as active hydrogen capture (hydrogen partitions from aqueous solution into palladium) due to formation of a solid solution (to form the α-phase or PdHx for x<0.5). This unique property combined with good room temperature mobility of hydrogen in palladium34, 35 make palladium membranes highly interesting for industry applications, for example in hydrogen purification. However, some palladium alloys perform better as they combine these benefits with a lower brittleness.36 In particular Pd75Ag25 has been widely employed.37

Here, a commercially available Pd75Ag25 alloy film with 10 μm thickness (compared to the pure palladium membrane with 25 μm thickness) is investigated. Figure 5A shows chronopotentiometry data comparing Pd75Ag25 without surface modification and Pd75Ag25 with additional electrodeposited Pd (for 60 s). A significant change occurs. The bare alloy membrane, although responding to the injection of 0.1 M formic acid into the fuel compartment, does not yield a significant equilibrium potential change. Only with the palladium electrodeposition can the activity of the membrane (as indicated by the equilibrium potential of close to −0.24 V vs. SCE) be observed. Chronoamperometry experiments with the bare Pd75Ag25 film using an applied potential of 0.1 V vs. SCE with 0.1 M formic acid injection (not shown) results in a current consistent with hydrogen oxidation, but the current remains very low at approximately 10 nA. However, after increasing the activity of the palladium catalyst by additional electrodeposition of Pd, chronoamperometry data using 0.1 M formic acid show a much higher current (ca. 60 μA, see Figure 5B, compare Figure 4C). This increase in current is likely to be linked (at least in part) to the much thinner membrane allowing a higher flux of hydrogen from the catalysis compartment into the electrochemical compartment.

Details are in the caption following the image

A) Zero current chronopotentiometry data in 10 mM H2SO4 for Pd75Ag25 alloy membranes bare (i) or modified on the catalysis side by palladium electrodeposition for 60 s (ii). Potential transients are observed after injection of aqueous 0.1 M formic acid (see arrows) into the catalytic compartment. B) Chronoamperometry data for the hydrogen oxidation response at E=0.1 V vs. SCE applied to the Pd75Ag25 alloy membrane with 60 s Pd electrodeposition. Injection of aqueous formic acid (see arrows) of 0.01 M (i), 0.1 M (ii), 1.0 M (iii), and 2.5 M (iv) concentration into the fuel compartment triggers the anodic current. SEM images of the Pd75Ag25 alloy membrane surface before (C) and after (D) electrodeposition of Pd for 60 s.

The effects of injecting different concentrations of formic acid into the catalysis compartment are investigated (Figure 5B). When concentrations higher than 0.1 M of formic acid were used, again a significant decay in the current with time was observed. Therefore, a concentration of 0.1 M formic acid was employed for further experiments. Scanning electron microscopy (SEM) images are shown for the alloy membrane before and after electrodeposition of palladium (Figure 5C and 5D). The electrodeposition of Pd during 60 s forms a rougher palladium-rich surface that improves the catalytic activity towards the formic acid.

2.4 Indirect Formic Acid Fuel Cell Power Responses

An indirect formic acid fuel cell prototype was realized by adding an oxygen gas diffusion electrode (see experimental, Figure 1B) to complement the palladium membrane (with circular area of 2 mm diameter exposed on both sides, mounted in lamination foil, see Figure 1C). Initially, the chronoamperometry experiment was performed with the gas diffusion electrode for oxygen reduction only (Figure 6A). In the presence of oxygen (1 bar) an equilibrium potential of 0.72 V vs. SCE is observed a little below that expected with Eequilibrium=0.989 – (0.059×1.8)=0.883 V vs. SCE. Next, chronopotentiometry experiments with the indirect fuel cell (Figure 6B) are performed in two steps. First, oxygen is passed through the cathode (see arrow) and second, aqueous 0.1 M formic acid is injected into the fuel compartment (see arrow) to start the hydrogen evolution reaction. The equilibrium potential for the cell reaches 0.95 V and 0.97 V for the palladium membrane with 60 s Pd electrodeposit (25 μm thickness) and for the Pd75Ag25 alloy with 60 s Pd electrodeposit (10 μm thickness), respectively. This is consistent with the sum of equilibrium potential data for anode and cathode separately, and indicative for active hydrogen production at both types of anodes.

Details are in the caption following the image

A) Zero current chronopotentiometry data in 10 mM H2SO4 for a carbon/Nafion gas diffusion electrode with commercial C/Pt(40 %) catalyst. When allowing oxygen gas (1 bar, see arrow) to interact with the electrode the equilibrium potential approaches 0.72 V vs. SCE. B) Chronopotentiometry for the Pd membrane (i) or the Pd75Ag25 alloy membranes (ii) modified on the catalysis side by palladium electrodeposition for 60 s. Potential transients are observed after first passing oxygen (1 bar, see arrow) and then injection of aqueous 0.1 M formic acid (see arrow) into the catalytic compartment. Voltammetry data (C) (from open circuit scanning towards 0.0 V, scan rate 10 mV s−1) showing indirect fuel cell power generation (current×voltage versus voltage) in the range from open circuit conditions to short circuit conditions.

Power generation in the indirect fuel cell was assessed by slow cyclic voltammetry. A voltage scan rate of 10 mVs−1 was applied starting from open circuit potential (OCP) and scanning towards short circuit conditions. Figure 6C shows data for the power generation (voltage×current). For both types of membranes, palladium-based and palladium alloy-based, a peak in power is observed. The plots show a maximum power at approximately 0.6 V for the palladium membrane and at 0.3 V for the Pd75Ag25 membrane. The maximum power readings are 8 μW and 28 μW (corresponding to 0.25 μW cm−2 and 0.89 μW cm−2) for palladium and for the thinner palladium alloy, respectively.

These observations are consistent with the palladium anode (not the gas diffusion electrode) being the rate limiting component in the indirect fuel cell with thinner palladium films providing higher power, and the current/power at the thinner palladium alloy membrane being higher. That is, hydrogen transport is the overall rate-limiting step. Flux of hydrogen across the membrane is important in power generation and a decrease in membrane thickness by a factor of 2.5 is likely to play a major role in the increase in power by a factor 3.5. The electrodeposition of palladium directly onto the palladium or the palladium alloy surface is important to increase and stabilise the current. In the future, both membrane thickness and catalyst coatings (on both sides of the films) will be key to further improvements in performance and application of these indirect fuel cells.

3 Conclusions

The concept of an indirect fuel cell has been introduced based on hydrogen being produced as an intermediate and converted immediately into electricity. Studies have been performed with half cells (in three-electrode configuration) and with an indirect full cell (in two-electrode configuration) to explore the rate limiting factors and options for improvements. The half-cell experiments revealed that interfacial modification of the hydrogen permeation membrane by catalyst coating (e. g. electrodeposition of palladium) is an essential factor to increase the current due to an enhanced rate of hydrogen generation (and an increase in the power output) and to maintain good current flow during cell operation. For the indirect fuel cell, a maximum power output of 0.89 μW cm−2 was obtained with aqueous 0.1 M formic acid. This could be increased in future with a higher formic acid concentration. However, further studies are necessary to better explore the mechanisms and the deactivation processes that affect the catalyst. The production of gaseous products such as CO and CO2 in the catalysis compartment need to be quantified. Better catalysts are required for both the catalysis compartment and the electrochemical compartment. Improvements in catalyst performance are also desirable for a more sustained operation in the presence of higher fuel concentrations. New catalysts are desirable for other types of fuels.

The distinctive architecture of the indirect fuel cell avoids problems of fuel cross-over towards the cathode and contamination of the electrolytic compartment with complex mixtures of reagents and products. The indirect fuel cell design also allows the use of a diverse range of fuels from gases to oils, acids to bases and without any requirements in terms of conductivity. However, cross-over of oxygen from cathode to anode still remains a problem and losses due to ambient oxygen are also significant in the catalysis compartment, where the hydrogen generation occurs. The overall rate limiting steps currently are in the catalytic hydrogen production and in the transport of hydrogen across the membrane.

The development of new and better catalysts could open up further interesting choices for fuels, particularly in view of bio-resources providing renewable energy fuels. Future exploration of waste as a source for hydrogen will be desirable. At the present state of study, there are many unanswered questions but also opportunities for improvements. Ideally, the type of membrane used in this work should be cheaper, thinner, and mechanically robust (maybe in the future based on graphene-type materials38). Recent work by Delima and coworkers already addressed this challenge.39 The amount of palladium or palladium alloy in the membrane could be lowered by using improved membrane fabrication techniques such as reactive ink-printing,40 or physical vapor deposition onto polymer-based substrate materials.13 This type of cell could be 3D-printed and adapted to applications for example in wearable devices and/or for refillable/rechargeable energy sources.

Acknowledgements

E.M., L.T.M., K.B., and F.M. thank EPSRC for support (EP/N013778/1). K.I.O., A.K.I., E.O., C.O.A, and F.M. thank EPSRC for the award of a global challenges research fund.

    Conflict of interest

    The authors declare no conflict of interest.