Proton-Exchange Membrane Fuel Cells with Low-Pt Content

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Glossary

Electrochemically active surface area (ECSA):

The surface area of Pt catalyst that is electrochemically active, requiring access to both protons and electrons. It is generally normalized to Pt mass (e.g., m²/g_Pt), and is the primary measure of Pt dispersion.

Fuel cell catalyst:

Materials that catalyze the electrochemical reactions. Pt or Pt alloy nanoparticles (3–5 nm in diameter) deposited on carbon blacks are commonly used with the goal of maximizing the available reaction site surface area per Pt mass.

Hydrogen PEMFC vehicle:

Vehicle that uses proton-exchange membrane fuel cell (PEMFC) as its primary power generator, commonly known as fuel cell electric vehicle (FCEV). It uses pure hydrogen gas fuel reacting electrochemically with oxygen gas from the atmosphere to generate electricity and emit only water. Generally requires Pt as electrocatalyst on both anode and cathode.

Ionomer:

Ion conducting polymer is used in the membrane and electrodes. In PEMFCs, the conducted ion is a proton, and the environment is strongly acidic with effective pH <1. Perfluorosulfonic acid (PFSA) such as Nafion® (DuPont tradename) is the most common.

Local transport loss:

Performance (i.e., voltage) loss due to the transport of oxygen and protons in close (<30 nm) proximity to the Pt reaction site. Characteristically, this loss is inversely proportional to the Pt roughness factor (i.e., low m²_Pt/m²_MEA) and is most prevalent at high-current density.

Membrane-electrode assembly (MEA):

The MEA is at the heart of the fuel cell where the electrochemical reactions occur. Hydrogen oxidation reaction (HOR) occurs in the anode. Oxygen reduction reaction (ORR) occurs in the cathode. The polymer membrane, sandwiched between the two electrodes, conducts proton across from the anode to cathode and acts as an electrical and reactant separator.

Oxygen reduction reaction (ORR):

O₂ is electrochemically reduced to water on the cathode. ORR is responsible for most of the overall voltage (i.e., efficiency) loss in a fuel cell even with heavy use of Pt catalyst. Therefore, research on high-activity ORR catalyst is of high priority. ORR kinetic activity is commonly expressed by either normalizing to its Pt mass (mass activity) or to its available Pt surface area (area-specific activity).

PGM:

Platinum group metals (Pt, Pd, Ir, Ru, Rh, and Os) and other precious metals (Au, Ag, Re) must be minimized or avoided to enable affordable fuel cells.

Pt roughness factor (r.f.):

The Pt surface area on an electrode for electrochemical reaction per MEA geometric area (m²_Pt/m²_MEA). This is a product of Pt ECSA (m²/g_Pt) and the MEA Pt loading (g_Pt/m²_MEA).

Definition of the Subject

Widespread commercialization of fuel cell electric vehicles (FCEV) relies on further reduction of PGM (platinum group metals) usage. Although enhancements in the activity and stability of the catalyst are necessary, those alone are insufficient. In a fuel cell with low PGM content, transport of reactants (oxygen and protons) to a small area of catalyst can cause large performance loss at high power. Because it is this high-power point that determines the required fuel cell area, these losses drive up the size, and thus the cost, of the fuel cell stack. This entry discusses fuel cell cost reduction with special focus on the challenges and opportunity associated with Pt reduction.

Introduction

PEM fuel cells offer a zero-emission tank-to-wheels solution for sustainable transportation, extending to a well-to-wheels solution when renewable hydrogen is used. Although a few automotive manufacturers, notably Hyundai, Toyota, and Honda, have begun to commercialize fuel cell electric vehicles (FCEV), their high cost limits market penetration. The availability of hydrogen stations is also limited. Yet, major fuel cell developers have defined plausible pathways to reduce the fuel cell vehicle total cost of ownership (vehicle plus fuel cost over life) to approach that of incumbent gasoline-engine vehicles in the long term. Two critical elements of this roadmap include decreasing material cost of the fuel cell system and reducing manufacturing cost through economies of scale.

An automotive fuel cell typically requires about 10 m² of electrochemically active area which is distributed over 250–400 individual cells in series, each with 400–250 cm² of active area. Each cell is a high-current (>500A) and low-voltage (~0.6 V) device, and cells are stacked in series to deliver high DC power required for vehicle propulsion. The trade-off defining the number of cells and active area is dictated by a cost optimization involving the power electronics that are used to interface the fuel cell stack with the high-voltage electrical system [1], a topic outside of the scope of this entry. Instead, this entry focuses on the issue of reducing overall electrochemically active area needed to produce a given power, an issue that is independent from the trade-off determining the selection of the number of cells.

State-of-the-art FCEVs use about 30 g of Pt [2, 3], the only PGM used in the fuel cell system. At today’s (June 2017) Pt price of $30 per gram, the cost of Pt metal itself is about $900, a small fraction of a vehicle cost. But it is significantly larger than what is used in the current clean light-duty internal combustion engine (ICE) vehicle catalytic converter (<5 g PGM, comprising Pt, Pd, and Rh) [4, 5]. Pt is rare, and because of its high resistance to corrosion, it is used in many applications. Of the 218 tons of platinum sold in 2014, 45% was used for vehicle emission control devices, 34% for jewelry, and 9% for chemical production and petroleum refining [6]. The remainder was consumed in other industries including electronics, glass manufacturing, and the medical and biomedical industries. Because Pt is such a well-established commodity, increase in demand will put pressure on its availability and price. Analysis in Fig. 1 shows a significant increase in Pt demand once FCEVs with 10 g_Pt penetrate the mass market (scenario B, assumes five million FCEVs/year, about 5% of global vehicle market), generating upward pressure on Pt price [7]. This supply-demand scenario indicates the need for technologies to reduce Pt usage well below 10 g/vehicle in the long run.

Fig. 1

Impact of fuel cell vehicles on Pt consumption (Reprinted with permission from Ref. [7]. Copyright 2013 Roland Berger LLC)

Figure 2 illustrates the relationship between cathode Pt loading and the cost of major fuel cell stack components assuming a 2016 state-of-the-art current-voltage curve [2]. Although the Pt cost is a large portion of the stack cost at 0.3 mg_Pt/cm² (~30 g_Pt/vehicle), reducing Pt loading below 0.2–0.1 mg_Pt/cm² results in only a marginally lower stack cost, and decreasing to 0.05 mg_Pt/cm² actually results in a stack cost increase. This is due to poor high-power performance of the low-Pt cathode, to be discussed in more detail below, making it necessary to increase stack area and overcoming the benefit of the Pt areal-loading decrease. Improvement in the high-power performance of the low-loaded cathode will minimize stack cost and decrease the Pt loading at which the minimum cost occurs. This analysis also indicates that further reduction of Pt loading below 0.1 mg_Pt/cm² (~10 g_Pt/vehicle) must be done with the objective of lowering the overall fuel cell stack cost. Whereas reduction below 0.1 mg_Pt/cm² with current fuel cell performance may not necessarily be a major vehicle-level cost saver, it would be important to mitigate demand on the global Pt market and thus enable high market penetration of automotive fuel cells.

Fig. 2

Effect of cathode Pt loading on stack cost. Anode Pt loading is kept constant at 0.025 mg_Pt/cm². Cost estimated using SA/DOE 2013 cost study, 90 kW_gross system, 500 k system/year [8] (Reprinted with permission from Ref. [2] Copyright 2016 American Chemical Society)

If 10 g_Pt/vehicle could suppress FCEV mass-market penetration due to supply-demand factors, what level is needed to eliminate this barrier? Reductions to a level comparable to the catalytic converter (~5 g PGM/vehicle) would largely neutralize the demand increase due to the corresponding decrease in catalytic converter demand. Thus, 5 g PGM/vehicle is a reasonable long-term target [2]. Other important factors could involve broader PGM market shifts. For example, an average wedding ring weighs about 5 g; thus, many married couples own approximately an FCEV equivalent of Pt. Consider this reference point and the fact that the world is consuming 34% of total Pt consumption for jewelry. Thus, a shift in consumer preference to use Pt to enable emission-free future transportation and to use other metals (tungsten, gold, etc.) for jewelry could have a large beneficial impact on Pt availability and cost for FCEV use.

The ultimate goal of fuel cell catalyst development is to entirely eliminate the need for PGM. Significant progress has been made by packing as many active sites as possible into carbon-nitrogen-based non-PGM catalysts [9,10,11]. However, their stability is currently unacceptably poor [12]. In addition, their useable power density is only about one-tenth of the PGM catalyst system, making vehicle packaging and cost (e.g., of other stack components that scale with surface area) impractical. Finally, if PGM use is successfully reduced to a level comparable to incumbent vehicle technology (~5 g_PGM/vehicle), the economic benefit of an alternative may not be favorable.

Alkaline membrane fuel cells (AMFC) operate under less corrosive conditions, and low-cost non-PGM cathode catalysts may be used [13, 14]. However, palladium is currently still required on the anode to achieve power density, and thus stack size reduction, approaching that possible with PEMFCs. Furthermore, the instability of AMFC membrane candidates at high temperature (80–100 °C) and the deactivation of its ionic carriers due to CO₂ in air are other major technology hurdles [15,16,17]. These performance, durability, and cost uncertainties have thus far prevented AMFC technology from mounting a serious challenge to PEMFC for automotive applications. AMFC technology status and trajectory is described in detail in a separate entry in this volume.

In this entry, we will provide an overview of the challenges and most promising research directions to develop automotive PEM fuel cell technology with sustainable Pt use.

PEM Fuel Cell Electrodes

The hydrogen oxidation reaction (HOR) on the anode is so fast that less than 1 g of Pt can suffice [18]. And when pure hydrogen is used, fast HOR rate and diffusion minimize the voltage loss on the anode. Therefore, the primary focus has been on improving the sluggish oxygen reduction reaction (ORR) on the cathode. Much progress has been made in improving the activity and stability of the ORR catalysts in the past 15 years. Many advanced catalysts (e.g., shape-/size-controlled alloy [19,20,21,22,23,24], Pt monolayer catalysts [25,26,27], etc.) have shown promising activity in rotating disk electrode (RDE) tests, although they have not shown comparable activities in fuel cell membrane-electrode assemblies (MEA). Fortunately, a more gradual improvement using a dealloying approach [28,29,30,31] on spherical PtNi and PtCo to control the “Pt skin” and subsurface composition has yielded materials that approach the required activity and stability in a fuel cell for a ~5 g_Pt/vehicle [2, 31, 32].

As shown in Fig. 3a, the ORR requires efficient delivery of oxygen, protons, and electrons at the same location. Facile transport of these species, which occur through different phases, to the active site is essential to allow high-power output. The electrodes are generally made of mixtures of proton-conducting polymers (ionomer, blue) and carbon-supported Pt nanoparticle (gray and black circles) catalysts. This design gives a porous layer (~60% porosity) that is good for gas transport and a large active area (roughness factors >30 cm²_Pt/cm²_MEA) for the reaction [33, 34].

Fig. 3

(a) Schematic of transport reaction in the fuel cell cathode. (DM = diffusion media; BP = bipolar plate). (b) Fuel cell polarization curves of PtCo/C catalyst at different cathode Pt loadings. Operating conditions in the order of anode/cathode: H₂/air, 94 °C, 65/65% RH, 250/250 kPa_abs,outlet, stoichiometries of 1.5/2. Single cell, 50 cm² active area (Reprinted with permission from Ref. [2]. Copyright 2016 American Chemical Society)

Carbon black is the preferred support to disperse Pt-based nanoparticles, thanks to its high electronic conductivity, high surface area for nanoparticle deposition, relatively high stability, and low cost. The morphology and properties of carbon play a critical role in determining the performance and stability of the catalyst [35,36,37,38,39]. Some popular carbons such as KetjenBlack (KB) possess a large number of internal micropores within its carbon particles, making it possible to achieve good Pt particle dispersion and thus high Pt surface area [40, 41]. Figure 4 shows a transmission electron micrograph of a Pt/KB catalyst. Tomography analysis on a section of the catalyst reveals the location of Pt particles in relation to the carbon particle. The red and green coloring represent Pt particles that are located on the surface and within the carbon, respectively.

Fig. 4

Transmission electron micrograph of Pt nanoparticles deposited on KetjenBlack carbon particles. Color inset shows the tomography of an area of the catalyst and the location of Pt particles. Red surface Pt. Green internal Pt (Reprinted with permission from Ref. [41]. Copyright 2010 American Chemical Society)

The presence of the internal Pt adds another set of reactant transport considerations within a fuel cell electrode (Fig. 5). The internal particles have been shown to be mostly electrochemically active and are believed to have access to proton and O₂ through small openings (1–5 nm) in the carbon micropores [40, 41]. But it is believed that the pores are too small for ionomer to intrude and form a direct proton conduction path to the Pt surface [39, 42,44,45]. Although it is hypothesized that condensed water can conduct protons in these pores, much remains unclear on the exact mechanism and the magnitude of the proton conductivity [46,47,48]. Additionally, ionomer can form a layer blocking the transport of O₂ and water at the opening leading to increased voltage loss [49, 50]. These complexities may make it appealing to use carbons that do not possess internal porosity. However, it has been shown that porous carbons could offer better Pt dispersion, Pt alloy quality, ORR activity, and catalyst stability [37,38,39]. Therefore, the best carbon support for each user may vary depending on their needs and materials-system trade-off.

Fig. 5

Illustration showing transport of proton, O₂, and water at different length scales in the cathode electrode (Reprinted with permission from Ref. [42]. Copyright 2011 Elsevier)

Performance of Low-Pt Fuel Cell

At higher power, transport phenomena (oxygen, proton, and electron) in a fuel cell will contribute to the voltage loss. As Pt loading and the available Pt area for ORR are reduced, higher O₂ and proton fluxes must be delivered to the Pt surface which can lead to noticeable and even severe voltage losses. This is particularly noticeable below 0.1 mg_Pt/cm²_MEA (Fig. 3b), corresponding to <10 g_Pt/vehicle. These internal losses in energy within the fuel cell are converted into waste heat which must be removed from the fuel cell using coolant and a radiator. The size of the radiator, and thus the amount of this heat removal, is limited by the frontal area of the vehicle. Depending on system design and requirement, cell voltage at the stack high-power (i.e., rating) point is generally required to be higher than 0.55–0.65 V to allow for waste heat removal from the vehicle and maintain sufficient hydrogen conversion efficiency (>50%) [1, 51]. Assuming that the fuel cell stack is sized at 0.58 V, one can estimate current densities from Fig. 3b of 1.65 and 2.0 A/cm² for 0.05 and 0.1 mgPt/cm²_MEA, respectively. This results in power densities at these points of 0.96 and 1.16 W/cm², translating for a 100 kW_gross stack to requirements of approximately 10.4 and 8.6 m² of fuel cell area, respectively. This case illustrates that although it is in principle desirable to reduce the Pt loading, worse fuel cell performance at the stack rating voltage translates to a 20% increase in overall stack size. Depending on the cost of the fuel cell components that scale with area (e.g., plates, membrane, and diffusion media), this can result in an increase in stack cost even with lower-Pt areal loading, as also shown in Fig. 2 at cathode Pt loadings less than 0.1 mg_Pt/cm².

A fuel cell performance mathematical model provides a useful tool to help understand the various internal voltage losses. One can build a model using known physics involved in a PEMFC that uses inputs from a number of in situ electrochemical diagnostics and ex situ characterization methods [34, 45]. Figure 6 illustrates the voltage loss terms estimated for various components as a function of current density. While a realistic fuel cell is operated under a wide range of conditions that vary over the fuel cell area, a simplified “differential cell” (i.e., high gas flow and constant temperature condition) is often used, and is modeled here, for diagnostic purposes. By far, ORR kinetic loss is the largest contributor. As current density increases, transport phenomena (oxygen, proton, and electron) contribute to the voltage loss. Ohmic loss (membrane protonic resistance and electronic resistance of other components), O₂ transport loss in the gas diffusion layers, and proton conduction loss in the electrodes are also noticeable. However, these losses do not change with Pt amount and are thus not the focus of this entry. More detail on these losses can be found elsewhere [34, 45].

Fig. 6

Voltage loss terms in a low-Pt PEMFC operated under differential cell conditions (i.e., high gas flow and constant temperature): H₂/air, 150 kPa_abs, 80 °C, and 100% RH. The symbols represent the experimental data. The lines are the thermodynamic equilibrium cell voltage (E_rev) subtracting various voltage losses calculated based on the measured component material and transport properties and electrode kinetics. MEAs: Pt/C anode and Pt/KB cathode (0.025 and 0.056 mg_Pt/cm², respectively) coated on an 18 μm thick composite membrane

As Pt loading and its surface area decrease, transport of protons and O₂ to the reaction sites becomes more challenging. Characteristically, this type of loss increases with decreasing Pt roughness factor and is called a “local transport loss,” postulated to be at or near the Pt surface [52]. As will be discussed in the next section, a portion of this loss can be attributed to the transport of O₂ through an ionomer thin film covering the Pt particle (yellow area). However, there remains voltage loss at high-current density that has not yet been accounted for by known physics or chemistry (orange area). Both losses grow rapidly as current is further increased or as Pt surface area drops further during long-term fuel cell operation. Therefore, they must be understood and minimized for the long-term Pt loading target to be met.

Local Transport Resistance

As discussed in Fig. 6, the local transport resistance can be divided into two components: (a) O₂ transport associated with ionomer thin film (yellow area) and (b) an unaccounted-for resistance (orange area). The latter component is hypothesized to result from proton and O₂ transport through nanometer-sized pores shown in Fig. 5. Some studies constructed “agglomerate” models, simulating impact of 50–500 nm diameter spherical elements filled with water and/or ionomer, in attempts to simulate the voltage loss [45, 49, 50, 53]. Indeed, there are examples indicating that this unaccounted-for loss (orange area) can be largely eliminated when nonporous carbons are used [39, 54,55,56]. However, due to the complex structural heterogeneity of the electrode (ionomer distribution [57, 58], Pt location [40, 41], carbon pore morphology [42, 44, 59], etc.) as well as engineering phenomena (e.g., localized water generation, drying due to local temperature increase), such models have been difficult to unambiguously test and validate. This area remains in need of improved diagnostics and modeling tools to definitively identify the source of this unaccounted-for loss.

The “O₂ local” component in Fig. 6 can be characterized by a resistance called R_O2^Pt, and this can be quantified by operating an MEA under O₂ transport-limiting conditions [52, 60]. In this so-called O₂ limiting-current measurement, both the Fickian (pressure-dependent component representing bulk gas transport, R_F) and non-Fickian (pressure-independent component representing Knudsen or through-film transport, R_NF) components of the O₂ transport resistance can be determined. Fickian transport represents bulk gas transports in gas-diffusion media and large pores in the microporous layer and electrode. Fickian transport resistance does not change with Pt loading. It is the R_NF that strongly correlates with high-current-density (HCD) performance of low-Pt electrodes. Physically, R_NF is made up of three transport resistances – one from the small pores in the microporous layer (MPL), another from the small pores in the cathode catalyst layer (CCL), and the third for a region close to the Pt surface [52]:

$$ {R}_{\mathrm{NF}}\approx {R}_{\mathrm{NF}}^{\mathrm{MPL}}+{R}_{\mathrm{NF}}^{\mathrm{CCL}}+\frac{R_{O2}^{\mathrm{Pt}}}{r.f.} $$

(1)

Figure 7a summarizes the R_NF as a function of Pt roughness factor (r.f.) for a variety of cathode catalysts [61]. Electrodes with low r.f. (low Pt loading) show high R_NF because more O₂ must be delivered to a smaller Pt surface resulting in a higher apparent electrode O₂ transport resistance. As shown in Eq. 1, one can determine the R_O2^Pt by plotting R_NF vs 1/r.f. (inset). In this case, the R_O2^Pt is determined to be 11.2 s/cm from the slope. In Fig. 7b, we summarize the fuel cell performance at 1.75 A/cm² as a function of roughness factor. The fuel cell voltage drops precipitously once the r.f. is below about 50, as the shrinking available surface area drives up the local reactant flux.

Fig. 7

(a) Non-Fickian O₂ transport resistance (R_NF) as a function of total Pt area on an MEA cathode (roughness factor is defined as the product of Pt loading and ECSA of the catalyst) for different catalysts. Inset is a plot of R_NF vs 1/r.f.. Pt/C (×), PtCo/C (○), Pt-ML/Pd/C (+), NSTF (■), NSTF with 2–4 nm ionomer coatings (▲) (Reprinted with permission from Ref. [2]. Copyright 2016 American Chemical Society (b) Voltage at 1.75 A/cm² showing the impact of r.f. and ionomer)

In contrast to the dispersed carbon-supported catalysts discussed above, the 3M nanostructured thin-film (NSTF, blue squares) catalyst shows impressive fuel cell performance despite its very low-Pt r.f. [62,63,64]. The majority of the 3M NSTF surface is free of ionomer and therefore relies on proton conduction on the Pt surface. However, when a thin (2–4 nm) ionomer film was coated on the NSTF surface, the catalyst exhibited similar behavior to that of the carbon-supported Pt nanoparticle-based (and presumably thin ionomer film coated) catalysts [61, 65, 66].

Ionomer Thin Film and Ionomer-Pt Interface

The above results strongly indicate that ionomer thin film (<5 nm) on the Pt surface contributes to the rise of the “local O₂” resistance (yellow shaded area in Fig. 6). In this section, we discuss potential mechanisms by which the ionomer induces this resistance. Known O₂ permeability of a thick membrane (e.g., a 10–20 μm membrane such as typically used in the fuel cell) cannot explain the large O₂ transport loss observed in the fuel cell electrodes, as there is a factor of 3–10 increase in the apparent resistivity of a thin film [52]. In bulk perfluorosulfonic acid (PFSA) membrane, the ionomer phase segregates into hydrophobic regions and water-containing domains with 2–5 nm diameter channels. This efficiently segregated two-phase morphology is believed to be an important feature responsible for the superior proton conductivity as well as increased water and O₂ transport rates as compared to non-PFSA membranes [67,68,69]. In a fuel cell electrode where ionomer exists as a 1–5 nm thin film on Pt and carbon, the dimensions are too small to allow development of the two-phase morphology present in thicker films. The resulting increased importance of interfacial properties and interactions with solid substrates are expected to induce changes in its structure and transport properties (Fig. 8).

Fig. 8

Thickness dependence and substrate interaction of ionomer thin film (Reprinted with permission from Ref. [70]. Copyright 2017 American Chemical Society)

When the ionomer film thickness approaches the characteristic domain size of the ionomer, structure and transport properties of the ionomer can change due to the so-called confinement effect. Many ex situ techniques such as X-ray scattering, neutron and X-ray reflectivity, TEM, XPS, AFM, and FTIR were employed to study this effect [57, 71,72,73,74,75,76,77,78]. Some effects include a formation of multilamellar nanostructure and reduced transport properties such as lower water uptake and uptake rate [59, 72, 74,75,76,77, 79,80,81,82]. These effects are highly dependent on treatment condition, substrate type, and operating environments [76, 83]. These findings corroborate the notion based on domain-size arguments that transport properties of ionomer in a fuel cell electrode are very different from those in bulk membranes. A comprehensive discussion can be found in the PFSA ionomer thin-film entry in this volume and a review by Kusoglu and Weber [70].

Sulfonate groups on the ionomer can adsorb on Pt and reduce ORR activity [84, 85]. The adsorption of the acid group immobilizes the ionomer chain reducing its degrees of freedom [86,87,88]. Electrodes with higher acid group concentration (lower equivalent weight ionomer) were shown to have higher apparent local O₂ resistance [89]. Some studies showed that different ionomer acid groups and ionic liquid additives can improve ORR activity [24, 90,91,92]. Furthermore, ex situ O₂ permeability measurement by Litster and coworkers showed that there was no such increase in thin ionomer O₂ resistance when as low as 50 nm thick ionomer films were coated on a polycarbonate substrate and placed in the diffusion path, but not in direct contact with the Pt surface [93]. This result indicates that ionomer interaction with the Pt surface is associated with the increase of the local O₂ resistance.

As an alternative consideration to sulfonate group interaction with Pt, molecular dynamics and DFT simulation by Jinnouchi et al. indicated that it is energetically preferable for the large number of CF₂ groups on the ionomer backbone to fold on the Pt surface [94]. Such a dense layer adjacent to Pt can reduce the O₂ concentration and may be a root cause for the local O₂ resistance.

Published studies using alternative ionomer structures in the electrodes are limited. In general, use of hydrocarbon ionomers results in poor fuel cell performance [95,96,97,98,99]. This is primarily due to their characteristically lower gas permeability which, although favorable when used as a membrane, is detrimental in the electrodes. Among the PFSA ionomers, decrease in the ionomer equivalent weight (increase in the acid group concentration per mass of ionomer) was shown to worsen HCD performance in one study [89]. Ionomers with short side chain or rigid backbone have been shown to mitigate reversible degradation of the electrode [92]. Some acid groups are found to adsorb less strongly to the Pt surface which might translate to higher activity and improved transport properties, although its HCD benefits were not confirmed in actual fuel cell performance [90]. The most significant impact observed is when a small cyclical ring is inserted into the ionomer backbone in order to create a sterically enhanced O₂ permeability through a more open structure (Fig. 9) [100,101,102]. This results in a substantial improvement in HCD voltage as shown in Fig. 7b (green triangle). These are very encouraging initial results. However, stability and processability concerns still remain.

Fig. 9

Schematics of hypothesized ionomer structure at the Pt surface for conventional ionomer (a) and high-oxygen-permeable ionomer (b) (Reprinted with permission from Ref. [2]. Copyright 2016 American Chemical Society)

Altering the process when forming the catalyst layer by changing the solvent system or mixing procedure was shown to improve HCD performance [103,104,105]. However, it is uncertain whether these enhancements are due to local transport (characteristically scales with Pt loading) or optimization of ionomer distribution in the electrode.

Recent studies on ionomer thin films have identified many changes in the microstructure and physical properties of the ionomer; however, large gaps still remain in understanding how these changes affect its transport properties (oxygen, proton, and water) within the cathode as well as how the ionomer structure might change with operating conditions in a fuel cell. These topics are critical future research areas that could have significant positive impact on cathode performance and durability.

Catalyst Roadmap

The mathematical model discussed above can also be used to create catalyst roadmaps that consider the impact of the local transport resistance, indicating critical ex situ measured parameter requirements needed to achieve performance targets. Figure 10 shows catalyst requirements, ORR mass activity, and Pt surface area (ECSA), needed for acceptable performance (>0.58 V) at 1.5 and 2.0 A/cm² for cathode catalyst loadings of 0.05 and 0.1 mg_Pt/cm². Requirements with (parabolic dashed lines) and without the local resistance (horizontal dotted lines) are shown. Parameters must be higher than the dashed parabolic lines in order to meet the associated requirement, and materials with values in the upper right-hand portion of the plot are most desirable. Some experimental data are also shown on the map – open symbols are representative MEA data before and after durability tests for four well-known catalyst families. The arrows show significant deleterious impact of degradation during operation, representing a large opportunity if catalyst behavior could be stabilized close to beginning-of-life values. The shaded ovals represent an optimistic expectation of what each approach may achieve in the next 10 years.

Fig. 10

ORR mass activity and Pt ECSA targets which enable the cathode to meet vehicle requirements (0.58 V) at the indicated current density and cathode catalyst loadings. Local resistance of 12 s/cm was used in the analysis. Dashed lines and dotted lines show the minimum values to meet the requirement with and without local resistance, respectively. Catalysts that are above and to the right of a given boundary are sufficient to meet or exceed the voltage requirement. Data points from MEAs before (arrow start) and after (arrow end) accelerated durability tests are shown for various catalyst systems [31, 106, 107], and shaded areas indicate estimate of achievable targets (Reprinted with permission from Ref. [2]. Copyright 2016 American Chemical Society)

Pt alloy catalysts are the most mature. The Toyota Mirai FCEV, introduced to the market in 2014, already uses this type of catalyst although at a higher Pt loading [3, 108]. The high ORR mass activity and moderate Pt surface area suggest that the current catalysts can meet requirement at 10 g_Pt/vehicle (red dashed line) but fall short of 5 g_Pt/vehicle (blue-dashed line) [31, 32]. Increase in both ORR activity and Pt surface area over operating life is required. Other promising advanced catalysts (e.g., shape-/size-controlled alloy [19,20,21,22,23,24], Pt monolayer catalysts [25,26,27], etc.) have shown some encouraging initial results but have generally not yet shown competitive stability in fuel cell MEA testing. Furthermore, they are still made in small quantities at lab scale, and processes amenable to high-volume production remain to be developed.

The Pt monolayer catalyst family pioneered by Brookhaven National Laboratory, shown toward the upper right-hand portion of Fig. 10, is conceptually appealing in that it places essentially all Pt atoms on the particle surface and does not waste Pt atoms in the particle core [25, 26]. This construction gives the highest Pt surface area possible, and representative data are shown in Fig. 11. On a Pt content basis, this catalyst outperforms other types of catalyst, thanks to its high Pt ECSA [27]. However, at the current stage, a relatively large amount of Pd is needed to form Pt monolayer shell, and Pd is subject to leaching out from the core resulting in destabilization of the Pt monolayer [26, 27, 109]. Furthermore, global tightening of automotive emissions standards has raised the price of Pd in recent years to the point that economical advantage of Pd over Pt is minimal. (Both are about $30/g in June 2017.) Some early work on platinum-monolayer shell on palladium-tungsten-nickel core catalyst (Pt-ML/PdWNi), in which half the Pd core has been replaced by less expensive materials, has shown promising performance and durability (middle of Fig. 10). Further work along these lines, pursuing reduction of Pd use and stabilization of the core materials, appears to be a worthwhile development direction.

Fig. 11

H₂/air polarization curves of different catalysts with Pt loadings of 0.05 mg_Pt/cm². Pt_ML/Pd catalyst contains 15 wt% Pt and 25 wt% Pd (Reprinted with permission from Ref. [27]. Copyright 2016 American Chemical Society)

Durability of Low-Pt Fuel Cell

Usable performance of the fuel cell must extend over the lifetime of the vehicle (>12 years). Automotive producers generally aim to allow less than 10–20% performance degradation over its life. In the context of Fig. 10, one will need at the end of expected life a catalyst that remains higher than the required ORR activity and Pt ECSA. Characteristics of the degradation of the catalyst and electrode can be observed through decrease in ORR activity and Pt ECSA and loss of transport properties. These losses can be either permanent or reversible. The major degradation mechanisms include Pt and transition metal dissolution, particle migration and coalescence, carbon corrosion, and contaminant adsorption. An extensive review of this topic is available in Ref. [110].

Although one might expect the degradation of a catalyst to be relatively independent of Pt loading, more systematic study is warranted. And even if degradation mechanisms and rates in a low-Pt electrode are the same as in a higher-Pt electrode, the impact on voltage loss will be higher in the low-Pt case. This is because, as discussed in Figs. 7b and 10, the voltage drops precipitously at low r.f. since the resistance (and thus loss) is inversely proportional to this factor. Therefore, research to develop and implement low-Pt catalysts that start and end life with high ECSA and specific activity is critical to the pursuit of affordable automotive fuel cells.

As the total Pt surface area is reduced in a low-Pt fuel cell, the electrodes become more susceptible to contamination. Some chemical degradation is reversible, and performance can be recovered during normal vehicle operating modes. On the anode, H₂ fuel quality is critical to allow Pt reduction (<0.025 mg_Pt/cm²), where contaminants such as CO and H₂S must be carefully controlled. On the cathode side, a chemical air filter may be required to remove potential contaminants (SO_x, H₂S, NO_x, volatile organic compounds, etc.) from intake air [111, 112]. In addition to external contaminants, degradation products (sulfate and organic compounds from the membrane [66, 113,114,115]; cobalt from the catalyst [116, 117]) from within the fuel cell MEA can also decrease fuel cell cathode performance. Development of more stable MEA components and strategies to mitigate degradation are needed to enable low-Pt fuel cells under real-world application.

Other Challenges

In this entry, we have focused on the performance of the catalyst and cathode electrode. However, for further improvement other components in the fuel cell must also be developed. Figure 12 breaks down the sources of the voltage loss at high-current density for the near-term target of 0.1 mg_Pt/cm² at 2 A/cm² (upper left, about 10 g_Pt/vehicle) and possible paths toward further improvement. For further Pt and cost reduction, one can reduce the Pt loading or increase the maximum current density. To decrease Pt loading (mg_Pt/cm²) at the same current density (move to right in Fig. 12), improvement in ORR activity and local transport are key enablers. To increase current density at the same Pt loading (move downward in Fig. 12), improvement in ORR activity and local transport remain of highest priority, but important issues broaden to include ohmic loss reduction, bulk O₂ transport, and cathode proton conduction in the cathode. This downward path can have higher cost reduction impact since smaller stack size will also reduce the amount of bipolar plate, membrane, electrode, and diffusion media material required.

Fig. 12

Two pathways to reducing cost further [1]: decreasing Pt loading and [2] increasing current density. Losses are estimated using a fuel cell voltage loss model

Finally, in order to realize a truly sustainable technology, more precious metals must be recycled. Although growing in recently years, recycling currently only accounts for 30% of the global PGM supply [6]. Pt global recycling rates (60–70%) are better than those of other PGMs thanks to its favorable recycling economics. However, in the automotive sector, the recycling rate only reaches 50–55% [118]. Recycling of PGM is not only technically feasible and environmental friendly but also profitable. Recycling rates can be increased through improved waste collection mechanisms, as well as shifting public perspective from “waste management” to “resource management” [119]. Management of PGM life cycle ultimately needs to expand across many markets and applications including automotive, jewelry, and electronics. For this to occur, strong support will likely be needed from policy makers at state, federal, and international levels.

Future Directions

Fuel cell electric vehicles with about 30 g of Pt are now on the road [2, 3], and next-generation FCEVs are expected to use about 10–25 g of PGM. These are significant accomplishments and encouraging progress toward commercializing this sustainable transportation technology. However, considering commercial factors as well as promising catalyst technologies early in the pipeline, a long-term PGM target is warranted at a level comparable to that used in automotive catalytic convertors (~5 g_PGM). Progress in Pt-based catalysts in recent years has been due to alloy optimization resulting in notable activity gains, but opportunities remain to achieve better Pt surface area (ECSA) and alloy stability over operating life. In addition to these structure and kinetics considerations, fundamental understanding of the origin of the local transport resistance is needed in order to optimally engineer the nanostructure near the catalyst active surfaces. Development of ionomer specifically designed for this purpose is a promising research direction, as encouraging early data exists. New issues will be encountered as the use of low-Pt roughness factor increases contamination susceptibility, and fundamental studies to conclusively identify poisoning mechanisms and mitigations approaches will also be needed. Judging from the steady progress made in the past decades, we are optimistic that the concerted efforts of materials developers and electrode designers can resolve these issues, enabling fuel cell vehicles that are affordable for the mass market.