Perspectives

Toward sustainable fuel cells

See allHide authors and affiliations

Science  16 Dec 2016:
Vol. 354, Issue 6318, pp. 1378-1379
DOI: 10.1126/science.aal3303

A quarter of humanity's current energy consumption is used for transportation (1). Low-temperature hydrogen fuel cells offer much promise for replacing this colossal use of fossil fuels with renewables; these fuel cells produce negligible emissions and have a mileage and filling time equal to a regular gasoline car. However, current fuel cells require 0.25 g of platinum (Pt) per kilowatt of power (2) as catalysts to drive the electrode reactions. If the entire global annual production of Pt were devoted to fuel cell vehicles, fewer than 10 million vehicles could be produced each year, a mere 10% of the annual automotive vehicle production. Lowering the Pt loading in a fuel cell to a sustainable level requires the reactivity of Pt to be tuned so that it accelerates oxygen reduction more effectively (3). Two reports in this issue address this challenge (4, 5).

Strategies for tuning the activity of Pt catalysts are based on the premise that pure Pt binds the oxygen-containing reaction intermediates too strongly. State-of-the-art fuel cells use Pt–Ni or Pt–Co cathode catalysts (2). The solute metal (Ni or Co) leaches out from the surface layer into the acid electrolyte, leaving behind a Pt overlayer. Ni and Co atoms are smaller than Pt atoms and therefore exert a compressive strain on the Pt surface atoms. Compression weakens the binding to the oxygen-containing reaction intermediates (6). A mild weakening increases the catalytic activity of Pt for oxygen reduction.

From model studies to real devices

Novel Pt-based catalysts, including those reported by Bu et al. and Li et al. in this issue, perform better than commercial pure Pt nanoparticles in model liquid half cells (15). The next challenge is to translate the full extent of this superior performance to fuel cells.

GRAPHIC: V. ALTOUNIAN/SCIENCE

On page 1410 of this issue, Bu et al. (4) report Pt–Pb nanoplatelets for use as oxygen-reduction electrocatalysts. The catalytic activity of the platelets is high (4.3 A/mg at 0.9 V; see the figure) in a liquid half cell (an electrochemical cell that allows catalysts to be probed under well-defined conditions). The superior performance of the nanoplatelets is somewhat counterintuitive because Pb atoms are larger than Pt atoms and should therefore exert tensile strain. According to conventional wisdom, such an effect should strengthen the binding to the reaction intermediates (6), impeding the kinetics of oxygen reduction. However, transmission electron microscopy images suggest that the core of the Pt–CPb nanoplatelets imposes tensile strain in some directions and compressive strain in others (4). As a result, some surface sites are under mild compression; these sites dominate the activity for oxygen reduction. This finding is somewhat analogous to our own research on alloys of Pt with rare earth metals. We have found that the larger rare earth atoms distort the compound away from a closely packed structure, inducing a compressive strain to the surface (7).

On page 1414, Li et al. report their investigations of Pt nanowires. Whereas earlier studies have reported partially dealloyed Pt–CNi catalysts (810), Li et al. completely leach Ni out of Pt–CNi nanowires, resulting in jagged nanowires of pure Pt. The surface area and surface-specific catalytic activity of the nanowires are both exceptionally high, yielding a record-breaking mass activity of 13.6 A/mg Pt at 0.9 V in a liquid half cell (see the figure). X-ray absorption spectroscopy measurements, supported by simulations, suggest that the Pt–CPt distance in the nanowires is shorter than in bulk Pt. The simplest explanation for the high oxygen reduction activity of the nanowires is that the surface is under compressive strain (8).

Both the Pt nanowires and the Pt–CPb nanoplatelets show negligible activity losses under accelerated degradation tests at room temperature. Such resistance to degradation is surprising, given that strained Pt structures are thermodynamically destabilized relative to unstrained Pt. It remains to be shown, however, how well these catalysts will operate in real fuel cells, as opposed to the idealized conditions of liquid half cells.

Liquid half cells are simple to optimize, require low amounts of catalysts, and are thus well suited for laboratory-scale studies. Catalyst activity is typically benchmarked at 0.9 V because at higher current densities, the reaction is limited by the poor O2 transport in liquid half cells. In the case of commercial pure Pt nanoparticles, which are relatively inactive, there is excellent agreement between liquid half cells and fuel cells at 0.9 V (11), but this is not always the case for more active catalysts (see the figure) (2).

For instance, Stamenkovic and co-workers' Pt–CNi nanoframes have a mass activity of 5.7 A/mg at 0.9 V in a liquid half cell; such high catalytic activity was unprecedented in 2014, when their work was published (8). Preliminary, unoptimized experiments on Pt–CNi nanoframes in a fuel cell yielded an activity of 0.76 A/mg (see the figure) (12); to the best of our knowledge, this is the highest activity reported for any catalyst in a fuel cell. The nanoframes clearly offer high activity in a fuel cell but, thus far, have not captured the superior performance promised by the experiments in liquid half cells. This illustrates the challenges in translating advances from model studies to technological application.

Furthermore, fuel cells are operated at high current densities in order to maximize power; this requires the overall cell potential to be below 0.9 V. The benefit of using very active catalysts at such high current densities is currently under debate. Recent reports suggest that a high power output is more easily sustained in a fuel cell loaded with catalysts with a large electrochemically active surface area, rather than those with an intrinsically high catalytic activity (2). To this end, the high surface area of Li et al.'s Pt nanowires may be particularly beneficial.

The reason for this phenonemon encountered close to maximum power is unclear. At high overpotentials (that is, distance from equilibrium potential), Pt-based catalysts seem to reach a limiting surface-specific current density of ∼0.1 A/cm2 (13). Kongkanand and Mathias have suggested that this is due to poor transport through the ionomer phase that covers each catalyst particle (2). We propose another possibility: At sufficiently large overpotentials, the barriers associated with electrochemical charge transfer processes should become negligible. Thus, the kinetics for the reaction cannot be accelerated further. At this point, the catalyst reaches a limiting value of the current density. The overpotential required to reach this current density falls with increased catalyst activity. Different catalysts operating beyond this limiting overpotential should exhibit the same surface-specific activity.

Improvements in catalyst design have already allowed Pt loadings in fuel cells to be lowered substantially. This progress has been reached on the basis of model studies that have elucidated fundamental bottlenecks at the electrochemical interface and shown how they can be overcome. Scientists should extend this approach to the investigation of well-defined catalysts at the high current densities at which fuel cells are operated (14). The resulting insights will help to translate the spectacular gains in activity and stability represented by the latest generation of fuel cell catalysts to real devices.

References and Notes

  1. All data shown are for 0.9 V versus a reversible hydrogen electrode. Experimental parameters such as temperature and scan rate may have differed between the experiments.
View Abstract

Stay Connected to Science

Navigate This Article