PEM Fuel Cells and Platinum-Based Electrocatalysts

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

Anode:

An electrode where the electrochemical oxidation reaction(s) occurs, generating free electrons that flow through a polarized electrical device and enter the cathode. In a fuel cell, the fuel oxidation reaction happens at the anode.

Cathode:

An electrode where the electrochemical reduction reaction(s) occurs, by consuming the electrons originated from the anode. In a fuel cell, the oxygen reduction reaction happens at the cathode.

Electrocatalyst:

A material that is applied on the surface of an electrode to catalyze half-cell reactions.

Normal hydrogen electrode (NHE):

Also known as the standard hydrogen electrode (SHE), it is a redox reference electrode which forms the basis of the thermodynamic scale of oxidation-reduction potentials. The potential of the NHE is defined as zero and based on equilibrium of the following redox half-cell reaction, typically on a Pt surface: 2H⁺(aq) + 2e⁻→H₂(g)

The activities of both the reduced form and the oxidized form are maintained at unity. That implies that the pressure of hydrogen gas is 1 atms and the concentration of hydrogen ions in the solution is 1 M.

Oxygen reduction reaction (ORR):

An electrode reaction, in which oxygen gas is reduced at the cathode of an electrochemical cell. The product of the reaction can be water molecules, hydroxyl ions (OH⁻), or sometimes hydrogen peroxide molecules. It is a very important and much-studied electrochemical reaction because it occurs at the cathode of practically all fuel cells.

Proton-exchange membrane fuel cells (PEMFC):

Also known as polymer electrolyte membrane fuel cells, these are a type of fuel cells that use proton-conducting-ionomer membrane as the electrolyte to separate anode and cathode. Their distinguishing features include low operating temperature (<80 °C), high power density, quick start-up, and quick match to shifting demands for power. They are being developed for transport applications as well as stationary and portable applications.

Pt mass activity:

The kinetic current of the oxygen reduction reaction normalized by the mass of Pt metal contained in the electrocatalyst.

Pt-specific activity:

The kinetic current of the oxygen reduction reaction normalized by the electrochemical surface area of the Pt metal contained in the electrocatalyst.

Reversible hydrogen electrode (RHE):

This differs from the NHE by the fact that the hydrogen-ion concentration of RHE reaction is the same as that in the actual electrolyte solution used for the working electrode. The potential of RHE is therefore (−0.059* (pH of the electrolyte)) V.

Definition of the Subject

H₂/air-powered PEM fuel cells are a future substitute for combustion engines as the green power source for transport application. In PEM fuel cells, because of their low operating temperature and low pH, both anode and cathode reactions are catalyzed by Pt or Pt-based electrocatalysts. Pt is a precious and expensive noble metal, and therefore its loading requirement plays a major role in determining the cost of fuel cells in mass production. The anode hydrogen oxidation reaction on Pt is intrinsically fast and requires very little Pt, while the cathode oxygen reduction reaction (ORR) is a very sluggish reaction that consumes about 90% of the total Pt content in PEM fuel cells. The current Pt loading in the most advanced fuel cell vehicles that use state-of-the-art Pt-based catalysts is about four- to eightfold higher than the target established for mass-produced fuel cell vehicles. Therefore, lowering the Pt loading at the cathode is the most critical mission for the PEM fuel cell development. To do that, significant depth of knowledge in understanding the ORR on Pt and Pt-based electrocatalysts’ surfaces is required; and the search for novel Pt-based electrocatalysts with enhanced ORR activity is seemingly the most productive pathway.

Introduction

Ever since 1839, when Sir William Robert Grove introduced the first concept of fuel cells [1, 2], researchers have been continuously trying to apply fuel cells for stationary and mobile power sources [3, 4], because of their high energy efficiency and low environmental footprint. Fuel cells can be customarily classified according to the electrolyte employed, with different electrolytes operating at different temperature ranges. The operating temperature then dictates the types of fuels and electrode materials that can be used in a fuel cell. For example, aqueous electrolytes are limited to operating temperatures of about 200 °C or lower because of their high water vapor pressure. At these temperatures the fuel is, in applications requiring high current density and low cost, restricted to hydrogen. To accommodate the slow kinetics of the electrochemical reactions at such low temperature in acid environments, platinum catalysts are required for both cathode and anode. In high temperature fuel cells, CO or even CH₄ can be used as the fuel, and the catalyst is not necessary to be noble metals, because of the inherently fast electrode reaction kinetics.

Among various types of fuel cells, proton exchange membrane fuel cells (PEMFCs) have attracted the most attention in recent years due to their low operating temperature (about 80 °C), high power density, quick start-up, and quick match to shifting demands for power [3]. Hydrogen/air-powered PEMFCs make them the primary candidate for the power source of light-duty vehicles and buildings. In a recent study by Thomas [5], the author compared fuel cell and battery as the power sources for all-electric vehicles, and found that for any vehicle range greater than 160 km (100 miles), fuel cells are superior than batteries in terms of mass, volume, cost, initial greenhouse gas reductions, refueling time, well-to-wheel energy efficiency (using natural gas and biomass as the source) and life cycle costs. PEMFCs also allow direct use of methanol without a processor, called DMFC. DMFCs are the primary candidates for portable electronic applications; low power densities and high Pt requirements have precluded their use in vehicles to date.

One difficult challenge to make PEMFC vehicles cost-competitive with traditional combustion engine cars is the high platinum catalyst loading and poor catalyst durability of the fuel cells [6, 7]. In the mid- to late-1980s and in the early 1990s, the fuel cell team at Los Alamos National Laboratory (LANL) succeeded in demonstrating high performance H₂/air fuel cells with a platinum loading less than 0.5 mg Pt/cm² per electrode, which was one magnitude lower than its previous level [8,9,10,11]. In that effort, Raistrick [8] was the first one to cast ionomer into the electrocatalyst layer by impregnating a Pt/C electrode into the ionomer solution before hot-pressing it onto the membrane, and thus greatly increased the electrocatalyst-electrolyte interfacial contact area. Wilson et al. [9,10,11] improved that process by mixing the Pt/C powder catalyst with ionomer solution before coating the electrode. Those research achievements at LANL partly led to the renaissance in PEMFCs in the past 2 decades [4].

In a PEMFC, if pure hydrogen is used as fuel, the anode reaction is then the hydrogen oxidation reaction (HOR) at the surface of the anode platinum electrocatalyst. The hydrogen oxidation reaction (HOR) and hydrogen evolution reaction (HER) are by far the most thoroughly investigated electrochemical reaction system [12]. Due to the fast electrode kinetics of hydrogen oxidation at platinum surface [12,13,14], the anode platinum loading can be reduced down to 0.05 mg Pt/cm² without significant performance loss [15]. The cathode reaction in a PEM fuel cell is the oxygen reduction reaction (ORR) at platinum surface in an acidic electrolyte. In contrast to HOR at the anode, the cathode ORR is a highly irreversible reaction even at temperatures above 100 °C at the best existing catalyst – the platinum surface [16,17,18,19]. Gasteiger et al. [20] found that 0.4 mg Pt/cm² was close to the optimal platinum loading for the air electrode using the state-of-the-art Pt/C catalyst and an optimized electrode structure. Further reduction of the cathode platinum loading will result in cell voltage loss at low current densities that follows the ORR kinetic loss. The high platinum loading at the cathode originates from the slow kinetics of (ORR) at platinum surface. To make the fuel cell vehicles commercially viable on the market, the platinum loading on the cathode has to be reduced significantly. As shown in Fig. 1, when Pt loading requirement is translated from the target set for power-specific Pt consumption in g Pt/kW, a fourfold Pt mass activity (activity per unit platinum mass) improvement is required, if combined with a 50% reduction in mass transport-related voltage loss [21]. Recent increases in Pt prices suggest that one should be striving for at least an eightfold improvement.

Fig. 1

Pt-mass-specific power density [g_Pt/kW] versus cell voltage, E _cell [V], based on a 50 cm² single-cell H₂/air performance. η_tx = cell-voltage loss caused by mass transport; Pt_cath = cathode-Pt loading. The cell was tested at T _cell = 80°C, 100% RH (relative humidity), at a total pressure of 150 kPa_abs and stoichiometric flows of s = 2.0/2.0. Catalyst-coated membrane (CCM) was based on a ca. 25 25 μm low-EW (equivalent weight = 900) membrane, and ca. 50 wt% Pt/carbon (0.4/0.4 mg Pt/cm² (anode/cathode)). It was assumed that the cell performance could be maintained at a reduced anode loading of 0.05 mg Pt/cm² (Reproduced from [21] with permission)

There are two ways one might think of that could help to reach that goal: further increasing the platinum dispersion (defined as the ratio of surface metal atoms to total number of atoms) by making finer platinum particles, if there is no decrease of Pt-specific activity (activity per unit Pt surface area); or alternatively, increasing the Pt-specific activity. One could also seek a combination of these two approaches.

This selected brief review will be focused on the research and development progress on ORR kinetics. The origin of the problem related with the low ORR activity of platinum will be discussed, followed by a review of recent progress in making more active, more durable platinum-based ORR catalysts. These include platinum alloy catalysts, platinum monolayer catalysts, platinum nanowire and nanotube catalysts, and the more recent shape- and facet-controlled platinum-alloy nanocrystal catalysts. The progress in the mechanistic understanding on the correlation between the activity and the electronic and structural properties of surface platinum atoms will be reviewed as well. The future direction of the research on platinum-based catalysts for PEM fuel cell application will be proposed.

Electrocatalysis of the ORR at Platinum Surfaces

It is widely accepted that the ORR on platinum surfaces is dominantly a multistep and four-electron reduction process with H₂O being the final product. However, the detailed mechanism of ORR still remains elusive [17]. The overall four electron reduction of O₂ in acid aqueous solutions is

$$ {\displaystyle \begin{array}{l}{\mathrm{O}}_2+4{\mathrm{H}}^{+}+4{\mathrm{e}}^{-}\to 2{\mathrm{H}}_2\mathrm{O};\\ {}{\mathrm{E}}_0=1.229\mathrm{V}\, \mathrm{vs}\, \mathrm{NHE}\, \mathrm{at}\, 298\mathrm{K}\end{array}} $$

Since the four-electron reduction of oxygen is highly irreversible, the experimental verification of the thermodynamic reversible potential of this reaction is very difficult. The irreversibility of ORR imposes serious voltage loss in fuel cells. In most instances, the current densities practical for kinetic studies are much larger than the exchange current density of ORR; therefore the information obtained from current-potential data are confined only to the rate-determining step (RDS). On the other hand, in the ORR kinetic potential region, the electrode surface structure and properties strongly depend on the applied potential and the time held at that potential, which makes the reaction more complicated. While the relationship between the overall kinetics and the surface electronic properties is not well understood, it is widely accepted that in the multistep reaction, the first electron transfer is the rate-determining step, which is accompanied by or followed by a fast proton transfer [16,17,18]. Two Tafel slopes are usually observed for ORR on Pt in RDE tests in perchloric acid, from −60 mV/decade at low current density, transitioning to −120 mV/decade at high current density. The lower Tafel slope of the ORR in perchloric acid at low current density has been attributed to the potential dependent Pt oxide/hydroxide coverage at high potentials [22,23,24,25,26].

Recently, by using Density Functional Theory (DFT), Norskov and coworkers [27] calculated the Gibbs free energy of ORR intermediates as a function of cathode potential based on a simple dissociative mechanism, i.e., with the adsorbed oxygen and hydroxide being the only intermediates. They found that oxygen or hydroxide is so strongly bound to the platinum surface at the thermodynamic equilibrium potential that proton and electron transfer become impossible. By lowering the potential, the stability of adsorbed oxygen decreases and the reaction may proceed. They suggested that these effects are the origin of the overpotential of the ORR on platinum surfaces.

By setting the reference zero potential to be NHE, the proton chemical potential in the electrolyte is related to the electrode potential. The authors made DFT calculations to get the bond energies of O* and HO* for a number of interesting metals. From this, they can evaluate the reaction free energies of the two basic steps: the hydrogenation of the two adsorbed intermediates O* and HO*. The larger one of the two reaction free energies was taken as the activation energy barrier of the rate-determining step (RDS) in ORR. By using a microkinetic model they constructed, the rate constant of the RDS, and therefore the ORR activity, can be evaluated based on the activation energy barrier. As shown in Fig. 2a, the model predicts a volcano-shaped relationship between the rate of the ORR and the oxygen adsorption energy, with platinum and palladium being among the best metals for electrocatalysis of ORR. Fig. 2b shows that the bonding energy of OH is roughly linearly correlated to that of O, indicating both are nearly equivalent parameters in determining the ORR activity.

Fig. 2

(a) Trends in oxygen reduction activity plotted as a function of the O-binding energy. (b) Trends in oxygen reduction activity plotted as a function of both the O and the OH-binding energy (Reproduced from [27]. With permission)

More recently, Wang et al. [28] derived an intrinsic kinetic equation for the four-electron (4e⁻) oxygen reduction reaction (ORR) in acidic media, by using free energies of activation and adsorption as the kinetic parameters, which were obtained through fitting experimental ORR data from a Pt(111) rotating disk electrode (RDE). Their kinetic model consists of four essential elementary reactions: (1) a dissociative adsorption (DA); (2) a reductive adsorption (RA), which yields two reaction intermediates, O and OH; (3) a reductive transition (RT) from O to OH; and (4) a reductive desorption (RD) of OH, as shown below [28] (Reproduced with permission from [28]).

In contrast to a conventional ORR kinetic model, in this work there is no single particular RDS assumed, as the authors believe that a single RDS assumption may not hold over a wide potential region since the reaction pathway may change with potential, or there may exist two RDSs with similarly high activation barriers. The results indicate that the first electron transfer in the RA reaction ($\Delta {G}_{\mathrm{RA}}^{\ast 0}=0.46\mathrm{eV}$, where $\Delta {G}_i^{\ast 0}$ is the activation energy of reaction i (at equilibrium potential)) is not the rate-determining step (RDS) for the ORR on Pt at high potentials, because dissociative adsorption ($\Delta {G}_{\mathrm{DA}}^{\ast 0}=0.26\, \mathrm{eV}$) provides a more active adsorption pathway. However, the reaction intermediates, O and OH, are strongly trapped on the Pt surface, requiring considerable overpotential to overcome the barriers for O to OH transition ($\Delta {G}_{\mathrm{RT}}^{\ast 0}=0.50\mathrm{eV}$) and OH reduction to water and desorption ($\Delta {G}_{\mathrm{RD}}^{\ast 0}=0.45\mathrm{eV}$).

Thus, the ORR on Pt is desorption-limited at high potentials, exhibiting a low apparent Tafel slope at those potentials. Wang et al. [29] further used this kinetic model to fit a typical IR-free polarization curve of a PEMFC, by adjusting the parameters to reflect the fuel cell-operating conditions at 80 °C. The results showed that the transition of the Tafel slope occurs at about the same 0.77 V that is the equilibrium potential for the transition between adsorbed O and OH on a Pt surface with low adsorbed O coverage [27].

Neyerlin et al. [30,31,32] investigated the ORR kinetics on high-surface-area carbon-supported platinum catalyst Pt/C in an operating PEMFC. By assuming the transfer coefficient α = 1 and using a single Tafel slope, i.e., −70 mV/decade at 80 °C, three kinetic parameters could be extracted through fitting the kinetic model to the fuel cell data: the exchange current density or current density at a constant IR-free cell potential, the reaction order with respect to oxygen partial pressure, and the activation energy. One may need to note that the lower limit of the electrode potential after IR and transport correction in this work is about 0.77 V, so that this single-Tafel-slope treatment can still be consistent with Wang et al.’s [29] result. Neyerlin et al. [30] were concerned about the accuracy of the kinetic current extracted from low potentials in RDE tests, since at these low potentials the experimental measured current is more than ten times lower than the kinetic current derived from it. The transport correction used in RDE analysis assumes perfect first-order kinetics, which is not strictly true, and errors from this imperfect correction could become large at low potentials. The authors also studied the relative humidity (RH) effects on ORR kinetics in PEMFCs [32]. They found that when RH is above 50–60%, the kinetics are independent of the RH, but they observed significant ORR kinetic losses at lower RH. The reduction of ORR kinetics at low RH was interpreted as most likely due to a lowering of the proton activity (therefore only indirectly related to the lowering of the water activity) via hydration of acidic groups and the sequestering of protons at low RH.

Another factor that plays an important role in determining the minimum loading of Pt catalyst required for PEMFCs is the Pt size effect on ORR, not only through altering the fraction of surface Pt atoms over the number of total Pt atoms, but also through changing the ORR kinetics per surface Pt atom (Pt specific activity). Earlier results by Blurton et al. [33] showed that in 20% H₂SO₄ at 70 °C, the highly dispersed Pt (with size of about 1.4 nm) supported on conductive carbon prepared through ion exchange on resin followed by pyrolysis has a Pt-specific activity 20 times lower than that of crystalline Pt black (with size of about 10 nm), though it is not clear to what extend the lower activity could be caused by contamination of Pt and/or a “burying effect“during the catalyst preparation. Blurton et al. [33] correlated the decreased ORR specific activity with the decreased coordination number of surface Pt atoms on smaller Pt particles, causing more severe oxidation of the Pt surface. Peuckert et al. [34] later investigated a series of Pt-on-carbon catalysts with Pt weight percents from 5% to 30% prepared by an impregnation method, with corresponding fractions of metal atoms on the surface from 1.0 (Pt size <1 nm) to 0.09 (Pt size <12 nm). The Pt-specific activity toward ORR in 0.5 M H₂SO₄ at 298 K on these catalysts was found to be constant for Pt particle sizes above 4 nm but to decrease by a factor of 20 as the particle size decreased from 3 to 1 nm. Taking into account the larger percentage of buried and these inactive Pt atoms in larger particles, this result suggested that the optimal Pt size for the maximum Pt mass activity is about 3 nm. Kinoshita [35] also reviewed and analyzed the particle size effect for ORR on Pt/C catalysts. Based on the literature data of ORR on Pt/C collected in H₃PO₄ [36,37,38] and H₂SO₄ [34] solutions, Kinoshita proposed the decrease of Pt-specific activity with decrease of Pt particle size is a consequence of the changing distribution of surface atoms at the (100) and (111) crystal faces. Recent literature data [18] in H₃PO₄ reported that when Pt particle size increases from 2.5 to 12 nm, there is about threefold of increase in Pt-specific activity and confirmed the optimal Pt particle size for maximum mass activity to be around 3 nm. Gasteiger et al. [21] investigated Pt/C and Pt black catalysts for ORR in HClO₄ solution at 60 °C, with the Pt particle size ranging from 2 nm to over 10 nm, and found that the magnitude of activity improvement is comparable to that in literature data [39, 40], although the absolute values are about ten times higher than those reported in Ref. [40], due in part to the use of a less-strongly adsorbing electrolyte.

It is well established that ORR on Pt single crystals is structure-sensitive, depending on the electrolyte. In H₂SO₄, the order of activity of Pt(hkl) increases in the sequence (111) <<(100) <(110) [19, 41]. The variation in H₂SO₄ originates from highly structure-specific adsorption of sulfate/bisulfate anions in this electrolyte, which has a strongly inhibiting effect on the (111) surface. Given that the dominant Pt crystal facets of high-surface-area Pt/C catalysts are {111} and {100}, the Pt size effect on ORR activity in H₂SO₄ can be explained by the structure-sensitive adsorption. In perchloric acid, the variation in activity at is relatively small between the three low index faces, with activity increasing in the order (100) < (111) (110) [42], owing to the structure sensitive inhibiting effect of OH_ads, i.e., a stronger inhibiting effect on (100) and smaller effects on (110) and (111) [41]. Norskov and other researchers [43,44,45,46,47,48], based on their DFT calculations, recently proposed using the concept of averaged d-band center energy to explain the reactivity of metal surface atoms, which was supported by numerous experimental data. According to them, when Pt particle size decreases, the average coordination number of surface Pt atoms decreases, causing the d-band center to move closer to the Fermi level and the reactivity of those atoms to increase. As a result of that, the Pt atoms bind oxygen-hydroxide stronger, and therefore, have a lower ORR activity. A stronger adsorption of OH species on Pt surface when the particle size is reduced to below 5 nm was reported in Ref. [49].

While the Pt particle-size effect on ORR suggests that a further increase of Pt dispersion by decreasing the particle size to much smaller than 3 nm will not improve Pt mass activity, it is worth noting that the Pt-size effect is not universally believed. Recently, Yano et al. [50] studied ORR on a series of carbon-supported Pt-nanoparticle electrocatalysts (Pt/C) with average diameters in the range of roughly 1–5 nm, combined with measurements on ¹⁹⁵Pt electrochemical nuclear magnetic resonance (EC-NMR) spectroscopy. They observed that ORR rate constants and H₂O₂ yields evaluated from hydrodynamic voltammograms measurements did not show any particle size dependency. The apparent activation energy of 37 kJ mol⁻¹, obtained for the ORR rate constant, was found to be identical to that obtained for bulk platinum electrodes. This was consistent with the negligible difference in the surface electronic properties of these Pt/C catalysts, revealed from the practically no change of surface peak position of ¹⁹⁵Pt NMR spectra and the spin-lattice relaxation time of surface platinum atoms with the particle size variation.

Nevertheless, facilitating the ORR kinetics by augmenting the Pt specific activity is important for the future of fuel cells. A recent perspective in Science by Gasteiger and Markovic [51] showed some promising advances in this aspect.

Pt-Alloy Electrocatalysts

Pt-alloy catalysts, predominantly Pt binary and ternary alloys with 3D transition metals, have been the main focus of catalyst research for PEM fuel cells in the past decades, as they confer enhanced ORR activities over those available from pure Pt catalysts. Great progress has been made in past decades in developing more active and durable Pt alloy catalysts and in understanding the mechanism of their activity enhancements. Two- to threefold specific activity enhancements versus pure Pt were typically reported in literature [21, 40, 52,53,54], while exceptions existed from earlier results, claiming either no enhancement [55,56,57], or over an order of magnitude enhancement [58]. As far as which alloy and what alloy compositions confer the highest ORR activity, there seems to be lack of general agreement. This is probably because the measured activity depends highly on the catalyst surface and near-surface atomic composition and structure, on impurities on the surface, and on particle size and shape, all of which could be affected by the preparation method, heat treatment protocol, and testing conditions. For example, to achieve the optimal alloy structure for maximum activity, different Pt alloy particles may require different annealing-temperature protocols to accommodate the distinctions between metal melting points and particle sizes [59, 60]. In a number of earlier papers [61,62,63] it was showed that Pt-Cr is the most active ORR cathode catalyst in phosphoric acid fuel cells, while some recent results reported that in the PEMFC-oriented settings, the most active Pt alloy ORR catalyst could be Pt-Co [64], Pt-Fe [58, 65], Pt-Cr [52], Pt-Ni [66], or Pt-Cu [67] at specific atomic ratios of Pt to the alloying elements. Several representative mechanisms have been proposed in the literature to explain the enhanced activities observed on Pt alloy catalysts: (1) a surface roughening effect due to leaching of the alloy base metal [68, 69]; (2) decreased lattice spacing of Pt atoms due to alloying [52, 61, 70];(3) electronic effects of the neighboring atoms on Pt, such as increased Pt d-band vacancy [52, 58, 65, 71] or depressed d-band center energy upon alloying [42, 64, 67, 72, 73]; and/or (4) decreased Pt oxide/hydroxide formation at high potential [52, 74,75,76]. The increased Pt surface roughness alone may help increase Pt mass activity but will not increase the Pt-specific activity. Other mechanisms are correlated with each other, for example, the decreased lattice spacing may affect the electronic structure of Pt atoms, which in turn may inhibit the Pt oxide/hydroxide formation. A more detailed discussion follows.

Jalan and Taloy [61] believed that the nearest-neighbor distance between Pt atoms plays an important role in the ORR, based on the reaction model proposed by Yeager et al. [16], i.e., the rate-determining step being the rupture of O-O bond via various dual site mechanisms. They proposed that the distance between nearest-neighbor atoms on the surface of pure Pt is not ideal for dual site adsorption of O₂ or “HO₂” and that the introduction of foreign atoms that reduce the Pt nearest-neighbor spacing would result in higher ORR activity. By testing a number of carbon-supported Pt-M alloy catalysts fabricated into gas diffusion electrodes, with various nearest-neighbor Pt atom distances determined from X-ray diffraction, a linear relationship was obtained between the activity and the distance, with Pt-Cr exhibiting highest ORR activity and smallest nearest-neighbor distance, as shown in Fig. 3. While the geometric distance between the neighboring Pt atoms is shorter in alloys, the surface electronic structure of Pt alloys is different from that of pure Pt as well, so it is difficult to separate the two factors. Yet other studies [55, 68, 69] claimed no activity enhancement was observed on Pt-Cr alloy over pure Pt, except for the increased surface roughness [68, 69], but this has not been supported by more recent literature [52, 70].

Fig. 3

Specific activity for the ORR versus electrocatalyst nearest-neighbor distance (Reproduced from [61] with permission)

Mukerjee and coworkers did a series of studies [52, 74, 77, 78] on Pt binary alloys for ORR applying in situ X-ray absorption spectroscopy (XAS) to electrochemical systems. The spectra consist of two parts, the near-edge part XANES (X-ray absorption near-edge structure), which gives chemical information and EXAFS (extended X-ray absorption fine structure) , which gives the structural information around the element of interest. In a related study, Mukerjee et al. [52] investigated five binary Pt alloys (PtCr/C, PtMn/C, PtFe/C, PtCo/C, and PtNi/C) supported on high-surface-area carbon for ORR in a proton exchange membrane fuel cell. The electrode kinetic studies on the Pt alloys showed a two- to threefold increase in ORR activity relative to a reference Pt/C electrocatalyst, with the PtCr/C alloy exhibiting the best performance. Contractions in the Pt-Pt bond distances were observed by both EXAFS and XRD. In addition, they found that in the double-layer potential region (0.54 V versus RHE), the alloys possess higher Pt d-band vacancies than Pt/C, while in the high potential region (0.84 V versus RHE), Pt/C shows higher d-band vacancy relative to alloys. This was interpreted by the adsorption of OH species at high potential on Pt/C but to a lesser extent on Pt alloys. Correlation of the electronic (Pt d-band vacancies) and geometric (Pt-Pt bond distance) with the electrochemical performance characteristics exhibits a volcano-type behavior with the PtCr/C alloy being at the top of the curve, shown in Fig. 4. They rationalized the enhanced activity on alloys on the basis of electronic and geometric effects, and of inhibition of OH adsorption. Similar observations were also reported from the ORR measurements in phosphoric acid [66], and in alkaline solution [79,80,81].

Fig. 4

Correlation of the ORR performance of Pt and Pt-alloy electrocatalysts in PEMFC with Pt-Pt bond distance (solid circles) and the d-band vacancy of Pt (empty circles) obtained from in situ XAS (Reproduced from [52] with permission)

Surface segregation of Pt has been experimentally observed in a wide range of Pt-M binary alloys, such as Pt-Fe [82], Pt-Ni [83], Pt-Co [84], Pt-Ru [85], and has also been reported in theoretical calculations [86, 87]. Furthermore, it has been reported that the topmost layer is composed of pure Pt while the second layer is enriched in the transition metal M for Pt-rich Fe, Co, and Ni alloys [72, 76, 88], produced by displacement of Pt and M atoms in the first two layers to minimize the surface free energy during annealing. Stamenkovic et al. [76] studied polycrystalline Pt₃Ni and Pt₃Co alloys for electrocatalysis of ORR in acid electrolytes using the rotating ring disk electrode (RRDE) method. Polycrystalline bulk alloys of Pt₃Ni and Pt₃Co were prepared in ultra-high vacuum (UHV) having two different surface compositions: one with 75% Pt (by sputtering) and the other with 100% Pt (by annealing). The latter was called a “Pt-skin” structure and is produced by an exchange of Pt and Co in the first two layers. Activities of Pt-alloys for the ORR were compared to those of polycrystalline Pt in 0.5 M H₂SO₄ and 0.1 M HClO₄ electrolytes. It was found that in H₂SO₄, the activity increased in the order Pt₃Ni > Pt₃Co > Pt; in HClO₄, however, the order of activities was “Pt-skin/Pt₃Co” > Pt₃Co > Pt₃Ni > Pt. The catalytic enhancement was greater in 0.1 M HClO4 than in 0.5 M H₂SO₄, with the maximum enhancement observed for the “Pt-skin” on Pt₃Co in 0.1 M HClO₄ being three to four times that for pure Pt. The activity enhancement was attributed to the inhibited Pt-OH_ad formation on an alloy, even one covered with pure Pt, relative to the surface of a pure-Pt electrode, and the ORR reaction mechanism (pathway) was found to be the same on alloys as on a pure-Pt electrode. In a more recent study, Stamenkovic et al. [64] investigated polycrystalline Pt₃M (M = Ni, Co, Fe, Ti, V) surfaces for ORR in 0.1 M HClO₄, for both “Pt-skin” and sputtered alloy surfaces. The activity was correlated to the d-band center energy obtained in UHV via ultraviolet photoemission spectroscopy (UPS). A “volcano behavior” was revealed with the Pt₃Co has the highest activity for both “Pt-skin” and sputtered surfaces. The “Pt-skin” surface was found to be more active than sputtered surface again for each Pt₃M. The electrochemical and post-electrochemical UHV (ultra-high vacuum) surface characterizations revealed that Pt-skin surfaces are stable during and after immersion to an electrolyte. In contrast, all sputtered surfaces formed Pt-skeleton outermost layers due to dissolution of transition metal atoms [89].

By using DFT calculations, Xu et al. [47] studied the adsorption of O and O₂ and the dissociation of O₂ on the (111) faces of ordered Pt₃Co and Pt₃Fe alloys and on monolayer Pt skins covering these two alloys. Results were compared with those calculated for two Pt(111) surfaces, one at the equilibrium lattice constant and the other laterally compressed by 2% to match the strain in the Pt alloys. The absolute magnitudes of the binding energies of O and O₂ follow the same order in the two alloy systems: Pt skin < compressed Pt(111) < Pt(111) < Pt₃Co(111) or Pt₃Fe(111). The reduced bonding strength of the compressed Pt(111) and Pt skins for oxygen was rationalized as being due to the shifting of the d-band center increasingly away from the Fermi level. They proposed that an alleviation of poisoning by O and enhanced rates for reactions involving O could be some of the reasons why Pt skins are more active for the ORR.

Chen et al. [88, 90] recently studied carbon supported Pt₃Co nanoparticles for ORR and correlated their activity with the chemical composition and structural information of individual particles. Conventional and aberration-corrected high-angle annular dark-field (HAADF), scanning transmission electron microscopy (STEM), and high resolution transmission electron microscopy (HRTEM) were used to obtain the particle compositional and structural information. For the acid treated Pt₃Co nanoparticles, they observed the formation of percolated Pt-rich and Pt poor regions within individual nanoparticles, analogous to the skeleton structure observed for sputtered polycrystalline Pt alloy surfaces after acid leaching [89]. The acid treated alloy nanoparticles yielded about two times the specific activity of pure-Pt nanoparticles. After annealing of the acid-treated particles, sandwich-segregation surfaces of ordered Pt₃Co nanoparticles were directly observed, with the topmost layer being pure Pt atoms. The specific activity of annealed nanoparticles was about four times that of pure Pt nanoparticles. The enhanced Pt-specific activity toward ORR was attributed to the reduced binding energy of oxygenated species, owing to combined two effects, i.e., the increased compressive strain in Pt atoms, and the ligand effect from underlying Co atoms.

Strasser et al. [59, 67, 91,92,93,94] recently applied a freeze-drying technique in the synthesis of Pt alloy nanoparticle catalysts with enhanced ORR activity. The Pt-Cu alloy catalyst after electrochemical dealloying was reported to have both mass and specific activities about four to six times those of a standard commercial Pt/C catalyst, in both RDE and MEA tests. The synthesis involved an impregnation/freeze-drying route followed by annealing. Preparation started with impregnation and sonication of a commercial 30 wt% Pt/C catalyst with an aqueous solution of a copper nitrate, with Pt:Cu atomic ratio of 1:3, followed by freezing in liquid N₂. The frozen sample was subsequently freeze-dried under a moderate vacuum (0.055 mbar). Reduction and alloying of Pt and Cu on the carbon support was thermally driven under a reductive H₂ atmosphere in a tube furnace. Electrochemical etching (voltammetric dealloying) was employed to remove the surface Cu atoms from Cu-rich Pt-Cu alloy precursors. Bulk and surface structural and compositional characterization suggested that the dealloyed active catalyst phase consists of a core-shell structure in which a multilayer Pt rich shell is surrounding a Pt-poor alloy particle core. This work constitutes significant progress on initial activity, since a fourfold of increase of Pt mass activity is the performance target for commercially viable fuel cell cathode catalyst [21]. Fig. 5a shows a schematic of the dealloying process, and Fig. 5b exhibits the Pt mass activities of Pt-Cu/C synthesized at different temperatures, compared with the Pt/C catalyst.

Fig. 5

(a) The schematic model of a Pt-Cu alloy particle before and after electrochemical dealloying of the near-surface Cu atoms (pink balls = Cu atoms; gray balls = Pt atoms); (b) Pt mass activities of Pt-Cu/C catalysts at various annealing temperatures compared to that of Pt/C catalyst (Reproduced from [67] with permission)

As to the mechanistic origin of the activity enhancement in dealloyed Pt-Cu catalyst, the authors believe geometric effects play a key role, because the low residual Cu near-surface concentrations make significant electronic interactions between Pt and Cu surface atoms unlikely. Therefore, they suspect that the dealloying creates favorable structural arrangements of Pt atoms at the particle surface, such as more active crystallographic facets or more favorable Pt-Pt interatomic distances for the electroreduction of oxygen, as predicted by DFT calculations [47]. A fourfold enhancement in Pt mass activity on monodispersed Pt₃Co nanoparticles with particle size of 4.5 nm was also reported recently [95].

Watanabe and coworkers carried out a series of studies on the mechanism for the enhancement of ORR activity on Pt-Fe, Pt-Ni, and Pt-Co alloys [53, 58, 65, 71, 96, 97]. By using X-ray photoelectron spectroscopy combined with an electrochemical cell (EC-XPS) [96, 97], they identified quantitatively oxygen-containing species adsorbed on electrodes of a pure Pt and a Pt skin layer (generated by acid treatment, not annealing, and therefore equivalent to the “skeleton” layers described by Stamenkovic et al. [89]) formed on Pt-Fe and Pt-Co alloys’ surface from N₂- and O₂-saturated 0.1 M HF solution. Four types of species were distinguished with binding energies at 529.6, 530.5, 531.1, and 532.6 eV; the first two were assigned to O_ad and OH_ad, while the latter two were assigned to the bilayer water molecules, H₂O_ad,1 and H₂O_ad,2. The XPS results showed that the Pt skin layer exhibited a higher affinity to O_ad but less to H₂O compared to pure Pt, particularly in the O₂-saturated solution. The enhanced ORR activity at the Pt skin/Pt-alloy electrode was ascribed to higher coverage of O_ad than that at pure Pt. They also found that such an enhancement is induced without changing the activation energy but the corresponding frequency factor value in the pre-exponential term from that of pure Pt. From the measurements in a flow channel with 0.1 M HClO₄, in the temperature range of 20–50 °C, a two to fourfold of ORR-specific activity enhancement was reported for those Pt alloys [53].

Pt metal dissolution in fuel cells has been reported to play a key role in Pt surface area loss and cell performance loss [98]. While Pt-alloy catalysts with fourfold enhancement in both Pt mass activity and specific activity relative to standard Pt/C catalyst seem achievable, as has been shown above, the long-term durability of the alloy catalysts is still a concern, due to dissolution of the base metal from the alloys [21, 99]. For Pt-Cu alloy catalysts, an additional possible risk is that the dissolved Cu may migrate from cathode through membrane to anode and deposit on the anode Pt surface, causing a poisoning effect on anode hydrogen oxidation kinetics, as the Cu redox potential is higher than the anode hydrogen redox potential. Fortunately, the durability of the cathode catalyst can be compensated by careful system control, mainly through lowering the cathode upper potential limit and narrowing its operating potential window [100, 101].

Pt Monolayer Electrocatalysts

Pioneered by Adzic et al. [102, 103], the idea of a Pt monolayer electrocatalyst has been one of the key concepts in reducing the Pt loading of PEM fuel cells in recent years. Pt submonolayers deposited on Ru nanoparticles had been earlier demonstrated to give superior performance with ultra-low Pt loading compared to commercial Pt/C or Pt-Ru alloy catalysts for the anode CO-tolerant hydrogen oxidation reaction [103,104,105,106,107]. More recently, Adzic and coworkers applied this concept in making novel Pt monolayer catalysts for the cathode ORR, which will be the focus of the review in this section. In general, the new method of synthesizing Pt monolayer catalysts involves underpotential deposition (UPD), a technique well known to produce an ordered atomic monolayer to multilayer metal deposition onto a foreign metal substrate [108,109,110]. Specifically, the method consists of two steps [111, 112]: first, a monolayer of a sacrificial less-noble metal is deposited on a more noble metal substrate by UPD, such as Cu UPD on Au or Pd; second, the sacrificial metal is spontaneously and irreversibly oxidized and dissolved by a noble metal cation, such as a Pt cation, which is simultaneously reduced and deposited onto the foreign metal substrate. The whole procedure can be repeated in order to deposit multilayers of Pt (or another noble metal) on the foreign metal.

The advantages of Pt monolayer catalysts include (1) full utilization of the Pt atoms that are all on the surface, and (2) that the Pt activity and stability can be tailored by the selection of the substrate metals. For example [102], when a Pt monolayer is deposited onto different substrate metals, as shown in Fig. 6, due to the lattice mismatch between the metals, it can experience compressive or tensile stress, which is known to affect the Pt activity by adjusting its d-band center energy [43, 47] and consequently its ORR activity.

Fig. 6

Models of pseudomorphic monolayers of Pt on three different substrates inducing compressive strain (Ru(0001) and Pd(111)) and tensile strain (Au(111)). (Reproduced from [102] with permission)

Pt monolayer deposits on Pd(111) single crystals (Pt/Pd(111)) and on Pd/C nanoparticles (Pt/Pd/C) have been studied for ORR [112], and improved activities compared to Pt(111) and commercial Pt/C, respectively, were reported. The ORR reaction mechanism of the monolayer catalysts was found to be the same as that on pure Pt surface. Pt/Pd(111) was found to have a 20 mV improvement in half-wave potential versus Pt(111), and the Pt/Pd/C had a Pt-mass activity five- to eight-times higher than that of Pt/C catalyst. If the total noble metal amount (Pt + Pd) is counted, the mass activity is about 80% higher than that of Pt/C catalyst [112]. The enhanced ORR activity is attributed to the inhibited OH formation at high potential, as evidenced from XAS measurements. In a real fuel cell test, 0.47 g Pt/kW was demonstrated at 0.602 V [113].

In order to understand the mechanism for the enhanced activity of a Pt monolayer deposited on Pd metal, the ORR was investigated in O₂-saturated 0.1 M HClO₄ solution on platinum monolayers supported on Au(111), Ir(111), Pd(111), Rh(111), and Ru(0001) single-crystal RDE surfaces [114]. A comparison of the polarization curves at 1600 rpm is shown as in Fig. 7. The trend of the ORR activities increase in the sequence Pt/Ru(0001) < Pt/Ir(111) < Pt/Rh(111) < Pt/Au(111) < Pt(111) < Pt/Pd(111).

Fig. 7

Polarization curves for O₂ reduction on platinum monolayers in 0.1 M HClO₄ solution on a RDE. Rotation rate = 1,600 rpm (Reproduced from [114] with permission)

The authors further correlated the kinetic activities of the “monolayers” with the Pt d-band center energies (not shown) and Pt-O binding energies, and found a “volcano” relationship, with the Pt/Pd(111) having the optimal d-band center, as well as Pt_ML-O-binding energy, for the maximum ORR activity. The “volcano” behavior was rationalized as being controlled by the two key steps in ORR, the O-O bond dissociation which is followed by the other step, the O-H bond formation. As shown in Fig. 8, the activation energies of the two steps both correlated linearly with the Pt_ML-O-binding energies (and with the d-band center energies, not shown), but in the opposite trend, indicating the Pt_ML-O binding can be neither too strong, nor too weak, for the best ORR activity.

Fig. 8

Kinetic currents (j_K; square symbols) at 0.8 V (versus RHE) calculated from Fig. 7 for ORR and the activation energies for O₂ dissociation (solid circles) and for OH formation (empty circles) on Pt/Ru(0001) (1); Pt/Ir(111) (2), Pt/Rh(111) (3), Pt/Au(111) (4), Pt(111) (5) and Pt/Pd(111) (6), as functions of the calculated binding energy of atomic oxygen (BEO) (Reproduced from [114] with permission)

For further fine-tuning of the monolayer Pt/Pd ORR activity, they further introduced mixed metal + Pt monolayer catalysts [115], which contained 0.2 monolayer of a foreign metal from selection of (Au, Pd, Rh, Ir, Ru, Os, and Re) combined with 0.8 monolayer of Pt co-deposited on Pd(111) or on Pd/C nanoparticles. The foreign metals have either a weaker M-OH bond (for the case of Au-OH), or a stronger M-OH bond (for the rest of the cases) than the Pt-OH bond. DFT calculations [115] showed that, in addition to altering the Pt d-band center energies, the OH_(−M) − OH_(−Pt) (or O_(−M) − OH_(−Pt)) repulsion plays an important role in augmenting the ORR activity, as shown below in Fig. 9. Instead of adjusting the composition of the top2most Pt monolayer, replacing the substrate Pd(111) with Pd₃Fe(111) to generate Pt/Pd₃Fe(111) was recently reported to also have an enhanced ORR activity [116].

Fig. 9

Kinetic current at 0.80 V as a function of the calculated interaction energy between two OHs, or OH and O. Positive energies indicate more repulsive interaction compared to Pt/Pd(111) (Reproduced from [115] with permission)

Another type of Pt monolayer catalyst showing improved ORR activities are Pt monolayers deposited on (noble metal)/(non-noble metal) core-shell nanoparticles [117]. The synthesis approach started with impregnation of high surface carbon into a mixed solution of noble metal precursor and non-noble metal precursor, followed by stir-drying in air. Core-shell metal substrates were formed by surface segregation of the noble metal at elevated temperature in a reductive atmosphere. A Pt monolayer was then deposited on the core/shell substrates by galvanic displacement of a Cu monolayer that was UPD-deposited onto the core-shell substrate particles. Three combinations were investigated: Pt/Au/Ni, Pt/Pd/Co, and Pt/Pt/Co. The enhancement of Pt mass activity of the best case (Pt/Au/Ni) was reported as being over an order of magnitude relative to the commercial Pt/C catalyst. The total noble metal mass activities were reported to be 2.5–4 times higher than that of Pt/C, with the Pt/Pt/Co having the highest number. The enhancement of activities was attributed to the geometric effect induced by the fine-tuning of the Pt lattice spacing with the substrate core-shell particles and to the inhibited PtOH formation because of lowering of the d-band center position relative to the Fermi level.

In a related study, Zhang et al. [118] investigated a partial monolayer of Au deposited on Pt(111) and on Pt/C nanoparticles for ORR. The catalysts were synthesized by Au displacement of a monolayer Cu that was deposited with UPD onto the Pt(111) or Pt/C substrates. Due to the valence-state difference between Au³⁺ and Cu²⁺, two thirds of a monolayer of Au was deposited onto each of the Pt surfaces. The Au atoms appeared to form clusters on Pt surfaces. While the ORR activities of the Au/Pt(111) and Au/Pt/C catalysts were slightly lower than those of pure Pt(111) and Pt/C catalysts, respectively; the stability of the Au/Pt/C was superior compared to that of Pt/C catalyst. Potential cycling tests were performed on RDE in 0.1 M HClO₄ solution, with the potential cycling window between 0.6 and 1.1 V, for 30,000 cycles. The catalytic activity of Au/Pt/C, measured as half-wave potentials on the O₂ reduction polarization curves obtained before and after potential cycling, showed only a 5 mV degradation in over the cycling period. In contrast, the corresponding change for Pt/C amounted to a loss of 39 mV. In situ X-ray absorption near-edge spectroscopy (XANES) with respect to the potential applied on the catalyst surfaces revealed the oxidation of Pt nanoparticles covered by Au at high potentials was decreased in comparison with the oxidation of Pt nanoparticles lacking such coverage.

More recently, Wang et al. [119] investigated the ORR on well-defined Pt/Pd and Pt/PdCo core-shell nanoparticles for the effects of particle size, facet orientation, and Pt shell thickness. The Pt shell was generated and the shell thickness was controlled by a novel method called the Cu-UPD-mediated electrodeposition method, in which repeated Cu-UPD/stripping, potential cycling, and Pt irreversible deposition occurred simultaneously in the same electrolyte, with the Pt deposition under diffusion control, until a desired thickness of Pt was achieved. Atomic level analysis using Z-contrast scanning transmission electron microscopy (STEM) coupled with element-sensitive electron energy loss spectroscopy (EELS) showed that well-controlled core-shell particles were obtained. ORR tests on RDE showed that Pt-monolayer catalysts on 4 nm Pd and 4.6 nm Pd₃Co cores exhibited 1.0 and 1.6 A/mg Pt mass activities at 0.9 V, respectively, about five- and ninefold enhancements over that of 3 nm Pt nanoparticles. Also, two- and threefold enhancements in specific activity were observed respectively, which were mainly attributed to the nanosize- and lattice-mismatch-induced contraction in (111) facets based on the DFT calculations using a nanoparticle model. Scale-up methods were developed for synthesis of the core-shell particles [120].

In summary, Pt monolayer catalysts show a promising pathway toward solving one of the major problems facing PEM fuel cells by enhancing the Pt-specific activity and the utilization of Pt atoms, and therefore reducing the cost of the cathode catalyst, although more fuel cell tests of durability are needed before the monolayer catalysts can be put in fuel cell vehicles. There is still a need for a reduction in the total noble metals in these catalysts.

Pt and Pt-Alloy Nanowire and Nanotube Electrocatalysts

As has been discussed in the introduction, low activity and limited durability are the two major issues related to the high-surface-area-carbon-dispersed Pt nanoparticle catalysts (Pt/C) in PEM fuel cells. Shao-Horn et al. [121] have given a comprehensive review on the instability considerations of Pt/C catalysts. Conventional PEM fuel cell catalysts typically consist of Pt nanoparticles in size of 2–3 nm that are supported on high-surface carbon with 20–50 nm primary carbon particles for electrical conductivity and high levels of catalyst activity. As it has been shown above, this kind of catalyst has already approached the maximum Pt mass activity as a pure Pt/C catalyst, as the Pt particle size is close to the optimal value. However, the 2–3 nm Pt nanoparticles are intrinsically not stable enough under PEM fuel cell operating conditions if no system-mitigation methods are applied. As an example, for a fuel cell short stack operating at steady-state open-circuit voltage (OCV, <0.95 V versus RHE), with H₂/air flows (stoichiometric reactant flows of s = 2/2) at 80 °C, fully humidified and 150 kPa_abs for 2000 h, the Pt surface area decreased from <70 m²/g Pt to <15 m²/g Pt, corresponding to an almost 80% of Pt surface area loss [98]. If accompanied by a similar percentage loss of activity, this is not acceptable for commercialization of fuel cell vehicles. In addition, corrosion of carbon support makes the situation even more serious [122, 123]. One way to address this issue is to lower the upper voltage limit of the fuel cell through system mitigation [100, 101], or alternatively, to improve the intrinsic Pt stability through catalyst design, such as using Pt or Pt-alloy nanowire/nanotube catalysts. The local curvature of the nanowire/nanotube Pt or its alloy is expected to be small (in at least one direction), and it consequently has a lower surface free energy and higher stability. On the other hand, the Pt-specific activity of the nanowire/nanotube is expected to be higher than its nanoparticle counterpart, as it should bind OH/O less strongly.

The 3M nanostructured thin film (NSTF) catalyst is such a non-conventional catalyst [124, 125]. The support particle is a crystalline organic pigment material, perylene red (PR), which is vacuum-deposited and converted to an oriented whisker phase by thermal annealing. The result is a uniquely structured thin film composed of highly oriented, densely packed crystalline organic whiskers [126]. A crystalline Pt coating can be vacuum-deposited on the whiskers. Fig. 10 shows SEM images of the NSTF catalyst-coated whiskers prior to incorporation onto the surfaces of a PEM to form a catalyst-coated membrane [127]. The cross section of the whiskers is on the order of 50 nm, and the lengths of the whiskers are controllable by the thickness of the as-deposited PR film, typically in the range of 0.5–2 μm. For a practical loading of <0.2 mg Pt/cm², the typical Pt crystallite size of the coated PR whiskers is 10–11 nm, and the specific surface areas of the NSTF-Pt catalysts are <10 m²/g Pt [124].

Fig. 10

Scanning electron micrographs of typical NSTF catalysts as fabricated on a microstructured catalyst transfer substrate, seen (top) in cross section with original magnification of ×10,000, and (bottom) in plain view with original magnification of ×50,000. The dotted scale bar is shown in each micrograph (Reproduced from [127] with permission)

Bonakdarpour et al. [128] investigated Pt/NSTF and Pt-Co-Mn/NSTF for ORR by using the rotating ring disk electrode (RRDE). The nominal chemical composition of the ternary alloy was calculated to be Pt_0.68Co_0.3Mn_0.02. The catalyst-coated whiskers were carefully brushed off of the original substrate web and applied onto the glassy carbon disk of the RRDE. The measurements were done in O₂-saturated 0.1 M HClO₄ at room temperature. The Pt-specific activity of the Pt/NSTF was found to be close to that of a Pt polycrystalline disk. A twofold gain of Pt specific activity was observed on Pt-Co-Mn/NSTF versus Pt/NSTF. In PEMFC measurements for the same loading of 0.2 mg Pt/cm², using GM recommended conditions for measuring specific and mass activity at 80 °C, saturated H₂/O₂ at 150 kPa_abs, at 900 mV, the NSTF Pt-Co-Mn catalyst-coated membrane (CCM) generated specific activities of 2.93 mA/cm²Pt, and mass activities of 0.18 A/mg¹ Pt. The specific activity is <12 times higher, and mass activity is about two times higher than those of TKK 47 wt% Pt/C.

To investigate the NSTF electrode stability under high voltage cycling, Debe et al. [129] tested a series of NSTF Pt and NSTF Pt-ternary catalysts along with Pt/C (Ketjen Black) and Pt/C (graphitic) types by scanning at 20 mV/s, between 0.6 and 1.2 V under saturated H₂/N₂ at 80 °C. The MEA’s contained the same 3 M-ionomer PEM having a 1000 equivalent weight (EW), and the same 3 M-coated GDL. The NSTF electrodes had loadings of 0.1 mg/cm², while the carbon- and graphitic-carbon-supported catalysts had loadings of 0.4 mg/cm². Fig. 11 compares the normalized surface area as a function of the number of CV cycles for all the samples. It is interesting to note that for the NSTF samples the Pt and ternary catalysts behave similarly and lose approximately 30% of the surface area out to 14,000 cycles. The Pt/C and Pt/graphitic carbon, on the other hand, lose substantially more surface area in significantly fewer cycles. The superior ability of the NSTF catalyst to withstand thousands of fast voltage scans over the potential range most critical for Pt dissolution and Pt agglomeration demonstrated a significant differentiating feature over carbon-supported catalysts [129].

Fig. 11

Normalized surface area versus number of CV cycles from 0.6 to 1.2 V for four NSTF catalyst samples and three Pt/carbon catalysts at 80°C. All MEAs used the same 3 M ionomer PEM and GDL (gas diffusion layer) (Reproduced from [129] with permission)

Recently, Chen et al. [130] developed supportless Pt nanotubes (PtNTs) and Pt-alloy nanotubes (e.g., Pt-Pd alloy nanotubes (PtPdNTs)) as cathode catalysts for PEMFCs. PtNTs were synthesized by a galvanic replacement of silver nanowires (AgNWs) by following a similar method developed by Xia and coworkers [131, 132]. The AgNWs were synthesized using a polyol method [133] and were subsequently heated at reflux temperature with Pt(CH₃COO)₂ in an aqueous solution. For the preparation of PtPdNTs, mixed aqueous Pt(CH₃COO)₂ and Pd(NO₃)₂ solutions were used. The diameter (Fig. 12a, b) and length of AgNWs are about 40 nm and 10 μm, respectively. After Pt replacement, the diameter, wall thickness (Fig. 12c, d), and length of the PtNTs are about 40 nm, 6 nm, and 10 μm, respectively.

Fig. 12

(a) SEM image of AgNWs. (b) TEM image and electron diffraction pattern (inset) of AgNWs. (c) SEM image of PtNTs. (d) TEM image and electron diffraction pattern (inset) of PtNTs (Reproduced from [130] with permission)

Chen et al. [130] tested the durability of these materials by cycling the electrode between 0 and 1.3 V versus RHE at a scan rate of 50 mV/s in argon-purged 0.5 M H₂SO₄ solution at 60 °C. As shown in Fig. 13a, the electrochemical surface area (ECSA) of the PtNTs decreased by about 20% after 1000 cycles, while the Pt-black and Pt/C catalysts lost about 51% and 90% of their platinum ECSA, respectively. Fig. 13b shows the comparison of typical ORR polarization curves of the respective catalysts obtained at room temperature in O₂-saturated 0.5 M H₂SO₄ using a rotating disk electrode (RDE) at 1600 rpm. At 0.85 V versus RHE, the mass activity of PtNTs was reported slightly higher than that of Pt/C, and 1.4 times higher than that of Pt-black catalysts. The specific activity of the PtNTs was 3.8 times and 1.8 times higher than those of Pt/C and Pt-black catalysts, respectively. For PtPdNTs, the mass activity was measured 1.4 times and 2.1 times higher than those of Pt/C and Pt-black catalysts, while the specific activity was 5.8 times and 2.7 times higher than those of Pt/C and Pt-black catalysts, respectively.

Fig. 13

(a) Loss of electrochemical surface area (ECSA) of Pt/C (E-TEK), platinum-black (PtB; E-TEK), and PtNT catalysts with number of potential cycles in Ar-purged 0.5 M H₂SO₄ solution at 60°C (0-1.3 V versus RHE, sweep rate 50 mV/s). (b) ORR curves (shown as current-voltage relations) of Pt/C, platinum black (PtB), PtNTs, and PdPtNTs in O₂-saturated 0.5 M H₂SO₄ solution at room temperature (1,600 rpm, sweep rate 5 mV/s). Inset: Mass activity (top) and specific activity (bottom) for the four catalysts at 0.85 V (Reproduced from [130] with permission)

The examples given above illustrate that it is possible to achieve both activity and durability goals of Pt-based catalysts through special morphology and structural design of the catalyst and electrode. There are many other important advancements not discussed here on nanostructured Pt and Pt-alloy catalysts/electrodes showing improvements in activity and/or durability for ORR, such as single-crystal-Pt nanowires grown on continuous carbon-layer-coated Sn-fiber 3D electrodes [134], Pt-Pd bimetallic nanodendrites [135, 136], faceted Pt nanocrystals [137], nanoporous Pt alloy electrodes [138,139,140].

Facet- and Shape-Controlled Pt-Alloy Nanocrystal Electrocatalysts

Stamenkovic and coworkers [42] demonstrated in HClO₄ that on RDE Pt₃Ni(111) single-crystal surfaces of <6 mm in diameter, the specific activity for ORR is about an order of magnitude higher than on the Pt(111) surface and is about 90 times higher than on Pt/C catalyst, while the other two low-index surfaces, [Pt₃Ni(100) and Pt₃Ni(110)] are much less active than Pt₃Ni(111). This result is very intriguing in that it suggests that if one can make Pt₃Ni nanocrystals with all exposed surfaces having {111} orientations, one can hope to gain an enhancement of specific activity by up to two orders of magnitude relative to state-of-the-art Pt/C catalysts. Recently, two interesting papers [141, 142] have shown progress on synthesizing such Pt alloy nanocrystals.

Wu et al. [141] recently reported an approach to the preparation of truncated-octahedral Pt₃Ni (t,o-Pt₃Ni) catalysts that have dominant exposure of {111} facets. The shape-defined Pt-Ni nanoparticles were made from platinum acetylacetonate [Pt(acac)₂] and nickel acetylacetonate [Ni(acac)₂] in diphenyl ether (DPE) using a mixture of borane-tert-butylamine complex (TBAB) and hexadecanediol as the reducing agents. Long-alkane-chain amines were used as the main capping agents, and adamantanecarboxylic acid (ACA) or adamantaneacetic acid (AAA) was used to affect the reaction kinetics. The population of truncated octahedral crystals could be adjusted by the types and amounts of the reducing agents and capping agents. Three sets of Pt₃Ni nanocrystals were generated with various truncated-octahedral crystal populations; see Fig. 14.

Fig. 14

TEM images of Pt₃Ni nanocrystals with truncated octahedron population of (a) 70%, (b) 90%, (c) 100%, and (d) HR-TEM image of a truncated octahedron showing the (111) lattice (Reproduced from [141] with permission)

While Fig. 14a, b contain 30% and 10% of Pt₃Ni cubes (with the remaining particles being truncated-octahedrons), respectively, Fig. 14c contains only truncated-octahedrons. The particle size is on the order of 5 to 7 nm. Only two types of facets are exposed of all the nanocrystals, i.e., the {111} and {100}. The fractions of the {111} surface area over the total surface area could be calculated based on the geometries of the shapes and the population statistics. The ORR kinetics of the nanocrystals were studied on RDEs in O₂-saturated 0.1 M HClO₄, at room temperature, at 1600 rpm, with a potential scan rate of 10 mV/s. Fig. 15 shows comparison of polarization curves, cyclic-voltammetry curves, mass activities, and specific activities of the Pt3Ni nanocrystals to the standard TKK Pt/Vulcan carbon catalyst. As shown in Fig. 15d, almost-linear correlations were obtained for both mass activities and specific activities versus the fraction of the (111) surface area over the total surface area. A tabulated kinetic activity comparison is shown in Table 1. The mass activity and specific activity comparisons were made at 0.9 V versus RHE. Roughly 4× mass-activity and specific-activity enhancements were observed on the 100% truncated octahedral Pt₃Ni nanocrystals over the Pt/C catalyst. While the {111} facets of the nanocrystals showed much higher specific activity than the {100} facets, as indicated in Fig. 15d, in agreement with trend found on bulk Pt₃Ni single crystal disks, the absolute values of the specific activities of the nanocrystals are still far below those observed on bulk single-crystal surfaces [42].

Fig. 15

(a) Polarization curves and CV curves (inset), (b) area (mA/cm²Pt), (c) mass (A/mg Pt) specific ORR activity for the t,o-Pt₃Ni and reference Pt catalysts; and (d) the correlations between specific activities and fractions of (111) surfaces of these Pt₃Ni catalysts. The ORR polarization curves were collected at 1,600 rpm (Reproduced from [141] with permission)

Table 1 ECSA, mass- and area- specific ORR activities of Pt₃Ni and Pt/catalysts (at 0.9 V versus RHE) (Reproduced from [141] with permission)

Another interesting report is from Zhang et al. [142] on the synthesis and ORR activity of Pt₃Ni nano-octahedra and nanocubes, with the two shapes of nanocrystals having only {111} facets and {100} facets exposed, respectively. The monodispersed Pt₃Ni nano-octahedra and nanocubes were synthesized via a high-temperature organic solution chemistry approach, which involved using mixed oleylamine and oleic acid at elevated temperature as the reducing agent and capping agent, and tungsten hexacarbonyl W(CO)₆ as the shape controlling agent. Detailed procedures for synthesis of the nanocrystals can be found in Ref. [142]. Fig. 16 shows the SEM and TEM images of those shape and size controlled nano-octahedral (a-e) and nanocube (f-j) crystals. The chemical compositions of the crystals were analyzed by using combined ICP-MS and EDS techniques (from both TEM and SEM), and the results suggested that the average molar ratio of Pt over Ni was 3:1.

Fig. 16

(a-e) Images for Pt₃Ni nano-octahedra. (f-j) Images for Pt₃Ni nanocubes. (a, f) Field-emission SEM images. (b, g) High-resolution SEM images. (c) 3D model of an octahedron. (d, i) TEM images. (e, j) High-resolution TEM images of single nanocrystals. (h) 3D model of a cube (Reproduced from [142] with permission)

Zhang et al. [142] further investigated ORR activities of the shape controlled nanocrystals by using RDE measurements. The ORR measurements were conducted in an O₂-saturated 0.1 M HClO₄ solution at 295 K. A characteristic set of polarization curves at 900 rpm for the ORR on Pt3Ni nano-octahedra, Pt3Ni nanocubes, and Pt nanocubes are displayed in Fig. 17a. After mass transport correction using Koutecky-Levich equation, and normalizing by the Pt surface area and mass, the kinetic activities (specific activity and mass activity) at 0.9 V were plotted in Fig. 17b. The Pt-specific activity of Pt₃Ni nano-octahedra were determined to be 5.1 times of that of the Pt₃Ni nanocubes and <6.5 times that of the Pt nanocubes, while the Pt mass activity of the Pt₃Ni nano-octahedra was <2.8 times of that of Pt₃Ni nanocubes and <3.6 times of that of Pt nanocubes. The significant shape dependence of ORR activity agreed with the observation from the extended Pt₃Ni single crystal surfaces, although the absolute values of specific activities observed on Pt₃Ni nanocubes and nano-octahedra were about four- to sevenfold lower than those reported in Ref. [42], respectively. One apparent puzzle in these reported results is that the Pt surface area or ECSA of the nanocrystals derived from the specific activity and mass activity (ECSA per unit mass of Pt = mass activity/specific activity) is 5–10 times lower than one would expect from the size of particles revealed by SEM and TEM images. The discrepancy may come from the low utilization of the surface area because of impurities or from overlap of the nonsupported nanocrystals. Another set of data for the high surface carbon supported those nanocrystals was reported in the online supporting information of the paper, and this showed a better agreement between measured ECSA and diameter to TEM, suggesting that particle aggregation caused the low area observed for the unsupported catalysts.

Fig. 17

(a) Polarization curves for ORR on Pt₃Ni nano-octahedra, Pt₃Ni nanocubes, and Pt nanocubes supported on a rotating glassy carbon disk electrode in O₂-saturated 0.1 M HClO₄ solution at 295 K; with scan rate = 20 mV/s; rotation rate = 900 rpm. Catalyst loading in terms of Pt mass: Pt₃Ni octahedra, 3.0 μg; Pt₃Ni cube, 2.0 μg; Pt cube, 1.1 μg. Current density was normalized to the glassy carbon geometric surface area (0.196 cm²). The arrow indicates the potential scan direction. (b) Comparison of the ORR activities on the three types of catalysts. Specific activity and mass activity were all measured at 0.9 V versus RHE at 295 K (Reproduced from [142] with permission)

In summary, the size- and shape-controlled synthesis of nanocrystal Pt-based electrocatalysts has shown a promising path to high Pt-specific activity, although the absolute number of the activity is still not comparable to that observed on extended Pt alloy single-crystal surface, probably due to the size effect, residual impurities and defects in the nanocrystal surface, and incomplete formation of smooth, segregated Pt layers on the facet surfaces. The Pt mass activity achieved for the best case is already about four times higher than state-of-the-art Pt/C catalyst. If the core of the nanocrystals can be replaced with some corrosion-resistant material but keeping the surface of the Pt alloy shell still in {111} facets, one could expect a significant further reduction of the Pt loading required for the cathode catalyst. In addition, the size of the particles can be larger to gain the benefits of the particle size effect, as the Pt consumption is determined by the thickness of the shell. The durability of such nanocrystals could be expected to be high because of the lack of low-coordination atoms in their surfaces.

Future Directions

Low platinum loading, high activity, and more durable catalysts still remain as critical challenges for PEFCs for automotive applications. Further fundamental understanding of the correlations between activity, stability, and structural properties at the atomic level are most desired from both theoretical and experimental perspectives. Studies of the connections between the activities of controlled-facet-orientation nanoparticles and extended single-crystal surfaces would be helpful. Structure- and surface-controlled syntheses of catalysts (Pt monolayer catalysts, nanostructured catalysts and electrodes, size- and facet-controlled Pt alloy nanocrystals, combined with core-shell structure) should provide a practical viable path to achieving fuel cell catalyst loadings required for large-scale commercialization.