What Can We Learn in Electrocatalysis, from Nanoparticulated Precious and/or Non-Precious Catalytic Centers Interacting with Their Support?

Abstract

This review is devoted to discussing the state of the art in the relevant aspects of the synthesis of novel precious and non-precious electrocatalysts. It covers the production of Pt- and Pd-based electrocatalysts synthesized by the carbonyl chemical route, the synthesis description for the preparation of the most catalytically active transition metal chalcogenides, then the employment of free-surfactants synthesis routes to produce non-precious electrocatalysts. A compilation of the best precious electrocatalysts to perform the hydrogen oxidation reaction (HOR) is described; a section is devoted to the synthesis and electrocatalytic evaluation of non-precious materials which can be used to perform the HOR in alkaline medium. Apropos the oxygen reduction reaction (ORR), the synthesis and modification of the supports is also discussed as well, aiming at describing the state of the art to improve kinetics of low temperature fuel cell reactions via the hybridization process of the catalytic center with a variety of carbon-based, and ceramic-carbon supports. Last, but not least, the review covers the experimental half-cells results in a micro-fuel cell platform obtained in our laboratory, and by other workers, analyzing the history of the first micro-fuel cell systems and their tailoring throughout the time bestowing to the design and operating conditions.

1. Introduction

Fuel cell technology is a promising strategy to provide power for areas where there is no access to the public or where the cost of wiring and electricity transfer network is important. These devices are usually employed as an energy source for purposes that require non-intermittent power stations such as power generation and energy distribution systems. The use of hydrogen as a fuel makes them clean energy systems with less noise, and pollution generation [1]. Currently, fuel cells are used in both small and large systems such as combined-heat-power (CHP) systems, mobile power, portable computers and communication equipment [2]. Despite all these advantages, there are some limitations for the use of fuel cells. For example, the short lifetime caused by impurities in the gas stream inlets, degradation and poisoning of the catalysts used as electrodes, slow kinetics to perform multi-electron charge transfer reactions, low density output power per unit volume and low accessibility are, to name a few, challenges to overcome. However, in the last years, a great progress in the development of this technology using novel catalysts and fuels aiming at optimizing the operation of fuel cells [2] has been made. Among the main fuel cells operation issues, we can find the high cost and availability of electrocatalysts, the degradation of the proton exchange polymeric membrane, and the oxygen reduction reaction (ORR) kinetics. This reaction takes place on the most active element, Pt, at an overpotential of around 0.3 to 0.4 V [3]. For this reason, fuel cells operate, in practice, at a much lower potential relative to the theoretically possible and with a very low efficiency [4]. The ORR is, in fact, one of the most important reactions in electrochemistry because of its central action in metal-air batteries, electro-synthesis of hydrogen peroxide and the manufacture of cathodes for fuel cells [4]. The molecular oxygen reduction is ideally performed by a transfer mechanism of four electrons. A process involving a single step in which molecular oxygen is reduced to water [5]. Any new ORR electrocatalyst should be investigated to determine the ORR kinetics, since the H₂O₂ produced often causes the degradation of the polymer electrolyte membrane in the heart of proton exchange cells [5]. Nowadays, an important variety of precious metals are used as ORR electrocatalysts.

Platinum supported on carbon is usually used as an anode to conduct the oxidation reaction of hydrogen (HOR) as well as at the ORR at the cathode. It is well stablished that Pt is costly and very sensitive to the presence of contaminants [6]. Numerous strategies have been explored to reduce the mass loading and improve its activity. One of these strategies is the formation of Pt-alloys with 3d transition metals. These kind of alloys can also inhibit the formation of adsorbed species such as hydroxyl, Pt-OH [7]. The transition metal alloys of Pt with, e.g., Ir, Ru, Rh, Re, Os, increase substantially the electrochemical activity, reduce the mass loading, induce lower overpotentials, and promote the reaction of four-e⁻ [8]. A study carried out by Min et al. [9] considers that the electrochemical property of Pt and its alloys is a result of electronic factors. They suggested that the ORR kinetics increases with the decreasing vacancies in the 5d band of Pt, since such hybridization is less favorable as there is an increase in particle size, implying a decrease of the adsorption strength of the oxygen species. Mukerjee et al. [7] concluded that, on PtCo/C, PtCr/C, PtMn/C, PtFe/C, and PtNi/C alloys, the catalytic activity is increased by an electronic combination which produces changes in the vacancies of the d band, and alloy network contractions. Thus, better ORR properties were obtained on PtCr/C. Markovic et al. [10] reported the synthesis of two Pt-based materials; PtFe and PtCo nanowires. The PtFe electrocatalysts presented higher activity than PtCo toward the ORR due to a greater Co leaching from the alloy. The mass activity and specific surface of the PtFe nanowires were higher than the commercial platinum electrocatalyst. Besides the effect obtained by alloying with early or late transition metals, the electronic properties of Pt NPs can also be modulated by interaction with the support. Indeed, in recent years, exhaustive research activities have been conducted for searching active Pt structures to reach the Department of Energy (DOE) targets [11], which led to the development of many different types of alloys, often identified as skin-layer-, core-shell-, and thin film- electrocatalysts. Greeley et al. [12] has identified, by a density functional theory (DFT) computational screening, a skin-type Pt_xY catalyst as a promising cathode material for the ORR. In this connection, Pt₃Y alloys were the most stable Pt₃M alloys able to bind oxygen species slightly less than pure Pt, therefore showing a higher ORR activity [13]. The high stability of Pt₃Y alloy was attributed to the almost half-filling of the metal-metal d-band of the two alloyed metals, which corresponds to the bonding states filling, whereas the antibonding states are empty.

Another kind of electrocatalysts are the transition metal chalcogenides. Indeed, the substitution in a cluster compound of Ru by Mo atoms in Mo₆Se₈ led to a ternary chalcogenide, Ru₂Mo₄Se₈, with interesting properties toward the ORR in acid media [14]. These type of materials are metal clusters with octahedral centers where the delocalization of electrons provides a high electronic conductivity acting as reservoirs of electrical charges, while it remains kinetically stable [15]. Transition metal chalcogenides have also the ability to provide binding sites for the reactants neighbors and intermediaries, they possess also the property to change the volume and distance during the electronic transfer. As a result of these studies, the Ru has been positioned as one of the main electrocatalytic alternatives for the ORR: it is less noble than Pt and is oxidized at a potential 0.2 V more negative than Pt, and the oxides and also hydrides formed on its surface can act as a chemical oxidant for the CO adsorbed at a potential between 0.3 and 0.4 V more negative than pure Pt, this prevents its poisoning as occurs in the case of Pt [16]. Furthermore, it has been shown that Ru possesses good electrocatalytic activity and that combining it with other elements they show greater stability [17,18,19,20]. In the Ru_xSe_y material, the Se modifies the electrical properties of the active sites by synergistic effects facilitating the electron transfer [21,22,23,24]. Other studies demonstrated that the presence of Mo and Ru atoms decreases the percentage of peroxide produced [25], the Sn reduces the ORR overpotential [26] and finally, the W facilitates the oxygen adsorption [27].

Another strategy to tailor the electrochemical ORR activity and reduce the amount of Pt is by replacing it with Pd, resulting in cheaper electrocatalysts with high ORR activity [28]. Pd has a valence electron configuration and a lattice parameter similar to Pt [29]. Furthermore, it is suggested that the oxygen reduction activity on Pt can be enhanced if Pd is used in combination with other elements which can improve the rate of the electro-reduction of oxygen [30] and the desorption kinetics of the ORR products. Pd is an element more abundant and also cheaper than Platinum; in addition, its stability in acid medium can be comparable with Pt. However, its performance to carry out the ORR is several times smaller in comparison to Pt and its alloys [31]. In order to enhance the catalytic activity of Pd catalysts for PEMFC (proton exchange membrane fuel cell), the particle size control [32], the structure [33], alloying metals [34] and supports [31] are crucial parameters to be considered. In general, the reactivity and selectivity of Pd catalyst for a certain catalytic reaction can be tailored by controlling their shape and thus the type of facets. Pd NPs perform the ORR in acid media and proceed preferentially via four-e⁻ [35], Pd-Ag, Pd-Cu alloys have shown catalytic activity for the ORR in acid medium with 2- and 4-electrons [36]. Recent reports have shown that Pd alloys such as Pd₃Fe(111) have excellent electrochemical properties [37,38]. Similar studies prove that PdCo/C alloys display comparable activity to Pt/C towards the ORR in acid medium [39]. It has been suggested that the increased oxygen reduction activity of PdCo alloys might be attributed to a Pd electronic stabilization due to the addition of this last element [40]. The electrochemical activity of PdCo/C containing 10%–30% of Co presented a half-wave potential value of 0.66 V/RHE (reversible hydrogen electrode) while Pt/C showed an E_1/2 = 0.68 V/RHE. Both electrodes were evaluated in a 0.1 M HClO₄ solution [41]. Several studies have shown that the catalysts formed by the Pt-Pd alloy present a comparable activity comparable with pure Pt [42] due to the facility to modify the length of the lattice parameter depending on the Pt amount in the alloy in order to favor the highest oxygen adsorption over the catalyst surface [43]. Among the electrocatalysts with different Pt-Pd ratios, the Pt₃₀Pd₇₀ presented an onset potential 17 mV more positive than Pt indicating a faster kinetics. The Pt₃₀Pd₇₀ electrocatalyst is thermodynamically more active for the ORR than pure Pt as revealed recently by the density functional theory (DFT) studies [44]. This scenario, aside from ORR, another interesting reaction exists where Pd-based catalysts play an important role in the oxidation of organic molecules (methanol (CH₃OH) [45], and formic acid (HCOOH)) [46]. Pd-based electrocatalysts have been employed to oxidize methanol; nevertheless, besides their good electro-activity, those catalysts can overcome the effect of carbon monoxide-poisoning [47] with a high performance as anode catalysts in alkaline DMFC’s. Pd-based catalysts show a higher tolerance than Pt to CO species however, the disadvantage of Pd-based catalysts is their lack of durability [48].

Under this scenario, many efforts have been made tailoring and improving the electrochemical activity and durability of Precious and Non-precious based catalysts to perform two of the most important electrochemical reactions in fuel cells; the hydrogen oxidation and the oxygen reduction reactions. In this work we emphasize the catalytic response of precious catalytic centers like Pt-, Pd-based electrocatalysts, and chalcogenides, synthesized via the carbonyl chemical route. We also present the catalytic performance of non-precious electrocatalysts produced via a free-surfactant chemical heating route. All these materials were tested towards the ORR and HOR. The selection of such materials was based on the experience that our research group has acquired in the synthesis methods and electrochemical evaluations. In addition, the electrocatalysts modification by the hybridizing process of some catalytic centers with some carbon and ceramic supports is discussed. Taking into account the most active materials synthesized in our laboratory, they were evaluated in a micro-fuel platform and the results were compared with the state of the art of the micro fuel cells domain analyzing the design, operating arrangements and conditions.

2. Materials Design

2.1. Precious Catalysts

Electrocatalytic materials in the nanodivided form can be obtained by the pyrolysis, in an organic solvent, of transition metal complexes [M_y(CO)_x] that contain a metallic cluster in their base structure, or using metallic salts and employing the carbonyl chemical route through the reactivity of carbon monoxide to achieve the chemical coordination of CO with a metal center in organic or aqueous solvents [49]. The physicochemical and catalytic properties of electrocatalysts depend directly on the chemical nature of the surface but also on their size and shape [50]. Due to its high feasibility and great potential to tailor electrocatalysts, the carbonyl chemical route has been extensively employed to produce active, and chemically stable electrocatalysts [50]. The carbonyl chemical route [51] involves the reactivity of metal carbonyls in solvents to obtain metallic clusters. It has been employed to synthesize Ru [49,52], Pt [49,53,54,55] and recently Pd-based [56,57] catalysts for electrocatalytic purposes.

2.1.1. Pt- and Pd- Based Electrocatalysts

The synthesis of carbonyl complexes was initiated by Longoni et al. [58]. These authors developed a process to produce the dianions [Pt₃(CO)₆]_n²⁻ (n = 10, 5, 4, 3) from the reduction of several platinum precursors (Na₂PtCl₆·6H₂O, H₂PtCl₆·xH₂O, Na₂PtCl₄, Pt(CO)₂Cl₂) using various reducing agents (alkali hydroxides under carbon monoxide, Fe(CO)₅, cobaltocene, alkali metals). The n value of the final product depends greatly on the nature of the reagents and the experimental conditions, the reductive carbonylation proceeds as follows [58]:

[PtCl₆]²⁻ → [Pt₃(CO)Cl₃]⁻ → [Pt₃(CO)₆]₁₀²⁻ (insoluble) → [Pt₃(CO)₆]₆²⁻ (olive-green)

(1)

[Pt₃(CO)₆]₅²⁻ (yellow-green) → [Pt₃(CO)₆]₄²⁻ (blue-green)

(2)

[Pt₃(CO)₆]₃²⁻ (violet- green) → [Pt₃(CO)₆]₂²⁻ (orange-red) → [Pt₃(CO)₆]²⁻ (pink-red)

(3)

The reactivity of these complexes depends on the value of n and generally, the trend was that increasing n results in an increase in the reactivity towards nucleophilic and reducing agents, while decreasing n results in an increase in reactivity towards electrophiles and oxidizing agents.

For Pt-based electrocatalysts, this method was modified in our group and can be shortly described as follows: the platinum carbonyl [Pt₃(CO)₆]₅²⁻ was prepared by mixing specific quantities of Na₂PtCl₆·6H₂O with sodium acetate (mol ratio of NaAc/Pt = 6) in methanol solution under CO atmosphere, for 24 h. Activated carbon Vulcan XC-72 was aggregated to the previous solution and stirred for another 12 h under nitrogen atmosphere. The product, nanostructured Pt/C, was washed, filtered and dried [50]. The above process is shown in the reaction (4) [51].

[Pt₃(CO)₆]₁₀²⁻ + M_yX_n → [Pt_xM_y(CO)_z] → Pt_xM_y

(4)

where M = Sn, Ni, Cr, Co, and X = Cl⁻.

Based on these previous studies, CO tolerant carbon supported anode bimetallic electrocatalyst Platinum-Tin (Pt₃Sn) was developed [59]. Figure 1a,b schematize the process established by Longoni and Chini [58], and by Alonso-Vante’s group [59], respectively. It is important to mention that the process followed by Longoni and Chini stops at step 4. The advantage of the reactivity of the cluster complex is that careful handling the platinum complex, can be taken to deposit the Pt nanoparticles over several supports (step 5). In this case, by depositing carbon black (Vulcan XC-72) Pt/C, can be obtained.

As a result of this synthesis modification, it was found that commercial Etek catalysts presented a broader particle size than the homemade Pt-Sn produced also by the carbonyl chemical route, Figure 2.

In the same line, further bimetallic nanoalloys, e.g., carbon-supported well dispersed Pt-Ni; and Pt-Cr alloys were obtained [54,55]. In this latter, CrCl₃ was used with a final heat treatment from 200 °C to 500 °C. The lattice parameter presents an inversely proportional relation with the content of Cr (Pt/C Etek 0.3923 nm, Pt:Cr (3:1) 0.3895, Pt:Cr (2:1) 0.3863 and Pt:Cr (1:1) 0.3817) [54]. Pt-Cr showed an enhancement factor of 1.5–3 in the mass activity (MA), and in the specific activity (SA), Figure 3; because of the changes in this lattice parameter and the Pt-Cr composition.

For Pt-Ni, the wide angle X-ray spectroscopy patterns, and the Debye function analysis showed that the as-prepared Pt-Ni electrocatalysts presented face-centered cubic structure and that the lattice parameter is reduced with the increase of Ni [55]. More important, both Pt-Cr and Pt-Ni alloys showed high tolerance to methanol than the Platinum electrocatalyst, Figure 4.

To explore the effect of the support material onto Pt NPs the synthesis of platinum supported on Vulcan-carbon (Pt/C) and multi-walled carbon nanotubes (Pt/MWCNTs), via the carbonyl chemical route, was done [61]. After 20 V cycles (CVs) from 0.05 to 1.2 V/RHE, an improved stability was obtained with Pt/MWCNT in comparison to Pt/C. This kinetics and stability improvement were attributed to a higher graphitization degree of MWCNT support, leading to an increased strength of the π-sites with the MWCNTs acting as anchoring centers for Platinum nanoparticles [62]. Furthermore, this interaction also limited the agglomeration of Pt nanoparticles.

Through a modification of the classic carbonyl chemical route, the synthesis of carbon supported Pt-Ti nanoparticles was performed [63]. For the chemical composition, a post heat treatment process on the distribution of the particle size, and also on the electronic properties of Ti and Pt atoms on the performance on the oxygen reduction, and the methanol tolerance were done. The catalytic activity and selectivity towards methanol showed a remarkable enhancement for the nanoalloyed structure. The improvement was caused by the annealing at 875 °C as a result of the alteration of the density of electrons present in the platinum d-orbitals modifying the strength of the oxygen adsorption. In consequence, the arrangement of a surface alloy with a high order degree is the responsible to mediate the methanol poisoning of platinum sites [63].

The carbonyl chemical route was further explored to synthesize, for the first time ever, carbon supported Pd NPs supported for the ORR [57]. Herein, K₂PdCl₆ and sodium acetate were used as precursors reacting with CO. All of them were dissolved in CH₃OH, a step which was the key in the modified process because this reaction has to be performed for 2 h in an ice-bath, in this way the reaction rate was decreased avoiding the agglomeration of particles during the complex formation. Thereafter, the process continued following the classical route. The synthesis process comparison for platinum- and palladium-based nanomaterials via carbonyl chemical route is contrasted in Figure 5.

As a result of this process, Pd samples showed metal mass-loading dependent morphologies. In 10 wt % Pd/C, Pd NPs were highly agglomerated on the carbon support. In Figure 6a, we can find many particles with an average diameter higher than 10 nm. For 20 wt % Pd/C, large amounts of Pd nanowires could be observed, see Figure 6b. In Figure 6c, it is possible to observe Pd nanorods which present an average diameter of 3 nm, these nanorods were well dispersed on the carbon support presenting a total composition of 30 wt % Pd/C. In Pd/C-4, the nanoparticles were almost agglomerated, Figure 6d. In comparison with platinum-based catalysts obtained by the carbonyl chemical route, the morphology of the palladium electrocatalysts changed depending on the metal mass loading [57].

The importance of this study lies in the description to control the morphology of palladium nanoparticles depending of the metal loading following the carbonyl route.

2.1.2. Transition Metal Chalcogenides

Transition metal chalcogenides are promising materials in the field of catalysis and photo-electrocatalysis, since they combine the properties of electrically conducting materials with the catalytic advantage of molecular metal-clusters [64]. In the 70s, various research groups developed the so-called Chevrel phases such as M₆X₈ (M = Mo; Re, X = S, Te and Se), having exceptional superconducting properties [65]. Perrin et al. [66] reported the synthesis and properties of Mo_xRe_ySe_z and Mo_xRe_yTe_z but also Mo_xRu_ySe_z, Mo_xRu_yTe_z and Mo_xRh_yTe_z [67]. The authors found that molybdenum sulfide phases and those with mono or divalent elements, a rhombohedric structure with a triclinic deformation predominated. They assumed that Mo and Re d-electron bands were responsible for the electrical superconductivity in these materials. All these samples were prepared by a direct synthesis from the elements in evacuated quartz ampoules, heated up to 1200 °C. The generated samples were ground and annealed at the same temperature for 24 h. To obtain the sulfide compounds, it was necessary to let the reaction take place near 1300 °C, and several subsequent heat treatments were necessary in order to get a pure sample. For selenide and the telluride clusters the synthesis was easier. In order to evaluate and manipulate the samples, the powders were converted into pellets and finally sintered [66,67].

The use of such transition metal chalcogenides (Mo_4.2Ru_1.8Se₈) as electrocatalysts for the oxygen reduction in acid medium was reported in 1986 [68]. The authors reported the first test of such materials to perform the ORR with an appreciable cost advantage with respect to platinum [68]. It was further assessed that the Mo_4.2Ru_1.8Se₈ cluster, in powder form fixed on supports with the help of of Nafion^® ionomer thin film, onto glassy carbon, Figure 7a, and on p-type gallium phosphide (p-GaP) semiconductor, Figure 7b, favored the ORR, and hydrogen evolution reaction (HER) under Visible illumination [69]. Mo_4.2Ru_1.8Se₈ clearly enhanced both processes. On p-GaP surface the photo- induced hydrogen evolution was maintained. The limiting currents were reduced by only 3% at −0.56 V (vs. SHE). This was quite remarkable and suggested (now a phenomenon well established) that hydrogen, besides protons, can penetrate the Nafion layer easily, curve 3, in Figure 7b. The flat-band potential of the unmodified electrode was determined to be 0.94 V vs. SHE and was not shifted as a consequence of attachment of the Nafion film and the cluster particles.

Summing up, the Mo_4.2Ru_1.8S₈ cluster employed as a model catalyst for multi-electron reduction processes for the ORR, was not suitable for multi-electron oxidation processes—the oxygen evolution reaction (OER), due to insufficient kinetic stabilization [69]. Based on previous works [70,71], a certain degree of kinetic stabilization can be accomplished with chalcogenides, which can permit a photoreaction of holes via transition-metal d-states which are not significantly mixed with chalcogen states. Regarding the electrochemical activity towards the ORR, Mo_4.2Ru_1.8S₈, Figure 8, a pseudo ternary cluster, outperformed the ternary clusters, e.g., Ni_0.85Mo₆Te₈, where Ni atoms occupies the free sites in the channels between the Mo-clusters. Two main positive effects on the pseudo ternary compound were claimed, namely, the increase of metal-metal distance in the cluster, thus possibly facilitating the breaking of the –oxygen–oxygen– bond liberating OH⁻ ions. The expected shift of the electronic levels upwards was caused by the increase of the charge in the cluster.

Because the catalytic properties and the high density of d states at the Fermi level can be related, this relation on Mo₄Ru₂Se₈, Mo₂Re₄Se₈, Mo₆Se₈, and the Mo metal was undertaken [72]. Indeed, Mo₆Se₈ showed metallic properties, whereas Mo₄Ru₂Se₈ and Mo₂Re₄Se₈ showed semiconducting properties. The explanation lies on the fact that the substitution of Mo by Re and/or Ru atoms, that contain a high valence electron number, causes an increment of the electrons from 20 to 24 in the cluster M₆X₈, resulting in a filling of the e_g band. This effect shifts upwards the Fermi level, as schematized in Figure 9.

These compounds were evaluated towards HER and ORR. As a result of the narrow energy band caused by the d-state density, some improvement was observed on the HER, with a significant influence on ORR. This study revealed that hydrogen species solely interact with individual metal atoms, the molecular oxygen interacts with the atoms M, and M’ in the (MM’)₆X₈ cluster [74], as shown in Figure 10, thus forming a surface transition metal complex. Based on the statistics of the calculated possible combinations and taking again into account the density of d states, the complex IV must play a major role in the ORR electrocatalytic materials activity.

Based on these Chevrel phase model compounds, prepared in quartz ampoule at high temperature [67], Mo-Ru-Se, in powder and/or in thin layer form, was synthesized via a soft chemical method; the carbonyl chemical route by reacting molybdenum hexacarbonyl, tris-ruthenium dodecacarbonyl and selenium powder in xylene [75]. The XPS (X-ray photoelectron spectroscopy) characterization revealed the presence of SeO₂ and MoO₃, Figure 11. After cathodic electrochemical treatment, in presence or absence of oxygen at 0.35 V during 5 min in 0.5 M H₂SO₄, the emission line for SeO₂, disappears and the MoO₃, emission line is strongly reduced, see curves (2) in Figure 11, indicating that the catalysts still contain a small amount of Mo. Exposed to air for more than 100 days, curves (3), the MoO₃ species is further reduced to a minimum in comparison to the shortly treated electrode, curves (2). The absence of SeO₂, species demonstrates a higher stability than Chevrel phase compounds against oxidation under oxygen atmosphere [13].

The Extended X-ray absorption fine structure spectroscopy (EXAFS) disclosed that chemically generated Mo-Ru-Se electrocatalysts were, structurally speaking, different to that of the Chevrel phase compound [64]. It was concluded that the Mo and Ru atoms, in Mo-Ru-Se, were located as independent phases: MoSe_x, MoO_x and RuSe_x [76]. They, however, showed high selectivity to perform the ORR, in a solution containing methanol, with a four-electron reaction pathway, with an activity nevertheless, lower than that of the standard Pt catalysts in methanol-free electrolyte [77].

A further characterization by X-ray diffraction at wide angles (WAXS) of Ru_x and Ru_xSe_y was carried out [78]. Indeed, chalcogenide particles produced from the reaction between Ru₃(CO)₁₂ and Se were more resistant to oxidation than monometallic Ru_x nanoparticles with similar size. This phenomenon was associated to the coordination of Selenium atoms over the ruthenium atoms surface. As revealed by EXAFS analysis [79,80] the closest Ru-Ru averaged distance was 2.64 Å (corresponding to the hcp phase of metallic ruthenium) and that of the metal chalcogenide Ru-Se was 2.43 Å. It is worth noting that the calculated Debye-Waller parameters were a factor of 2 higher for the Ru-Se coordination distance than for Ru-Ru. Considering a particle diameter of 17 Å a pure ruthenium cluster-like model particle was developed containing 153 atoms (Ru₁₅₃). This is the third species in a series of hcp closed shell clusters with “magic” numbers 13, 57, 153, 323, etc. Therefore, within the hcp cluster model (17 Å particle size, 153 atoms), the number of ruthenium atoms available for electrocatalytic purposes at the surface of such cluster can be 153 − 57 = 96. Because x = 2 and y = 1 in the chalcogenide material, Ru_xSe_y, the cluster model contains 99 Ru atoms and 54 Se atoms, with an increased disorder of selenium atoms, deduced from EXAFS calculations [79], almost all selenium atoms are coordinated to ruthenium atoms on the cluster in a statistical way, as shown in Figure 12. In this way, the Ru-clusters surface coverage by Se atoms induces the necessary free active sites (Ruthenium sites) to perform an efficient coordination with oxygen/water [78].

The use of selenium atoms to coordinate catalytic center surface atoms, as well as impinging on it electronic modification was demonstrated for ruthenium in Ru_xSe_y [81]. This concept was further applied onto platinum, generated via the carbonyl chemical route [82]. The surface Pt NPs in Pt/C was modified with Se (selenization process). Platinum supported on carbon and SeO₂ were mixed in isopropanol solution with specific quantities and stirred for 12 h at room temperature. This mixture (20 wt % Pt_xSe_y/C) was heat-treated (200 °C) for 1 h under a nitrogen atmosphere. In order to account the optimal amount of Se atoms, to allow the adsorption and subsequent reduction of O₂ molecules, Se was concomitantly stripped from the Pt surface using Chronoamperometry at different stripping times, at an established anodic potential. Employing a stripping time of 20 min at 1.1 V/RHE, the ORR half-wave potential (E_1/2) shifted to positive values, whereas the OCP (open circuit potential) remained constant. The optimal Se coverage (see arrow in Figure 13a) kept to a maximum the ORR in the presence of methanol [83]. The partial leaching of Se atoms from the Pt surface improved the activity of Pt_xSe_y/C towards the oxygen reduction, with a high tolerance as compared to Pt/C, Figure 13.

2.2. Non-Precious Catalysts

Despite the efforts carried out in the production of novel and efficient electrocatalyst, platinum-based materials are still one of the most efficient oxygen reduction catalysts employed in PEMFCs [84]; however, their high cost, low abundance and poor durability are a main obstacle in the development of PEMFCs [85]. To overcome these issues, the search of non-precious metal ORR catalysts has been extensively conducted in recent years [86]. Pt-free catalysts such as transition-metal chalcogenides have shown promising activities towards fuel cell reactions [51]. Co and Fe are also promising cathodic catalyst centers [87]. In recent studies, researchers have focused their efforts on two kinds of non-precious metals as catalytic centers: Co and Fe, and transition-metal chalcogenides with application in the ORR [88].

The development of new non-precious electrocatalyst dates back to the seventies with Co₃S₄ and Me^(a)Me₂^(b)X₄ (Me^(a) = Mn, Fe, Co, Ni, Cu or Zn, Me^(b) = Ti, V, Cr, Fe, Co or Ni, X = O, S, Se or Te). This material showed the highest catalytic activity for the oxygen reduction in a 1 M H₂SO₄ solution, and presented an open circuit potential of ca. 0.8 V vs. HRE [87]. However, Co₃S₄ was not close to the catalytic activity of Pt. Based on this work other Co-based materials [89,90,91,92,93] came into light. Feng et al. [94] synthesized three Co₃S₄ (-01, -02 and -03) compounds following different methodologies. Co₃S₄-01 was produced from a Co₂(CO)₈ decomposition in a mixture solution which contained tri-octylphosphane oxide and oleic acid as surfactants, just after sulfur and cobalt particles were reacted and deposited over the carbon substrate [91]; Co₃S₄/C-02 was prepared following the same synthesis procedure for Co₃S₄-01 but without the use of surfactants. Co₃S₄/C-03 was synthesized by means of a microwave-assisted process [94]. Co₃S₄/C-02 and Co₃S₄/C-03 synthesized without surfactant showed the highest activity toward the ORR, Figure 14. This work suggested that the surfactants, used in the synthesis of Co₃S₄/C-01, block the electrocatalyst active sites. These results demonstrated that routes without the surfactants were effective methods to produce non-precious catalysts supported onto carbon [94].

In modifying and upgrading the Co-Se electrocatalysts synthesis, CoSe₂ nanoparticles with different nominal loading from 20 wt % to 50 wt % were produced [95] via conventional heating process [94]. In short, for the synthesis of 20 wt % CoSe₂/C, Co₂(CO)₈ and Vulcan carbon were well dispersed in p-xylene under a nitrogen atmosphere at 25 °C for a half-hour. This suspension was heated and refluxed, and then cooled down. Thereafter, a specific amount of selenium was dispersed in p-xylene and stirred half-hour. Both suspensions were mixed, heated, and refluxed for 10 min. The final product was filtered, washed, and dried under vacuum at 25 °C. All CoSe₂/C electrocatalysts with different nominal ratio were heat-treated at 300 °C under high purity N₂ for 3 h.

The oxygen reduction cathodic current obtained for 20 wt % CoSe₂/C was in agreement to the results reported in previous works for non-precious metallic electrocatalysts (ca. 2.5 mA·cm⁻² at 0.4 V at 1600 rpm) [96,97,98]. Some data from the literature were 2.0 mA·cm⁻² for Co_1−xSe/C (50 wt % Cobalt on Vulcan carbon) [96], and 2.5 mA·cm⁻² for (Co, Ni)S₂ thin films [98] at 2000 rpm. The OCP value of 0.81 V in oxygen-saturated 0.5 M H₂SO₄ surpassed the Co_1−xSe/C [96], and FeS₂ (0.78 V) [97]. The OCP was quite similar to that of NiS₂ (0.80 V), and CoS₂ (0.82 V) [98], but not better than that of (Co, Ni)S₂ (0.89 V) [98].

The electrode loading effect of CoSe₂ (from 11 μg·cm⁻² to 44 μg·cm⁻²) was investigated via the rotating ring disk electrode (RRDE) technique, Figure 15a. In this experiment, the disk current increased at an electrode potential of 0.5 V from 2.5 to 3.8 mA·cm⁻² by the increase of the catalyst loading, with a concomitant decrease of H₂O₂% generation (at ca. 0.32 V). The variation of the mass loading as a function of the peroxide production can be interpreted in terms of an increase of the active sites in the electrocatalyst. In other words, the H₂O₂ has additional time to react to water in a thicker layer. As a result of this effect, an increase in the total number of electrons, and a decrease in the hydrogen peroxide was detected at the ring electrode. Similar results on galena and pyrite materials were observed by Ahlberg et al. [99]. The authors found 25%–30% H₂O₂ production at the disk electrode. On the other hand, in Figure 15b we can observe the influence of 20 wt %, 35 wt %, and 50 wt % CoSe₂/C systems on the oxygen reduction activity and the hydrogen peroxide generation. Since the hydrogen peroxide generated on the surface of the ring electrode decreases from 30% to 15% when the loading rate is changed from 20 wt % to 50 wt %, there must exist a limiting catalytic site density. A similar trend was observed in other systems, e.g., 20 wt % Ru_xSe_y/C, which overpasses the current density of 70 wt % Ru_xSe_y/C [100], including Pt/C [101].

Besides the ORR activity on the Co-Se system, the methanol tolerance was further investigated in alkaline medium [102]. Without surprise, the CoSe₂ nanocatalyst performance in alkaline medium is better than that in acid medium revealing further that the carbon-supported cubic phase of CoSe₂ is a promising non-precious electrocatalyst for alkaline fuel cells [103]. Furthermore, the oxygen reduction activity of 20 wt % CoSe₂/C in 0.1 M KOH was slightly higher than that of Co₃O₄/rmGO (reduced mildly oxidized graphene oxide) and comparable to that of Co₃O₄/NrmGO (nitrogen-doped mildly oxidized graphene oxide) as reported by Liang et al. [104]. Apropos methanol tolerance, Figure 16 contrasts the ORR curves of 20 wt % CoSe₂/C with that of 20 wt % platinum supported on carbon in oxygen-saturated 0.1 M KOH containing methanol. For Pt/C, the onset potential shifts negatively from 1.0 to 0.6 V, with an oxidation peak observed at ca. 0.65 V when the methanol concentration increased from 0 to 5 M. For 20 wt % CoSe₂/C, the onset potential remains unchanged and no oxidation peak is observed testifying immunity to methanol. At j = −3 mA·cm⁻², however, the potential shifts from 0.68 V to 0.62 V for 20 wt % CoSe₂/C catalyst, suggesting that methanol species in alkaline medium perturb the ORR. Nevertheless, the results herein show that 20 wt % CoSe₂/C presented an improved selectivity to perform the oxygen reduction reaction in presence of methanol than Pt/C electrocatalyst in alkaline medium. However, CoSe₂ is not hundred percent tolerant to methanol.

3. Fuel Cell Reactions

3.1. Hydrogen Oxidation Reaction—HOR

The electrocatalysis of hydrogen is a central issue in low temperature membrane fuel cells (PEMFC). H₂ is oxidized at the anode electrode:

H₂ ⇋ 2H⁺ + 2 e^-

(5)

The HOR largely depends on the strength of the intermediate interaction and bonding of metal and adsorbed hydrogen atoms [105]. As a consequence, the binding energies of the adsorbed hydrogen to the surface should always be in a range of an adsorption energy. A stronger bond would hamper the catalysis process by reducing surface dangling states and active sites, while a weaker bond could possibly indicate an inert electrode surface character with low number of dangling states and unsuitability of electrode for the catalysis [106]. Equation (5) can be represented by the following microscopic steps:

H₂ ⇋ 2H_ad  Tafel

(6)

H₂ ⇋ H_ad + e^- + H⁺  Heyrovsky

(7)

H_ads ⇋ H⁺ + e^-  Volmer

(8)

So, that the overall reaction can be revealed knowing some parameters, such as the exchange current density, activation energy, charge transfer coefficients, and Tafel slope. In PEMFCs the mass loading reduction of the expensive Pt is required to catalyze the HOR. Significant efforts have been made on the potential use of partial or complete replacement of Pt electrocatalyst at the cathode [107,108,109].

For the replacement of the most common based on platinum group metals (PGMs = Pt, Pd, Ru, Ir, Rh and Os [110]) catalysts employed so far for the HOR in acidic media [43,111,112,113,114,115,116,117], anionic exchange membrane fuel cells (AEMFCs) are also very interesting because their applicability of non-Pt metal catalysts in alkaline medium, where most of the transition non-precious metals presents high stability [118,119]. While research on oxygen ORR catalysts in alkaline medium has been ongoing for many years [120], studies on hydrogen HOR catalysts for AEMFCs constitute practically a new field of investigation. In contrast to the fast kinetics of the HOR on PGMs catalysts in acidic media, it was found the HOR activity of several PGMs in alkaline electrolyte is around two orders of magnitude slower than the activity of these materials in acid electrolyte, all this as a result of an intrinsic HOR kinetic barrier [121,122].

Contrary, to the simple Volmer reaction performed in acidic media, Equation (8), in alkaline medium, Equation (10) is required to complete the HOR that comprise two different sub-processes of hydrogen adsorption on the metallic surface and water formation in alkaline medium [123].

H₂ + OH^- ⇋ H_ad + H₂O + 1e^-  Heyrovsky

(9)

OH^- + H_ad → H₂O + 1e^-  Volmer

(10)

Recent attempts to reduce the amount of Pt in the design and production of electrocatalysts for the HOR were carried out by Gu et al. [124]. Using a hydrothermal method, the synthesis of Pt@Ru in the face-centered cubic (fcc) phase with a metastable phase for Ru under ambient conditions was performed. The fcc Pt@Ru nano crystals were produced by a one-step hydrothermal procedure, in which K₂PtCl₄ and RuCl₃ were reduced by formaldehyde (HCHO) at 160 °C in the presence of sodium oxalate (Na₂C₂O₄), HCl, and poly(vinyl-pyrrolidone) (PVP). The samples were referred to as Pt_x@Ru_100−x, in which x was the molar percentage of Pt in the feedstock.

As revealed by density functional theory (DFT) calculations, a preferentially epitaxial growth of Ru atom layers on the non-closest-packed facets of hetero fcc metal seeds led to the formation of fcc Ru shells to form Pt@Ru tetrahedrons/C, Figure 17; these morphological arrangements showed enhanced electrocatalytic activity with an order of magnitude more efficient toward the hydrogen oxidation reaction (HOR) in acidic electrolyte as compared with hydrothermally synthesized Ru/C [124]. The sample composed by tetrahedrons shows an average size of 8.3 nm, Figure 17b,c, which are in agreement with the models of fcc tetrahedrons from each zone axes, explicitly indicating that these particles are fcc tetrahedral single crystals enclosed by (111) facets. High angular annular dark field-scanning transmission electron microscopy image and their corresponding contrast line profile are shown in Figure 17d,e.

The HOR polarization curves obtained on fcc Pt@Ru, Ru/C (as-prepared), and commercial Pt/C loaded on glassy carbon RDE with a rotation speed of 2500 rpm in H₂ saturated 0.1 M HClO₄ solution are presented in Figure 18. A significant HOR activity was achieved on fcc Pt@Ru as compared to Ru/C. Pt₁₀@Ru₉₀/C showed the highest activity among the core-shell samples, i.e., 0.19 A·mg⁻¹ and 0.30 mA·cm⁻². The mass, and area activities are higher, by one order of magnitude, than the as-prepared Ru/C (0.016 A·mg⁻¹ and 0.017 mA·cm⁻²) and comparable with those of commercial platinum (0.23 A·mg⁻¹ and 0.27 mA·cm⁻²).

In search of new materials for HOR, palladium phosphide electrocatalyst with a small particle size was synthesized. Its improved stability, in acid medium, and the increased electrocatalytic activity towards the HOR with respect to commercial Pd/C was obtained [125]. Transition metals alloyed with phosphorus produced metal-phosphide, resistant to corrosion of the underlying transition metal, presenting metallic conductivity, or low electrical resistivity as low-band gap semiconductors [126]. Furthermore, they can be synthesized following a variety of routes and presenting high stability in basic and acidic media in comparison to the unalloyed metals [127,128,129]. The catalytic activity of transition metal phosphides and metal-phosphide materials namely, Ni [130,131,132,133], Co [134], Mo [135] and W [136] towards the HER has been studied. From these experiments it has been determined that the HOR performance of Pd-phosphide is comparable to that of palladium supported on carbon (Figure 19). The materials show a similar curve for the HOR, although there is a reduction in the limiting current value for PdP₂/C. Materials containing phosphide showed a weak wave before the limiting current which it is associated with the hydrogen oxidation on the phosphide lattice. This indicates that H₂ diffusion could be higher in the particles than in the palladium precursor. However, a faster onset for the HOR than the Pd/C precursor was visualized on the phosphorus material by dividing the current density by the corresponding limiting current.

Regarding the HOR in alkaline conditions, a pioneering study investigating non-PGM catalysts for HOR in AEMFCs, was carried out by Lu et al. [137]. In this work, Ni nanoparticles were decorated with Cr. This system generated an initial maximum power density of ca. 70 mW·cm⁻², operating at 60 °C. Another recent report has described nickel-based anode catalysts in alkaline media to oxidize hydrazine [138].

In a recent contribution Ni was combined with Mo and Co in ternary alloys: NiMoCo, which were tested for HOR in alkaline conditions [119]. This latter showed an improved HOR activity versus Ni alone; however, the recorded high HOR activity could be sustained only for anodic potentials below of 0.1 V, apparently because of the excessive surface oxidation above this potential. The HOR on both CoNiMo and Pt increases with applied overpotential below ~0.1 V, Figure 20. While the HOR increases beyond 0.1 V on Pt, on CoNiMo this reaction decreases certainly due to the passivation of Ni on the catalyst surface as suggested [119,139].

Recent papers have proposed the use of various nanoscale electrocatalysts, such as Ni films partially covered by a palladium layer which provide one of the most active materials for the HOR [140]. These Pd/Ni electrodes were synthesized by an electrolytic deposition of palladium over a polycrystalline Ni surface varying the composition of the surface. The catalytic performance of the Pd/Ni electrodes towards the HOR was evaluated in alkaline medium, showing an improvement for this bimetallic catalyst. Limiting currents at 0.55 V in a H₂ saturated KOH solution, as well as the current exchange values are shown in Figure 21. Ni material did not present any catalytic activity in these conditions; this could be the result of a Ni surface covered with oxygenated species which block the surface and avoid the hydrogen adsorption. Nevertheless, a minimal palladium amount (0.5%) deposited onto Ni, activates the compound to perform the HOR. An increase of the coverage up to 17% in palladium shows two interesting phenomena; the kinetic current experiences a pseudo-linear increase from 1.5% to 17% palladium coverage. Contrary to the kinetic current increase, the onset potential presents a decrease as a function of the palladium loading. Both parameters did not present further improvement with a Pd deposition above 17%, Figure 21.

From this study the authors determined that the 17% Pd coverage corresponds to 7 atomic layers (approximately a thickness of 1.5 nm). It was also determined that Pd/C catalysts, under the same work conditions conveyed values below 0.1 mA/cm²_Pd at an overvoltage η = 0.05 V with a palladium loading of 5 μg/cm² [141], while Pd/Ni with 0.9 μg/cm² reaches 0.30 mA/cm²_disc and 0.57 mA/cm²_Pd at an overvoltage η = 0.05 V [120]. For Ni covered by a high amount of Pd, the exchange current density value reached a very similar one to that of Pd/C. This value < 0.1 mA/cm²_Pd is the usual value for palladium in alkaline medium, this was corroborated in a similar study where a gold surface was covered by palladium [142]. Coming back to the Pd/Ni study, the highest exchange current values were obtained with a low Pd coverage, even above 1.5% on Ni (Figure 21 (red dots)) [140].

To replace a portion of the expensive platinum metal without compromising the HOR electroactivity in alkaline medium, the synergistic ligand and strain effects were put into advantage by alloying Pt NWs (nanowires) to metals such as: Fe, Co, Ru, Cu, Au [143,144]. The synthesis of such materials was done by a solution-based technique [145]. Pt NWs were synthesized using a hexachloroplatinic acid hydrate solution combined with cetyltrimethylammonium bromide (CTAB) in chloroform. Subsequently, water was added to this mixture to be stirred for an additional 30 min at 1600 rpm. After 30 min reaction, a sodium borohydride solution was poured in, as the reducing agent, and the reduction process was confirmed by a color change from gray to black. The entire mixture was stirred for an additional 20 min at 1600 rpm, subsequently centrifuged and the supernatant was discarded. The binary NWs were analogously synthesized using commercial precursor. To generate a series of PtRu, PtFe, PtCu, PtCo, and PtAu ultrathin nanowires, ruthenium (III) chloride hydrate, iron(III) nitrate nonahydrate, cupric chloride, cobalt chloride and hydrogen tetrachloroaurate (III) hydrate were used, respectively. A cleaning protocol was used to electrochemically eliminate the CTAB surfactant and activate the electrode surfaces [146]. Figure 22a shows the performance of as-synthesized Pt NWs with that of binary alloy catalysts made of 70 atom % Pt and 30 atom % Fe, Co, and Ru, respectively. Pt₇Ru₃ NW catalyst was able to achieve a diffusion-limited current at lower potentials, indicative of an improved HOR kinetics. Pt alloyed with both Au and Cu (30 at. %) are depicted in Figure 22b. Specifically, the Pt₇Cu₃ NW catalyst attain the diffusion-limited current density at ~0.2 V, whereas the Pt₇Au₃ NW catalyst only attained this value at ~0.25 V. This observation would indicate relatively slower HOR kinetics for this particular Pt₇Au₃ composition, with respect to other binary catalysts. The experimental HOR exchange current densities obtained at 2500 rpm with respect to the calculated hydrogen binding energy (HBE) values are found in Figure 22c [147]. Summing up, the exchange current densities have been used to evaluate catalytic activity, close to the reversible potential, since this is the potential where HOR is kinetically limited [148].

There is no kinetic data in alkaline medium for rhodium electrodes. In this context, a study with nanostructured Rh electrodes with application in the HOR was carried out through the evaluation of the elementary kinetic parameters of the Tafel-Heyrovsky-Volmer (THV) mechanism [149]. By means of the sputtering of Rh on a glassy carbon substrate an electrode was prepared, this process was carried out in an inert atmosphere [150]. The analysis of the limiting current at an overpotential of η = 0.3 V as a function of the rotation rate, under the kinetic mechanism of THV, is shown in a Koutecky-Levich plot, Figure 23. At this overpotential, one can read a kinetic current density of 3.44 mA/cm².

To establish the THV-mechanism, Equations (3), (6) and (7), certain constrains that imply the Koutecky-Levich expression were considered [151]. We can observe how in the potential range of interest, the HOR followed preferentially the Tafel-Volmer route [149]. The equilibrium reaction rate value obtained from the correlated experimental results was j₀ = 1.18 × 10⁻⁴ A·cm⁻², this value was significantly lower than that obtained for Rh in acid medium (6.77 × 10⁻⁴ A·cm⁻²) [150].

3.2. Oxygen Reduction Reaction—ORR, and Carbon Supports

In addition to the precious metals, required to improve the kinetics of the ORR, the oxygen reduction reaction is also affected by the ionic species contained in the acid or alkaline media, where the electrocatalysts are evaluated [152]. The kinetics, and the electroreduction mechanism for oxygen is carried out through an adsorption of the oxygen molecules on the catalysts surface. This process can be affected by blocking species composed by the anions of the electrolyte or oxygen containing molecules (OH-ads). The presence of these species affects the coordination and the adsorption energy of the molecular oxygen over the catalyst surface, thus influencing negatively the reaction ORR mechanism [153]. Previous studies demonstrated that this deficiency of the electrochemical exchange is due to the effect of the blocking species: OH⁻ and anions present in electrolyte solutions. Such chemical species induce even a higher blocking effect on Pt-based electrocatalysts [154].

Various studies [155,156] have demostrated that the decrease in the ORR activity in sulfuric acid solutions (H₂SO₄) is caused by a substantial adsorption of sulfate (SO₄²⁻) or bisulfate (HSO₄⁻) onto the Pt surface which cause an adsorption mixed potential between the anions H₂SO₄⁻/SO₄²⁻ and the molecular oxygen. Another kind of species which affect the oxygen adsorption onto the catalysts surface are the phosphate anions (PO₄³⁻) present in the phosphoric acid solution (H₃PO₄). Thus, it was determined that prechloric acid (HClO₄) is an ideal electrolytic solution to perform the ORR in acid medium since the perchlorate anions (ClO₄⁻) practically do not interfere in the oxygen adsorption process at the catalysts surface [157].

On the other hand, the ORR kinetics is carried out in a more efficient manner in alkaline medium than acid medium [158]. This allows the use of catalysts with a lower price than Pt, such as silver, palladium, copper, nickel, among others [159]. The main technical disadvantages presented by the ordinary alkaline fuel cells (AFC’s) are the durability of the electrodes, in highly caustic medium, the adequate water supply, at the electrodes to avoid both the anodic flooding, and the properly cathode drying, and the progressive electrolyte carbonatation, which is caused by the CO₂ present in the atmosphere or produced from the oxidation of the fuel used [160]. The carbonates formation has a particular importance since these species precipitate over the electrodes, damaging and blocking the microporous structure of the catalysts and therefore the catalytic active sites reducing progressively the fuel cell performance [1]. To avoid the aforementioned issues, it was developed AFC’s with flowing electrolytes streams, compared with the AFC’s systems with stationary flow, these novel fuel cells present a better performance and lifetime [160,161,162].

Regarding catalysts supports, for Pt and non-precious metal based catalytic centers, carbon materials and oxides are most commonly used and investigated materials. Low and high graphitic carbon are, e.g., carbon black (CB), single-walled or multi-walled carbon nanotube (CNT), graphene oxide (GO) [55,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182] and carbon nitride (CN) [183,184,185,186] among others. The semi-conducting oxides, such as TiO₂, doped-TiO₂ and doped-SnO₂, have been reported as support for catalytic centers [187,188,189,190,191,192,193]. The conductivity of oxides can be improved by synthesizing oxide/carbon composites [194]. The supports of this kind of composite materials are widely used for Pt or other catalytic centers [194,195,196,197,198]. The utilization of supporting materials can affect the nanoparticle (NPs) morphology, dispersion, and create a support-metal interaction (SMSI) [199,200,201,202]. These facts can favor the catalytic process, and hence, the support has a great impact on the design of composites based catalysts, aiming at improving the kinetics and stability for the electrocatalytic processes in low temperature fuel cells.

TiO₂ + 2H₂ + 3Pt → TiPt₃ + 2H₂O

(8)

The SMSI effect is directly related with the electrochemical activity of novel electrocatalysts. This interaction was firstly found in metal NPs supported by oxide [199,200]. Tauster et al. demonstrated the loss of H₂ and CO absorption capacity on TiO₂ supported group 8 metals (Ru, Rh, Pd, Os, Ir and Pt) [200]. This phenomenon revealed the formation of a chemical bond between the metal and TiO₂, cf. Equation (8) [200].

Later, similar effect was discovered in, such as La₂O₃ [203], and ZrO₂ [204] supported noble metals. In addition to oxides, noble metals supported onto carbon-based materials showed also SMSI, as revealed by the electron transfer at Pt-C interface [201]. For non-precious metals, it is well known that Co–N chemical bonds form from cobalt-oxides supported N-doped carbon materials [205,206,207]. The SMSI effect was also reported in Ni/Al₂O₃, [208], Ni/HZSM-5 (zeolite) [209] composites. Recently, Alonso-Vante et al. [195,197,198,210] reported SMSI in Pt and rare earth oxide interface, in these experimentations different from classical SMSI, the SMSI was identified as structural surface modification of Pt NPs. Therefore, from this point of view, the SMSI widely exist in metal-support composites, induced by the surface/structure modification of the catalytic surfaces. The catalytic performance can thus be improved via tailoring SMSI, via, the use of oxide, oxide-carbon composites, and highly graphitic carbon supports [163,198,211,212], as revealed through the use of various analytical tools of data collected from XRD (X-ray diffraction), XPS or/and electrochemical CO-stripping [163,198,211,212]. A way to probe the SMSI for Pt surface, and structure modification can be brought into light from XRD data, since the Pt diffraction peak broadening is affected by crystallite size, micro-strain and stacking fault [211], on the other hand, the Pt XPS spectra may reflect the charge transfer at the interface, induced by chemical bonding [163]. The SMSI on Pt NPs surface can be probed further via CO-stripping [213].

3.2.1. Low Graphitic Carbon

With respect to low graphitic carbon materials, commercially available carbon black from Vulcan or Ketjen is usually applied to support nanoparticles (NPs) [164,214,215,216,217,218,219]. They were reported as supports for Pt, alloyed Pt-M (M = 3d transition metals), Pd and alloyed Pd-M NPs. For instance, commercial Pt NPs supported onto carbon black, have been widely used as a benchmark for electrocatalytic process in half-cells or fuel cells [195,197,198,220,221]. Pt-M nanoalloys supported by carbon black were developed to investigate the dealloying effect of M atoms on Pt surface towards ORR activity in acid medium [215,219]. Recently, supported by theoretical DFT (density functional theory) calculation, Chorkendorff’s group reported experimental facts showing that the Pt-RE (RE = rare earth elements) alloys were the most active and stable Pt-based electrocatalysts for ORR [12,222,223,224]. Therefore, the synthesis of highly and homogeneously dispersed Pt-RE nanoalloys is an interesting phenomenon in the domain of energy conversion. However, the synthesis is extremely difficult via a soft chemical route, since the reduction potential of metallic REs ca. < −2.0 V/RHE [197,198]. This is the reason why a layer of RE oxides usually cover the surface of the obtained Pt NPs, even after heat-treatment at 300–900 °C [195,197,198,210,225,226]; surprisingly, the ORR activity on RE oxides modified Pt surface was improved with respect to pure Pt surface. Therefore, it is very interesting to study the effect of RE oxides on Pt surface towards ORR, using carbon black as supports. As shown in Figure 24a–d, for Pt-M₂O₃ (M = Y, Gd and Sm) supported onto carbon black, the particle size and morphology were almost the same for both Pt and Pt-M₂O₃ NPs, whereas the ORR, Figure 24e, activity was enhanced on Pt-M₂O₃ NPs. Among oxide modified samples, the oxygen reduction activity on Pt-Y₂O₃/C overcame that on Pt-Gd₂O₃/C and Pt-Sm₂O₃/C samples.

As indicated above, cf. Figure 6, different Pd NPs morphologies could be generated by adding different amounts of carbon support, showing the impact of carbon black support towards the nucleation of Pd NPs [57]. Moreover, the morphological impact of Pd NPs towards ORR and fuels (methanol, and formic acid) oxidation were identified [56,57]. In this regard, Co₃O₄ NPs, with different morphology deposited onto carbon black, showed different catalytic activity towards ORR [227]. The synthesis of Co₃O₄ NPs was done in two steps: (1) dissolving Co³⁺ precursor into a solution of water and DMF (dimethyformamide); and (2) adding ammonia into the solution to form Co(OH)₃ followed by a heat-treatment at 330 °C [227]. Adjusting the ratio of water and DFM to 1:0 (Co₃O₄-10), 3:1 (Co₃O₄-31), 1:1 (Co₃O₄-11) and 1:3 (Co₃O₄-13), a different morphology of Co₃O₄ NPs was, indeed, generated, Figure 25a–d [227]. This phenomenon indicates that nucleation of Co(OH)₃ onto carbon support was controlled by the ratio of aqueous and organic solvents during the synthesis [227]. The ORR Tafel plot, Figure 25e further shows the morphological impact of Co₃O₄ NPs towards catalytic activity [227].

Concerning the carbon nitride supports, Di Noto et al. [183] described the synthesis of a series of Pd-based catalysts with a general formula K_n[Pd_xCo_yC_zN_lH_m] and a molar ratio y/x > 1 and y/x < 1. The most relevant results rely on the ORR activity presented by the Pd-based catalysts with y/x > 1 annealed at 700 °C and 900 °C which overpassed the activity presented by a commercial Pt/C electrocatalyst. With respect to HOR and methanol tolerance no-improvement against Pt/C was obtained. To explain this ORR catalytic improvement, the authors suggested that a high concentration of cobalt atoms provokes the Pd-rich domains coalescence improving the mass utilization efficiency of the active sites. Ongoing with the incorporation of active metals in a carbon nitride matrix, this group synthesized two Pd-based electrocatalysts [184]. The materials PdNi-CN_h 900 and PdCoAu-CN_h 600 were produced by a pyrolysis method. The samples were physical-chemically characterized and evaluated towards the ORR with and without the presence of methanol and HOR. The main physical-chemical results revealed the formation of a disordered graphitic carbon nitride domain in PdNi-CN_h 900. On the other hand, PdCoAu-CN_h 600 presented a poorer nitrogen atoms incorporation due to a less graphitic carbon nitride matrix. No enhancement towards in the ORR and the HOR against the commercial Pt reference was observed. However, both electrocatalysts showed a better methanol tolerance than commercial Pt. Di Noto el al. [186] also worked in the production of Pt-based-carbon nitride (CN) core-shell electrocatalysts, and their application as cathodes in a single fuel cell system. PtNi-CN and PtFe-CN were synthesized via pyrolysis process and decorating a hybrid organic-inorganic material. In general, both PtNi-CN and PtFe-CN improved the fuel cell performance in comparison with the commercial Pt assembly. PtNi-CN presented a higher selectivity and a four-electron pathway to perform the ORR than the PtFe-CN. This enhanced activity was the result of a higher density of neighboring platinum active sites in the PtNi-CN surface [186].

Continuing in the same context, carbon nanohorns (CNH) is another interesting low graphitic carbon material, showing horn shape carbon particles self-assembling to a dahlia-like morphology [228,229,230]. Our group reported recently N-doped (NCNH) supported CoSe₂, Figure 26a, for ORR in alkaline medium [229]. The ORR activity on CoSe₂/NCNH was improved with respect to CoSe₂/CNH, CoSe₂/C and NCNH [229]. The XPS results revealed a charge transfer from N atom to CoSe₂ nanocrystals in the CoSe₂/NCNH sample [229], thus indicating a support catalyst center interaction in CoSe₂/NCNH, responsible for the ORR kinetics enhancement [229].

3.2.2. High Graphitic Carbon

High graphitic carbon materials, such as carbon nanotubes (CNT) and graphene oxides (GO) are widely used as supports for both precious and non-precious catalytic centers [104,171,177,231,232,233,234,235,236,237,238,239,240,241] The high graphitic carbons are less corrosive than low graphitic carbon as shown in Figure 27a [164]. Taking CNT as an example, the Raman spectra show that the oxidized multi-walled CNT (ox.MWCNT) has dominant graphitic sp² domains, as compared to carbon (Vulcan XC-72) which shows highly disordered sp³ domains, Figure 27b. The surface ratio (I_D)/(I_G) domain was 2.1 and 1.2 for carbon Vulcan and ox.MWCNT, respectively.

When a catalytic center, e.g., Pt, is chemically deposited onto highly graphitic carbon the interaction takes place with both sp³ and sp² domains, Figure 28a,b, shows the close Pt particle size and morphology [201]. However, the surface chemistry is totally different as depicted in Figure 28c. There is a sole CO-oxidation peak on the Pt/Vulcan sample, whereas there are two peaks on the Pt/MWCNT-m sample [201]. Based on Raman, XPS and DFT calculation results one can conclude that the peak I should be related to monoxide carbon oxidation on the platinum surface interacted with the sp² domain of the support, and peak II attributed to oxidation on Pt site interacted with the sp³ domain [201]. The ORR stability, Figure 28d,e, in acid medium was improved on CNTs support with respect to carbon, and attributed to the interaction between Pt NPs and sp² domains of CNT.

Graphene, a highly graphitic carbon material, has been used for catalytic centers support. An increased ORR activity of Pt NPs deposited onto N-doped reduced graphene oxide (NRGO) was reported [242]. In this work, Pt NPs supported by GO, graphite, carbon black (Vulcan XC-72), reduced graphene oxide (RGO) and NRGO were synthesized and studied for ORR [242]. According to XPS results, Figure 29a,b, the electron transfer from Pt to N moieties was identified, and responsible for modifying the electronic state of Pt atoms, supported by DFT calculation [242], and the increased ORR activity, see Figure 29d. The finger print, onto the various supports, of Pt NPs are contrasted in Figure 29c.

The support also interacts with non-precious metal centers. This phenomenon is clearly shown for CoSe₂ in Figure 30a,b, where the ORR activity in alkaline medium was improved when the chalcogenide was supported on functionalized CNT (f-CNT) as compared to that on carbon black [243]. In acid medium, the f-CNT supported CoSe₂ was the most active catalyst for ORR, as compared to carbon black, TiO₂-C composite and NCNH supported CoSe₂ [243]. In alkaline medium, however, CoSe₂/NCNH showed the highest ORR activity, followed by CoSe₂/f-CNT [243]. One should note that the NCNH support was very active towards ORR in alkaline medium [229]. Therefore, it is not surprising that the oxygen reduction activity was further increased on CoSe₂/NCNH.

An important contribution in the metal-support interaction domain was reported by Lee et al. [244]. They developed three novel Pd-Pt electrocatalysts supported onto three different carbon supports, namely, defective graphene nanosheets (GNSs), herringbone graphite nanofibers (GNFs), and single-walled carbon nanotubes (SWCNTs). The synthesis of the electrocatalyst Pd₃Pt₁ supported onto such different supports was carried out by chemical reduction of Pd and Pt precursor deposited on the various carbon supports. Among other characterization techniques, Raman spectrum was one of the most interesting tools to determine that GNS presented the highest disordered carbon structure with a D/G ratio of 1.23 over the values of 1.12 and 0.61 for Pd₃Pt/GNF and Pd₃Pt/SWCNT, respectively. Pd₃Pt/GNS presented the highest activity toward ORR in acid medium producing the highest percentage of water (70.35%). Because the use of the same active metal supported onto various carbon supports, the authors attribute the enhanced performance of Pd₃Pt/GNs to a metal-support interaction. Two years later this group reported the synthesis of a Pd-based electrocatalyst supported onto reduced graphene nanosheets (RGN) which were produced by an electrostatic method (RGNSECM) but also via a chemical reduction method (RGNCM) [245]. They deposited cubic Pd nanoparticles (10 nm) onto these two RGN substrates by an electrostatic deposition process. XPS analysis reveals one of the main characteristics to understand the metal-support interaction effect; the carbon contents in RGNSECM and RGNCM were 95.9% and 77.2%, respectively. Additionally, the oxygen content for these supports were 4.1% for RGNSECM and 22.8% for RGNCM. This analysis confirms the higher number of physical defects on RGNSECM resulting in more oxygen-containing functional groups which improve the supports vacancies to perform a better oxygen reduction. The electrochemical evaluations carried out in alkaline medium pointed out a higher mass activity for Pd cubes/RGNSECM but also a best catalytic response in the ORR overcoming the electrocatalytic response of RGNCM and a commercial Pd/C electrode. Furthermore, RGNSECM produced the less amount of hydrogen peroxide showing an oxygen reduction tendency of four-electrons but also a higher stability. These excellent properties of RGNSECM were related to the substrate defects, which induce a strong binding trap for Pd, a higher charge transfer, and a downshift in the d-band of the Pd clusters.

3.2.3. Oxide and Oxide-Carbon Composites

Oxides, such as semiconducting TiO₂ have been used as support materials for Pt-based catalytic centers, since much more corrosion-resistant than carbon materials, cf. Figure 27a. To favor the charge transfer between the oxide support and catalytic center, the conductivity of Titania must be improved via doping. Recently, Mo- and Ru-doped TiO₂ (Ti_0.7M_0.3O₂, M = Mo and Ru) were reported as support for Pt NPs [193,246]. In both cases, the enhanced electron transfer between oxide and Pt was identified, leading to different Pt surface chemistry, catalytic activity and stability [193,246]. As depicted in Figure 31a, the X-ray absorption spectra show that the white-line intensity in X-ray Absorption Near Edge Structure (XANES) zone of Pt/Ti_0.7M_0.3O₂ was lower than Pt/C and PtCo/C [193]. In Figure 31b, the Pt/Ti_0.7M_0.3O₂ sample showed the lowest number of unfilled d-states, followed by PtCo/C, Pt/C and Pt foil, indicating that the electron transfer is carried out from the support to platinum, known as SMSI effect, induces the d-band vacancy of platinum. The effect is more significant on Ti_0.7M_0.3O₂ than on Pt. The ORR kinetics, in acid medium, shown in Figure 31c, is enhanced on oxide supported Pt with respect to Pt/C and PtCo/C. These results show that increased SMSI favors catalytic activity on Pt surface towards ORR.

In addition, oxide-carbon composites are good candidates as catalytic center supports [194,196,197,198,210,247]. Our group reported Pt NPs supported by TiO₂-C, and Y doped TiO₂-C [194]. It was found that TiO₂ became more conducting after doping with Yttrium (Y:TiO₂-C). This material was as conducting as the carbon black, ensuring, in this way, the efficiency of electron transfer between the catalyst center and the support. The SMSI effect between Pt NPs and support was improved on photo-deposited Pt NPs onto oxides with respect to Pt/TiO₂-C prepared via a soft chemical route [194]. The SMSI effect can be easily visualized in the CO-stripping process, see Figure 32a, where one can observe that the photo-deposition (PD) favors the interaction between Pt and sp² domain of carbon support, whereas the carbonyl chemical route (CCR) favors the Pt-sp³ interaction. Such an observation demonstrated a random dispersion of Pt NPs via CCR, while PD is more selective to deposit Pt NPs onto sp² domains of carbon. On TiO₂-C and Y:TiO₂-C, cf. Figure 32a, the Pt NPs interact with the sp² domain of carbon, and oxide. Concerning the ORR activity, in acid medium, there was a slight enhancement on Pt/Y:TiO₂-C, with respect to platinum samples without Y, Figure 32b. The oxygen reduction stability on Pt/Y:TiO₂-C, studied via an accelerated stability test (AST), from 0.6 to 1.0 V/RHE, is close to that of Y-free samples. However, when cycling from 0.6 to 1.2 V/RHE, the stability of Pt NPs was clearly put in evidence on the oxide, Figure 32c–f. Such an observation indicated that the SMSI of Pt-Y:TiO₂ stabilized the Pt NPs on the support.

4. Micro-Fuel Cell Platform

Micro-Laminar Flow Fuel Cells

In the power generation field, as novel energy sources for residential and industrial applications, fuel cells have become essential devices. The fast advancement in the technology demands multi-functional, smaller and lighter electronic devices. This has stimulated manufacturers to look for alternative solutions [248]. Portable electronics, mobile phones, and computing gadgets are among the largest and fastest growing devices nowadays [249]. The statistics reveal that worldwide the number of mobile phone users increased from about 500 million in the year 2000 to about 5000 million in the year 2011 [250]. Recent data show a quasi-exponential growth of these values. The International Telecommunication Union’s (ITU) latest report estimates that more than 7 billion mobile phone subscriptions were globally conducted by the end of 2014, with a population penetration rate of 97% [250]. Portable electronics currently rely on batteries as a power source, but the need for constant replacement and recharging, interrupted operation, disposal issues, and increasing power requirements has prompted the development of new power sources [251]. Thereby, micro fuel cells have become promising devices to satisfy the global power requirements [252] thanks to their easy operation, competitive efficiency and environmental friendly performance. Even nowadays, the fabrication of such devices confronts many fabrication challenges, namely, design fabrication, the gas diffusion layer (GDL), electrocatalysts improvement and the exchange membrane assembly (MEA) [253]. The development of such devices is documented in the reference [254]. A recent modification of micro fuel cell systems are the silicon-glass-bonding systems by Mex et al. [255]. Silicon-glass-bonding systems retain the Nafion^® membranes disadvantage, methanol crossover, therefore, extra barrier layers have to be employed. A new thin MEA consisted of sputtered catalysts deposited onto porous silicon supports, ceramics or foils. One of the main characteristic of those films is the low permeation rate and chemical and thermal stability.

On the other hand, researchers from the Mechanical Technology Incorporated (MTI) MicroFuel Cells institute carried out the development of a simple direct methanol fuel cell design that can be scaled to applications for a wide range of portable devices. This novel device design does not require the use of pumps and water collectors; these modifications reduce the manufacture cost but also allows the use of concentrated fuels [256]. On the same line, people from the Lawrence Livermore National Laboratory developed the prototype of a miniature fuel cell [257]. This device incorporated a thin-film and a microfluidic fuel processor into a common package. Methanol from replaceable cartridges was used as fuel, the power generation device provided three times longer operating time than rechargeable batteries. A design configuration in a planar array [258,259] a ‘‘flip-flop’’ arrangement which allows a fully continuous electrolyte [259] is shown in Figure 33. The series combination performance of the two configurations is shown in Figure 34.

The fabrication and evaluation of silicon wafers miniature fuel cells was reported by Yu et al. [260]. These devices consisted of a MEA between two silicon substrates. The design included feed holes and fuel distributors (channels in the silicon wafers). Pt/C electrocatalysts were used as electrodes (Pt loading = 4 mg/cm²). To form the MEA, two electrodes with an area of 5 cm² were hot-pressed to a Nafion 112 piece. The results showed that the fuel cell performance was increased as a function of the composite thickness (silicon wafer layers). The best performance was presented by the miniature fuel cell with smaller size channels, these results are presented in Figure 35 [260].

Until 2004, researchers were focused on tailoring, and miniaturizing, polymer exchange membrane fuel cells—PEM related issues [261]. It was not until 2005 that Choban et al. [262] worked on the first membraneless micro FC. This novel design eliminates the fuel crossover and cathode flooding issues presented in fuel cells. In this research, the membraneless micro FC performance was studied in acid, alkaline and also mixed-reactant media. In this last condition the cathode is in acid medium, while the anode is in alkaline medium or vice versa. In these flow conditions, the aqueous streams are merged into a microchannel and they continue flowing in a laminar flow through the channels configuration (Figure 36a).

In this experiment, unsupported Pt/Ru nanoparticles and Pt black were used as anode and cathode, respectively [262]. The individual electrodes polarization curves in different media are shown in Figure 36b, the results presented in this work are in agreement with the literature [263,264]. After several days, without operational interruption, any performance drop was appreciated. Also, any carbonate formation issue was found in this LFFC. In the mixed-reactant mode (anode-alkaline, cathode-acid), Figure 36c region II, we can observe how at constant cathodic potential the current potential value increased until the anodic value became the limiting factor.

The limiting behavior of mass-transfer is observed for the oxidation of methanol in these experimental conditions [265]. A higher overall cell potential, OCP and current density was obtained for the mixed-reactant configuration than those registered for all acidic and alkaline conditions.

The fuel cell operation in alkaline medium presents positive advantages in the kinetic of the anodic and cathodic reactions. However, the called “mixed-reactant” conditions allows the opportunity to increase the maximum OCP achievable in simple alkaline reactant conditions. Additionally, the design simplicity for this condition allows experiments to be carried out consecutively in a single fuel cell without modifying the system and changing only the pH of both electrolytic streams [262]. It is well known that the flow channel configuration [266] can induce the forced water removal inside the fuel cell [267]. According to this, Hsieh et al. [268] studied the flow fields configuration and effects under fixed operating conditions. Four flow fields were studied herein; the so-called interdigitated flow channel presented the most uniform transient current in comparison with the rest of channels designs: Parallel, mesh and serpentine. Finally, in this study it was developed a correlation of the water accumulated in the flow channels according to the time [268].

The development of a miniature PEM fuel cell stack with carbon bipolar plates was described in a previous contribution [269]. In Figure 37 is shown a representation of the PEM fuel cell stack. The stack is assembled in the following order: end plate, rubber gasket, bipolar end plate, rubber gasket, MEA, rubber gasket, bipolar plate, rubber gasket and end plate as depicted in the final fuel cell stack shown also in Figure 37. The way to keep together all those components is using screws which align the cell components and also allow the compression pressure control.

In this way, the stack was tested at different temperatures, Figure 38. The current-potential curves showed how the voltage increased as a function of the temperature increment. The optimal temperature conditions for this design was 60 to 80 °C and employing external hydration, because the dehydration of the membrane a cell drop performance occurred above 80 °C [270].

Regarding the methanol crossover issue, and the methanol tolerant electrocatalysts development for the ORR, Whipple et al. [271], worked on the miniaturization of fuel cells tailoring the design and simplifying the fuel cell operation, Figure 39.

For this experiment two different cathodic streams without methanol and containing the same amount of methanol at anode were used in the operation of the laminar flow fuel cell. In the same way, an acid solution (0.5 M sulfuric acid) containing 0.1, 1, 2, 5, 7, 10, or 15 M CH₃OH was used as anolite. Figure 40 shows the individual cathodic and anodic polarization curve. These curves show the high depolarization of the Pt cathode in the presence of a stream containing methanol, Figure 40a2, this effect does not affect the same Pt electrode in the free-methanol stream, Figure 40a1. As we can observe, this depolarization effect does not occur when a Ru_xSe_y electrode is used as cathode; on the contrary, we can observe a slightly enhancement. This result proves that Ru_xSe_y electrocatalyst presents high tolerance to CH₃OH and does not present a depolarization effect caused by a mixed potential as occurs with Pt [82,272,273,274].

The Ru_xSe_y performance increase in presence of methanol can be explained as a result of a change in the oxygen diffusivity and solubility (Figure 40b2). Itoe et al. [275] showed that these two last parameters change in function of the CH₃OH–H₂SO₄ concentration ratio. This research group found that upon addition of CH₃OH to 0.5 M H₂SO₄, the initial oxygen diffusivity jumps by a factor of 2.5 to 3; after this, this value decreases until the initial value.

This study demonstrated that the activity of the ruthenium chalcogenide (Ru_xSe_y) cathode increases in presence of CH₃OH concentration by 30 to 60% in the concentration range of 1 to 7 M [271].

The flow fields design of micro PEMFCs was reported by Lu et al. [253]. In this study, the authors evaluated the effect of different flow field topologies in the μ-PEMFCs performance, Figure 41. The electrochemical tests were performed at reactant flow rates of 15, 30 and 50 mL·min⁻¹. As a result of this experiment it was found that the μ-PEMFCs with different flow patterns present a similar tendency at flow rates of 50 mL·min⁻¹. On the contrary, the μ-PEMFC performance is rapidly deteriorated at low flow rates, this decrease of performance is caused by the micro channels flooding. The design with mixed and long multichannel yielded the best performance [253].

In Figure 42 we can observe the maximum power density value reached by each flow field (D1-D4) at a current density value of 175 mA·cm⁻². At a flow rate of 15 mL·min⁻¹, the power density value is in the range from 76 mW·cm⁻² to 105 mW·cm⁻² (38% discrepancy), 21% for the flow rate of 30 mL·min⁻¹, and finally 13% for the flow rates of 50 mL·min⁻¹.

Since we know that cluster-like compounds such as the ruthenium chalcogenide Ru_xSe_y presents a high methanol tolerance, and selectivity to perform the ORR in presence of methanol [82,271,272,273,274], Gago et al. reported on the electrochemical measurements of 20 wt % Ru_xSe_y/C as cathode for the oxygen reduction in the presence of formic acid in a microfluidic formic acid fuel cell (μFAFC) [276]. A commercial Pt/C electrode was used also for comparison purposes. It was found that at formic acid concentration of 0.5 M, Ru_xSe_y/C and Pt/C reached maximum current density values of 11.44 mA·cm⁻² and 4.44 mA·cm⁻², respectively. Meanwhile the power density values were similar for both materials, ca. 1.9 mW·cm⁻². On the other hand, at a concentration of 5 M formic acid the power density value for Ru_xSe_y outperformed 9 times the power density value obtained by Pt/C, Figure 43, attesting that the Ru_xSe_y electrocatalyst is more suitable than Pt/C to be used as cathode in a μFAFC [276].

In 2012, Brushett et al. [160] performed a very interesting experiment on the alkaline microfluidic H₂/O₂ fuel cells field. They used a micro fuel cell to evaluate the electrochemical activity of Pt/C-based electrodes in alkaline medium. Their studies demonstrated that the behavior of such alkaline fuel cells (AFCs) is restricted by carbonated species presented in the electrolyte. These species affected only the performance of the anode, whereas the cathode performance remains practically unchanged. The carbonate formation (CO₃²⁻/HCO₃⁻) is one of the major technical issues in AFCs, this species formation reduces the concentration of OH⁻ in the electrolyte, hence it reduces the solution conductivity and the kinetic of the electrode reaction. One strategy to mitigate the water management and carbonate issues in alkaline FCs is with a recirculating electrolyte, the flow of the electrolyte helps the water management and removes carbonates resulting in a carbon dioxide tolerance up to ~100 ppm [160]. In this work, experiments to study the tolerance to CO₂ poisoning were performed as described as follows; in the first condition, CO₂ was injected in the anodic side of the μLFFC in stationary electrolyte conditions (3 M KOH). In order to determine a baseline, the first evaluation of the cell performance was carried out without the presence of CO₂. After this, a flow of 15 sccm of CO₂ was injected to the anode for 10 min meanwhile a nitrogen flow of 15 sccm was flowed in the cathode. After the gas exposure, the device was electrochemically evaluated to determine any change in performance. Afterwards, a second CO₂ exposure (10 min) followed by the electrochemical evaluation were performed. Subsequently, a rinse of the chamber was done by flowing KOH solution (10 min). Finally, the fuel cell was tested in a stationary mode in order to determine the KOH rinse effect. To perform a second study, the research group followed the same conditions of the first tests but introducing the CO₂ at the anode with a flowing of 3M KOH solution at a continuous flowing rate (0.3 mL/min) [277].

Figure 44a shows the electrochemical evaluation of the fuel cell operating with a stationary electrolyte. As a result of the CO₂ exposure the fuel cell suffers a power density decrease after the first exposure stage, and this power density reduction drops even further after the second CO₂ exposure. Nevertheless, the maximum power density value of 103 mW·cm⁻² was regained after the KOH rinse, Figure 44a. The reduction of current density registered was attributed to precipitants which can be eliminated by the effective KOH rinse. On the other hand, in Figure 44b we can observe the polarization and power density curves for the flowing electrolyte conditions. From these curves we can observe how CO₂ does not affect the fuel cell performance because any carbonate species formed inside the fuel cell is immediately flushed. This study confirms that the fuel cell electrodes in flowing electrolyte conditions are not affected by the CO₂ exposure [277].

Coming back to the transition metal chalcogenides and the membraneless fuel cells, two novel catalysts; carbon-supported Pt_xS_y and CoSe₂ were tested in a novel μLFFC [278]. A schematic image of the fabrication and assembly of the micro fuel cell components is shown in the Figure 45, detailed information about the components and operation can be consulted in reference [278].

In this study, the authors reported how such chalcogenide materials can be employed for developing cheaper, simpler, and smaller micro devices [278]. In Figure 46, we can observe how at a fuel concentration of 5 M CH₃OH the power density curves reached maximum values of 6.5 mW·cm⁻² for Pt, 4 mW·cm⁻² for Pt_xS_y and 0.23 mW·cm⁻² for CoSe₂ when these material were used as cathodes in the micro fuel cell assembly. Moreover, when the micro fuel cell was operated in mixed-reactant mode, the maximum value for the Pt power density decreased by 80%; and 35% for Pt_xS_y; nevertheless, CoSe₂ did not present any change in performance [278].

These results suggested further that a μLFFC with selective electrodes can operate in mixed-reactant mode (one electrolyte + fuel stream). This configuration makes easier the fuel delivery and the pumping system reducing the manufacturing costs. Additionally, the crossover issue is no longer a major problem on this kind of micro-devices [278,279].

The use of methanol tolerant electrocatalysts such as Pt_xTi_y and Pt_xSe_y in an air-breathing methanol microfluidic fuel cell was further reported [280]. This study presented the improvement of a mixed-reactant micro fuel cell (MRFC) performance in comparison to a laminar flow fuel configuration (LFFC). We can observe that for the LFFC configuration, the anolyte and the catholyte contain 0.5 M H₂SO₄, Figure 47a. The 5M CH₃OH fuel is contained only in the anodic side. In the case of MRFC configuration, Figure 47b, the fuel (5M CH₃OH) and the supporting electrolyte (0.5 M H₂SO₄) are contained in only one reservoir [280].

The individual anodic and cathodic polarizations for the Pt/C, Pt_xTi_y/C and Pt_xSe_y/C electrocatalysts are presented in Figure 48a–c. In the mixed-reactant condition a mixed potential between the oxygen reduction and the methanol oxidation is formed at the surface of the non-tolerant cathode. Because an inhibition effect caused by the Ti atoms on the Pt surface, the Pt_xTi_y/C overcame the potential value obtained for Pt/C [63]. Interestingly, the Pt_xSe_y/C presented a similar but enhanced effect, this electrocatalyst presented a selectivity to reduces de oxygen instead of oxidizing the methanol contained in the electrolytic solution. Thus, it allows the minimization of the methanol poisoning by the use of Se coordinated atoms onto the Pt surface [280].

Furthermore, recent studies showed that is possible to improve the electrocatalytic activity of non-precious CoSe₂ electrocatalysts evaluated in alkaline medium, with different carbon supports [229]. From this study, it was determined that a CoSe₂/NCNH electrocatalyst showed a lower onset overpotential than CoSe₂ supported on carbon, and carbon nanohorns (CoSe₂/C and CoSe₂/CNH, respectively). To validate the half-cell results, a self-breathing μLFFC was constructed and the CoSe₂/NCNH was used as a methanol tolerant cathodic electrocatalyst. The highest maximum power density value of 10.05 mW·cm⁻² was achieved with CoSe₂/NCNH (5 M methanol as the fuel), Figure 49. This enhancement presented by CoSe₂/NCNH suggests a modification of the CoSe₂ active center by an interaction with the NCNH support [229].

Nanostructured palladium electrocatalysts supported on Vulcan carbon XC-72 were synthesized via the carbonyl chemical route and studied in micro-laminar flow fuel cells [56]. When these materials were evaluated to perform the oxidation of organic molecules, they presented a relationship between the electrocatalysts morphology and their electrochemical activity. A 20 wt % metal loading Pd/C electrocatalyst was selected to be used as anode in a direct formic acid μLFFC. As we can observe in Figure 50a, the power density in the direct formic acid μLFFC reached 14.7 mW·cm⁻² with Pd/C-2, while its commercial counterpart Pd/C presented a power density value of 6.01 mW·cm⁻². The individual polarization curves for the anode and cathode shown in Figure 50b, compares the performances of the anodes. While the cathodic polarization curves do not present variation. It turns out that the enhanced μLFFC performance is the result of a better oxidation reaction on the Pd/C-2 anode.

Pd/C-2 was also selected to perform the methanol oxidation reaction in a direct methanol μLFFC. Similarly, to the previous formic acid oxidation test, a commercial Pd/C electrode was used as reference material. From Figure 51a we can observe how the micro fuel cell configuration with Pd/C-2 as anode, presented a maximum power density value of 1.3 mW·cm⁻² whereas the commercial counterpart reached only 0.45 mW·cm⁻². In Figure 51b we can observe that the power density difference is the result of the anodic polarizations [56].

5. Concluding Remarks

An attempt regarding the development of electrocatalytic materials has been outlined in this review. The activity and stability towards the ORR on Pt and Pd nanoparticles can be attained through ligand and support effects. The former focuses on synthesis of Pt or Pd nanoalloyed with a transition metals (ex. Co, Ni, etc.). The latter is tailoring the SMSI phenomenon, to attain the electronic, surface and structural modification of the catalytic center. The main trend in this direction is the application of highly graphitic carbon, carbon-oxide composites, and conducting/semiconducting oxide as supports. In this sense, significant progress in catalytic process on the catalytic center has been so far achieved. As pointed out herein, low temperature synthetic routes, e.g., carbonyl chemical or photo-deposition, was successfully employed.

The photo-deposition is particularly interesting for Pt-based NPs anchored onto high graphitic carbon, semiconducting oxide, and oxide-carbon composites. Strong metal support interaction (SMSI) between platinum and support could be reinforced, increasing the stability of catalytic center for ORR process. In addition, by means of Ti, Se or/and Cr as ligands, the tolerance of Pt was enhanced, favoring their application in a fuel cell using small organics, such as methanol as a fuel. Non-platinum metal chalcogenides, such as Ru_xSe_y and CoSe₂, are methanol tolerant electrocatalysts for ORR. Compared with Pt or Pd-based catalysts, the non-precious metal ones showed less catalytic activity, but an enhanced tolerance to methanol leading to the use of mixed-reactant systems, since cross-over effect in direct methanol fuel cell can be reduced, as described through the micro-fuel cells platform. Pt- or Pd-based electrocatalysts are still the most active materials for the hydrogen oxidation reaction (HOR). Nevertheless, non-precious metal nanoalloy, such as CoNiMo, was identified as active for HOR. Research on such kind of electrocatalyst has still a long way to go.

Concerning the oxidation process of organic fuels, e.g., methanol, and formic acid, Pd-based catalysts showed increased activity with respect to Pt ones. Chalcogenide materials containing Pt and Pd were measured for ORR and various fuel oxidation in a micro-fluid fuel cell. The non-precious metal catalysts become more and more attractive. The mechanism for the catalytic activity of such centers is still to be developed. It seems that the morphological impact plays an important role on the electrocatalytic activity. Thus, efforts should be further devoted to this issue.