Challenges

Currently there are some obstacles which need to be overcome before large scale commercialization of DMFC: (a) the high cost of Nafion membrane in the range of US$ 800-2000/ m²; (b) the reduction of oxygen on cathode is also low though the problems are not so serious as with aqueous mineral acid electrolytes; (c) the permeability of the current

perfluorosulfonic acid membranes (Nafion) to methanol, which allow considerable crossover of methanol from anode region to cathode region. This leads both to degradation of performance, since mixed potential develops at the cathode, and to deterioration of fuel utilization. Methanol vapor also appears in the cathode exhaust, from which it would have to be removed. (d) and perhaps of greatest concern at the moment is the low activity and high cost of anode electrocatalyst, the anode reaction has poor electrode kinetics, particularly at lower temperatures, making it highly desirable to identify improved catalysts and to work at as high a temperature as possible. With regard to new DMFC anode catalysts, there are two major challenges, namely, the performance, including activity, reliability and durability, and cost reduction.

2.6 Anode catalysts of DMFC

For DMFC anode catalyst performance improvement, the exploration of new catalyst materials including noble and non-noble metals is necessary. In this respect, an alloying strategy is one of the R&D directions. With the help of fast activity screening, a breakthrough could be accelerated to meet the requirements for DMFC commercialization. The other is a support strategy. Rapid development of nanotechnology, especially in the area of the synthesis of carbon nanostructured materials, will create more stable and active supported catalysts.

Nanoparticle supported catalysts are believed to be the most promising materials for catalysis in DMFCs. In recent years, significant progresses have been made to improve the performance of DMFCs. And the research activities are mainly focused on the development of the materials including the electrocatalysts and the electrolyte membranes, which are important in DMFC.

Regarding cost reduction, for early DMFC commercialization, DMFC anode catalyst loadings must drop to a level of <1.0 mg cm⁻² from the present 2.0–8.0 mg cm⁻², depending

on applications. Loading reduction through increasing Pt utilization is one of the R&D directions. Alloying and nanoparticle supporting strategies could dramatically reduce the Pt content in the catalysts without performance compromise. Non-noble catalyst development is the other approach for catalyst cost reduction. However, at this current stage, non-noble DMFC anode catalysts are not yet feasible. A more thorough exploration is needed in this area.

Real breakthroughs in DMFC anode catalysis are necessary with respect to performance and cost. When such a situation arises, it is important to open up new avenues for making low cost and effective catalysts. Until now many avenues that are already opening up are that of i) novel nanostructures of Pt, ii) new cost-effective synthesis routes, iii) binary or multiple catalysts, and iv) new catalyst supports to replace the generally used activated carbons, besides ideally the Pt-based catalysts should be replaced with abundant, non-precious materials. In the following section, we will present a brief literature review on the development of DMFC anode catalysts.

2.6.1 Pt nanostructures

To lower the cost of Pt catalysts, great efforts have focused on the development of nanostructures of Pt catalysts with a high surface area to achieve high catalytic performance and utilization efficiency. A variety of Pt nanostructures such as nanoparticles [39], nanowires [40, 41], nanocubes [42], 3D nanoflowers [43], multipods [44] dendritic [45] and nanotubes [46] have been proposed.

Recently, it is shown that enhanced electrocatalysts of Pt can be obtained as hollow nanospheres without changing catalyst loading [3]. The individual Pt nanosphere (diameter ca.

24 nm) is composed of a porous shell consisting of 2 nm-Pt nanoparticles (Fig. 2-3). These features endow the Pt hollow nanospheres with a high surface area which contributes to the high catalytic activity towards methanol electrooxidation. The result suggests a simple route

to enhance the catalytic efficiency of Pt catalysts by a simple improvement of the morphology.

In the search for novel nanostructures of Pt, high-index facets such as {730}, {210}, and {520}

surfaces of tetrahexahedral Pt nanocrystals have been pointed out having high catalytic activity for electrooxidation due to the large density of atomic steps and dangling bonds. [47]

The large scale synthesis method of them still remains a vital task.

Figure 2-3 TEM images of Pt hollow nanospheres [3].

2.6.2 New cost-effective synthesis routes

Another way to lower the cost of Pt catalyst is to develop cost-effective routes for making more efficient Pt catalysts. Recently nanoporous catalysts have attracted great interest in the field of catalysis. Most syntheses of nanoporous materials reported so far have focused on template-assisted bottom-up processes, including soft templating and hard templating methods, which are relatively complicated. Recently, the conversion Li storage mechanism occurring in transition metal compounds has been developed into a template-free up-down method of wide applicability for the synthesis of well-crystallized materials with favorable nanoporous structures [48]. Based on this strategy, nanoporous Pt can be obtained from

submicrometer PtO2 by electrochemical lithiation followed by dissolving the Li2O in acidic aqueous solution or even water (Fig. 2-4). The synthesis is relatively simple (starting from micrometre-sized transition metal oxides), yet very effective. Owing to the high surface area (142m² g^-1), the presence of various pore sizes (2–20 nm) and the pronounced stability of the nanoporous Pt, the so-prepared Pt shows outstanding properties when used as an electrocatalyst for methanol oxidation [48].

Figure 2-4 The simple template-free electrochemical lithiation synthesis of nanoporous structures [48].

2.6.3 Binary and Multiple Catalysts

Considerable attention has been paid to based binary catalysts (generally Pt alloys, Pt-M), ternary catalysts (generally Pt-M1-M2), and multiple catalysts (generally Pt-M1-M2- M3O) because the systems can not only reduce the cost but also improve the catalytic performance, such as mitigating CO poisoning, lowing overpotential, and suppressing Pt dissolution.

According to Wasmus et al. [49], it is virtually undisputed that Pt-Ru is better than Pt and there is a consensus about the fact that, for the methanol oxidation, Pt-Ru is the best material among the Pt-based bimetallic electrocatalysts [49, 50–54]. These two metals, i.e. Pt and Ru, have close electronegativities (Fig. 2-5) [55] and similar bulk Wigner-Seitz radii (Fig. 2-6) [56]. Pt when alloyed with Ru strongly segregates while Ru strongly anti-segregates (Table 2-1), thereby a Ru site should be surrounded with some Pt sites, what is in agreement with the three to five Pt atoms necessary to activate the adsorption of methanol and the single Ru atom

necessary to activate water [57].

Figure 2-5 shows the electronegativities (eV) of the transition metals [55].

Figure 2-6 shows the Bulk Wigner-Seitz radius [a.u.] of the transition metals [56].

Table 2-1 Compilation of calculated segregation energies on the closest packed surface of all binary combinations of the transition metals [55].

Table 2-2 Shifts in d-band centres of surface impurities (A) and overlayers (B) relative to the clean metal values (bold) [55].

Moreover, the Pt electronic structure should be changed by the presence of neighbouring Ru.

According to Table 2-2, the Pt d-band centre shifts down when Pt is alloyed with Ru, what suggests weaker Pt-adsorbate bonds, while the Ru d-band centre slightly shifts up. In other words, when Pt and Ru are alloyed, the adsorption of adsorbates is weaker on the Pt sites and stronger on the Ru sites. These tendencies may explain the enhanced activity of Pt-Ru, which is attributed to both bifunctional mechanism and electronic effect, where the bifunctional mechanism involves the adsorption of OH species on Ru atoms thereby promoting the oxidation of CO to CO2 [58, 59]. In this way, the Pt poisoning by the CO-like species would be decreased because these species would more weakly adsorb on Pt and Ru would provide the necessary OH species, which would permanently be available on the Ru surface sites, more strongly adsorbing the OH species. Elements as alternative to Ru were investigated.

Antolini et al. [52] worked on Pt3Co1 and Pt3Ni1 (atomic ratio 3:1). The authors chose Co and Ni since their presence lowered the electronic binding energy in Pt and so promoted the C–H cleavage reaction at low potentials and, moreover, they provided OH species necessary for the CO oxidation. The performances of fuel cells with Pt3Co1 or Pt3Ni1 as anode catalysts were slightly worse than that of the fuel cell with Pt. The performance of Pt3Co1 was slightly better than that of Pt3Ni1. These two metals, i.e. Co and Ni, have similar electronegativities (Fig. 2-5) and bulk Wigner-Seitz radii (Fig. 2-6).

According to Table 2-1, Pt when alloyed with Co strongly segregates and Pt when alloyed with Ni moderately segregates, as already reported [60]. For both Pt-Co and Pt-Ni, the Pt d-band centre shifts down. The d-d-band centres of Co and Ni shift up when they are alloyed with Pt. This analysis of the surface modifications suggests the following remarks: compared with Pt-Ru, (i) the adsorption of OH species on the Co or Ni sites should be weaker; (ii) the lower segregation of Pt in either Pt-Co or Pt-Ni is an indication of a higher dilution of Pt and so a decrease in the number of Pt surface sites. Therefore, Pt-Ru should be a better catalyst than both Pt-Co and Pt-Ni, and as the Pt segregation is more severe with Co, Pt-Co should be a better catalyst than Pt-Ni with a larger number of Pt surface sites. However, as underlined by

Antolini et al. [50], conflicting results regarding the Pt-Co and Pt-Ni alloys were reported in the literature. Interestingly, the authors showed that the methanol oxidation activity on Pt-Ni and Pt-Co was improved or unchanged or decreased in relation to pure Pt. It was observed an opposite effect of the Co/Ni presence in going from low-contents (negative effect on the methanol oxidation) to high contents (positive effect) and the decreased activity in the presence of low Co/Ni contents was ascribed to the dilution of Pt, hindering the methanol adsorption, while the positive effect was related to several reasons, namely the electronic effect, an enhancing of the CO oxidation and the presence of oxide species [50]. This study unfortunately shows the limitations of the theoretical understanding of the catalysts behaviors but it fortunately stresses on the essential side of the experiments. Choi et al. [61] observed that the current density produced by methanol oxidation over Pt2-Rh1 (atomic ratio 2:1) was larger than that over pure Pt but lower than that over Pt-Ru. The authors concluded that the enhanced activity of Pt2-Rh1 was mainly due to an intrinsic improvement in catalytic activity and not to an improvement in CO oxidation. Ru and Rh display similar electronegativities and bulk Wigner-Seitz radii (Figs. 2-1 and 2-2). Pt moderately segregates when alloyed with Rh while it very strongly segregates when alloyed with Ru. The Pt d-band centre shifts down when alloyed with both though the down shift is more important with Ru. These reflections suggest that Pt-Ru should be better with higher surface concentration of Pt sites and lower strength of the adsorption of CO-like species over the Pt sites.

More recently, Choi et al. [62] reported that the Pt-Au alloy and pure Pt showed almost the same activity. Au itself was inactive for methanol oxidation and was not helpful for removing COads on the Pt surface. In fact, when alloyed with Au, Pt strongly anti-segregates (Table 2-1). The surface concentration in Pt sites is then reduced. Moreover, Au has one of the lowest d-band centres and the presence of Pt shifts down it, what suggests very poor abilities for the adsorption of OH species. The Pt d-band centre shifts up when alloyed with Au and thereby the adsorption of CO-like species becomes stronger, what favours the Pt sites

poisoning.

Table 2-3 Possible occurrence of segregation and possible shift in d-band centre for the elements of the trimetallic Pt-Ru-M alloy [5] with metals (M) as Mo, W, Co, Fe, Ni, Cu, Sn and Au [55].

Consequently, Au is not interesting for the preparation of active Pt-based bimetallic alloys devoted to methanol oxidation. Pt-Ru-based trimetallic electrocatalysts were envisaged as well [57]. The addition of a third element, i.e. Au, Co, Cu, Fe, Mo, Ni, Sn and W, gives promising results. The following classification enabled to note that Mo,W, Co, Fe and Ni improved the Pt- Ru activity towards the oxidation of methanol, the best promoter being Mo:

Pt-Ru-Mo > Pt-Ru-W > Pt-Ru-Co > Pt-Ru-Fe > Pt-Ru-Ni > Pt-Ru-Cu > Pt-Ru > Pt-Ru-Sn >

Pt-Ru-Au. Even if the concepts of Nørskov and co-workers [63–65] do not consider the alloys with three metals, it could be tried out proposing few trends regarding the segregation and the d-band centres variations. Table 2-3 proposes the possible occurrence of the segregation and the shifts in d-band centres for the different metals, i.e. Pt, Ru, Mo, W, Co, Fe, Ni, Cu and Au.

The analysis of the data given by Table 2-1 suggests that the best trimetallic catalysts are the ones for which the third metal M antisegregates while both Pt and Ru segregate. Furthermore, it seems that the d-band centres of Pt and Ru should shift down. The trimetallic material for which the d-band centres of Pt and Ru shift up is the worst alloy. Unfortunately, these types of data are not available for Mo, W and Sn what would have been useful to completely validate such observations. Consequently, the improvement of the Pt-Ru alloys would require a third metal with which Pt and Ru would segregate and their d-band centres would shift down. All of the previous observations could be used as criteria to select, from Tables 2-1 and 2-2,

bimetallic catalysts that could be active for the methanol oxidation. The first criterion could be the d-band centre shift. It should be close to that of Pt and Ru when alloyed together, i.e.

about −2.9 and about −1.3 eV, respectively. Thereby, the analysis of Table 2-2 provides Pd-Ni.

The second criterion could be the segregation. Hence, Table 2-1 confirms Pd-Ni as Pd has a tendency to strongly segregate (like Pt) while Ni moderately antisegregates (like Ru).

Therefore, the Pd-Ni alloy might provide an activity similar to that of Pt-Ru. Nevertheless, for the methanol oxidation, as it is remarked by a large number of investigations, the best current way to improve the anode electrocatalyst would be the addition of a third metal to Pt-Ru [57].

Many investigations about DMFC have more or less regarded all the mono- and bimetallic electrocatalysts displaying catalytic abilities towards methanol oxidation. It seems then that the remaining tracks to follow are the ternary or quaternary compositions and especially the Pt-Ru-based ones. A literature analysis shows that some of the best matrices for such catalyst are the ternary or quaternary compositions: Pt–Ru–Os, Pt–Ru–Ir, and Pt–Ru–Os–Ir [66, 67]. The addition of Ir (cheaper, but less active than Pt) into the conventional Pt–Ru catalyst can suppress Ru dissolution, which is a significant challenge for DMFC anode catalysts utilizing Ru. Dissolution of Ru from the Pt–Ru anode catalyst, followed by the Ru ions crossing over the membrane [68] and depositing on the cathode, can result in degradation of both the anode and cathode catalyst and a decrease in the fuel cell performance. The addition of Ir, however, does not significantly increase the activity of the Pt–Ru catalyst.

As mentioned above, the many favorable effects have been found in Pt-based binary and multiple catalysts, such as the bifunctional effect, the electronic effect, and the hydrogen spillover effect. With progressive understanding of these effects, low-cost Pt catalysts with tunable performance will be developed in the future.

2.6.4 New Catalyst Supports

To date, the most promising catalytic materials used for methanol oxidation at room temperature are supported Pt based catalysts. The dispersion and utilization of the catalysts depend on surface area, pore characteristics, and surface functionalities of the supports.

Although the widely used catalyst support is activated carbons like Vulcan XC 72R, the search for new efficient supports is still underway. Carbon nanotubes (CNTs) have exceptional mechanical and electronic properties and are attractive support materials for Pt-based catalysts in DMFCs. There mainly are two approaches to achieve high Pt dispersion.

One approach is by modifying either CNTs or Pt catalysts. The former has been extensively employed by chemical-oxidation treatments to modify the CNT walls for introducing more defect sites for Pt loading, while the latter has rarely been used. Recently, we have reported a novel process to prepare well-dispersed Pt nanoparticles on CNTs by modifying Pt nanoparticles with organic molecule triphenylphosphine (PPh3) [55]. In contrast to CNTs, the functionalization of Pt nanoparticles was facile and effective under much more benign conditions, such as no harsh acid and room temperature, more efforts are expected.

Another approach is by in situ growth methods. Zheng et al. [69] reported a one-step in situ method to disperse Pt nanoparticles on CNTs, using H2PtCl6 as a Pt source, ethylene glycol as a reducing agent, and dimethyl formamide (DMF) as a solvent. By this method, well-dispersed Pt nanoparticles can be directly loaded onto the CNT walls (Fig. 3d). So far, many in situ methods have been used, but preparing well-dispersed Pt on CNTs still remains a challenge. The thesis is mainly focused on research of electrocatalysts, which are the important materials in DMFC that is a power generator to convert a chemical energy to electrical energy.

2.7 Pulse electrodeposition

2.7.1 Pulse waveform

Generally, the pulse waveform was divided into two groups: (1) unipolar, where all the pulses are in one direction (with no polarity) and (2) bipolar, where anodic and cathodic pulses are mixed. There are many variants on these [70], but the number of variables increases with complexity of the waveform, which makes it more difficult to understand how a particular waveform affects the deposition. Typical waveforms include: (1) cathodic pulse followed by a period without current (or an anodic pulse), (2) direct current (DC) with superimposed modulations, (3) duplex pulse, (4) pulse-on-pulse, (5) cathodic pulses followed by anodic pulses-pulse reverse current (PRC), (6) superimposing periodic reverse on high frequency pulse, (7) modified sine-wave pulses and (8) square-wave pulses (potential/current).

Among these, the square wave pulses (potential/current) have the advantage of an extensive duty cycle range [71]. In our experiments, we use a potentiostatic bipolar pulse plating (PBPP) to prepare Pt nanoparticles, 2D and 3D nanostructures. The more detailed explanation of PBPP is given in following section:

2.7.2 Potentiostatic bipolar pulse electrodeposition

In the potentiostatic pulse plating the potential of the test electrode is controlled, while the current, the dependent variable, is measure as a function of time. The potential difference between the test electrode and the reference electrode is controlled by a potentiostat (shown in Fig. 2-7). The input function, a constant potential, and the response potential i = f(t), are shown in Fig. 2-8.

In PBPP [72–74] the potential is alternated swiftly between two different values. This results in a series of pulses of equal amplitude, duration and polarity, separated by zero current. Each pulse consists of an ON-time (ton) during which potential and/current is applied, and an OFF-time (toff) during which zero current is applied as shown in Fig. 2-9. It is possible to control the deposited film composition and thickness in an atomic order by regulating the

pulse amplitude and width [75, 76].

Figure 2-7 shows the schematic diagram of apparatus for potentiostatic measurements; E, controlled potential; e1, test electrode; e2, reference electrode; e3, counter electrode.

Figure 2-8 shows the variation of current with time during potentiostatic electrolysis.

Figure 2-8 shows the variation of current with time during potentiostatic electrolysis.