1. G ENERAL I NTRODUCTION
1.3. O XYGEN R EDUCTION R EACTION
The Achilles heel of atomically dispersed catalysts is their catalytic stability. Industrially important catalytic processes are often conducted at significantly high temperatures under oxidative or reductive reaction conditions. Under such harsh conditions, the agglomeration of metal particles or coking on the catalyst surface are unavoidable, which undermines the superb catalytic properties of atomically dispersed catalysts. In this context, it is of paramount importance to develop atomically dispersed catalysts that are stable under the harsh reaction conditions conventionally used in industrial processes. To obtain highly stable atomically dispersed catalysts, Yan and co-workers fabricated Pt SACs firmly anchored onto the internal surface of mesoporous Al2O3.90 The Pt/m-Al2O3 catalysts withstood 60 cycles of an CO oxidation stability test performed between 100 and 400 °C for over one month, suggesting a method to address the stability issues of atomically dispersed catalysts. The Hosono group stabilized single Pt atoms within the sub-nanometer cavities of 12CaO7·Al2O3 (C12A7).91 The interactions between the metal anions and positively charged surface cavities of C12A7 permit the stable operation of the selective hydrogenation of nitroarenes under reducing conditions up to 600 °C.
Zhang and co-workers reported the use of coke-resistant atomically dispersed Ni catalysts in the dry reforming reaction of methane (DRM).92 Due to selective activation of the first C–H bond in methane with negligible carbon deposition, the DRM could be operated for 100 h.
is of critical importance.
The ORR can take place two different pathways: the 4 e− pathway ORR, where oxygen is converted to H2O by complete reduction and the 2 e− pathway ORR involving partial reduction of O2
to generate H2O2.94 The 2 e− pathway ORR has been regarded as an adverse side reaction that impedes the efficient 4 e− pathway ORR in PEMFCs, as it degrades the performance of PEMFC by destroying the Nafion membrane.95,96 However, the selective 2 e− pathway ORR has recently garnered a surge of interest as a means of the electrochemical production of H2O2.97,98 The annual global production of H2O2
is estimated to reach a value of ~ 6 billion US dollars by 2023.99 Ninety-five percent of the current H2O2
production uses the anthraquinone process,97 which undesirably requires high-pressure H2 and expensive Pd-based catalysts, large infrastructures, and energy-intensive distillation steps. This process typically produces H2O2 in high concentrations in a large volume, with attendant safety risks related to the storage and transportation of H2O2. The electrochemical H2O2 production via 2 e− pathway oxygen reduction has emerged as a promising alternative to the anthraquinone process.97–99 Electrochemical H2O2 production allows continuous, on-site H2O2 production with dilute H2O2, mitigating the drawbacks of the anthraquinone process. Thus, the development of the selective catalysts for 2 e− pathway oxygen reduction has also been a significant issue.
Figure 1.12. Schematic illustrations of (a) associative and (b) dissociative mechanism of the ORR on metal catalysts.
Two reaction mechanisms have been considered for the ORR (Figure 1.12): (i) associative mechanism where oxygen is adsorbed as an end-on configuration onto metal surfaces which can produce both H2O via reaction (1) ~ (3) and H2O2 through reaction (1) and (4);
*O2 + (H+ + e−) → *OOH (1)
*OOH + (H+ + e−) → *O + H2O (2)
*O + (2H+ + 2e−) → H2O (3)
*OOH + (H+ + e−) → H2O2 (4)
(ii) dissociative mechanism in which side-on adsorption of oxygen occurs, yielding only H2O as a product via reaction (5) and (6).100,101
(a)
4 e− ORR
2 e− ORR
4 e− ORR (b)
: Hydrogen : Oxygen : Metal
*O2 → 2*O (5)
2*O + (4H+ + 4e−) → 2H2O (6)
These reaction mechanisms are corresponding to the ORR in acidic media. In alkaline conditions, the associative and dissociative reactions identically occur, but protons are provided by the dissociation of H2O, and products are converted from H2O and H2O2 to OH− and HO2−. 4 e− pathway oxygen reduction can take place through both associative and dissociative mechanisms, whereas 2 e− pathway oxygen reduction is able to occur only by the associative mechanism. Hence, the catalyst should be reasonably designed for the purpose of target applications. In this dissertation, design approaches of heterogeneous metal catalysts for the 2 e− pathway and 4 e− pathway ORR are respectively introduced.
1.3.2 Electrocatalysts for 2 e− Pathway Oxygen Reduction Reaction
Figure 1.13. Schematic illustration of ORR pathways depending on the geometric structure of catalysts and three major approaches for isolating active metal sites. Reprinted with permission from ref. 103.
Copyright © 2020 Elsevier B.V.
It is pivotal to rationally design the electrode catalysts that can promote the selective 2 e− pathway ORR while suppressing the competing 4 e− pathway ORR for electrochemical H2O2 production.102 To promote the 2 e− pathway ORR, ensemble or hollow sites that facilitate the side-on adsorption of O2
should be eliminated by isolating the surface metal atoms. There have been three major catalyst preparation strategies for isolating the active metal atoms to promote electrochemical H2O2 production (Figure 1.13): (i) alloying the active metal with an inert metal to provide an isolated geometry of active metal atoms, (ii) surface poisoning to block the exposed ensemble sites with inert species, and (iii) preparing atomically dispersed catalysts.103
The most straightforward technique for designing electrocatalysts with isolated metal sites is alloying an active metal with an inert metal. The inert metal species can induce geometric isolation of
: Inert Site : Active Metal Site
: Hydrogen : Oxygen
Presence of Ensemble Sites
Absence of Ensemble Sites
H2O2 H2O
Alloying Poisoning Atomic Dispersion
the active metal, which can alter the adsorption geometry of the O2 molecule. This approach was initially exploited in heterogeneous thermocatalytic reactions. One notable example is PdAu alloy catalysts for the gas-phase direct synthesis of H2O2 from H2 and O2, where high H2O2 selectivity was achieved by isolating the active Pd atoms with inactive Au atoms.104–107 Inspired by this, Jirkovský and co-workers introduced the alloying concept in electrocatalytic H2O2 production.108 They first screened alloy combinations to identify an optimum catalyst composition by DFT calculations, showing that an isolated Pd, Pt, or Rh site on the surface of Au enhanced H2O2 production compared with that achieved with pure Au. Pd content-controlled PdAu alloys were prepared to validate the DFT results. With 8%
Pd, no surface segregation of Pd occurred, leading to 95% H2O2 selectivity. Higher Pd loading led to a decline in the selectivity by forming Pd ensemble sites that preferentially dissociate the oxygen bond.
The Siahrostami group also screened new alloy catalysts for electrochemical H2O2 production using DFT calculations, demonstrating the optimal performance of the PtHg4 alloy comprising active Pt atoms surrounded by inert Hg atoms.109 Based on the DFT calculations, core-shell NPs comprising a Pt core and a Pt–Hg shell were prepared via electrodeposition of Hg on Pt. The resulting Hg-modified Pt NPs provided more than 90% selectivity at 0.3–0.5 V. Importantly, the mass activity of this catalyst was 26 A gnoble metal−1 at an overpotential of 50 mV, which was the best mass activity achieved for H2O2
production at that time. Later, the same group extended this approach to the Pd–Hg alloy, whose mass activity was five-fold that of the Pt–Hg alloy, with 95% H2O2 selectivity.110 Deiana and co-workers later scrutinized the structure of the Pd–Hg alloy catalyst with high-angle annular dark-field scanning transmission electron microscopy and energy-dispersive X-ray spectroscopy, revealing that the Pd–Hg particle comprised a Pd core and Pd–Hg ordered alloy shell with isolated Pd atoms, clarifying the origin of the high catalytic activity for H2O2 production.111
Another notable strategy for modifying metal catalysts is partial poisoning of the active metal surfaces with catalytically inert molecules. Markovic and co-workers demonstrated this concept by adsorbing poisonous ions on the metal surfaces, and subsequent desorption of the adsorbed ions by sweep voltammetry.112,113 On the basis of the concept, the Markovic group covered a Pt(111) electrode with a self-assembled monolayer of calix[4]arene molecules to lower the number of Pt ensemble sites.114 As the coverage of the calix[4]arene molecules increased, the available Pt ensemble sites decreased, reducing the ORR activity, with concomitant formation of H2O2. Similarly, Choi and co-workers coated Pt NPs on carbon (Pt/C) with amorphous carbon layers by acetylene chemical vapor deposition.115 The carbon coating layer in the resulting amorphous carbon-coated Pt/C catalysts promoted the end-on adsorption of O2 molecules while suppressing their side-on adsorption, thereby promoting the 2 e− ORR pathway. The H2O2 selectivity was proportional to the thickness of the carbon layer, with a maximal selectivity of 41%. The carbon coating layer further suppressed decomposition of the produced H2O2, which produces water via disproportionation or reduction reactions, by hindering access to the produced H2O2. Selenium, sulfur, or an ammonium ion has also been used as a modifier for Pt catalysts to promote
the 2 e− pathway ORR.116–119
The most efficient method for eliminating metal ensemble sites is the preparation of atomically dispersed catalysts. Lee and co-workers prepared TiN-supported Pt (Pt/TiN) catalysts by an incipient wetness impregnation method.89 At low Pt loading (0.35 wt%), atomically dispersed Pt sites were exclusively obtained. The H2O2 selectivity of the resulting 0.35 wt% Pt/TiN catalyst reached 65%, which was much higher than that of the 2 wt% Pt/TiN (30%) comprising NPs as well as atomically dispersed sites. The Lee group further investigated the effects of supports on the H2O2 production selectivity with the TiN and TiC supports, in which 0.2 wt% Pt/TiC exhibited higher ORR activity and H2O2 selectivity than 0.2 wt% Pt/TiN.86 DFT calculations suggested that the more favorable adsorption energy and energy profiles of Pt/TiC toward the 2 e− pathway ORR led to its superior activity and selectivity for H2O2 production. Atomically dispersed sites inevitably have high surface energy and readily agglomerate during preparation, making their syntheses with high metal loading challenging.
To circumvent this problem, Choi and co-workers used zeolite-templated carbon possessing a high concentration of sulfur (HSC) as a support, where the ultrahigh surface area of HSC (2770 m2 g−1) and numerous S sites with strong anchoring ability facilitated the formation of atomically dispersed Pt sites up to a loading of 5 wt%.42 This is markedly higher metal contents than that of atomically dispersed catalysts reported at that time (less than 1 wt%). The Pt/HSC catalyst showed 95% H2O2 selectivity in the potential range of 0.1–0.7 V, which is a dramatic improvement compared with the H2O2 selectivity of 28% achieved with Pt NP-based catalysts (Pt/zeolite-templated carbon). Li and co-workers prepared
∼15 wt% atomically dispersed Pt catalysts using sulfur-containing support, CuSx; the resulting h-Pt1- CuSx catalyst afforded ≥90% H2O2 selectivity in the potential range of 0.05–0.7 V.120 Importantly, the h-Pt1-CuSx catalyst could generate H2O2 with a yield of 546 mol kgcat−1 h−1, which is among the highest values reported for atomically dispersed electrocatalysts and thermocatalysts for H2O2 production.
Other atomically dispersed metal catalysts comprising Ni, Au, or Fe as the metal center were prepared and exploited to facilitate selective electrochemical H2O2 production.121–125
1.3.3 Electrocatalysts for 4 e− Pathway Oxygen Reduction Reaction
During the 4e− pathway ORR, the strong oxygen double bond is cleaved. For an efficient 4e− pathway ORR, the catalysts should possess an optimal oxygen binding energy that is neither too weak nor too strong for the facile adsorption of O2 and desorption of H2O, respectively. Among the various transition metal catalysts studied, Pt is placed on the top of the volcano plot of oxygen binding energy (Figure 1.14a).126 However, Pt is considerably expensive and scarce. In existing fuel cell technology, a Pt loading of ~22.5 g is required to operate vehicles powered by fuel cells.127 Due to such a high Pt loading, fuel cell-based vehicles cannot be produced at a reasonable cost. Consequently, tremendous efforts have been devoted toward enhancing Pt usage via nanostructuring or alloying. In addition, there have been several attempts to completely replace Pt catalysts with non-precious metal-based
catalysts.128
The major design approaches to Pt-based catalysts are as follows: The preparation of i) core–
shell structures constructed using a cheap transition metal-based core and Pt shell, ii) Pt alloys with transition metals, and iii) Pt-based shape-controlled NPs. The first core–shell approach can maximize the utilization efficiency of Pt, while significantly reducing the amount of expensive Pt used. Adzic and co-workers prepared core–shell AuNi@Pt, PdCo@Pt, and PtCo@Pt catalysts using a galvanic replacement method. The resulting core–shell catalysts surpassed the ORR activity observed for a commercial Pt/C catalyst by utilizing a significantly small amount of Pt.129 Similarly, the Chen group also fabricated a Ni@Pt core–shell catalyst, which exhibited superior ORR activity when compared to Pt/C.130 The Xia group carried out atomic layer-by-layer deposition of Pt on Pd nanocubes using one to six-layers of Pt. The Pd nanocubes coated with monolayer Pt exhibited the highest mass activity, which was almost a three-fold enhancement when compared to a commercial Pt/C catalyst.131
Figure 1.14. (a) Trends in oxygen reduction activity plotted as a function of the oxygen binding energy.
(b) A volumetric current density of the best non-precious metal-based catalysts for the ORR at that time.
Reprinted with permission from refs. 126 and 156. Copyright © 2004 American Chemical Society.
Copyright © 2009 American Association for the Advancement of Science.
Alloying Pt with other transition metals is perhaps the most popular approach to boost ORR activity. Based on DFT calculations and experimental results, Nørskov and co-workers established that alloying Pt with transition metals can modulate the oxygen binding energy by modifying the electronic structure of Pt, thereby affecting the ORR activity.132,133 Numerous subsequent studies have demonstrated the high ORR activity of Pt alloy catalysts, including PtFe,134 PtCo,135,136 PtNi,137 PtPb,138 PtPd,139 PtZn,140 and PtLn.141 Control over specific crystallographic facets in Pt-based alloys can further improve the ORR activity. Most notably, Marković and co-workers revealed that the Pt3Ni (111) surface exhibits 10-fold higher ORR activity than the Pt (111) surface and 90-fold higher activity than a commercial Pt/C catalyst.142 Inspired by this, several research groups have prepared Pt3Ni octahedral NPs that expose the Pt3Ni (111) surface, which can substantially improve the ORR activity.143–146 Huang and co-workers demonstrated that doping Mo onto Pt3Ni octahedra further improved the ORR
(a) (b)
activity.147 In addition, Pt alloy-based nanoframes148,149 and nanowires150,151 with high surface areas and reinforced strain effects yielded unprecedented high ORR activities. However, the significantly high ORR activity obtained in half-cell measurements have not yet been fully translated under full-cell operation conditions.152
Several classes of non-precious metal-based catalysts have been investigated with the aim of replacing expensive Pt-based catalysts. In particular, M–N/C catalysts comprised a transition metal (Fe and Co), nitrogen, and carbon have been of considerable interest due to their high cost-effectiveness and outstanding ORR activity. The possibility of M–N/C catalysts was first realized by Jasinski, who discovered that cobalt phthalocyanine showed electrocatalytic activity in the ORR.153 Subsequently, Jahnke and co-workers recognized the importance of heat treatment during the synthesis of metal-N4
chelates toward enhancing the ORR catalytic activity and stability.154 An early stage of this research was focused on the preparation of M–N/C catalysts via heat treatment of metal-nitrogen (M–Nx)- coordinated macrocyclic compounds in an effort to prepare M–N/C catalysts. In 1989, the Yeager group successfully demonstrated that active M–N/C catalysts can also be prepared through the heat treatment of a combination of metal, nitrogen, and carbon precursors instead of macrocyclic compounds, which established up a new strategy for designing M–N/C catalysts.155 Despite steady progress in the field of M–N/C catalysts, their ORR activity still remains considerably lower than those of Pt-based catalysts.
In 2009, Dodelet and co-workers reported significant enhancements in the activity of a Fe–N/C catalyst via the optimization of its synthesis (Figure 1.14b), indicating the possibility of replacing platinum catalysts.156 In 2011, the Zelenay group reported the stable long-term operation of a Fe–N/C catalyst in a full-cell test using polyaniline as a precursor, thereby accelerating research on M–N/C catalysts.157 For the rational design of M–N/C catalysts, the primary research direction focuses on the elucidation of their active sites. As a result of discussions over the past few decades, there appears to be a consensus that atomically dispersed M–Nx sites are major active species responsible for high ORR performance.54,72,158,159 The on-going challenge for M–N/C catalysts is their poor durability. During the first 100 h of testing, the fuel cell performance observed using a M–N/C catalyst as the cathode material typically degrades by 40–80%.160–162 Using online inductively coupled plasma mass spectrometry and differential electrochemical mass spectroscopy coupled to a modified scanning flow cell system, Mayrhofer and co-workers revealed that carbon oxidation has mainly adverse effects on the ORR activity of M–N/C catalysts.163,164 However, there have not yet been any clear solutions to prevent carbon oxidation during the operation of fuel cells. Therefore, further investigations are required for the reliable operation of fuel cells using non-precious metal catalysts.
1.3.4 Atomically Dispersed Catalysts for Oxygen Reduction Reaction
Atomically dispersed catalysts have been utilized for both the 2e− and 4e− pathway ORRs;
however, there have been some controversies in terms of their ORR catalytic properties. As described
above, the Lee, Choi, and Li groups reported that atomically dispersed Pt catalysts exhibit a selective 2e− pathway ORR with low activity.42,86,89 On the contrary, the Sun165 and Wang166 groups reported an efficient 4e− pathway ORR using atomically dispersed Pt catalysts with high ORR activity. These contradictory results continue to the present day. Han and co-workers demonstrated that atomically dispersed catalysts supported on TiC exhibited the 2e− pathway ORR,124 whereas the Ma group reported that single-atom Pt supported on MoC and MoN exhibited a 4e− pathway ORR.167
Regarding the 4e− pathway ORR exhibited by non-precious metal-based atomically dispersed M–N/C catalysts, geometrically isolated metal sites do not always guarantee the preservation of the oxygen bond during the reaction. From a thermodynamic perspective,94,168 ORR selectivity is determined by the oxygen binding energy of catalysts in the associative mechanism due to the scaling relationship among oxygen-related intermediates. As the oxygen binding energy of the catalyst reduces, the catalyst tends to generate H2O2. Conversely, H2O is preferentially produced when the oxygen binding energy of the catalyst becomes stronger. Thus, depending on the type of metal and ligand used, the oxygen binding energy of atomically dispersed catalysts can be modulated. Consequently, the ORR catalytic properties of atomically dispersed catalysts can be varied. However, this hypothesis is only based on thermodynamic considerations. Thus, relevant experimental studies are necessary to resolve the discrepancies among previous results and understand the distinct catalytic trends of atomically dispersed catalysts. This dissertation presents our efforts to reveal the origin of the unique catalytic behavior of atomically dispersed catalysts in the ORR.