Heteroatom-Doped Carbon Supports with Enhanced Corrosion Resistance in Polymer Electrolyte Membrane Fuel Cells

: Polymer Electrolyte Membrane Fuel Cells (PEMFC) are currently considered the most advanced fuel cell technology. However, the industrial implementation of PEMFCs is strongly hindered by deﬁcient durability, especially that of the carbonaceous materials commonly used to support the platinum-based catalyst nanoparticles, which are prone to electrochemical corrosion at the cathode, resulting in a serious performance loss of the entire cell. In the attempt to overcome this issue, many research groups have tried to introduce heteroatoms (N, S, B, P) into the carbon lattice, thus trying to make the electrode corrosion-resistant. Newly developed heteroatom-doped carbons were subjected to corrosion tests in half-cell and single-cell systems to evaluate their stability. This paper reviews the recent studies devoted to corrosion research of heteroatom-doped carbon supports for Pt-based catalysts in PEMFCs. In particular, an overview on N, B, and S dopants and their effects on carbon corrosion is provided.


Introduction
Today, polymer electrolyte membrane fuel cells (PEMFC) are among the most promising alternative energy conversion technologies, showing relatively high efficiency along with zero emission of polluting gases. The energy policy of developed countries is increasingly shifting towards fuel-cell-based energy production systems and vehicles in the effort to minimize their CO 2 emissions [1,2].
The electrochemical reactions in commercial PEMFC are currently catalyzed by Ptbased nanoparticles supported on high-surface-area carbon (Pt/C). Carbon blacks such as Vulcan XC-72 and BP-2000 have been widely used to support the Pt nanoparticles [3,4], and the fuel cell performance, especially that related to the system durability, seems to strongly depend on the carbon support. The support material needs to provide not only sites for metal nanoparticles and prevent their agglomeration, but also pathways for electron and mass transfer by forming continuous porous, and possibly hierarchical, channels.
Unfortunately, carbonaceous materials are susceptible to electrochemical corrosion under the operating conditions in the PEMFC cathode (high potential, high acidity, high relative humidity, high oxygen concentration, and moderate temperature), which is mostly governed by their low oxidation potential, as shown in the following equations: C + 2H 2 O → CO 2 + 4H + + 4e − E • C/CO 2 = 0.207 V vs. NHE (T = 298 K) C + H 2 O → CO + 2H + + 2e − E • C/CO = 0.518 V vs. NHE (T = 298 K) CO + H 2 O → CO 2 + 2H + + 2e − E • CO/CO 2 = 0.103 V vs. NHE (T = 298 K) The reaction in Equation (1) is the primary reaction that expresses carbon corrosion, since oxidation to carbon monoxide (Equation (2)) is thermodynamically less favored [5]. Hence, the reaction described in Equation (1) is thermodynamically allowable at the potentials at which the fuel cell cathode operates (0.6-0.7 V under steady-state operating conditions), but it is believed to be negligible due to its sluggish kinetics in this potential range. However, corrosion rate becomes significant at electrode potentials higher than 1.0-1.1 V vs. normal hydrogen electrode (NHE), which are very relevant to transient operating conditions such as fuel cell startup and shutdown [6]. Under these conditions, the local cathode voltage can increase up to~1.5 V when the air/fuel boundary occurs on the anode side during the startup or shutdown of the cell, resulting in a significant increase in the corrosion rate of the cathode carbon support [7,8]. Considering that PEMFCs in automotive applications undergo an estimated 10,000-40,000 startup/shutdown cycles over their service lives, depending on the vehicle type [9], in long-term usage, carbon support corrosion becomes a significant issue, reducing the stability and durability of the electrode and catalyst materials.
Carbon corrosion can lead to detachment of Pt nanoparticles from the support or aggregation of Pt; both will result in a decrease in the electrochemically active surface area (ECSA) of the catalyst and overall performance loss [10]. Figure 1 depicts the corrosion process of a carbon support for a typical Pt/C catalyst. Carbon is oxidized to CO 2 , which leaves the system as a gas, forming holes in the support. As a result, some Pt nanoparticles remain dethatched, and can be washed out from the cell. Figure 2 shows a scanning transmission electron microscope (STEM) image of different zones at the cathode catalyst layer after an accelerated stress test (AST) experiment, simulating multiple startups/shutdown cycles of the fuel cell. It is noticeable that due to carbon corrosion, voids with no carbon support (white circles in Figure 2) were formed in the catalyst layer [11]. In addition, electrochemical carbon oxidation leads to the collapse of the porous structure and loss of hydrophobicity resulting in diminished oxygen diffusion and severe performance loss [12,13]. Hence, the development of stable carbon supports is essential to meet the durability requirements and fulfil the full potential of PEMFCs.    In effort to improve the corrosion resistance of the support and the stability of the catalyst/support system, researchers have been developing new carbon-based [14] and non-carbon-based [15][16][17][18][19][20][21][22][23][24][25] support materials to replace the commonly used carbon black.
One path that has been explored in recent years is the introduction of heteroatoms such as N, S, B, and P, into the carbon lattice. This has become a hot topic among researchers, as it seems like a promising approach to overcome the current problem of carbon support corrosion in fuel cells. Although there are several review articles on various innovative support materials for PEMFC cathodes [26][27][28], there is no article that exhaustively reviews the subject of heteroatom-doping as a strategy to improve the corrosion resistance of carbon carriers. Herein, we present a complete review of recent studies devoted to corrosion research of heteroatom-doped carbon supports for Pt catalysts in PEMFCs.

Carbon Corrosion in PEMFCs-Protocols and Testing
There are two main approaches for studying the durability of PEMFC components: (1) life tests, (2) and accelerated stress tests (AST). Although the life test most reliably reflects the degradation that occurs in a fuel cell during its real-time operation [29], such experiments are difficult to implement on an ongoing basis due to their high cost and long duration. Hence, ASTs were designed to simulate, in a relatively short time, the long-term exposure of fuel cell components to the extreme physical conditions that the cell is exposed to throughout its entire service life. Each key constituent in PEMFC has an individual AST protocol designed specifically to determine its durability (electrocatalyst, catalyst support, MEA, membrane, air compressors, humidifiers, etc.). The catalyst support AST was developed to evaluate the support's durability and set the performance metrics. During unmitigated startups and shutdowns of a fuel cell system, an H 2 /air front can be formed in the anode, while the cathode compartment is filled with air. Due to the fact that the anode potential is determined by the reactant/s in the anode, the presence of oxygen at the anode, which will be reduced to water through the oxygen reduction reaction ORR), will result at the formation of a local voltage at the H 2 /air front at the cathode, which can be as high as 1.5 V. This high potential can lead to a severe carbon corrosion reaction at the cathode. This process stops as soon as the anode is completely filled with either air or hydrogen and the H 2 /air front disappears [30].
Since high potential at the cathode is usually the main cause for carbon corrosion, applying a high potential to the cathode is a common stressor in support corrosion ASTs. For example, in an earlier United States department of energy (U.S. DOE) catalyst support testing protocol, each test cycle consisted of voltage held at 1.2 V for 24 h, followed by a polarization curve and ECSA measurements. The suggested total test duration was 200 h. Material showing less than a 60% loss of initial catalytic activity at the end of the test, can be considered as acceptable catalyst support for the cathode [31]. Although slightly outdated, some researchers from the field still prefer to study the stability of carbon supports using this protocol. A newer U.S. DOE catalyst support durability protocol, from 2016, recommends using a rapid triangle wave voltage cycle to accelerate catalyst support corrosion while minimizing degradation of the catalyst itself. The test includes 5000 cycles of cathode voltage sweep between 1.0 and 1.5 V [32], since this voltage range more realistically represents the conditions at the cathode during an unmitigated system startup or shutdown. Most researchers seem to follow this AST protocol in their catalyst support corrosion studies [33][34][35][36]. Usually, carbon support durability tests are carried out while nitrogen is flowing at the cathode to investigate electrochemical corrosion separately, without the effects of chemical corrosion in the presence of air. In addition, the inert nitrogen atmosphere eliminates the effect of some parameters, such as oxygen concentration and flow field design on the results, making it convenient to compare with others.
Measuring the CO 2 content in the electrode exhaust during the AST is considered the primary way for carbon corrosion evaluation. Various methods for CO 2 detection and quantification can be used. For example, fuel cell fixture connected to a mass spectrometer (FC-MS) allows online monitoring of the gases emitted from the electrodes [37]. Figure 3 schematically shows FC-MS system configuration for catalyst support corrosion ASTs. The cathode outlet is connected to a mass spectrometer that quantifies the CO 2 concentration in the exhaust during the test, making it possible to determine the severity of support corrosion in operando.

FOR PEER REVIEW 5 of 16
schematically shows FC-MS system configuration for catalyst support corrosion ASTs. The cathode outlet is connected to a mass spectrometer that quantifies the CO2 concentration in the exhaust during the test, making it possible to determine the severity of support corrosion in operando. The ability of mass spectrometer to detect extremely low concentrations of carbon dioxide (at the ppm level) has made it one of the instruments of choice for studying support corrosion in half-cells as well [38]. In particular, a differential electrochemical mass spectrometer (DEMS), consisting of an electrochemical half-cell, a polytetrafluoroethylene (PTFE) membrane, and a vacuum system including the mass spectrometer, is widely used to characterize carbon corrosion [10,33].
Another useful method for carbon support degradation studies is infrared (IR) spec- The ability of mass spectrometer to detect extremely low concentrations of carbon dioxide (at the ppm level) has made it one of the instruments of choice for studying support corrosion in half-cells as well [38]. In particular, a differential electrochemical mass spectrometer (DEMS), consisting of an electrochemical half-cell, a polytetrafluoroethylene (PTFE) membrane, and a vacuum system including the mass spectrometer, is widely used to characterize carbon corrosion [10,33].
Another useful method for carbon support degradation studies is infrared (IR) spectroscopy. In this method, the effluent gases from the fuel cell cathode, including CO 2 , enter an infrared gas analyzer connected in line to the electrode exhaust. Unlike mass spectrometry, IR equipment does not include a vacuum system, which allows more gas to be sampled at the outlet, making the quantification of carbon loss more accurate [12,39,40]. Nevertheless, some research groups have reported the possibility of IR chamber flooding and disruption of CO 2 signal measurement by water in the gas stream. The necessity for water to be condensed before gas is supplied to the infrared sensors complicates this method for long-term tests [41][42][43].
A much less popular method that has been used for the study of CO 2 evolution is gas chromatography. For example, Paul, T.Y. et al. and Q. Li et al. used a gas chromatograph equipped with a methanizer to collect the CO 2 emitted from the cathode exhaust, and convert it into CH 4 to measure the CH 4 concentration as equivalent CO 2 concentration [44,45].
In summary, there are several ways to detect CO 2 emitted from the fuel cell as a consequence of carbon corrosion there, which can be used in operando to help tie the emitted gas to the applied potential and other operating conditions. These methods differ a little by accuracy, and the selection between them seems to be mainly governed by availability of instrumentation. There does not seem to be a standard method in the protocols, and as long as it is reliable, accurate and can be used in operando, it should be suitable for such studies.

N-Doped Carbons
The integration of N atoms into the carbon lattice is relatively convenient due to the similarity of the atomic radii of N (70 pm) and C (77 pm), which allows the formation of strong covalent N-C bonds [46]. This has prompted many researchers to try to improve the corrosion resistance of carbon and extend the durability of the catalyst by modifying the carbon with nitrogen. This section reviews the latest developments on this topic. N-doped carbon materials have long proven themselves in a number of applications such as batteries and supercapacitors [47][48][49]. Nitrogen has five valence electrons for bonding with carbon. As a result, it can be present in the carbon lattice in various forms that differ in their effect on the electronic and crystalline properties of the carbonaceous materials. For example, considering that nitrogen incorporation is accompanied by defects in graphene networks, quaternary N induces an n-type doping, while pyridinic N results in a p-type configuration [50,51]. Moreover, the amount of nitrogen affects the physicochemical properties of carbon. For instance, low-to-moderate nitrogen content (~4-8 wt%) has been reported to increase the electric conductivity of mesoporous carbons, but a higher nitrogen concentration leads to lower conductivity since the abundance of nitrogen interrupts the continuity of the graphitic structure [51]. A prerequisite of any electrode material in general, and catalyst supports in particular, is electronic conductivity. Later comes the surface area, hierarchical structure, catalyst-support interaction, wetting, etc., all of which can have significant influence on the performance of the fuel cells. Hence, it is important to maintain the electronic conductivity of the support after doping it with nitrogen, and thus it is not advised to reach concentrations higher than 8%, and in some cases even 5%.
Density functional theory (DFT) calculations propose several possible mechanisms for carbon corrosion reactions in N-doped carbons. It has been found that, based on their specific electronic structures, different nitrogen configurations (i.e., pyrrolic, pyridinic, and quaternary N) favor different mechanisms. Pyridinic N-doped carbon exhibits the best corrosion resistance due to the unstable structure of surface oxides which are intermediates in the CO 2 formation [52].
Despite the progress that has been made in experimental studies, the results are still inconclusive. For example, in the case of nitrogen-modified carbons synthesized by heat treatment with ammonia [53][54][55] nanoparticles deposited on N-doped Vulcan carbon supports which were prepared at different ammonolysis temperatures 400 and 800 • C (referred to as N-V-400 and N-V-800). Carbon support ASTs were conducted in a single-cell for the Pt/N-V-400 and Pt/N-V-800 cathode catalysts according to the 2016 U.S. DOE protocol. As illustrated in Figure 4a, the beginning-of-life (BOL) performance is similar for all three Pt/V and Pt/N−V catalysts, indicating that N presence does not greatly affect the initial activity. However, the carbon support ASTs results showed that the voltage loss at 0.6 A cm −2 is more dramatic after the first 200 cycles for the Pt/N−V samples, while Pt/V showed a more moderate potential drop (Figure 4b), Pt/N-V-400 and Pt/N-V-800 suffered a voltage loss of 115 and 66 mV, respectively. The authors attribute the inferior performance of the N-doped carbons to defects, in the form of nitrogen atoms, that destabilize the carbon structure [56]. best corrosion resistance due to the unstable structure of surface oxides which are inter mediates in the CO2 formation [52].
Despite the progress that has been made in experimental studies, the results are stil inconclusive. For example, in the case of nitrogen-modified carbons synthesized by hea treatment with ammonia [53][54][55], E. Hornberger et al. studied the activity of Pt nanopar ticles deposited on N-doped Vulcan carbon supports which were prepared at differen ammonolysis temperatures 400 and 800 °C (referred to as N-V-400 and N-V-800). Carbon support ASTs were conducted in a single-cell for the Pt/N-V-400 and Pt/N-V-800 cathode catalysts according to the 2016 U.S. DOE protocol. As illustrated in Figure 4a, the begin ning-of-life (BOL) performance is similar for all three Pt/V and Pt/N−V catalysts, indicat ing that N presence does not greatly affect the initial activity. However, the carbon sup port ASTs results showed that the voltage loss at 0.6 A cm −2 is more dramatic after the firs 200 cycles for the Pt/N−V samples, while Pt/V showed a more moderate potential drop (Figure 4b), Pt/N-V-400 and Pt/N-V-800 suffered a voltage loss of 115 and 66 mV, respec tively. The authors attribute the inferior performance of the N-doped carbons to defects in the form of nitrogen atoms, that destabilize the carbon structure [56].  In contrast, V. Golovin et al. came to the opposite conclusion in their research. They modified commercial KetjenBlack (KB) carbon with nitrogen-containing compounds: acetonitrile and pyridine, resulting in two N-doped carbons, ANKB and PNKB. Half-cell carbon degradation ASTs were carried out for the newly manufactured carbons. Tracking the evolution of cyclic voltammograms (CV) and assuming that the oxidation of the sample occurs mainly by increasing the number of carbon surface oxides, they concluded that ANKB and PNKB have higher corrosion resistance than pure KB [57].
N-doped carbons synthesized in situ, that is, by direct carbonization of C-N containing compounds [58,59], are also of interest to scientists seeking to improve the durability of PEMFCs. S. Shrestha et al. synthesized nitrogen-functionalized highly ordered mesoporous carbon (NOMC) from pyrrole using an SBA-15 template. Stability experiments showed that NOMC was more resistant to corrosion than Vulcan XC-72R, which the authors attribute to the strong C-N bonds and the higher graphitization degree of NOMC [60]. Recently Z. Qiao et al. reported a three-dimensional porous graphitic carbon (PGC) derived from a polymer hydrogel based on polyaniline and polypyrrole composite, as Pt catalyst support. Doped nitrogen was observed in PGC samples due to the precursors used. Measurements of carbon corrosion in a fuel cell cathode showed that the carbon loss from commercial Pt/C was much higher than that of a PGC-supported Pt catalyst. The authors suggested that the higher resistance to corrosion is more likely related to the improved degree of graphitization in PGC than to the presence of nitrogen [61]. Similarly, a higher degree of graphitization was considered to be the reason for better stability of nitrogen-modified carbon composite support when compared to conventional carbon black in fuel cell support corrosion tests [62].
In summary, it can be argued that there is no unequivocal agreement in the experimental literature that nitrogen doping inhibits the electrochemical corrosion of carbon. Generally, in cases where N-modified carbon demonstrates higher stability, it is not necessarily related to nitrogen presence, but can also be attributed to other structural properties of the material such as the graphitization level. However, the beneficial effect of N on Pt catalyst characteristics is not controversial. Nitrogen incorporation in carbon supports improves the dispersion of Pt nanoparticles and prevents agglomeration, strengthens the Pt-support interactions and suppresses Pt dissolution during fuel cell operation [61,63]. Moreover, the existence of nitrogen functionalities, and their influence on the carbon support's electronic configuration, enhances the ORR activity, though the exact mechanism is not fully understood yet [60,64].

B-Doped Carbons
Boron is one of the most interesting doping elements in the context of carbonaceous materials. Incorporation of boron into a carbon framework has proven to be an effective tool for changing some of the physical properties of these materials. The unique bonds formed between neighboring C and electron-deficient B atoms can minimize the mechanical stress, enhance the electronic conductivity, and tailor the electrochemical features of the host carbon [65,66]. Therefore, materials of this type are considered useful for various applications such as supercapacitors and hydrogen storage devices [67]. For high temperature applications, boron compounds have long been used in surface coatings to inhibit the oxidation reaction of carbon in air at temperatures above 400 • C [68][69][70]. In the case of electrochemical oxidation of B-doped carbons, there is a very limited number of publications, and unfortunately, none of them reports experiments in a fuel cell system. For example, C.K Acharya et al. studied the durability of Pt electrocatalysts deposited on pristine and boron-doped carbon supports under multiple potential cycles (0.05-1.20 V vs. NHE) [71]. The corrosion resistance of the carbons was determined by the current peak at 0.35 V vs. saturated calomel electrode (SCE), which was associated with the formation of carbon oxide surface species. In the case of Pt/C, this peak increases throughout the experiment, as shown in Figure 5a, whereas, in the case of Pt deposited on the boron-modified carbon (Pt/BC), almost no carbon oxidation current was observed (Figure 5b). Accordingly, the authors concluded that the boron-doped carbon support undergoes less electrochemical oxidation than pure carbon. It was attributed to the formation of boron oxides that passivate the carbon surface, inhibiting the corrosion of carbon.
Another example is presented in the work of A.B. Bose et al., who carried out cyclic voltammetry experiments with boron-doped carbon microspheres (BCMS). In contrast to Vulcan XC-72, no carbon surface oxides for BCMS were detected during the test, suggesting its higher electrochemical stability [72]. This is also supported by the work of R. Yao et al. who studied boron-doped ECP600 carbon black as a support for a Pt catalyst. Half-cell AST in high potential range (1.0-1.6 V vs. reversible hydrogen electrode, RHE) showed that the Pt/B-ECP600 exhibits a lower ECSA loss than the Pt/ECP600 at the end of life (EOL). This result was attributed to the better corrosion resistance of B-ECP600 carbon, although more thorough corrosion measurements have not been conducted [73]. Similar observations were reported by J. Wang et al. They investigated the durability of Pt catalyst supported on a composite of boron-doped carbon nanorods (BCNR) grown on carbon paper. In the case of Pt/BCNR electrode, higher percentage of the initial ECSA remained after the AST when compared to Pt/CNTs. Measurements of ORR activity before and after AST also suggested a higher stability of the Pt/BCNR system. Similar to the previous examples, the authors assume that boron oxides passivate the carbon surface, which leads to inhibited carbon corrosion and higher stability of Pt/BCNR [74].  Almost all of the publications reporting Pt/B-doped C systems unanimously indicate that the Pt-support interaction is improved in the presence of boron in the carbon carrier. Higher binding energies were found for Pt atoms adsorbed on B-doped carbons when compared to pristine carbons. Most likely, this is due to a strong hybridization between Pt d-orbitals and B p-orbitals, leading to direct chemical bonding between the platinum and boron atoms [75,76]. For example, DFT calculations for Pt clusters adsorbed on pristine and on B-doped fullerene showed that the adsorption energy per Pt atom in the Pt clusters was greater with the B-doped fullerene than with the pristine fullerene by as much as~13 kcal/mol. (Figure 6) [75]. Despite the few attempts to investigate the electrochemical durability of Pt/B-doped carbon catalysts, the scientific literature is still deficient in information on these very interesting supports. Currently, there is no literature on corrosion ASTs for B-doped carbon supports in fuel cells. All publications mentioned above assess the corrosion resistance based on half-cell ASTs only. Extensive testing of B-doped carbon in a fuel cell is required to reliably study the effect of B-doping on such a complex phenomenon as catalyst support corrosion.

S-Doped Carbons
Sulfur, having a relatively large atomic radius, was reported to cause strain and introduce defect sites into the carbon network when used as a dopant [77]. S-doping affects the electronic density of carbon due to the lone electrons' pair of sulfur, acting as an electrons donor. Additionally, as such, S-doped carbon materials proved to be of interest for applications such as sensors, superconductors and Li-S batteries [78,79]. S-doped carbons are a popular subject of study as platinum group metal (PGM)-free electrocatalysts for ORR [80][81][82]. However, as a support material for Pt catalysts, S-containing carbons are not as popular, mainly due to concerns about Pt-poisoning by sulfur contaminants such as SO2, COS, and H2S [83,84]. According to results reported by V. Perazzolo et al., the use of S-doped carbon in fuel cells does not result in improved catalytic activity. In their study, Pt deposited on sulfur-doped mesoporous carbon (Pt/S-MC) showed low ORR performance in a PEMFC, attributed to a lack of optimization of the Pt/S-MC for MEA configuration, leading to high ohmic and oxygen transport resistances [85]. P. Yin et al. conducted experiments on the durability of sulfur-doped carbon (S-C) supports with high sulfur loadings (14 %wt). As shown in Figure 7, their Pt/S-C cathode showed satisfactory initial ORR activity, but it could not tolerate the support degradation during high-voltage cycling (1.0-1.5 V), resulting in dramatic decline in fuel cell performance. The corrosion of S-C has not been monitored, but its high surface area (1200 m 2 /g) measured after the AST, suggests it has been severely corroded [86].

S-Doped Carbons
Sulfur, having a relatively large atomic radius, was reported to cause strain and introduce defect sites into the carbon network when used as a dopant [77]. S-doping affects the electronic density of carbon due to the lone electrons' pair of sulfur, acting as an electrons donor. Additionally, as such, S-doped carbon materials proved to be of interest for applications such as sensors, superconductors and Li-S batteries [78,79]. S-doped carbons are a popular subject of study as platinum group metal (PGM)-free electrocatalysts for ORR [80][81][82]. However, as a support material for Pt catalysts, S-containing carbons are not as popular, mainly due to concerns about Pt-poisoning by sulfur contaminants such as SO 2 , COS, and H 2 S [83,84]. According to results reported by V. Perazzolo et al., the use of S-doped carbon in fuel cells does not result in improved catalytic activity. In their study, Pt deposited on sulfur-doped mesoporous carbon (Pt/S-MC) showed low ORR performance in a PEMFC, attributed to a lack of optimization of the Pt/S-MC for MEA configuration, leading to high ohmic and oxygen transport resistances [85]. P. Yin et al. conducted experiments on the durability of sulfur-doped carbon (S-C) supports with high sulfur loadings (14 %wt). As shown in Figure 7, their Pt/S-C cathode showed satisfactory initial ORR activity, but it could not tolerate the support degradation during high-voltage cycling (1.0-1.5 V), resulting in dramatic decline in fuel cell performance. The corrosion of S-C has not been monitored, but its high surface area (1200 m 2 /g) measured after the AST, suggests it has been severely corroded [86]. In a different work, the chemical stability of non-platinized S-MC was studied. Halfcell ASTs based on 10,000 cycles at a potential range of (−0.25) -1.00 V vs. SCE were conducted. The intensity of carbon surface oxides peak during cycling has not changed for S-MC when compared to the pristine carbon without the sulfur. The authors attribute this to the presence of thiophenic functional groups, which prevents the oxidation of the carbon [87]. This is further supported by studies with sulfurized carbon xerogel supports (S-CXG), showing that the post-degradation ORR performance loss has improved from 45% for the un-doped xerogel (Pt/CXG) to 38-39% for the doped support (Pt/S-CXGs) [88].
Many publications have reported that there is a strong metal-support interaction between Pt nanoparticles and S atoms in the carbon supports, which enhances the physical and electrochemical stability of Pt [89][90][91][92]. It is important to note that, despite the numerous works on degradation of S-doped carbons, to date, none of them has included corrosion measurements. Further experiments with focus on quantification of the corrosion process and its kinetics will be very valuable to understanding the merits of this group of materials.

Other Heteroatom-Doped Carbons
Interesting attempts to increase the durability of PEMFC cathodes have also been made using phosphorus (P) and fluorine (F)-doped carbons [93,94]. Phosphorus-containing carbon demonstrated good corrosion resistance as protective coating, of bipolar plates [95]. Z. Li et al. synthesized Pt/F-C catalyst using fluorinated carbon black as a support. AST cycling (0.6-1.0 V vs. RHE) showed that the stability of the catalyst was significantly enhanced through the fluorination of the carbon. This was partly attributed to the presence of stronger C-F bonds versus the original C-O and C-C bonds [96]. Carbonaceous materials doped with more than one type of heteroatom also seem to be an attractive venue for research [97,98]. For example, nitrogen and fluorine co-doped graphite nanofibers (NF-GNFs) were investigated as Pt support. Under harsh corrosive conditions, Pt/NF-GNF catalyst presented superior long-term stability when compared to the commercial Pt/C In a different work, the chemical stability of non-platinized S-MC was studied. Halfcell ASTs based on 10,000 cycles at a potential range of (−0.25)-1.00 V vs. SCE were conducted. The intensity of carbon surface oxides peak during cycling has not changed for S-MC when compared to the pristine carbon without the sulfur. The authors attribute this to the presence of thiophenic functional groups, which prevents the oxidation of the carbon [87]. This is further supported by studies with sulfurized carbon xerogel supports (S-CXG), showing that the post-degradation ORR performance loss has improved from 45% for the un-doped xerogel (Pt/CXG) to 38-39% for the doped support (Pt/S-CXGs) [88].
Many publications have reported that there is a strong metal-support interaction between Pt nanoparticles and S atoms in the carbon supports, which enhances the physical and electrochemical stability of Pt [89][90][91][92]. It is important to note that, despite the numerous works on degradation of S-doped carbons, to date, none of them has included corrosion measurements. Further experiments with focus on quantification of the corrosion process and its kinetics will be very valuable to understanding the merits of this group of materials.

Other Heteroatom-Doped Carbons
Interesting attempts to increase the durability of PEMFC cathodes have also been made using phosphorus (P) and fluorine (F)-doped carbons [93,94]. Phosphorus-containing carbon demonstrated good corrosion resistance as protective coating, of bipolar plates [95]. Z. Li et al. synthesized Pt/F-C catalyst using fluorinated carbon black as a support. AST cycling (0.6-1.0 V vs. RHE) showed that the stability of the catalyst was significantly enhanced through the fluorination of the carbon. This was partly attributed to the presence of stronger C-F bonds versus the original C-O and C-C bonds [96]. Carbonaceous materials doped with more than one type of heteroatom also seem to be an attractive venue for research [97,98]. For example, nitrogen and fluorine co-doped graphite nanofibers (NF-GNFs) were investigated as Pt support. Under harsh corrosive conditions, Pt/NF-GNF catalyst presented superior long-term stability when compared to the commercial Pt/C [99]. In another study, Choi C. H. et al. synthesized a ternary-doped carbon (B, P, N-doped carbon (BPNDC)) as a Pt-free catalyst and studied its stability in an acidic environment. A degradation test (200 cycles, 1.12-1.52 V vs. RHE), showed that the decrease in the ORR activity of BPDNC was much smaller than that with Pt/C. However, a specific connection between the types of dopants and the superior stability of BPNDC was not established [100].

Concluding Remarks
Carbon, being the primary support material for Pt-based electrocatalysts in PEMFC, suffers from corrosion due to high potentials generated at the cathode during operation, in addition to the acidic and oxidative environment that exist in these fuel cells. Therefore, development of durable, corrosion-resistant supports is critical for the mass adoption of FC technologies. In recent years, heteroatom doping of carbons has become a popular approach to address the corrosion issue, as it significantly alters the physico-chemical characteristics of the pristine carbon. This review surveys the recent work done on the topic of heteroatom-doped carbon support corrosion in PEMFCs, starting with the explanation of the proper methodology for carbon corrosion investigation. In the case of N-doping, there is no unequivocal agreement in the literature that nitrogen presence inhibits carbon corrosion, whereas for boron and sulfur dopants, it is evident that there is a deficiency of corrosion studies conducted in a single cell, and more extensive testing of these carbons in a fuel cell are required to reliably examine the effect of doping on the support corrosion. Conclusive insights cannot be achieved with a half-cell system alone, and such studies are the prevailing part of the literature in the field. The literature unanimously states that N, B, S, P, and F dopants improve the Pt-carbon interaction, leading to more homogenous dispersion and better stability of Pt nanoparticles. Hence, there is still much work to be done in the field in order to fully understand the role of the various dopants and their contribution to or adverse effects on the durability of the support.
Author Contributions: A.K. was responsible for literature collection and writing. A.K. and L.E. were in charge of structuring and conceptualization. Q.L. and L.E. edited the final version. All authors have read and agreed to the published version of the manuscript.