Native Ligand Carbonization Renders Common Platinum Nanoparticles Highly Durable for Electrocatalytic Oxygen Reduction: Annealing Temperature Matters

Current protocols for synthesizing monodisperse platinum (Pt) nanoparticles typically involve the use of hydrocarbon molecules as surface‐capping ligands. Using Pt nanoparticles as catalysts for the oxygen reduction reaction (ORR), however, these ligands must be removed to expose surface sites. Here, highly durable ORR catalysts are realized without ligand removal; instead, the native ligands are converted into ultrathin, conformal graphitic shells by simple thermal annealing. Strikingly, the annealing temperature is a critical factor dictating the ORR performance of Pt catalysts. Pt nanoparticles treated at 500 °C show a very poor ORR activity, whereas those annealed at 700 °C become highly active along with exceptional stability. In‐depth characterization reveals that thermal treatment from 500 to 700 °C gradually opens up the porosity in carbon shells through graphitization. Importantly, such graphitic‐shell‐coated Pt catalysts exhibit a superior ORR stability, largely retaining the activity after 20 000 cycles in a membrane electrode assembly. Moreover, this ligand carbonization strategy can be extended to modify commercial Pt/C catalysts with substantially enhanced stability. This work demonstrates the feasibility of boosting the ORR performance of common Pt nanoparticles by harnessing the native surface ligands, offering a robust approach of designing highly durable catalysts for proton‐exchange‐membrane fuel cells.


Introduction
As one of the most promising energyconversion devices, proton-exchangemembrane fuel cells (PEMFCs) have been attracting much attention for many years. [1][2][3] It is known that the sluggish cathodic kinetics of the oxygen reduction reaction (ORR) greatly limits the conversion efficiency of PEMFCs. [4,5] To date, Pt (or Pt-based alloys) nanoparticles (NPs) loaded on a carbon support are still the most promising ORR catalysts due to their intrinsically high activity. [6] However, Pt catalysts typically suffer from the poor durability during long-term operation, [7] which is one of the critical issues hampering the widespread applications of PEMFCs. [8,9] Under the harsh operation conditions of PEMFCs, Pt catalysts can easily lose the electrochemical surface area (ECSA) due to multiple reasons, including the Ostwald ripening and agglomeration of NPs, [8,10] detachment of NPs from the carbon support, [11] dissolution of Pt, [12] and corrosion of the carbon support. [13] In the past few decades, many efforts have been dedicated to addressing the stability issue of Pt catalysts. [14][15][16] Coating Pt NPs with a porous and electrochemically stable encapsulating layer has proven to be a promising approach to enhance the ORR durability, [17] as this can effectively prevent the agglomeration of NPs while simultaneously suppressing the dissolution of Pt, due to the combined barrier and confinement effects. [18] Among various encapsulating materials explored, carbon shows great promise because of its high electrical conductivity, electrochemical stability, and relatively benign corrosion products. [18,19] For instance, Kwon et al. reported the synthesis of carbon-coated Pt NPs on carbon nanofibers using Pt-aniline complexes as the precursor. [18] Hyeon et al. presented the synthesis of intermetallic ordered FePt NPs coated with N-doped carbon shells derived from polydopamine, where the carbon shell thickness can be tuned by controlling the polymerization duration. [20] Indeed, these carbon-coating strategies can dramatically enhance the durability of ORR catalysts as indicated by both rotating disk electrode (RDE) and membrane electrode assembly (MEA) tests. Despite the improved stability, the Current protocols for synthesizing monodisperse platinum (Pt) nanoparticles typically involve the use of hydrocarbon molecules as surface-capping ligands. Using Pt nanoparticles as catalysts for the oxygen reduction reaction (ORR), however, these ligands must be removed to expose surface sites. Here, highly durable ORR catalysts are realized without ligand removal; instead, the native ligands are converted into ultrathin, conformal graphitic shells by simple thermal annealing. Strikingly, the annealing temperature is a critical factor dictating the ORR performance of Pt catalysts. Pt nanoparticles treated at 500 °C show a very poor ORR activity, whereas those annealed at 700 °C become highly active along with exceptional stability. In-depth characterization reveals that thermal treatment from 500 to 700 °C gradually opens up the porosity in carbon shells through graphitization. Importantly, such graphitic-shell-coated Pt catalysts exhibit a superior ORR stability, largely retaining the activity after 20 000 cycles in a membrane electrode assembly. Moreover, this ligand carbonization strategy can be extended to modify commercial Pt/C catalysts with substantially enhanced stability. This work demonstrates the feasibility of boosting the ORR performance of common Pt nanoparticles by harnessing the native surface ligands, offering a robust approach of designing highly durable catalysts for proton-exchange-membrane fuel cells. www.advmat.de www.advancedsciencenews.com presence of a protective layer at the particle surface can inevitably block some surface sites and restrict mass transport, [21] thereby adversely influencing the catalytic performance of Pt catalysts. Therefore, the ability to fine-tune the thickness and microstructure of carbon shells to achieve an optimized tradeoff between activity and stability, although challenging, is critical to taking full advantage of this carbon-coating strategy. [19,22] Moreover, most previously reported carbon-coating methods involve the incorporation of a foreign species (usually a polymer) as the carbon source, [20,21] which does not necessarily have a preferential affinity to Pt, thus making it difficult to obtain complete and uniform carbon shells coated on individual NPs.
On the other hand, using uniform Pt NPs is an effective approach to inhibit the Ostwald ripening during the ORR. [8,[23][24][25] To date, monodisperse NPs are typically obtained by colloidal synthesis, involving the use of hydrocarbon ligands to control the nucleation and growth kinetics of NPs. [24] Typical ligands include oleylamine (OAm) and oleic acid (OA). [24][25][26] Such ligand-capped NPs are highly soluble in nonpolar solvents, [25] which is also beneficial for subsequent solution processing. Despite these advantages, however, these long-chain organic ligands are highly insulating and can act as a dense barrier to restrict the accessibility of NPs, [27] thus having a detrimental influence on their catalytic applications. As a consequence, to harness Pt NPs as ORR catalysts, the native surface ligands have to be replaced or removed to expose the surface sites. [28,29] To this end, a number of techniques have been developed to treat Pt NPs, such as multiple cycles of washing with nonsolvents, [30] UV-ozone cleaning, [24] and ligand exchange with small molecular (or ionic) species. [25,31] Alternatively, the hydrocarbon ligands can also be removed by calcination in air. [32] Although Pt NPs become active toward the ORR upon post surface treatment, the catalytic performance depends highly on the surface treatment methods. In addition, whether the native ligands are completely removed by these treatment methods remains an open question, [25] which accounts for the performance inconsistency commonly encountered.
In contrast to previous methods where attempts are made explicitly to remove or replace the surface ligands, [24][25][26][27][28][29][30][31][32][33] we report herein that the native ligands can be exploited to render common Pt NPs highly durable for the ORR. We show that simple thermal annealing can convert the hydrocarbon ligands into ultrathin, conformal carbon shells, without sacrificing the size monodispersity of Pt NPs. We highlight that the annealing temperature has a profound influence on the ORR performance of the resulting carbon-coated Pt catalysts. Specifically, thermal treatment of Pt NPs at 500 °C leads to highly dense carbon shells severely restricting the accessibility of surface sites, whereas increasing the annealing temperature triggers the reconstruction of carbon shells through graphitization, gradually opening up the transport channels for accessing the particle surface. In particular, thermal annealing of Pt NPs (supported on the carbon support) at 700 °C leads to bilayered graphitic shells with optimized porosity, yielding graphiticshell-coated Pt catalysts with a superior ORR activity and stability. Importantly, this in situ ligand carbonization strategy is robust and broadly applicable for all Pt or Pt-based alloy NPs and can be further extended to boost the stability of commercial Pt/C catalysts. Our work demonstrates that the native surface ligands, previously considered detrimental for catalytic applications, can be harnessed to develop highly stable ORR catalysts for high-performance PEMFCs.

Results and Discussion
Scheme 1 illustrates the preparation of the graphitic-shellcoated Pt catalysts. Monodisperse Pt NPs (≈5 nm in diameter) are synthesized via a standard thermal decomposition approach using OAm and OA as the capping ligands. [34] Prior to ligand carbonization, the as-synthesized Pt NPs are first loaded onto a carbon support such as Ketjenblack (KB, ECP-600JD) by sonication (Pt@KB). Subsequently, the Pt@KB powder is subjected to thermal annealing in a N 2 atmosphere at 500 °C for 1 h, which converts the OAm/OA ligands into conformal carbon shells. To obtain graphitic-shell-coated Pt catalysts, such carbon-coated Pt NPs are further annealed at various temperatures (600-700 °C) for another 1 h, which triggers the reconstruction of carbon shells to afford porous, graphitic layers, without affecting the shell thickness. [34] The resulting catalyst is labeled with Pt-T, where T is the annealing temperature. The typical Pt loading of Pt-T catalysts is ≈35 wt%, determined by thermogravimetric analysis (TGA) and inductively coupled plasma mass spectrometry (ICP-MS). www.advmat.de www.advancedsciencenews.com Figure S1 (Supporting Information) shows a transmission electron microscopy (TEM) image of the as-synthesized 5 nm-Pt NPs capped with OAm and OA ligands. The highresolution TEM (HRTEM) image of Pt-500 catalysts (Figure 1a) reveals the presence of an ultrathin carbon shell conformally coated on each Pt NP (Figure 1e), presumably arising from the carbonization of the OAm/OA capping ligands. A closer inspection of the HRTEM images indicates that the carbon shells are quite uniform in thickness, with the majority comprising two carbon layers. TEM also reveals that such carboncoated Pt NPs are highly sintering-resistant and can maintain their size upon thermal annealing up to 700 °C (Figure 1a-c), although apparent particle agglomeration was observed when annealed at 800 °C ( Figure 1d). Notably, no substantial variation in carbon shell thickness was observed from Pt-500 to Pt-700 (Figure 1e-g), whereas many Pt particles in Pt-800 are no longer coated by carbon shells (Figure 1h). These results demonstrate that Pt NPs supported on KB can withstand thermal annealing without sacrificing the size monodispersity when the annealing temperature is below 700 °C, which is remarkable considering the high Pt loading (≈35 wt%). Presumably, the excellent thermal stability of the Pt NPs is attributed to the ligand-derived carbon shells, which effectively prevent the coarsening of NPs during thermal annealing. Control experiments show that the same Pt NPs supported on KB, upon ligand removal by repeated washing with ethanol, underwent severe sintering when thermally treated at 700 °C ( Figure S2, Supporting Information). This illustrates the important role of the ligand-derived carbon shells in preventing the sintering of Pt NPs.
The size variation of Pt NPs with respect to the annealing temperature was further investigated by X-ray diffraction (XRD). Figure 2a shows the XRD patterns of various Pt-T catalysts, where the three major peaks located at 39.8°, 46.3°, and 67.5° were ascribed to the (111), (200), and (220) planes of Pt. [35] The crystalline size of Pt NPs, calculated based on the Scherrer equation, [36] was essentially unchanged from Pt-500 to Pt-700, and the significantly reduced peak width of Pt-800 was a result of the particle agglomeration at 800 °C. To reveal the characteristics of the ligand-derived carbon shells, the Raman spectra of Pt-500 and Pt-700 were comparatively analyzed. As shown in Figure S3 (Supporting Information), both Pt-500 and Pt-700 show the typical D and G bands at 1300 and 1590 cm −1 , respectively, with the I D /I G ratio decreasing from 1.32 to 1.01, suggesting that the graphitization degree of Pt-T increases with the increased annealing temperature. [37] However, we note that the KB support may dominate the Raman signals of Pt-T, and as a consequence, further characterization is needed to reliably evaluate the graphitization degree of carbon shells. As shown in Figure 2b, the TGA curves of both Pt-500 and Pt-700 show the major weight loss from 300 to 450 °C, which corresponds to the oxidation of the KB support. For Pt-500, an additional weight loss of ≈3.8 wt% appears in the lower temperature region between 160 and 200 °C (as indicated by the arrow in Figure 2b). This weight loss, presumably ascribed to the oxidation of the less-graphitized carbon shells, [38,39] was not observed for Pt-700. This result suggests that as the annealing temperature increased from 500 to 700 °C, the graphitization degree of carbon shells was increased to a level comparable to that of the KB support. As indicated by N 2 adsorption-desorption measurements ( Figure S4, Supporting Information), the surface area of Pt-700 is 495 m 2 g −1 , slightly lower than that (649 m 2 g −1 ) of Pt/C.
To verify that the carbon shell is transformed from the OAm/ OA ligands rather than the unstable species in the carbon support, [40] Pt NPs were also loaded onto noncarbonaceous supports such as hollow SiO 2 spheres followed by thermal annealing under otherwise identical conditions. The product  www.advmat.de www.advancedsciencenews.com was named as Pt@SiO 2 -T, depending on the annealing temperature (i.e., 500-700 °C). Similar to the Pt NPs loaded on KB, heat treatment of Pt@SiO 2 also led to carbon-coated Pt NPs (Figure 2c), thus confirming that the carbon shell indeed originates from the surface ligands. Moreover, the absence of carbon supports further allows us to disentangle the fine structural features of carbon shells with the varied annealing temperatures. Figure 2d shows the C K-edge X-ray absorption near-edge structure (XANES) spectra of Pt@SiO 2 -500 and Pt@SiO 2 -700, where the characteristic π-π* (≈285 eV) and σ-σ* (≈291 eV) resonance peaks elucidate the typical graphene-like features exhibited by the ligand-derived carbon shells. For all Pt@SiO 2 -T samples, the resonance absorption peak at 288.3 eV suggests the presence of heteroatoms (e.g., O) in carbon shells. [41] In addition, a weak shoulder at around 283.2 eV is also observed in both cases. This feature is attribute to the surface defects, such as dangling bonds and amorphous carbon. [42] With the increased annealing temperature, the pre-edge feature intensity of Pt@ SiO 2 -T slightly increases, suggesting that more defect sites have been created in the carbon shells. [43] The X-ray photoelectron spectroscopy (XPS) spectrum of Pt@SiO 2 -500 shows the presence of Pt, C, O, and N (Figure 2e), where N atoms (≈2.3 at%) should originate from the OAm ligands. As indicated by the high-resolution N 1s spectrum (Figure 2f), N mainly exists in the forms of pyridine N (≈398 eV) and pyrrole N (≈400 eV). The high-resolution C 1s spectrum indicates the presence of CN species in carbon shells (Figure 2g), [44] indicating that thermal annealing at 500 °C converts the native OAm ligands into N-doped carbon shells. Notably, the N signals of Pt@SiO 2 -700 were essentially undetectable by XPS (Figure 2e), suggesting that further annealing at 700 °C nearly completely removed N atoms from carbon shells, in addition to increasing the graphitization degree. For both Pt@SiO 2 -500 and Pt@SiO 2 -700, the Pt 4f  www.advmat.de www.advancedsciencenews.com signals can be deconvoluted into two doublets assigned to Pt 0 and Pt 2+ states (Figure 2h,i), respectively, indicating that Pt NPs remained unchanged during annealing. [20,45] In addition, XPS studies were also carried out on Pt-700 and Pt/C catalysts, and the signals ascribed to oxidized Pt species (Pt 2+ ) can be seen in both cases ( Figure S5, Supporting Information), suggesting the occurrence of surface oxidation of Pt NPs. Also, Pt-700 and Pt/C have a similar intensity ratio of Pt 2+ /Pt 0 , indicating that the content of oxidized Pt species in Pt-700 is comparable to that in Pt/C. Therefore, the thermal annealing treatment does not increase the oxidation extent of Pt NPs. Further observations reveal that the binding energy of the two Pt 4f 5/2 and 4f 7/2 peaks in Pt-700 is shifted positively by ≈0.78 eV relative to Pt/C ( Figure S5, Supporting Information). This positive shift in binding energy is indicative of an enhanced electronic interaction between Pt NPs and carbon shells, which has been reported to be capable of improving the stability and catalytic activity of Pt catalysts. [21] The ORR performance of various Pt-T catalysts was first investigated using RDE in O 2 -saturated 0.1 m HClO 4 . For comparison, commercial Pt/C catalysts (J&M, HISPEC4000, particle size ≈ 3 nm, Pt loading ≈ 35 wt%) were tested as a standard (Table S1, Supporting Information). [46] As expected, the assynthesized, ligand-capped Pt NPs (Pt@OAm) loaded onto KB were essentially inactive as indicated by linear sweep voltammetry (LSV, Figure 3a). The ORR inactivity of Pt@OAm was presumably caused by the surface hydrocarbon ligands, which effectively block the accessibility of surface sites, consistent with previous results. [25] Following the thermal annealing, all of the Pt-T catalysts demonstrated the ORR activity, with the degree of activity varying depending on the annealing temperature that was used. Specifically, both Pt-500 and Pt-600 catalysts exhibited a relatively poor ORR activity, though the latter is slightly better in terms of the onset and half-wave potentials (Figure 3a). Strikingly, when the annealing temperature reaches 700 °C, the resulting Pt-700 Adv. Mater. 2022, 34, 2202743 www.advmat.de www.advancedsciencenews.com catalyst became highly active (Figure 3a), with the half-wave potential a little lower than that of commercial Pt/C (0.896 V vs 0.902 V). Further increasing the annealing temperature to 800 °C decreased the ORR activity. These results suggest that increasing the annealing temperature from 500 to 700 °C effectively activates Pt NPs. This annealing temperature-dependent ORR activity was also evidenced by cyclic voltammograms (CV). As shown in CV curves (Figure 3b), Pt-700 catalysts showed the well-defined peaks of the monolayer hydrogen adsorption and desorption, whereas these characteristic features were barely observable for other three catalysts. Likewise, the ECSA of Pt-700 (34.82 m 2 g −1 Pt ) is much higher than that of Pt-500 (3.1 m 2 g −1 Pt ) and Pt-600 (8.8 m 2 g −1 Pt ) (despite all three having similar Pt particle sizes), indicating the greater exposure of surface sites as the annealing temperature varied from 500 to 700 °C. [21] More importantly, in addition to the excellent ORR activity, Pt-700 catalysts also exhibited superior long-term stability. Figure 3c shows the polarization curves of Pt-700 according to the number of accelerated durability test (ADT) cycles. The halfwave potential of Pt-700 was nearly unchanged after 20k ADT cycles, indicative of the negligible activity degradation, which was also verified by CV ( Figure S6, Supporting Information). The ECSA, mass activity (MA, at 0.9 V), and specific activity (SA) of Pt-700 were largely retained as well after 20k cycles (Figure 3e,f), further confirming the negligible decrease of the accessible active sites. [47] In comparison, the half-wave potential of commercial Pt/C was decreased by ≈50 mV after 20k cycles (Figure 3d), along with an intensive degradation in ECSA, MA, and SA (Figure 3e,f).
The poor stability of commercial Pt/C is typical for Pt-based catalysts, which is presumably caused by the agglomeration and/or Ostwald ripening of Pt NPs during the ORR. To further explore the degradation mechanism, we investigated the structural change of the same NPs during cycling by ex situ TEM by using the positioning gold grids (see Supporting information for details). [48] TEM shows that after 20k ADT cycles some irregular large particles emerged for commercial Pt/C ( Figure S7a,b, Supporting Information), whereas Pt-700 catalysts remained essentially unchanged with no particle agglomeration and detachment ( Figure S7c,d, Supporting Information). In addition, we also evaluated the dissolution of Pt NPs by analyzing the Pt species dissolved in electrolytes after 20k cycles by ICP-MS, which showed that the dissolved Pt content for Pt-700 catalysts was substantially lower than that for Pt/C (0.232 ng vs 40 ng). [49] The Pt particles from the Pt/C catalyst are ≈1.7× smaller than those from the Pt-700 catalyst (≈3 nm vs 5 nm), and it is well known that smaller Pt particles are more prone to dissolution than larger ones. [50][51][52] However, the antidissolution improvement of ≈170× far exceeds what would be expected based purely on differences in the particle size, and suggests that a further electronic effect may be largely responsible for the difference in Pt dissolution. Nonetheless, these results emphasize the great role of the ligand-derived graphitic shells in preventing migration, agglomeration, and dissolution of Pt NPs during the ADT.
Having established the importance of ligand carbonization for improving the ORR activity and stability of Pt NPs, we turn our attention to investigating the annealing temperature-dependent catalytic activity exhibited by Pt-T catalysts.
As mentioned above, Pt-800 showed an inferior ORR activity, which is expected considering the severe agglomeration of Pt NPs at 800 °C (Figure 1d). However, the drastically different ORR activity exhibited by Pt-500, Pt-600, and Pt-700 is somewhat surprising (Figure 3a), because these three catalysts comprise the same Pt NPs coated with carbon shells having a similar thickness (i.e., bilayer carbon). [53] The fact that Pt-500 catalysts has a very small ECSA (≈3.1 m 2 g −1 Pt ) despite no apparent change in Pt particle size indicates that the carbon shell resulting from annealing at 500 °C is very dense in nature, which severely blocks the accessibility of surface sites despite its ultrathin thickness. Conversely, the enhanced ECSA and ORR activity exhibited by Pt-600 and Pt-700 catalysts suggests the accessibility of Pt NPs was gradually improved with the annealing temperature increasing from 500 to 700 °C. Given the essentially unchanged carbon shell thickness, we speculate that the initially dense carbon shell undergoes structural reconstruction and becomes progressively porous with the increased annealing temperature, as illustrated in Scheme 1.
As it is difficult to directly visualize the proposed porosity in carbon shells by HRTEM (Figure 1e-g), we designed poisoning experiments to better reveal the subtle carbon-shell difference between various Pt-T catalysts. [20] Polystyrene with a terminal thiol group (denoted as PS-SH), with a molecular weight of 12 600 (the molecular length of PS-SH is estimated to be ≈28 nm), [54] was synthesized as a molecular probe. The thiol group is a well-known deactivator of ORR due to its strong binding affinity to Pt. [55] The three catalysts (Pt/C, Pt-500, and Pt-700) were treated with PS-SH in a tetrahydrofuran solution for 12 h. As reference, the small thiol-containing molecule, dodecanethiol (C 12 H 25 SH), was also used to treat catalysts under the same conditions. As expected, for Pt/C without carbon coating, both C 12 H 25 SH and PS-SH can easily poison Pt NPs as indicated by the fully deactivated ORR activity ( Figure S8a, Supporting Information). Likewise, both Pt-500 and Pt-700 were completely deactivated upon treatment with C 12 H 25 SH ( Figure S8b,c, Supporting Information), indicating that their carbon shells are permeable to C 12 H 25 SH molecules under the aforementioned treatment conditions. In the case of poisoning with PS-SH, however, Pt-500 and Pt-700 showed quite different anti-deactivation behaviors. Specifically, Pt-500 largely retained the ORR activity after PS-SH treatment ( Figure S8b, Supporting Information), indicating that the carbon shells are impermeable to the bulky PS-SH molecules. In contrast, treatment of Pt-700 with PS-SH led to an apparent degradation in the catalytic activity ( Figure S8c, Supporting Information). This implies that the carbon shells of Pt-700 have wider porous channels compared to Pt-500, which allow the partial penetration of PS-SH molecules. These results not only reveal the porosity difference of carbon shells between various catalysts, but also demonstrate that the porosity of the ligand-derived carbon shells, which is a key factor dictating the ORR activity of Pt NPs, can be tailored by simply controlling the annealing temperature. It is likely that with the increasing of the annealing temperature the graphitic and rigid carbon shells conformally coated on Pt NPs can no longer maintain a complete and continuous spherical geometry due to the high curvature, with multiple discontinuous segments becoming favorable as depicted in Figure S8d (Supporting Information). Such fractures and defects could www.advmat.de www.advancedsciencenews.com provide porous channels for accessing the encapsulated Pt NPs, which accounts for the higher ORR activity exhibited by Pt-700.
To test the stability of the graphitic-shell-coated Pt catalysts under the realistic operating conditions for PEMFCs, MEAs were fabricated and tested. [56,57] As shown in Figure 4a, the performance of the MEA with Pt/C degraded rapidly with the increased ADT cycles, whereas the Pt-700 cell demonstrated an exceptional durability (Figure 4b), consistent with RDE tests. [58] Interestingly, the Pt-700 cell showed slightly improved performance after the first 5k cycles (both under air and 100% O 2 ). While it may be anticipated that this was due to oxidation of the carbon shell leading to more exposed Pt sites, [18] there was no noticeable increase in ECSA following 5k cycles. Thus, the increase in performance is more likely a result of surface oxidation leading to enhanced proton transport within the catalyst layer. The variation of the maximum power density during the ADT was also compared. While the Pt-700 cell displayed a smaller initial maximum power density relative to the Pt/C cell due to the lower ECSA (Table S1, Supporting Information), it could maintain a more stable power density, which started to exceed that of the Pt/C cell after 10k cycles (Figure 4c). Furthermore, the Pt-700 cell exhibited a small voltage loss of only 11 mV at an operating current density of 0.8 A cm −2 after 20k cycles (Figure 4d), whereas the cell voltage of Pt/C decreased by 75 mV under the same cycling conditions. The polarization curves under 100% O 2 are also shown in Figure S9 (Supporting Information), which help to highlight that the performance loss is kinetic (as would be expected from this particular ADT) for both Pt-700 and Pt/C cataltysts, and the superior durability of Pt-700 versus Pt/C.
It is well known that decreasing the size of Pt NPs will dramatically increase the ECSA, which is advantageous for enhancing the ORR activity. [52] However, as the particle size decreases below ≈3 nm, the catalyst particles begin to demonstrate unfavorable ORR kinetics as a result of unfavorable crystal structures dominating the surface ("particle size effect"). Unfortunately, Pt particles of this size are also still prone to rapid dissolution. [50] This reality has historically meant that there is an unfortunate inverse relationship between mass activity and stability that must be balanced by MEA designers when selecting a catalyst. To determine whether the ligand carbonization strategy presented here is applicable to small Pt particles, we synthesized OAm-capped Pt NPs with a diameter of ≈3 nm ( Figure S10a, Supporting Information). Using the same protocol developed for 5 nm-Pt NPs, we obtained 3 nm-Pt-700 catalysts by thermally treating 3 nm-Pt NPs supported on KB (Pt loading ≈ 20 wt%) at 700 °C under N 2 protection. As evidenced by TEM ( Figure S10b, Supporting Information) and XRD ( Figure S10c, Supporting Information), no apparent particle agglomeration was observed in 3 nm-Pt-700 catalysts. HRTEM reveals that the 3 nm-Pt NPs are also homogeneously coated with approximately two graphic layers derived from the OAm ligands ( Figure S10b, Supporting Information, inset). Importantly, the 3 nm-Pt-700 catalyst exhibited a high ORR activity and stability as indicated by polarization curves (Figure S10d, Supporting Information) and CV (Figure S10e, Supporting Information). As expected, the ECSA of 3 nm-Pt-700 is much higher than that of its 5 nm-Pt counterpart (79.22 m 2 g −1 Pt vs 34.82 m 2 g −1 Pt , Table S1, Supporting Information), due to the smaller size of Pt NPs. Moreover, 3 nm-Pt-700 only showed small degradation in ECSA and MA (at 0.9 V) after 20k cycles ( Figure S10f, Supporting Information). It is thus confirmed that our in situ ligand carbonization strategy is general and can be exploited to boost the ORR stability of small-sized Pt catalysts,  www.advmat.de www.advancedsciencenews.com which could greatly expand the design abilities for commercial MEA developers. In addition to the small Pt NPs, this in situ ligand carbonization strategy is also applicable to Pt-based alloy NPs. As shown in Figure S11 (Supporting Information), thermal annealing of PtPd NPs (≈4 nm in diameter) supported on KB at 700 °C produced graphitic-shell-coated PtPd catalysts with an excellent ORR activity and stability (Table S1, Supporting Information). [59,60] Moreover, this ligand carbonization strategy can further be applied to improve the stability of commercial Pt/C catalysts. Surface modification was realized by mixing Pt/C catalysts (J&M, HISPEC3000, particle size ≈ 3 nm, Pt loading ≈ 20 wt%) with OAm ligands in isopropanol followed by stirring overnight. Subsequent ligand carbonization at 700 °C produced Pt/C-700 catalysts. TEM (Figure 5a,b) and XRD (Figure 5c) verify that the size of Pt NPs was almost unchanged upon annealing, while HRTEM reveals that the original bare Pt NPs was uniformly coated with bilayered graphitic shells derived from the OAm ligands (Figure 5b, inset). The ECSA of the pristine Pt/C catalyst was reduced by ≈10% upon ligand carbonization (62.72 m 2 g Pt −1 vs 56.45 m 2 g Pt −1 , Table S1, Supporting Information), indicating that over-coating Pt NPs with bilayered graphitic shells only blocked a fraction of surface sites. [61] Moreover, compared with the unmodified Pt/C ( Figure S12, Supporting Information), Pt/C-700 catalysts exhibited significantly improved stability as indicated by the negligible activity degradation after 20k cycles (Figure 5d,e). The ECSA and MA loss of Pt/C-700 catalysts was also provided in Figure 5f, where it is apparent that the ligand-derived graphitic shell greatly reduces the rate of loss in ECSA and MA compared with the untreated Pt/C.

Conclusion
We have developed a facile and effective method for preparing stable graphitic-shell-coated Pt catalysts by harnessing the native ligands tethered to the surface of colloidal Pt NPs. Simple thermal annealing of Pt NPs converts the hydrocarbon ligands into bilayered, conformal graphitic shells, which are sufficiently robust to protect Pt NPs against migration, agglomeration, and dissolution while largely maintaining the accessibility of surface sites. The ORR activity of the resulting carbon-coated Pt catalysts depends heavily on the thermal annealing temperature, which dictates the microstructure of graphitic shells. In particular, the graphitic-shell-coated Pt catalysts resulting from annealing at 700 °C show superior long-term stability to those of commercial Pt/C catalysts. Importantly, this in situ ligand carbonization strategy is general and can also be applied to improve the stability of commercial Pt/C catalysts, thus opening a new route of developing highly durable ORR catalysts for PEMFCs.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.