Zhuojie
Xiao
a,
Hao
Wu
a,
Huichi
Zhong
a,
Ali
Abdelhafiz
*bc and
Jianhuang
Zeng
*ad
aSchool of Chemistry and Chemical Engineering, South China University of Technology; Guangdong Key Lab for Fuel Cell Technology, Guangzhou 510641, China. E-mail: cejhzeng@scut.edu.cn
bDepartment of Nuclear Science and Engineering, Massachusetts Institute of Technology, Massachusetts 02139, USA. E-mail: ali_m@mit.edu
cDepartment of Materials Science and Engineering, Massachusetts Institute of Technology, Massachusetts 02139, USA
dZhongke of Hydrogen Power Co. Ltd, Nansha District, Guangzhou 511458, China
First published on 12th July 2021
In electrochemical reactions, interactions between reaction intermediates and catalytic surfaces control the catalytic activity, and thereby require to be optimized. Electrochemical de-alloying of mixed-metal nanoparticles is a promising strategy to modify catalysts’ surface chemistry and/or induce lattice strain to alter their electronic structure. Perfect design of the electrochemical de-alloying strategy to modify the catalyst's d-band center position can yield significant improvement on the catalytic performance of the oxygen reduction reaction (ORR). Herein, carbon supported PtCu catalysts are prepared by a simple polyol method followed by an electrochemical de-alloying treatment to form PtCu/C catalysts with a Pt-enriched porous shell with improved catalytic activity. Although the pristine PtCu/C catalyst exhibits a mass activity of 0.64 A mg−1Pt, the dissolution of Cu atoms from the catalyst surface after electrochemical de-alloying cycling leads to a significant enhancement in mass activity (1.19 A mg−1Pt), which is 400% better than that of state-of-the-art commercial Pt/C (0.24 A mg−1Pt). Furthermore, the de-alloyed PtCu/C-10 catalyst with a Pt-enriched shell delivers prolonged stability (loss of only 28.6% after 30000 cycles), which is much better than that of Pt/C with a loss of 45.8%. By virtue of scanning transmission electron microscopy and elemental mapping experiments, the morphology and composition evolution of the catalysts could clearly be elucidated. This work helps in drawing a roadmap to design highly active and stable catalyst platforms for the ORR and relevant proton exchange membrane fuel cell applications.
Interestingly, the shift of the electronic band structure of alloy nanoparticles and weakening of the chemisorption energy of oxygenated species can also be achieved by using electrochemical de-alloying for forming a Pt-rich shell to exhibit compressive strain, due to the dissolution of non-noble metal atoms from the alloy surface layer. For instance, Strasser et al. reported that the de-alloyed PtCu nanoparticles demonstrated higher ORR activity originating from the controllable compressive strain formed in Pt-enriched surface layers via potential cycling.17 Stimulated by this finding, numerous studies have been dedicated to developing de-alloyed Pt-based nanoparticles, especially PtCu nanoparticles,22–25 in which Cu is removed from the alloy surface layers, forming a Pt-rich shell and a PtCu core structure with enhanced ORR activity compared with pure Pt nanoparticles. Generally, the electronic structure of a Pt-rich surface can be perturbed by a PtCu alloy core with differing composition or structure, which can further affect the adsorption energy of oxygenated species.26 Although, immense efforts have been previously devoted to improving the ORR performance of de-alloyed PtCu nanoparticles, the structural and componential observation, especially the change of lattice strain in the process of electrochemical de-alloying and accelerated durability test (i.e. ADT), is not perfectly clear to the best of our knowledge. Thus, the correlation among the lattice strain, composition and enhanced catalytic activity of the de-alloyed PtCu catalyst was investigated after different potential cycling in perchloric acid electrolyte.
Herein, we report the synthesis of de-alloyed PtCu/C catalysts with high electrocatalytic performance and low Pt-loading, which can be prepared via a simple polyol method followed by an electrochemical de-alloying treatment for improving the ORR activity. The de-alloyed PtCu/C-10 catalyst demonstrated enhanced ORR performance (1.19 A mg−1Pt) with 28.6% loss after 30000 cycles at 30 °C, surpassing the DOE 2020 target in both Pt mass activity and durability, compared with the Pt/C-10 catalyst (0.24 A mg−1Pt) with 45.8% loss. For PtCu/C catalysts, the compositional and morphological evolution was determined after potential cycling. A Pt-rich shell is formed and covers the PtCu alloy core. Moreover, it was found that the de-alloying treatment results in structural and lattice parameter changes which can be observed in the shift of the diffraction peak in the X-ray powder diffraction patterns. This study provides visual and valuable insight into the rational design of de-alloyed Pt-based electrocatalysts with a modified electronic structure.
Single cell tests were performed on a fuel-cell test system (Scribner Associates Inc., Model 850e). The catalyst ink (2 mg mL−1) was prepared by blending the catalyst, Nafion solution, H2O and isopropanol at an ionomer/carbone weight ratio of 0.5/1 and an isopropanol/H2O volume ratio of 1/4, and then sprayed on a Nafion 211 membrane.27 The active area was 25 cm2. The anode catalyst was 20 wt% Pt/C-Pristine and the cathode catalyst was Pt/C-10 or PtCu/C-10. The Pt loading of the anode and the cathode was fixed at 0.10 mgPt cm−2 and 0.20 mgPt cm−2, respectively. The single cell electrochemical characterization was analyzed at a cell temperature of 80 °C, H2/Air (Stoichiometric of 1.2/2.5) with 100% relative humidity (RH) and 150 kPa of back pressure. The accelerated durability tests (ADTsc) of single cells were conducted in the potential range of 0.6–1.1 V with the sweep rate of 100 mV s−1 for 10000 cycles at a condition of cathode feed of fully humidified N2.
TEM images (Fig. S1†) revealed that the PtCu/C nanoparticles were well dispersed on the carbon supports, similar to the commercial Pt/C catalyst. EDX elemental mapping showed a high entropic alloying of PtCu nanoparticles, where both Pt and Cu dispersion were even across the whole nanoparticle. This further indicated that Cu atoms were infused within the Pt lattice to form a uniform PtCu alloy, which agrees with XRD results as discussed earlier. Fig. 1b and c show atomic resolution STEM images of PtCu/C and Pt/C nanoparticles, where the lattice parameter exhibits lattice fringes of 0.220 and 0.221 nm, where the lattice parameter of PtCu is further compressed by 0.45% compared to the monometallic Pt/C catalyst.30,31
The chemical states of the surface component were determined by XPS to observe changes in the d-band center position, as an indication on strain induced on Pt atoms.32–34Fig. 1d shows that the Pt 4f7/2 peak shifts towards a higher binding energy (71.5 eV) compared with pure Pt (71.3 eV),35,36 revealing the downshift in the center of the d-band for PtCu/C-Pristine catalysts. As shown in Fig. 1e, the obvious satellite peak located at the higher binding energy of 942.1 eV corresponding to that of Cu 2p3/2 peaks, suggesting the presence of Cu(II) species.37 In addition, the Cu 2p peaks can be fitted by two peaks with a close binding energy, which are assigned to Cu(I) or Cu(0).38 Therefore, the Auger Cu LMM spectra were used to confirm the presence of Cu(0), Cu(I) or Cu(II) in the pristine PtCu/C-Pristine catalyst.39Fig. 1f shows the Auger spectrum of Cu LMM, where the peak deconvolution resulted in three sub-peaks at 913.4, 917.3 and 918.0 eV, corresponding to Cu(II), Cu(I) and Cu(0) states. Auger analysis indicated the co-existence of metallic and oxide Cu states. Cu oxide is believed to form naturally due to oxidation under ambient conditions, when Cu atoms at the surface were exposed to air. Thus, the thickness of the native Cu oxide layer was too thin to be detected by XRD.40 Based on different electrochemical dealloying conditions (i.e., no# cycles), we found that PtCu/C-10 showed the highest electrocatalytic activity, as displayed in Fig. S2 and Table S1.† Thus, the characterization data shown in the manuscript hereon will be dedicated to PtCu/C-10, as a representative sample for dealloyed catalysts. Fig. 1g shows Pt 4f duplets, where Pt 4f7/2 peaks shift towards a higher binding energy (71.7 eV) compared to that of the pristine PtCu/C-Pristine catalyst, implying an improvement in ORR performance due to the downshift of the Pt d-band center.41 Meanwhile, Cu 2p peak intensities became significantly weak, owing to the dissolution of Cu surface atoms under electrochemical de-alloying conditions. It is worth noting that the absence of satellite peaks and the Auger peak of Cu(II) and Cu(I) species indicates the formation of a Pt-rich shell, implying the dissolution of Cu at the surface under electrochemical de-alloying conditions. Electrochemical dealloying treatment did not affect the particles’ size or morphology, as can be seen in Fig. S1.†
In addition, Fig. 2 displays the HAADF-STEM image and the corresponding elemental mapping of the PtCu/C-10 catalyst. Imaging clearly revealed the formation of core–shell nanostructures with a Pt-rich shell and a PtCu alloy core. Compared with PtCu/C-Pristine, the composition of PtCu/C-10 was further determined by EDS spectra. As displayed in Fig. S3,† the results of the molar ratio in both Pt and Cu indicate that the Pt composition concentered at the surface of metallic nanoparticles, suggesting the presence of a Pt-rich shell.
The catalytic performance of PtCu/C-Pristine for the ORR was investigated in an O2-saturated 0.1 M HClO4 solution.42 As can be seen in Fig. 2b, the MA of Pt/C-Pristine is 0.23 A mg−1Pt which is three times lower than the MA of PtCu/C-Pristine (0.64 A mg−1Pt). Tafel slopes represented in Fig. 2c are −62 and −96 mV per decade in low and high overpotential regions, respectively, which implies that the charge transfer and removal of oxygenated intermediates are the rate-limiting steps.43,44 As shown in Fig. S2a,† under different de-alloying conditions, the CV scans of PtCu/C-5, recorded in a N2-purged 0.1 M HClO4 solution, displayed a weaker hydrogen under potential deposition wave (Hupd), compared with that of PtCu/C-10. This indicated the formation of a Pt-rich shell in PtCu/C-10 due to the dissolution of Cu atoms on the surface. The electrochemical surface area (ECSA) of all catalysts, calculated by integrating the area of Hupd peaks, is listed in Table 1. Furthermore, the calculated ratio of ECSACO:
ECSAHupd (Fig. S3†), a descriptor of surface structure in the case of the Pt alloy, is 1.47 on the de-alloyed PtCu/C-10, implying the nature of the Pt-rich shell structure.45 As is evident from Fig. 2e, the ORR activity of PtCu/C-10 manifests higher than that of Pt/C-10 (1.19 versus 0.24 A mg−1Pt). However, the MA of Pt/C before and after de-alloying, almost, remained intact. The de-alloyed PtCu/C-10 catalyst delivers a higher SA of 1.69 mA cm−2, with a 6.5-fold enhancement relative to the Pt/C-10 catalyst (0.26 mA cm−2). Furthermore, the ORR polarization curves of PtCu/C-10 demonstrated the half-wave potential (E1/2) at 30 mV more positive compared to Pt/C-10. In Fig. 2f, the higher kinetic current on the de-alloyed PtCu/C-10 catalyst indicated that the PtCu/C-10 catalyst maintains a higher ORR performance across a wide range of potentials.46 Such an enhanced electrocatalytic activity is attributed to the modified electronic structure of the Pt-rich shell and the optimal lattice compression. Both effects would affect the bonding strength between Pt and oxygenated species.47,48 Compared with PtCu/C-Pristine and PtCu/C-5, a large number of active sites were formed on the surface of PtCu/C-10 (higher ECSA) after cycling. However, gradual cycling resulted in the further loss of Cu, so that the lattice parameter of PtCu was similar to that of Pt/C, indicating the degradation of catalytic performance. To further understand the effects of morphological changes on ORR activity, the loss ratio of Cu of PtCu alloy nanoparticles was investigated after different voltage cycles. ICP-OES experiments were conducted to determine the atomic ratio of Cu, as shown in Table S2.† At the 5th voltage cycle, the Cu fraction in PtCu/C dropped significantly from 58.2% to 37.2% due to the dissolution of outer layer's Cu atoms, in agreement with the XRD and XPS analyses. After further electrochemical de-alloying and stability tests, a slight decay in Cu atomic% was observed, indicating that interior Cu atoms can be greatly preserved intact from dissolution.
Samples | ECSA (m2 g−1) | ORR at 0.9 V | ||
---|---|---|---|---|
H upd | MA (A mg−1Pt) | MA loss (%) | SA (mA cm−2) | |
Pt/C-10 | 90.3 | 0.24 | N/A | 0.27 |
Pt/C-10 after 5 K | 77.0 | 0.23 | 4.2 | 0.30 |
Pt/C-10 after 15 K | 69.9 | 0.19 | 20.8 | 0.27 |
Pt/C-10 after 30 K | 49.2 | 0.13 | 45.8 | 0.26 |
PtCu/C-10 | 70.5 | 1.19 | N/A | 1.69 |
PtCu/C-10 after 5 K | 52.0 | 1.13 | 5.0 | 2.17 |
PtCu/C-10 after 15 K | 41.5 | 1.05 | 11.8 | 2.53 |
PtCu/C-10 after 30 K | 39.9 | 0.85 | 28.6 | 2.13 |
Apart from excellent ORR activities, the stability of catalysts is a crucial factor affecting PEMFCs’ lifetime. Consequently, the accelerated durability tests were performed to evaluate the stability of the dealloyed PtCu/C-10 and Pt/C-10 catalysts. In Fig. 3a, CV scans obtained using PtCu/C-10 showed a significant loss of ECSA after 30000 cycles (from 70.5 to 39.9 m2 g−1Pt), likely due to the aggregation of PtCu nanoparticles.49 Nevertheless, no obvious change was observed in LSV polarization curves (a slight negative shift of E1/2 from 0.90 to 0.89 V), as displayed in Fig. 3b. It should be noted that after 30
000 cycles, a slight change in the amount of copper was observed for PtCu/C-10 (from 35.8% to 26.9%), suggesting that a Pt-rich shell can perhaps protect the interior Cu atoms from leaching.50 After 30
000 cycles, the MA of the Pt/C-10 catalyst showed a loss of 45.8%, which is much worse than that of 28.6% for the PtCu/C-10 catalyst (listed in Table 1).
![]() | ||
Fig. 3 Cyclic voltammograms and ORR polarization curves for PtCu/C-10 (a–b) and Pt/C-10 (c–d) during ADT. Insert: Pt mass activities and specific activities at 0.90 V (vs. RHE). |
TEM analysis was conducted to observe the morphological and compositional changes to obtain more insight into the reasons for the electrocatalytic activity loss on the nanoscale.51 It can be observed from Fig. 4 that the morphology of all catalysts after ADT showed obvious aggregation, which agrees with ECSA analyses discussed earlier. Most remarkably, TEM images of Pt/C after ADT showed a dramatic increase in nanoparticles size compared to the PtCu/C-10 catalyst. Furthermore, the change in the composition of PtCu/C-10 catalysts after ADT is revealed in Fig. 5, and the atomic ratios of Pt and Cu were also detected and are displayed in Fig. S4.† From the elemental mapping analysis and EDS spectra, it can be seen that there is an obvious change in the atomic ratio of Cu. In Table 2, the atomic concentrations of Pt in the surface composition increased after ADT, while the atomic concentration of Cu decreased. The observation in composition suggests that the Pt-rich shell gradually thickened during ADT. Fig. 6a shows a schematic image of the formation of a thicker Pt-rich shell. Therefore, it can be supposed that the decrease in electrocatalytic performance of PtCu/C-10 catalysts mainly resulted from the aggregation and relieved lattice compression due to the thicker Pt-rich shell.
![]() | ||
Fig. 4 TEM images of PtCu/C-10 after 5 K (a), 15 K (b) and 30 K (c), as well as Pt/C-10 after 5 K (d), 15 K (e) and 30 K (f), Insert: the corresponding particle size distribution histogram. |
![]() | ||
Fig. 5 HAADF-STEM overview images and EDS-HAADF-STEM element mapping images of PtCu/C after 5 K (a–d), 15 K (e–h) and 30 K (i–l). |
Samples | Atomic concentration (at%) | |
---|---|---|
Pt | Cu | |
PtCu/C-Pristine | 17.2 | 82.8 |
PtCu/C-10 | 58.5 | 41.5 |
PtCu/C-10 after 5 K | 60.3 | 39.7 |
PtCu/C-10 after 15 K | 67.5 | 32.5 |
PtCu/C-10 after 30 K | 67.6 | 32.4 |
Single cell performances of PtCu/C-10 and Pt/C-10 catalysts before and after durability tests are shown in Fig. 6. In Fig. 6b, the PtCu/C-10 catalyst provided higher catalytic performance than Pt/C-10. This suggests that the existence of Cu in the cathode catalyst layer may enhance the proton conductivity of the electrode, thus improving the charge transfer performance of the cell.52 The PtCu/C-10 catalyst exhibited a maximum power density of 1.00 W cm−2, whereas the Pt/C-10 showed 0.95 W cm−2. At 0.8 V, the current densities were 0.35 and 0.30 A cm−2 for PtCu/C-10 and Pt/C-10 catalysts before ADTsc, respectively. Consistent with the observation from half-cell tests, the PtCu/C-10 catalyst delivered higher current density compared to Pt/C-10. After ADTsc, the current density and maximum power density were 0.33 A cm−2 and 0.85 W cm−2 for PtCu/C-10, and 0.29 A cm−2 and 0.83 W cm−2 for Pt/C-10. The loss of the maximum power density for PtCu/C-10 is 15.0%, which is much worse than that of 12.6% for the Pt/C-10 catalyst. Furthermore, a clear drop in the polarization curve of the PtCu/C-10 catalyst after ADTsc was observed in the low current density region. This suggests that the dissolved Cu ion may reduce the proton conductivity due to contaminating the membrane,53 thus indicating the necessity of the de-alloying process. The performance of Pt/C-10 exhibited a sharp decline in the high current density region, presumably due to the aggregation of Pt nanoparticles.54 Therefore, de-alloyed PtCu/C catalysts provide an ideal pretreatment method to improve the activity and durability of Pt based alloy catalysts for PEM fuel cell applications.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1nr02820k |
This journal is © The Royal Society of Chemistry 2021 |