Abstract
In this study, to enhance the stability of the cathode platinum (Pt) catalyst in polymer electrolyte membrane fuel cells, cerium oxide (CeOx) was deposited by plasma-enhanced atomic layer deposition (PEALD) process on the Pt catalyst sputtered on the cathode. A change in the peak power density loss after an accelerated stress test (AST) during I-V measurement of the membrane-electrode assembly according to the number of cycles was observed, which confirmed stability improvement. In polymer electrode membrane fuel cells (PEMFCs), free radicals lead to degradation of the performance and stability of catalysts; we used CeOx to prevent these problems. CeOx acts as a free radical scavenger through the redox reaction of Ce3+/4+ ions in the cell test and prevents oxidative hydroxyl and hydroperoxyl radical attack created in the reaction between hydrogen peroxide and released cations. By preventing oxidation, the stability was improved without decreasing the performance. Therefore, the improvement of stability through plasma-enhanced atomic layer deposition CeOx encapsulation can be considered a promising strategy for PEMFC catalysts.
1. Introduction
Polymer electrolyte membrane fuel cells (PEMFCs), spotlighted as next-generation power generators, are energy conversion devices that convert chemical energy into electrical and thermal energy through electrochemical reactions. This energy can be utilized for transportation mobility, such as in automobiles, ships, and drones. The advantages of these PEMFCs are their ecofriendliness, high power density, efficient energy conversion, low exhaust emission, and fast start-up time [1–4]. The performance of PEMFCs is greatly influenced by the oxygen reduction reaction (ORR) at the cathode, which is the slowest reaction [5]. The cathode of a PEMFC uses nanoscale platinum- (Pt-) based catalysts. Nevertheless, easy aggregation is a problem because Pt has a high surface energy and operates under high oxygen concentrations, high potential (0.6–1.2 V), and low pH (<1). Accordingly, the electrochemical surface area of the Pt catalyst is reduced, resulting in the deterioration of the catalyst performance and long-term stability. Therefore, a method to prevent such aggregation of the cathode Pt catalyst in PEMFCs is necessary [5–10].
Many studies have been conducted to improve the stability of Pt catalysts, particularly, by adding various metal oxide-based materials such as zirconium oxide, titanium oxide, and cerium oxide, or by manufacturing a core-shell type catalyst [11–17]. Gatto et al. reported that the treatment of the catalyst layer with TiO2 not only suppressed carbon corrosion but also minimized the extent of reduction in the ECSA [18]. In addition, Shroti and Daletou reported that a Pt-Co alloy electrocatalyst exhibited improved performance and stability and improved ORR activity [19]. Coating the catalyst surface with metal oxide is a suitable method to effectively stabilize the catalyst because it lowers the surface energy of the catalysts [20–23]. The protective layer should be very thin to avoid interference with the activity of the catalyst; the effect can be maximized when the porous electrode is uniformly coated. In general, after adding a metal oxide material, it must undergo transformation into a porous structure again through high-temperature heat treatment; however, this is disadvantageous as the process is complex and expensive. If the surface treatment of the catalyst is conducted through an ultrathin protective layer through atomic layer deposition (ALD), a uniform protective layer can be effectively combined with Pt without additional heat treatment [7, 24–26]. ALD has the advantage of controlling the thickness and composition of the coating layer by atomic layer through a self-limiting reaction and uniformly depositing the coating layer over a large area along a complex shape [27–31]. CeOx acts as a free radical scavenger through the redox reaction of Ce3+/4+ ions by preventing radical attack from hydrogen peroxide (H2O2), oxidative hydroxyl (HO·), and hydroperoxyl (HOO·). Hence, stability is improved through effective oxidation prevention for Pt catalysts [7, 32–36]. In this study, ALD CeOx encapsulation was formed on a porous Pt catalyst electrode in a PEMFC to confirm the stabilization effect of the Pt catalyst. In addition, various structural component analyses such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), and X-ray photoelectron spectroscopy (XPS) were conducted. Furthermore, we attempted to elucidate the optimal ALD process conditions for maximizing fuel cell performance via membrane electrolyte assembly (MEA) and single-cell fabrication and testing.
2. Methods
2.1. Manufacturing of Pt-CeOx Cathode
Pt was deposited on a carbon material gas diffusion layer (GDL) with an area of (Sigracet 39BB, SGL Carbon, Germany) by DC sputtering (Korea Vacuum Tech., South Korea). The DC electrical power was set at 31 W, with atmospheric pressure of 90 mTorr, and Pt was deposited for 23 min. Thereafter, CeOx fine particles were uniformly deposited on sputtered Pt using PEALD which was performed in a customized reactor (ICOT ALD, South Korea). Tris(isopropylcyclopentadienyl)cerium(III) (Ce(iPrCp)3) (Ichems, South Korea) was selected as the precursor of Ce. Also, O2 plasma treatment was performed as the oxidant. The chamber temperature was 200°C, and the precursor was vaporized at 165°C. The O2 plasma power was set to 250 W. An Ar gas flow of 15 sccm was used as the transport and purging gas, and the chamber base pressure was set at 10 mTorr. Three types of Pt-CeOx cathodes were manufactured by experimenting with this process in units of 10 cycles up to 30 cycles. Then, to prepare the electrolyte solution, Nafion ionomer suspension (EW1100, 10 wt%, DuPont, USA) and isopropyl alcohol (IPA) were mixed in a volume ratio of 1 : 19, stirred with a magnetic stirrer at 600 rpm for 30 min, and then again stirred in a sonication bath at 60 Hz for 30 min. We applied 40 μL of the prepared ionomer suspension to the GDL with micropipette on which sputtered Pt-CeOx was deposited. Then, it was dried in an oven at 50°C for 24 h to complete the electrode fabrication process, which is illustrated in Figure 1.

2.2. Manufacturing Pt Anode
The Pt anode was manufactured by DC sputtering using the same Pt deposition method as for the Pt cathode; 40 μL of the ionomer suspension was dropped by pipetting on the sputtered Pt-deposited GDL. Subsequently, it was dried in an oven at 50°C for 12 h to complete the electrode fabrication process.
2.3. Assembling Single Cell
MEA comprised an anode electrode, electrolyte, and cathode, and a Nafion® 212 membrane (DuPont Co., USA) was used as the electrolyte membrane. The active area of MEA is 0.8075 cm2. For single-cell production, four cathodes were used: bare Pt and Pt-CeOx, 10 cycles, 20 cycles, and 30 cycles. For manufacturing a single cell after MEA manufacture, a torque driver (Tohnichi, Japan) was used to increase the torque by 50 cN·m and tighten it with a force of 200 cN·m.
2.4. Fuel Cell Measurements
To test the electrochemical characteristics of the manufactured MEA, current-voltage (IV) and cyclic voltammetry (CV) were measured, and an accelerated stress test (AST) was performed using a fuel cell test station (CNL Energy, South Korea) and the ZIVE SP2 (Alpha Omega Electronics, Spain) electrochemical workstation. For the single-cell test, the cell was set to 80°C and relative humidity 100% (RH 100%). For the I-V measurements, 200 sccm of ultrahigh purity gas (>99.999%) was supplied to both H2 and O2, the scan rate was set to 5 mV/s, and the current change in the voltage range from 0.1 V to OCV was measured. Cell activation was performed by repeatedly sweeping 0.55 V to 0.6 V until the current was saturated. For the CV test, 100 sccm of H2 was supplied to the anode, and 5 sccm of N2 was supplied to the cathode, and the current change was measured under the conditions of voltage sweeping from 0.05 V to 1.5 V. In the case of the AST, voltage cycling of 0.6 V for 3 s and 0.95 V for 3 s was conducted for 30 k cycles according to the US-DOE electrocatalyst AST protocol [37]. At this time, 100 sccm of H2 and 30 sccm of N2 were injected to the anode and cathode sides, respectively. Performance tests for the MEA have been performed before AST, with , , and cycles (with denoting 1,000).
The morphology of the Pt-CeOx catalyst electrode fabricated using the PEALD process was examined by TEM, SEM (S-4800, S-4800, Hitachi, Japan), and scanning transmission electron microscopy (STEM; NEO ARM, JEOL, Japan). The electrode surface composition was analyzed by XPS (K-Alpha+, Thermo Fisher Scientific, USA). In addition, the surface area, pore volume, and pore size were analyzed using a Brunauer–Emmett–Teller (BET) analyzer (3Flex, Micromeritics, USA), and the loading amount of Pt catalyst was measured by thermogravimetric analysis (TGA; Q600, TA Instrument, USA). The performance and stability test of MEA was conducted in the atmosphere of 80°C at 10°C/min.
3. Results and Discussion
Figure 2 shows STEM EDS composition mapping of the Pt-CeOx catalysts for each ALD CeOx cycle, which confirms that ALD CeOx was uniformly deposited along the surface of Pt in proportion to the number of cycles. In particular, we confirmed that the thickness of the CeOx film was uniformly coated on the pore surface up to the lower part of the Pt column. As shown in Figures 2(a) and 2(b), CeOx uniformly penetrates the lower part of the porous Pt layer up to 20 cycles of ALD. Conversely, per Figure 2(c), we confirm that the pores of the Pt layer are blocked and local agglomeration occurs in ALD for 30 cycles or more. Table S1 shows the atomic% of Ce and Pt in the ALD cycle. Hence, we can see that the Ce content increases in proportion to the ALD cycle.

(a)

(b)

(c)
Figures 3(a)–3(c) present HRTEM images depicting the shape of ALD CeOx deposited on the Pt catalyst layer. From Figure 2, we can confirm once again that the ALD CeOx layer was uniformly deposited along the Pt shape. In addition, the growth rate was found to be approximately 0.28 nm/cycle, which is within the range of 0.25–0.3 nm/cycle, which is the growth per cycle (GPC) of ALD CeOx deposition using the previously reported Ce(iPrCp)3 precursor [38]. The ALD GPC of ALD CeOx is shown in Figure S1. The fast Fourier transformation (FFT) depicted (Figure 3(d)) was obtained from the HRTEM image using the Gatan digital micrograph program. When the distance is measured from the diffraction patterns, we can see that 0.31 nm 1/nm and 0.27 1/nm are consistent with the lattice of the HRTEM image [39].

(a)

(b)

(c)

(d)
The analysis of the CeOx layer composition was conducted by using XPS. As shown in Figures 4(a) and 4(c), it is possible to compare the distribution of electron binding energy in the Ce 3d orbital according to ALD CeOx 10–30 cycles. The Ce3+6 peaks and Ce4+4 peak spectra are consistent with the XPS reference for Ce 3d [40]. The ratio of Ce3+ to Ce4+ ions can be used to confirm the free radical scavenging effect of CeOx. In general, when the concentration of Ce3+ ions is higher than that of Ce4+ ions, surface oxygen vacancies increase in the lattice [32], as shown in Figure 4 ((a) 56% : 44%, (b) 51% : 49%, and (c) 38% : 62%). This indicates that oxygen vacancies are relatively low in the ALD CeOx 30 cycles. Figure 4(d) presents the XPS survey data showing the impurity concentration when ALD CeOx was deposited on a silicon wafer. As such, we can see that the impurities (F1s, S2p) from ALD CeOx were deposited with a high purity of 5% or lower.

(a)

(b)

(c)

(d)
Figure 5 shows the I-V curves with different numbers of ALD CeOx cycles at an operating temperature of 80°C. In the case of the bare sample, the peak power density (PPD) was confirmed to be approximately 600 mW/cm2, and it was confirmed to exhibit little decrease in performance after 20 cycles of CeOx deposition. However, compared with the bare sample at approximately 500 mW/cm2, for 30 cycles or more of CeOx, we see that the initial performance decreased significantly by 17%. This indicates that the thicker the CeOx protective layer, the greater the deactivation of the Pt catalytic region leading to decrease in performance. Through the STEM mapping shown in Figure 2, we could confirm that CeOx penetrated and deposited uniformly at lower cycles, but at 30 cycles or more, the pores were blocked and local agglomeration occurred. This is also observed in the BET analysis shown in Table S1. In addition, as shown in Figure 5, compared with the initial performance of the bare sample after AST 30 k cycles, the PPD decreased by approximately 42% (Fresh PPD: 599.1 mW/cm2, AST 30 k PPD: 350.5 mW/cm2), with the PPD loss of 20% was achieved after over 20 cycles of CeOx deposition (fresh PPD: 595.4 mW/cm2, AST 30 k PPD: 451.7 mW/cm2). The PPD and AST loss values for each sample are listed in Table S2. Based on these results, we confirmed that 20 CeOx deposition cycles are the optimal processing condition to ensure stability and output maintenance. In addition, as shown in Figure S2, the EIS data before and after 30 k AST cycles of the bare sample and the CeOx 20 cycle sample were analyzed. The polarization resistance of the bare sample was 0.45 and 0.89 Ω·cm2, while that of the CeOx-treated sample was 0.39 and 0.66 Ω·cm2 before and after the AST, respectively. The difference between the polarization resistance of the two samples before the AST was not significant. However, after the AST, the polarization resistance of the bare sample increased by 97%, whereas that of the CeOx-treated sample increased by 69%. Thus, the CeOx treatment was effective in improving the PEMFC durability.

(a)

(b)

(c)

(d)
Figure S3 illustrates the hydrogen monolayer adsorption analysis to obtain the ECSA (m2/gpt) of the bare sample and CeOx deposition over 20 cycles [41]. Thus, we can see that the normalized ECSA after >20 CeOx deposition cycles improved by approximately 11% after AST 30 k cycles compared with that of the bare sample (Figure S3(d)). This confirms that CeOx encapsulation can effectively stabilize Pt catalysts. This is also consistent with the BET surface area trend data presented in Table S3.
Figure 6 shows images of (a) bare Pt and (b) 20 cycles Pt-CeOx after Pt sputtering on the GDL, indicating the appearance of particles in which Pt agglomeration occurs before and after AST and depicts Pt dissolution in the electrode after AST. Figure 6(a) shows that in the bare case, extremely severe agglomeration and dissolution occurred as compared to the CeOx 20 cycles shown in Figure 6(b). Thus, ALD CeOx treatment prevents agglomeration of Pt and prevents dissolution through the structural anchoring effect. In addition, for quantitative analysis, TGA was performed, as shown in Figure 6(c). TGA analyzed the difference in the degree of Pt dissolution by analyzing the difference in the amount of Pt loading in (a) and (b) before and after AST 30 k, as shown in Figure S4. We see that the bare Pt loading decreased from 0.157 mg/cm2 to 0.128 mg/cm2 after AST 30 k, whereas in the case of the electrode treated with CeOx 20 cycles, the decrease was relatively small from 0.156 mg/cm2 to 0.145 mg/cm2 after AST. In the case of the catalyst treated with CeOx 20 cycles, the catalyst dissolution rate was only 7% after AST 30 k, but in the case of the bare catalyst, the catalyst dissolution rate was 18.5%. These results confirm that ALD CeOx treatment can effectively prevent Pt dissolution.

(a)

(b)

(c)
4. Conclusion
In this study, we performed 10–30 cycles of ALD CeOx encapsulation to enhance the stability of the Pt cathode. By analyzing TEM, the ALD CeOx layer formed to a thickness of 10 nm or less on the porous structure of the Pt catalyst. MEA was fabricated by sputtered Pt treated with ALD CeOx. Performance and stability test was performed through a single-cell test. As a result, we found that the performance of the fuel cell catalyst was maintained until the number of ALD cycles reached 20 and decreased beyond 30 cycles. As confirmed by HRTEM, the catalyst was deactivated by blocking the porous structure of Pt in ALD CeOx for 30 cycles or more. However, the stability of all the electrodes treated with CeOx encapsulation was improved. This is because ALD CeOx does not improve the catalyst performance, but as it is a free radical scavenger, the oxidation numbers of Ce3+ ions and Ce4+ ions change reversibly. This is because the stability is improved by preventing radical attack by H2O2, HO·, and HOO·. Thus, the ALD treatment was confirmed to be effective in stability evaluation using the AST method. As a result, after 30 k cycles of AST, power density of untreated MEA decreased by 42%, while the ALD CeOx 20 cycles sample showed only a 20% decrease in power density.
Data Availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Disclosure
A preprint has previously been published [42].
Conflicts of Interest
The authors declare no competing financial interest.
Authors’ Contributions
Dong Joon Kim and Heon Jun Jeong contributed equally to this work.
Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science and ICT, South Korea) (No. NRF-2019R1A2C2003054). Also, this work was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean government (Ministry of Trade, Industry and Energy) (20213030030040)
Supplementary Materials
Table S1: EDX quantitative analysis of PEALD Pt-CeOx 10, 20, and 30 cycles. Figure S1: thickness of CeOx layer by the number of ALD cycles. Table S2: peak power density and AST loss of bare and PEALD Pt-CeOx 10, 20, and 30 cycles. Figure S2: Nyquist plot in the bare sample and the CeOx 20 cycle sample under 0.6 V (a) before AST and (b) after 30 k AST cycles. Figure S3: results of the electrochemical characterization of MEAs by AST cycles: CV curve of MEA before and after AST 30 k cycles (a) bare, (b) CeOx 20 cycles, and (c) comparison of the change in the ECSA of bare MEA, 20 cycles MEA according to the AST cycles. (d) Normalized ECSA calculated from the hydrogen electroabsorption current of bare and CeOx 20 cycles MEA from the in situ CV curve. Table S3: BET surface area analysis of bare, ALD 10, 20, and 30 cycles. Figure S4: TGA profile of the electrode according to Pt loading in accordance with AST cycles of the (a) bare samples and (b) Pt-CeOx 20 cycles. (Supplementary materials)