Noble-Metal based Metallic Glasses as Highly Catalytic Materials for Hydrogen Oxidation Reaction in Fuel Cells

Electro-catalyst design with superior performance and reduced precious metal content (compared to state-of-the-art Pt/C) has been a challenge in proton exchange membrane fuel cells, preventing their widespread adoption. Metallic glasses have recently shown promising performance and large electrochemical surface area in catalytic reactions. The electro-catalytic behavior of recently developed Pt-, Pd-, and Pt/Pd-based metallic glasses was evaluated in this study using scanning electrochemical microscopy. The influence of chemistry and electronic structure on catalytic behavior was studied using scanning kelvin probe technique. The work function for the metallic glasses was lower by 75 mV to 175 mV compared to pure Pt. This resulted in higher catalytic activity for the amorphous alloys, which was attributed to the ease of charge transfer on the surface. The binding energy for the metallic glasses, measured using X-ray photoelectron spectroscopy, was higher by 0.2 eV to 0.4 eV. This explained easier removal of adsorbed species from the surface of amorphous alloys. The synergistic effect of Pt and Pd in alloys containing both the noble metals was demonstrated towards hydrogen oxidation reaction.


noble-Metal based Metallic Glasses as Highly catalytic Materials for Hydrogen oxidation Reaction in fuel cells Vahid Hasannaeimi & Sundeep Mukherjee
electro-catalyst design with superior performance and reduced precious metal content (compared to state-of-the-art pt/c) has been a challenge in proton exchange membrane fuel cells, preventing their widespread adoption. Metallic glasses have recently shown promising performance and large electrochemical surface area in catalytic reactions. the electro-catalytic behavior of recently developed pt-, pd-, and pt/pd-based metallic glasses was evaluated in this study using scanning electrochemical microscopy. The influence of chemistry and electronic structure on catalytic behavior was studied using scanning kelvin probe technique. The work function for the metallic glasses was lower by 75 mV to 175 mV compared to pure Pt. This resulted in higher catalytic activity for the amorphous alloys, which was attributed to the ease of charge transfer on the surface. the binding energy for the metallic glasses, measured using X-ray photoelectron spectroscopy, was higher by 0.2 eV to 0.4 eV. This explained easier removal of adsorbed species from the surface of amorphous alloys. The synergistic effect of Pt and Pd in alloys containing both the noble metals was demonstrated towards hydrogen oxidation reaction.
Fuel cells offer very high energy density among energy storage/conversion devices and are classified primarily based on their electrolyte or operation temperature 1 . Proton exchange membrane fuel cells (PEMFCs) are particularly attractive because they use hydrogen as fuel and operate at low temperatures (50 to 100 °C). PEMFCs have emerged as a promising power source for transportation and portable energy storage and are likely to replace aging alkaline fuel-cell technology used in space shuttles 2 . However, one of the major bottlenecks in widespread use of PEMFCs has been the high cost of platinum group metals (PGM) used as catalysts. Pt-and Pd-based alloys including Pt-Ru 3 , Pt-Sn 4 , Pt-Pd 5 , Pd-Co 6 , Pd-Au 7 , Pt-Ru-Mo 8 and Pt-Ru-Ni 9 have shown significantly higher electro-catalytic activity in both acidic and alkaline media compared to pure noble metal catalysts, allowing for reduced PGM loading. This has been attributed to the reduction in binding energy when the noble metals are alloyed with more oxophilic elements. Nevertheless, it should be noted that either too strong or too weak M-H (M stands for metal surface) binding energy may result in poor catalytic behavior 10 . Furthermore, there is limited understanding and open questions related to the mechanism of oxidation/reduction and its relationship with electronic characteristics such as binding energy and work function (WF) 11 . According to the free energy of activation (ΔG*) for proton (H + ) reduction, work function maybe directly correlated with the strength of M-H bond. However, the reported trends are inconsistent or even contradictory 12,13 .
A major concern in PEMFCs is the degradation and dissolution of catalysts and carbon support under fuel cell operating conditions 14 , which determines a fuel cell performance in addition to catalyst chemistry. The majority of investigations on electrocatalysts have been limited to crystalline materials 5,15,16 . Only recently, bulk metallic glasses (BMGs) have attracted attention as highly active electro-catalysts owing to the high density of low-coordination sites on their surface, metastable amorphous structure, and high durability [17][18][19][20] . However, the effect of different amorphous chemistries on catalytic activity is not well understood and has not been reported. Catalytic reactions mainly consist of diffusion, adsorption, and desorption of chemical species along with charge transfer process. Work function (WF), which is the minimum energy required for removing an electron from a material's surface, plays a key role in the charge transfer process and shown to be correlated with catalytic activity 21,22 . Scanning kelvin probe (SKP) may be used for measuring the contact potential difference (CPD)

Results and Discussion
The experimental setup and a schematic showing the array of amorphous alloys and pure Pt and Pd mounted side by side in a non-conductive resin are shown in Fig. 1(a). The tip was moved laterally over the surface of the specimen at a fixed distance of 50 µm from the surface. Energy diagrams of the tungsten tip and BMG substrate separated at a certain distance (d tip/s ) are shown schematically in Fig. 1(b). The vacuum levels are the same while the Fermi levels are different for each material 28 . When the tip and BMG sample are connected electrically, electron flow results in Fermi energy alignment and the system reaches equilibrium. However, a contact potential difference (V CPD ) is formed between the tip and sample as shown in Fig. 1(c). The external opposite potential (V SKP ) that nullifies V CPD ( Fig. 1(d)) is proportional to the relative work function difference between the tungsten tip and BMG sample (V SKP ~ Ф BMG -Ф tip , in which Φ represents the work function).
SKP analysis was performed with a constant tip to specimen distance of 50 µm. The potential map and a single line scan of V SKP are shown in the Fig. 2(a). Positive values for the BMGs indicate their higher work function there is no electrical contact between the tip and substrate; (c) tip and the substrate are connected electrically while a contact potential difference (V CPD ) is generated; and (d) an external electrical bias is applied to nullify V CPD , which is the SKP potential (V skp ). E v is the vacuum energy level, and E F(t) and E F(BMG) are Fermi levels of the tip and the BMG, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ . Pure Pt showed the highest work function followed by pure Pd and the amorphous alloys. The BMGs with higher fraction of noble metal showed higher work function. A roughly linear correlation between composition and work function has been reported for binary alloys of AgAu 29 , PtRh and NiCu 30 , showing an increasing trend with increase in the fraction of higher work function element. A similar trend was observed for the BMGs, as shown in Fig. 2(b). The alloys containing both Pt and Pd showed the lowest work function values, ~150 mV lower SKP potential compared to pure Pt. The rate of metal-catalyzed reactions in solid oxide fuel cells was shown to depend exponentially on average catalyst WF over a wide range of work functions (0.3-1 eV) 31 . In the present study, WF for all BMGs was found to be lower compared to pure Pd and Pt. Higher electronic conductivity results in higher reaction rates over the catalyst surface due to easier electron transfer 32 . It has been reported that the work function of contaminated Ni films decreased noticeably with hydrogen adsorption on its surface which resulted in increase of conductivity 33 . Similarly, the ability of Pt-based nanoparticle catalysts to activate molecular oxygen has recently been shown to have an inverse correlation with WF 26 . As mentioned in the Materials and Methods section, surface roughness does not affect the work function values from SKP measurement because of surface topography correction. A high resolution topography history of the samples was initially recorded using constant height mode of the SKP microscope. The results were calibrated using the topography history to ensure that they were attributed only to the work function difference between the tip and the sample. The average surface roughness (R a ) was calculated using Scanning Probe Microscopy to be 322.5 ± 20.5 nm for all specimens.
To determine the possible correlation between lower work function of BMGs and their binding energies, the chemical valence state of the samples were obtained using XPS. The recorded Pt 4 f and Pd 3d core levels are shown in Fig. 3. The core levels for Pt and Pd are separated into Pt 4f 7/2 , Pt 4f 5/2, Pd 3d 5/2 , and Pd 3d 3/2 , respectively, due to spin orbital splitting. The binding energy (BE) values at 71.3 eV and 74.6 eV correspond to Pt 4f 7/2 and Pt 4f 5/2 , respectively, which are attributed to Pt°. The absence of other Pt valence states such as Pt 2+ indicates that Pt is not present in the form of oxide or hydroxide on the surface, but in its zero-valent state. Interestingly, the BE of Pt for the Pt 57 -BMG showed a positive shift of + 0.4 eV with respect to pure Pt. The change in BE of Pt in the BMGs is likely related to the difference in electronegativity of the alloying elements (Cu, Ni and P). Moreover, it has been shown that the shift in BE is typically correlated to the change in d-band center, Fermi energy, and work function 34 . The lower work function for the BMGs correspond to the positive shift of Pt bond strength in these metallic glasses. The upshift of bond strength in Pd-and Pt-based metallic glasses is indicative of a downshift in d-band center 34 . Any shift in d-band center for the Pt surface would result in a surface core-level shift in the same direction 35 . The reduction in density of states at the Fermi level due to electron transfer from less electronegative elements to Pt or Pd has been reported to reduce the bond strength of Pt and adsorbed chemical species, leading to improved electro-catalytic activity of Pt-M (M = Ru, Sn) alloys 36 . In the present study, the chemisorption of protons on the Pd-and Pt-based amorphous alloy surface would be weakened since a higher d-band center corresponds to weaker M-H bonding 37 . Majority of the Pt is present in zero-valent state for all the BMG specimens. This may result in enhanced catalytic activity as zero-valent Pt provides more active sites 38 . A small difference (~0.2 eV) in Pt BE is seen among the different BMG compositions (the inset of Fig. 3(a)). A similar trend for Pd 43 Table 1. The BMGs showed higher binding energy of Pt and Pd in addition to reduced work function.
To investigate the electro-catalytic activity of the amorphous alloys with respect to pure noble metals, SECM imaging was employed in an electrolyte of 0.01 M H 2 SO 4 mixed with argon saturated 0.1 M Na 2 SO 4 . Hydrogen www.nature.com/scientificreports www.nature.com/scientificreports/ oxidation reaction was studied as it is the primary anodic reaction in PEMFC. Cyclic voltammetry at the Pt UME was recorded in bulk solution in the potential range of −1.2 V to +0.1 V (vs Ag/AgCl). At potentials higher than −0.4 V, the tip current approached zero (Fig. 4(a)). Since, hydrogen evolution reaction (HER) potential is roughly around −0.25 V 39 , hydrogen would be oxidized in this potential range (V > −0.4 V). Due to large over-potential of Pt tip for HER, significant reduction of proton was not observed until the potential reached values below −0.8 V. The tip current reached a steady state in the potential range of −0.9 V to −1.2 V (vs. Ag/AgCl), with mass transport current reaching a value of −900 nA. Further negative tip potentials (V < −1.2V) led to the reduction of bulk water followed by formation of H 2 bubbles. Approach curves were measured at various tip/substrate separations over the electrically conductive substrate and neighboring non-conductive resin. The tip and substrate were held fixed at −1.2 V and open circuit potential (OCP), respectively. The measured current was normalized with respect to the tip current measured in bulk solution. A positive feedback was seen at the UME tip ( Fig. 4(b)) and the i tip /i bulk increased as the tip approached the BMG surface due to catalytic oxidation of locally generated hydrogen (H 2 ). In contrast, a negative feedback was observed when the Pt tip approached the nonconductive resin surface. The reactivity of BMGs towards HOR was imaged by performing cyclic voltammetry. Depending on the tip/substrate distance, the samples showed different feedback towards HOR (Fig. 4(c)). At large tip/sample separation (shown schematically as d 1 in Fig. 4(c)), proton reduction occurred at the tip and there was no feedback from the substrate. With reduction of tip-substrate separation (shown schematically as d 2 in Fig. 4(c)), hydrogen oxidation from Pt 57 -BMG surface resulted in positive feedback. Hydrogen oxidation enhanced proton concentration in the vicinity of the Pt UME and increased the measured current at the tip as shown in Fig. 4(d). The recorded tip currents were normalized with respect to the measured values in bulk solution (Fig. 4(d)). A uniform tip response was observed when the tip was far from the substrate (d > 500 μm), with a nearly fixed current at the bulk diffusion-limited value = ( ) . At this separation, no feedback from the substrate was observed. The tip current significantly increased as the tip/substrate separation was reduced due to positive feedback from the BMG surface. Maximum tip current was recorded when the tip/substrate distance was ~5 µm. About twelve-fold increase in the tip current was observed for Pt 57 -BMG at 5 µm tip/sample separation as compared with bulk solution. The increase in tip current response was similar for Pd 43 -BMG, but the magnitude was lower (nine-fold increase in this case as shown in Fig. 4(e)). Although the electronic properties of Pd and Pt are comparable, hydrogen adsorption is stronger on Pd as compared with Pt, resulting in lower catalytic currents for Pd and its alloys 10 . The cyclic voltammetry plots for the other amorphous alloys were similar and are not shown.
To measure the kinetics of catalytic reaction, the Pt tip polarized at −1.2 V was scanned over all the alloys at a fixed tip/substrate separation of 5 µm. No over-potential was applied to the catalysts during the experiments. The variation in Pt tip current is proportional to the catalytic rate constant of the amorphous alloys. More negative values correspond to higher reactivity of the substrate. The plotted SECM map and the corresponding Pt tip current line scan are shown in Fig. 5(a). The variation in UME tip current was solely from the difference in chemistry of the BMGs since all the alloys were fully amorphous and had the same surface finish. Pure Pt and Pd specimens showed lower catalytic reactivity towards HOR compared to the amorphous alloys. Both the Pd 43 -and Pt 57 -BMGs exhibited higher activity compared to their pure counterparts, with the UME tip reduction current increase of 30 nA and 20 nA, respectively. The amorphous alloys containing both Pd and Pt exhibited the highest activity towards HOR among the systems studied. Higher specific activity (kinetic current per unit surface area of a catalyst) has recently been shown for Pd-Pt nano-dendrites towards ORR, which was related to the preferential exposure of more active crystalline facets (Pt(111) is more favored in the presence of Pd) on the Pt branches as compared with small Pt nanoparticles in Pt/C 40 . This synergistic effect has been attributed to the modification of Pt d-band center and weakening of electronic interaction between Pt atoms and oxygenated species. Most of the reports on synergistic effects of Pt/Pd towards different catalytic reactions have been attributed to crystallographic modification of Pt. However, BMGs have a homogenous amorphous structure without any crystalline facets 41 . The high density of low coordination sites in the amorphous alloys likely resulted in enhanced surface reations and higher catalytic activity 42 .
In order to evaluate the stability of the amorphous catalysts towards HOR, chronoamperometry was performed for all the samples. All the alloys reached steady-state current densities within ~200 sec (Fig. 5(b)). However, the current densities for the amorphous alloys containing both Pd and Pt were significantly higher compared to others, which may be attributed to the synergistic effect between these elements. The trend in chronoamperometry results were along the same lines as the SECM activity map. Modification of electronic structure may account for the better catalytic performance of Pt/Pd-based BMGs. Work function values for BMGs containing both Pd and Pt were lower compared with Pd 43 -and Pt 57 -based BMGs, promoting easier electron transfer. This also explains the shift in binding energies to higher values reducing the chemisorption of absorbed species for Pt/Pd alloys 34 .

Catalysts
Binding energy (eV) Pt 4f 7/2 4f 5/2 Pd 3d 5/2 3d 3/2 www.nature.com/scientificreports www.nature.com/scientificreports/ conclusions Scanning electrochemical microscopy was utilized to evaluate the catalytic performance of several Pd-and Pt-based amorphous alloys towards hydrogen oxidation reaction, the main anodic reaction in PEM fuel cells. The metallic glasses showed higher catalytic activity compared to pure Pt and Pd, which was attributed to their  electrochemical measurements. To measure relative work function of the specimens, SKP analysis was performed in dry lab air. A pure tungsten microprobe (150 µm in diameter) was used as reference electrode vibrating at a frequency of 80 Hz with an amplitude of 30 µm normal to the surface. A backing potential was applied to minimize the capacitance between probe and working electrode. A constant tip to substrate distance of 50 µm was used for the experiment. A high resolution topography history of the samples was initially recorded using constant height mode (CHM) of the SKP microscope (VersaSCAN-Princeton Applied Research) with step size of 20 µm over an area of 2 mm × 25 mm. The SKP potential map was recorded over all the samples with a fixed step size, potential gain, time constant and sensitivity. Finally, the results were calibrated using the topography history to ensure that they were attributed only to the work function difference between the tip and the amorphous alloys.
For measuring the electrocatalytic activity of the BMGs, high resolution SECM (VersaSCAN-AMETEK) was used in redox competition mode. The key components of the system included a potentiostat (VersaSTAT 3 F) coupled with a positioning system, an ultramicroelectrode (UME) and BMG sample as the working electrode. The tip/sample separation was optimized at 5 µm, and kept constant for all catalytic reaction studies using software based tilt correction. The catalytic performance measurement was done over an area of 2 mm × 25 mm, with step size of 10 µm at a fixed tip/substrate separation of 5 µm. Ag/AgCl/3 M KCl (E = 0.196 V vs SHE) and a Pt wire were used as the reference and counter electrodes, respectively. A Pt UME with a radius of 10 µm encased in borosilicate glass tube was used as SECM tip. The RG ratio (where RG is the ratio of insulating sheath radius divided by the Pt tip radius) of the probe was ~10 with a disk-in-plane geometry. The UME was polished gently with Alumina slurry, carefully washed with distilled water, and pre-treated with potential cycling in 0.1 M H 2 SO 4 before use. All cyclic voltammetry experiments were performed at a potential sweep rate of 50 mVs −1 and room temperature.
For determining the optimum distance between the tip and substrate, approach curves were plotted using Pt UME as the working electrode, while the insulator resin and conductive BMG alloys were employed as substrates. The distance at which the tip current changed to 75% of bulk solution current was chosen as the desired tip/ substrate separation. All electrochemical experiments were conducted in aqueous solution containing H 2 SO 4 and Na 2 SO 4 , deaerated with argon gas prior to measurement. To investigate the stability of the alloys towards HOR, chronoamperometry was performed in a three electrode cell set up with Pt wire and Ag/AgCl electrodes as counter and reference, respectively, at a scan rate of 10 mV/s for 1200 sec. The measured current was normalized by the surface area of the specimens.

Data Availability
All data generated or analysed during this study are included in this published article.  Table 2. The composition, glass transition temperature (T g ), crystallization temperature (T x ), and the difference between glass transition and crystallization temperatures (ΔT) for the BMGs.