Stabilising Oxide Core – Platinum Shell Catalysts for the Oxygen Reduction Reaction

J. C. Daviesa, B. E. Haydena,b* and L. Offina

aIlika Technologies Ltd, Kenneth Dibben House, Enterprise Road, University of Southampton Science Park, Chilworth, Southampton, SO16 7NS, UK.

brian.hayden@ilika.combChemistry, University of Southampton, Southampton, Hampshire, SO17 1BJ, UK.

beh@soton.ac.uk

Keywords: Core-Shell; Fuel Cells; Oxygen Reduction Reaction; Platinum; Doped Metal Oxides; High-throughput

Abstract

Thin film planar models of core-shell catalysts for the oxygen reduction reaction in PEM fuel cells, have been synthesised with tin doped titania as the support for platinum. High-throughput methods have been used to investigate the effect that the level of tin doping in the titania core supporting material and the degree of crystallisation have on the loading of platinum at which bulk platinum like oxygen reduction behaviour is achieved. At high effective coverages of platinum, the oxygen reduction activity is always similar to that of a polycrystalline platinum electrode. This suggests that a continuous platinum thin film can always be formed at higher loadings. As the loading of platinum is decreased on all of the support materials studied, the activity eventually decreases, corresponding to effective platinum coverages that form nucleated particles on the support. The effective platinum coverage at which a continuous layer of platinum is formed, and exhibits the high activity, depends strongly on the support. The critical effective coverage, below which the oxygen reduction activity decreases, is ca. 5ML on anatase, and slightly higher on the amorphous phase of titania. Doping the titania with tin 10 at.% > Sn > 0 % initially results in an increase in this critical effective coverage. At higher tin concentrations this critical coverage reduces again, and for supports with 23 at.% < Sn < 40 at.% , only ca. 1.6ML is required. Therefore, these catalysts have the highest mass activity. Stability cycling of these catalysts also shows that they are the more stable supported systems. It is also shown that the doped titania support material itself is stable in a sulphuric acid environment at 80°C for Sn < at 28%. These results

1

suggest that active and stable titania supported catalysts can be formed on tin doped supports with concentrations in the range 28 at.% > Sn > 23 at.% .

*Corresponding Author

beh@soton.ac.uk; +44 (0)2380 592776

Chemistry, University of Southampton, Southampton, Hampshire, SO17 1BJ, UK.

2

1. Introduction

Platinum and platinum rich alloys remain the most efficient catalysts for facilitating the oxygen reduction reaction (ORR) at the cathode of polymer electrolyte membrane fuel cell (PEMFC). The low abundance and high cost of platinum necessitate a high degree of dispersion as particles, traditionally on a carbon support, in order to maximise the reactive surface area and hence its mass normalised ORR activity. One of the limitations of this approach is that the platinum surface specific ORR activity decreases with particle size, as observed on both high area supported, and model catalysts: This results in an optimum in the ORR mass activity at platinum particle diameters of 3-4nm [1-4].

It also appears that as the size of platinum particles decreases, a larger loss in the electrochemical surface area of platinum is observed during potential cycling [5]. This may be associated with the intrinsic instability of the small platinum particles, or the oxidation of the carbon support under fuel cell operating conditions, leading to degradation of the catalyst and limiting the lifetime of the fuel cell [6]. Metal oxides have previously been investigated for use as alternatives to carbon as fuel cell catalyst supports for platinum particles [7-19]. Metal oxides are less prone to oxidative corrosion and have demonstrated improved stability over carbon in a fuel cell environment [7]. Disappointingly, the intrinsic activity of, for example, titania (and doped titania) supported platinum is reduced, resulting in a large increase in the overpotential for the ORR [8, 20]. This may be associated with the spill over of oxygen containing species from the support to the platinum, blocking active sites for oxygen reduction [8, 21]. Complete coverage of the titania by a continuous layer of platinum in an effective core shell structure would prevent spill over of oxygen and prevent poisoning of ORR on the platinum particle.

Since a continuous surface of Pt (probably the Pt(111) surface) appears to exhibit the highest specific ORR activity [22], and as optimising the surface area to mass of platinum remains a necessity, core shell structures offer a potential solution, and a range of core shell platinum catalysts are reported in the literature [18, 19, 23-34]. In addition to the increased mass activity resulting from a thin shell, structural modification of the platinum by the core has also been suggested as a tool for a further increase in mass activity through an increase in the specific activity of the platinum [35, 36]. Many of these systems contain another precious metal in the core material, such as Au or Pd [23-26, 28-30, 32]. These systems may offer a reduction in platinum loading, and a facile route to the electrochemical deposition of the platinum shell, but do not contribute significantly to an overall cost reduction. Alternatively, base metal cores such as Cu [31] offer a cost reduction, but are significantly less stable in the oxidising environment of the fuel cell cathode.

Metal oxide cores for platinum potentially provide a stable and cheaper solution to increase the utilisation of precious metals without sacrificing activity or stability. While such structures are claimed in a number of patents [18, 19, 27] for fuel cell applications, the claims are not exemplified. To the best of our knowledge, there are limited reports of their successful application in the electrochemical environment in the literature. For example, SnO2@Pt [27] core-shell structures for ethanol oxidation, and Fe2O3@Pt core-shell structures for methanol [34] and ORR [37], have been

3

reported. The major challenges are to identify oxides that are stable in the electrochemical environment, provide sufficiently electrical conductivity to ensure connectivity of the catalyst, and exhibit surfaces that can be designed to interact strongly (wet) with platinum to ensure a continuous thin and stable shell.

We describe here the development and optimisation of a doped metal-oxide core / platinum shell system employing combinatorial synthesis and screening methods on a model planar catalyst: The catalyst is optimised for the minimum platinum shell coverage that sustains high activity and stability for ORR. This approach allows the assessment of how the concentration of a dopant, in this case Sn in the anatase phase of titania, influences core stability and conductivity, and the ability to stabilise ultrathin continuous layers of platinum which exhibit high ORR activity. A schematic of the planar thin film model, and the corresponding core-shell is shown in Figure 1. There is selfevidently a limitation in using the planar model system for a core shel where the core will exhibit a curviture and hence exhibit a range of facets and edge sites for a crystalline oxide. Nevertheless, the polycrystalline planar oxides synthesised in this model investigation also exhibit a range of these facets and defects, and therefore represents an excellent starting point for optimisation. The advantage of this model approach is that the oxide substrate, the interface and the platinum coverage are well controlled, and high-throughput methods can be applied to determine optimal compositions for active and stable structures. A schematic of the 10x10 electrochemical screening chip on which the library of electrocatalysts are synthesised and electrochemically screened is also shown in Figure 1A.

Figure 1 Schematic of (A) the electrochemical screening chip, (B) the thin film planar catalyst structure showing the compositional gradient of the Sn dopant in TiO2 (core) and the platinum

coverage variation (shell) designed to model (C) the core-shell catalyst.

Traditionally, dopants have been added to metal oxides in order to increase conductivity by creating defects (such as oxygen vacancies). Whilst Sn is not a dopant in the traditional sense when incorporated in titania (i.e. both Ti and Sn can form oxides in the +4 oxidation state) it has been previously demonstrated that SnO2

has an inherent conductivity due to the fact that it readily forms oxygen vacancies [38]. Tin oxide is also known to easily form oxygen vacancies at the surface [38]. Initial nucleation of small clusters of platinum appear to be sensitive to adsorption sites on TiO2(110) [39] and oxygen defects may therefore play a role in nucleation. However, this remains unclear since nucleation appears insensitive to the state of

4

reduction of vicinal surfaces of rutile [40]. The introduction of the tin dopant was used to potentially create surface oxygen defects, and hence induce more extensive nucleation, leading to a surface which will encourage platinum to form a thin encapsulating layer at a low coverage. We suggest that a continuous (encapsulating) thin film of platinum should approach the high activity of bulk platinum surfaces in oxygen reduction, in contrast to nucleated nano-particles on titania surfaces which exhibit low activity [8].

In this work, we use a thin film catalyst as a model for a core shell structure, and demonstrate that tin doping of the titania (core) can be optimised to promote the wetting of the titania surface, resulting in a continuous thin film growth of platinum at just a few monolayers of the metal. Bulk platinum-like activity for the ORR reaction is achieved for these ultra-thin layers of platinum.

2. Experimental

Thin film samples were deposited by elemental evaporation using a multi-source, high-throughput PVD (Physical Vapour Deposition) system described in more detail elsewhere [41]. Fixed wedge shutters partially shadowing individual sources are used to create compositional gradient thin films through the simultaneous deposition of the constituent elements. Tin doped titania was synthesised by evaporation of titanium (Ti slugs 99.98 %, Alfa Aesar) from an electron gun (E-gun) source and tin (Sn slugs 99.995 %, Alfa Aesar) from a Knudsen cell (K-cell) source. Oxygen was incorporated from a molecular oxygen source at ca. 5 x 10-6 Torr simultaneous to the atomic metal deposition. The compositional gradient thin-film tin doped titania oxides were subsequently heated in a tube furnace at 450 °C under a flow of oxygen for 6 hours in order to produce the crystalline anatase phase.

Platinum (99.95 % Umicore) was deposited onto the TiSnxOy compositional gradient thin films from an E-gun source. Samples were first heated up to, and kept at, 200 °C prior to and during platinum deposition in order to ensure de-hydroxylation of the doped oxide surface. In order to achieve a variation of platinum coverage across the sample, a shadow shutter near the substrate was moved during deposition to vary the effective deposition time (and hence effective coverage) in one direction on the sample. This method is described in more detail elsewhere where it was used to nucleate low coverages of platinum to form particle size variations over oxides [8, 20]. The direction of platinum coverage variation was chosen to be orthogonal to the compositional gradient of the dopant (Figure 1). The effective coverage of the deposited platinum was calibrated by first depositing thicker films onto silicon substrates, and measuring the coverage of the films by AFM (Atomic Force Microscopy) and producing a calibration curve against deposition time [20]. The equivalent coverages of platinum deposited were converted from nanometres to monolayers (ML) of platinum for reporting. An assumption that the platinum formed a face centred cubic structure was made and therefore one monolayer of atoms would have a coverage of 0.227 nm.

A range of oxide samples were synthesised on different substrates depending on the requirements of subsequent characterisation and electrochemical screening. These included continuous or masked thin films on silicon wafers (LA-ICPMS, XRD and

5

XPS), carbon-coated grids (TEM) and 10x10 electrode arrays of gold electrodes on an electrochemical high-throughput screening chip [42] for electrochemical measurements.

The relative composition of the metal components (dopant level) of the oxide compositional gradient film was determined using a Laser Ablation Inductively Coupled Plasma Mass Spectrometer (LA-ICPMS; New Wave 213 nm laser & Perkin Elmer Elan 9000). Composition measurements were made on samples deposited onto silicon wafers, and typically involved the measurement of an array of 10x10 points over the compositional gradient. The same deposition conditions were subsequently used to deposit onto equivalent electrochemical arrays.

X-Ray diffraction (XRD) patterns were obtained using the Bruker D8 Discover diffractometer, with two-circle goniometer with independent stepper motors and optical encoders for the Theta and 2 Theta circles. The D8 diffractometer system was equipped with a GADDS detector operating at 45 kV and 0.65 mA and a high intensity X-ray IμS Incoatec source (with Cu Kα radiation) allowing high intensity and collimated X-rays to be localised on thin film materials providing efficient high-throughput structural analysis. This analysis was carried out on oxide films deposited onto Si substrates. Equivalent XRD measurements have been carried out on samples deposited onto electrochemical array substrates (additional diffraction peaks of the Au are also observed on these substrates) and the crystallisation has been shown to be the same on the two substrates.

Stability tests have been carried out on the oxide films deposited on Si substrates. The samples were immersed in 200 mL of 0.1 M H2SO4 at 80°C for a period of 24 hours. Compositional analysis using LA-ICPMS was also carried out on these samples before and after this testing.

Transmission Electron Microscopy measurements were performed using a JEOL (JEM-3010) TEM with an accelerating voltage of 300 kV. All images shown are at a magnification of 2 x 105 and all scale bars represent a length of 20 nm. The samples were prepared for analysis by depositing various coverages of Pt onto fresh carbon coated TEM grids (for comparison with the oxide materials), and oxide covered TEM windows which had previously been treated at 450 °C in oxygen for 6 hours.

The high-throughput electrochemical screening chip (Figure 1A) incorporating 100 independently addressable electrodes arranged in a 10x10 array is employed for electrochemical measurements using multi-channel current followers and potentiostatic control. This allows a parallel screening of electrochemical behavior (cyclic voltammetry, potential step and steady-state measurements) of all the electrodes simultaneously in a single electrochemical environment [2, 42]. The geometric area of each individual working electrode on the electrochemical array is 1.0 mm2. The design of the electrochemical cell and socket assembly provides a clean electrochemical environment with control of the temperature during experiments. In the experiments described, the temperature was maintained at 25°C. The reference electrode was mercury/mercury sulphate (MMSE). The potential of the MMSE was measured vs. a hydrogen reference electrode prior to each high-throughput electrochemical screening experiment. All potentials are quoted vs. the reversible hydrogen electrode (RHE). The counter electrode was a

6

platinum mesh contained in a glass compartment that was separated from the working electrode compartment by a glass frit.

The electrolyte used for all experiments was 0.5 M HClO4 prepared from concentrated HClO4 (double distilled, GFS) and ultrapure water (ELGA, 18 M cm). The gases used (Ar, O2 and CO) were of the highest commercially available purity (Air Products). Unless stated otherwise, experiments were performed under an atmosphere of argon. Oxygen reduction experiments were performed under an atmosphere of O2. During potential step measurements, oxygen was bubbled through the electrolyte. Unless noted otherwise, the maximum potential applied to the electrodes was 1.2 V vs. RHE.

3. Results and Discussion

3.1 Anatase and Amorphous Titania Supported Platinum

To compare amorphous and crystalline un-doped titania as a support for Pt, separate electrochemical arrays were prepared. The as-deposited titania was amorphous. The titania was crystallised by heating in a tube furnace at 450°C for 6 hours under a flow of oxygen. These conditions were found to be optimal for the crystallisation of anatase, which was confirmed by XRD.

The activity of the array electrodes for the oxygen reduction reaction were determined by cycling the electrodes between 0 and 1.2 V vs. RHE in oxygen saturated 0.5 M HClO4 at 5 mv s-1. Some example (negative going) sweeps of the CVs for different equivalent coverages of Pt supported on amorphous titania are shown in Figure 2. As the equivalent coverage of Pt decreases below a critical equivalent coverage, the reduction wave is shifted negative in potential. This suggests that the electrodes are less active for the oxygen reduction reaction, i.e. a higher overpotential is required for the reaction to occur.

Figure 2 O2 reduction CVs in O2 saturated 0.5 M HClO4 at 5 mV s-1 (negative going sweep shown) for 10 different equivalent coverages of Pt supported on amorphous titania.

7

Figure 3A shows the voltammetry for one electrode during the oxygen reduction reaction with 0.7 ML equivalent coverage of Pt supported on amorphous titania, the two currents at which ignition potentials [43] have been measured are indicated (black horizontal lines). Two reduction features are seen, a small feature with a high onset potential and a larger feature with a low onset potential. These two features are probably due to inhomogeneity within the sample, i.e. some large islands of platinum leading to a high ignition potential and some smaller discrete particles leading to the lower ignition potential. The ignition potential measured at 1 µA raw current gives an indication of the onset of the first reduction feature and the ignition potential measured at 3 µA raw current gives an indication of the onset of the second reduction feature.

Figure 3B shows the potential at which 1 µA raw current is reached (on the negative going sweep, defined as the ignition or onset potential) during the oxygen reduction slow cycling experiment (i.e. cycling at 5 mV s-1 in oxygenated 0.5 M HClO4) for various equivalent coverages of Pt (average results taken across an array row (10 electrodes) for identical Pt equivalent coverages) for Pt supported on amorphous and anatase titania. This gives a measure of the onset potential of the oxygen reduction reaction (the higher the onset potential the more active the catalyst for the oxygen reduction reaction). At high equivalent coverages of Pt the ignition potential remains fairly constant with decreasing equivalent coverage of Pt, and similar for both supports. However, between 5 and 6 ML (monolayers) equivalent coverage of Pt, the ignition potential starts to decrease on both support materials. The effect on the amorphous material is more significant. Under 6 ML equivalent coverage for both supports, the ignition potential decreases further (on the amorphous titania at low equivalent coverages there is a large amount of scatter in the data). Figure 3C shows the potential at which 3 µA raw current is reached in the oxygen reduction slow cycling experiment for the same set of data. At this current it is even clearer that the reduction wave is shifted more significantly negative on the amorphous support, suggesting that the anatase support provides better wetting for the platinum and therefore higher activity down to a lower equivalent coverage (and therefore loading) of Pt.

These results suggest that the platinum shows slightly better wetting on the anatase titania allowing bulk platinum like oxygen reduction activity to a lower equivalent coverage than on amorphous titania.

Figure 3 A) A typical O2 reduction slow cycling CV (0.7 ML of Pt supported on amorphous TiOx) indicating the two different potentials where ignition potentials have been extracted. B) Potential at

which 1 µA raw current is reached during the oxygen reduction slow cycling (5 mV s-1 in oxygenated

8

0.5 M HClO4) experiment for Pt supported on amorphous and anatase titania. C) Potential at which 3 µA raw current is reached during the oxygen reduction slow cycling experiment for amorphous and anatase titania. The error bars correspond to ± one standard deviation averaging over 10 identical

supported catalysts on the 10x10 array.

3.2 Tin Doped Anatase Supported Platinum

Compositional gradient TiSnOx films were deposited at room temperature, and subsequently heated to 450 °C for 6 hours under a flow of oxygen in a tube furnace: These conditions were found to be optimal for the formation of the anatase phase of TiO2, and were distinct from those used to produce doped rutile titania [20]. Figure 4 shows a series of XRD patterns obtained from compositional gradient films of TiSnOx

with varying Sn content deposited onto a silicon substrate (the background signal from the silicon substrate has been subtracted) after heating in the furnace in oxygen, and compared to a standard anatase TiO2 XRD pattern. At the limit of 0at.% Sn, the diffractogram corresponds to that of the anatase crystal structure. As the concentration of Sn in the film increases, the peaks associated with the anatase TiO 2

structure decrease in peak intensity, with concomitant broadening: The anatase structure is retained at lower Sn concentrations, with some evidence of a reduction in crystallite size. For films with a Sn content greater than ca. 25 at.% Sn, additional broad peaks at approximately 2θ=26.6° and 2θ=33.9° are observed; These correspond to the (110) and (101) indices, respectively, of rutile cassiterite SnO2

[44].

Figure 4 XRD patterns of TiSnOx films deposited on silicon (Si background signal subtracted) deposited at room temperature and heated for 6 hours at 450 °C in oxygen with varying Sn content. A

reference anatase TiO2 XRD pattern is also included.

Figure 5 shows TEM images for an equivalent coverage of 3.2ML of platinum on anatase titania, and tin doped anatase titania (ca. 28at.% Sn). While it is relatively straight forward to image the growth of nucleated platinum particles on the anatase

9

surface, the Sn doped support results in significantly lower contrast of any nucleation behaviour. While it is difficult to conclude on the basis of these TEM images alone, we believe that these differences are associated with an increased nucleation density and a tendency towards layer by layer growth on the Sn doped substrates.

Figure 5 TEM images for an equivalent coverage of Pt of 3.2ML on A) anatase titania and B) tin doped anatase titania (ca. 28 at.% Sn). The scale bar corresponds to 20nm.

Figure 6 (A, B, C) shows a comparison of oxygen reduction slow cycling voltammetry (the second negative going sweep in O2 saturated 0.5 M HClO4 at 5 mV s-1) for the as-prepared catalysts for a series of equivalent coverages of platinum on titania supports with (A) high (37.3 at.%), (B) low (6.2 at.%) and (C) no tin in the supporting titania. With high equivalent coverages of platinum, all three catalysts exhibit an onset of oxygen reduction at ca. 0.9VRHE, i.e that expected for a continuous platinum layer. As the platinum equivalent coverage is reduced, the onset potential for oxygen reduction moves to lower potential (i.e. a greater overpotential for oxygen reduction). The platinum coverage at which this takes place, however, varies significantly between the catalysts: For the un-doped titania support (C) this starts at 4.7ML of platinum, and a significant further reduction is observed for 1.6ML of platinum. In the case of the 6.2 at.% Sn doped titania support, this reduction in potential is already evident at 7.9ML of platinum, with large further reduction for lower coverages. In the case of the 37.3 at.% Sn doped titania, only a very small lowering of the onset potential is observed, and that is at 1.6ML of platinum. Incorporation of these higher concentrations of Sn leads to a catalyst which at very low equivalent coverages of platinum exhibits the behaviour of a continuous supported platinum layer: effective wetting of the titania based substrate has apparently been induced by the incorporated tin.

To assess the stability of these titania supported catalysts, 200 cycles between 0.025 and 1.2 V vs. RHE at 100 mV s-1 were carried out in Ar purged 0.5 M HClO4. Figure 6 (A’, B’, C’) shows a comparison of oxygen reduction slow cycling voltammetry (the second negative going sweep in O2 saturated 0.5 M HClO4 at 5 mV s-1) for these stability cycled catalysts for the same series of equivalent coverages of platinum on titania supports with (A’) high (37.3 at.% Sn), (B’) low (6.2 at.% Sn) and

10

(C’) no tin in the supporting titania. It is striking that in the case of the highly doped titania support (A’), the onset for oxygen reduction has not changed from the as prepared catalysts (A) for all of the equivalent coverages of platinum, including the very low equivalent coverage of 1.6ML: Incorporation of these higher concentrations of Sn leads not only to a catalyst which at very low equivalent coverages of platinum exhibits the behaviour of a continuous supported platinum layer, but also induces a stability in this layer of platinum. This is in strong contrast to the lower Sn doped support (B’) or the un-doped support (C’) which, after stability cycling, exhibit significantly lower onset potentials in comparison to the as synthesised catalysts (B, C) for all but the very highest platinum equivalent coverages: platinum layers synthesised on these supports are much less stable, and appear to undergo restructuring, probably nucleation of the platinum into particles on the support, which are much less active than the continuous platinum layers.

Figure 6 The second negative going sweep of the oxygen reduction slow cycling experiment (in O2

saturated 0.5 M HClO4 at 5 mV s-1) for selected TiSnOx compositions with varying Pt equivalent coverage before stability cycling (A, B, C) and after stability cycling (A’, B’, C’).

The results in Figure 6, and similar sets of oxygen reduction slow voltammetry data corresponding to the complete set of catalysts (measured on three 10x10 electrochemical arrays) with varying Sn doping concentrations in the titania support, and for various platinum equivalent coverages, can be summarised to show the lowest equivalent coverage of Pt necessary to achieve a continuous platinum oxygen reduction activity. The latter is defined as catalysts that exhibit an ignition potential of 8.5VRHE at 1 µA raw current. These results are shown in Figure 7 for both the as synthesised catalysts (open circles), and for catalysts that have been stability cycled (200 cycles between 0.025 and 1.2 VRHE at 100 mV s-1 in Ar purged 0.5 M HClO4)

11

(filled circles). For both the synthesised and stability cycled catalysts, it is those which have a Snat.% > 23 which have formed an active (and continuous) layer of platinum at the lowest effective coverages measured (1.6ML). Higher coverages of platinum are required to achieve this activity for both the un-doped and Sn doped titania with Snat.% < 23. Stability cycling this latter set of catalysts also leads to an increase in the effective platinum coverages necessary to reach the high activity: We associate this change to a nucleation and growth of platinum particles during stability cycling, and a concomitant disruption of the continuous platinum layer.

Figure 7 The lowest Pt equivalent coverage necessary for the catalyst to exhibit an oxygen reduction ignition potential of 8.5VRHE at 1 µA raw current. The results are shown for platinum supported on TiSnyOx supports as a function of Sn at.% concentration. The open circles are the data for the as-

synthesised catalysts, and the filled points are for catalysts that have been stability cycled (200 cycles between 0.025 and 1.2 VRHE at 100 mV s-1 in Ar purged 0.5 M HClO4).

In order to assess the long-term potential stability of the Sn doped titania substrates, a range of titania thin film supports containing varying amounts of Sn (ca. 0 to 40at.% Sn) were combinatorially deposited on Si substrates. The samples (two 10x10 arrays) were subsequently immersed in 200 mL of 0.1 M H2SO4 at 80°C for a period of 24 hours. Photographs of the samples were obtained after 0, 2, 4, 6 and 24 hours (not shown). Only the supports with the very highest concentrations of Sn (> 32 at.%) exhibited a colour change of the material.

To quantify any changes in the support material, LA-ICPMS measurements were performed prior to and after the acid exposure to assess any compositional change resulting from preferential dissolution of either Sn or Ti. The results are summarised in Figure 8 and show the Sn concentration in the support before and after 24 hours exposure to 0.1 M H2SO4 at 80 °C. The results show that there is no relative loss of Sn versus Ti for TiSnOx supports with up to 28 at.% Sn. Above this value there is evidence of a slight relative loss of Sn: This corresponds well with the changes observed visually for supports with concentrations > 32 at.% Sn.

12

Figure 8 Comparison of the tin content (measured by ICP-MS) of tin doped titania supports deposited on silicon wafers before and after 24 hours of exposure to 0.1 M H2SO4 at 80 °C.

By combining the data from the electrochemical measurements of oxygen reduction activity, which show that the greatest mass activity is achieved for tin doped titania supported platinum where Sn > 23 at.%, and the results of the accelerated stability tests on the support which show that tin doped titania is stable when Sn > 28 at.% ,

one can conclude that a stable and highly active catalyst should be formed by supporting platinum on tin doped titania with tin concentrations in the range 28 at.% > Sn > 23 at.% .

4. Conclusions

Thin film planar models of core-shell catalysts have been synthesised with varying platinum loading and titania supports with varying crystallinity and levels of tin doping. These thin film models have been tested for their activity towards the oxygen reduction reaction, and their stability in an acid environment.

At high effective coverages of platinum, the ignition potential is always similar to that of a polycrystalline platinum electrode suggesting a continuous platinum thin film later can always be formed. As the loading of platinum is decreased on all of the support materials studied, the ignition potential for the oxygen reduction reaction eventually starts to decrease (i.e. the overpotential for the oxygen reduction reaction starts to increase, or the electrodes are less active for the oxygen reduction reaction). This corresponds to effective platinum coverages that form nucleated

13

particles on the support which, as in the case of amorphous titania [8] and doped rutile [20] supported platinum, have a low oxygen reduction activity.

The effective platinum coverage at which a continuous layer of platinum is formed, and exhibits the high activity, is shown to depend strongly on the support. The critical effective coverage, below which the oxygen reduction activity decreases, is ca. 5ML on anatase, and slightly higher on the amorphous phase of titania. Doping the titania with tin 10 at.% > Sn > 0 at.% initially results in an increase in this critical effective coverage. At higher tin concentrations this critical coverage reduces again, and for supports with 23 at.% < Sn < 40 at.% , this critical coverage is ca. 1.6ML. These are therefore the catalysts with the highest mass activity. Stability cycling of these catalysts also shows that they are the more stable supported systems. It is also shown that the doped titania support material itself is stable in a sulphuric acid environment at 80°C for Sn < 28 at.% . These results suggest that active and stable titania supported catalysts can be formed on tin doped supports with concentrations in the range of 28 at.% > Sn > 23 at.% .

The challenge now is to find a methodology of synthesising cores of TiSnOx and subsequently coating them with a few monolayers of platinum in a reducing environment, reproducing the oxide termination provided in this model study. Initial attempts to synthesise cores of ca. 10nm diameter using flow hydrothermal methods appear successful, with TiSnOx exhibiting comparable XRD patterns to that observed here on the model system as a function of Sn content.

Acknowledgements

This work was partly funded by the Carbon Trust.

14

References:

[1] M. Peuckert, T. Yoneda, R.A.D. Betta, M. Boudart, Oxygen Reduction on Small Supported Platinum Particles, Journal of The Electrochemical Society, 133 (1986) 944-947.[2] S. Guerin, B.E. Hayden, C.E. Lee, C. Mormiche, J.R. Owen, A.E. Russell, B. Theobald, D. Thompsett, Combinatorial Electrochemical Screening of Fuel Cell Electrocatalysts, Journal of Combinatorial Chemistry, 6 (2003) 149-158.[3] H.A. Gasteiger, S.S. Kocha, B. Sompalli, F.T. Wagner, Activity benchmarks and requirements for Pt, Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs, Applied Catalysis B: Environmental, 56 (2005) 9-35.[4] Z. Xu, H. Zhang, H. Zhong, Q. Lu, Y. Wang, D. Su, Effect of particle size on the activity and durability of the Pt/C electrocatalyst for proton exchange membrane fuel cells, Applied Catalysis B: Environmental, 111–112 (2012) 264-270.[5] R.K. Ahluwalia, S. Arisetty, J.-K. Peng, R. Subbaraman, X. Wang, N. Kariuki, D.J. Myers, R. Mukundan, R. Borup, O. Polevaya, Dynamics of Particle Growth and Electrochemical Surface Area Loss due to Platinum Dissolution, Journal of The Electrochemical Society, 161 (2014) F291-F304.[6] F.A. de Bruijn, V.A.T. Dam, G.J.M. Janssen, Review: Durability and Degradation Issues of PEM Fuel Cell Components, Fuel Cells, 8 (2008) 3-22.[7] S.-Y. Huang, P. Ganesan, S. Park, B.N. Popov, Development of a Titanium Dioxide-Supported Platinum Catalyst with Ultrahigh Stability for Polymer Electrolyte Membrane Fuel Cell Applications, Journal of the American Chemical Society, 131 (2009) 13898-13899.[8] B.E. Hayden, D. Pletcher, J.-P. Suchsland, L.J. Williams, The influence of support and particle size on the platinum catalysed oxygen reduction reaction, Physical Chemistry Chemical Physics, 11 (2009) 9141-9148.[9] S.-Y. Huang, P. Ganesan, B.N. Popov, Electrocatalytic activity and stability of niobium-doped titanium oxide supported platinum catalyst for polymer electrolyte membrane fuel cells, Applied Catalysis B: Environmental, 96 (2010) 224-231.[10] E. Antolini, E.R. Gonzalez, Ceramic materials as supports for low-temperature fuel cell catalysts, Solid State Ionics, 180 (2009) 746-763.[11] Y. Chen, J. Wang, X. Meng, Y. Zhong, R. Li, X. Sun, S. Ye, S. Knights, Atomic layer deposition assisted Pt-SnO2 hybrid catalysts on nitrogen-doped CNTs with enhanced electrocatalytic activities for low temperature fuel cells, International Journal of Hydrogen Energy, 36 (2011) 11085-11092.[12] X. Cui, F. Cui, Q. He, L. Guo, M. Ruan, J. Shi, Graphitized mesoporous carbon supported Pt–SnO2 nanoparticles as a catalyst for methanol oxidation, Fuel, 89 (2010) 372-377.[13] D.-J. Guo, J.-M. You, Highly catalytic activity of Pt electrocatalyst supported on sulphated SnO2/multi-walled carbon nanotube composites for methanol electro-oxidation, Journal of Power Sources, 198 (2012) 127-131.[14] J. Ye, J. Liu, Z. Zou, J. Gu, T. Yu, Preparation of Pt supported on WO3–C with enhanced catalytic activity by microwave-pyrolysis method, Journal of Power Sources, 195 (2010) 2633-2637.

15

[15] J.W. Weidner, B.L. Garcia, Electrocatalyst support and catalyst supported thereon US 2009/0065738 A1, United States, 2009.[16] M. Cai, Y. LU, Z. WU, L.L. Feng, M.S. Ruthkosky, J.T. Johnson, F.T. Wagner, Electrocatalyst Supports for Fuel Cells US 2007/0037041 A1, United States, 2007.[17] T.B. Do, M. Cai, M.S. Ruthkosky, Mesoporous electrically conductive metal oxide catalyst supports WO 2009/152003 A2, 2009.[18] R. Adzic, J. Zhang, M. Vukmirovic, Electrocatalyst for oxygen reduction with reduced platinum oxidation and dissolution rates US 2006/0263675 A1, United States, 2006.[19] R. Adzic, M. Vukmirovic, K. Sasaki, Synthesis of metal-metal oxide catalysts and electrocatalysts using a metal cation adsorption/reduction and adatom replacement by more noble ones US 7,704918 B2, United States, 2010.[20] I. Cerri, T. Nagami, J. Davies, C. Mormiche, A. Vecoven, B. Hayden, Innovative catalyst supports to address fuel cell stack durability, International Journal of Hydrogen Energy, 38 (2013) 640-645.[21] B.E. Hayden, D. Pletcher, J.-P. Suchsland, L.J. Williams, The influence of Pt particle size on the surface oxidation of titania supported platinum, Physical Chemistry Chemical Physics, 11 (2009) 1564-1570.[22] N. Markovic, H. Gasteiger, P.N. Ross, Kinetics of Oxygen Reduction on Pt(hkl) Electrodes: Implications for the Crystallite Size Effect with Supported Pt Electrocatalysts, J. Elecrochem. Soc., 144 (1997) 1591-1597.[23] K. Hartl, K.J.J. Mayrhofer, M. Lopez, D. Goia, M. Arenz, AuPt core–shell nanocatalysts with bulk Pt activity, Electrochemistry Communications, 12 (2010) 1487-1489.[24] Y. Ma, H. Zhang, H. Zhong, T. Xu, H. Jin, X. Geng, High active PtAu/C catalyst with core–shell structure for oxygen reduction reaction, Catalysis Communications, 11 (2010) 434-437.[25] K. Sasaki, H. Naohara, Y. Cai, Y.M. Choi, P. Liu, M.B. Vukmirovic, J.X. Wang, R.R. Adzic, Core-Protected Platinum Monolayer Shell High-Stability Electrocatalysts for Fuel-Cell Cathodes, Angewandte Chemie International Edition, 49 (2010) 8602-8607.[26] W. Shuangyin, K. Noel, J. Sanping, W. Xin, Controlled synthesis of dendritic Au@Pt core–shell nanomaterials for use as an effective fuel cell electrocatalyst, Nanotechnology, 20 (2009) 025605.[27] J.C.M. Silva, R.F.B. De Souza, L.S. Parreira, E.T. Neto, M.L. Calegaro, M.C. Santos, Ethanol oxidation reactions using SnO2@Pt/C as an electrocatalyst, Applied Catalysis B: Environmental, 99 (2010) 265-271.[28] R. Wang, H. Wang, B. Wei, W. Wang, Z. Lei, Carbon supported Pt-shell modified PdCo-core with electrocatalyst for methanol oxidation, International Journal of Hydrogen Energy, 35 (2010) 10081-10086.[29] R. Wang, H. Li, H. Feng, H. Wang, Z. Lei, Preparation of carbon-supported core@shell PdCu@PtRu nanoparticles for methanol oxidation, Journal of Power Sources, 195 (2010) 1099-1102.

16

[30] W. Wang, R. Wang, S. Ji, H. Feng, H. Wang, Z. Lei, Pt overgrowth on carbon supported PdFe seeds in the preparation of core–shell electrocatalysts for the oxygen reduction reaction, Journal of Power Sources, 195 (2010) 3498-3503.[31] Z.D. Wei, Y.C. Feng, L. Li, M.J. Liao, Y. Fu, C.X. Sun, Z.G. Shao, P.K. Shen, Electrochemically synthesized Cu/Pt core-shell catalysts on a porous carbon electrode for polymer electrolyte membrane fuel cells, Journal of Power Sources, 180 (2008) 84-91.[32] Y.-N. Wu, S.-J. Liao, Z.-X. Liang, L.-J. Yang, R.-F. Wang, High-performance core–shell PdPt@Pt/C catalysts via decorating PdPt alloy cores with Pt, Journal of Power Sources, 194 (2009) 805-810.[33] M. Lopez, M. Lennartz, D. Goia, C. Becker, S. Chevalliot, Core/Shell-type catalyst particles comprising metal or ceramic core materials and methods for their preparation, WO2008/025751, (2008).[34] Y.-T. Liu, Q.-B. Yuan, D.-H. Duan, Z.-L. Zhang, X.-G. Hao, G.-Q. Wei, S.-B. Liu, Electrochemical activity and stability of core–shell Fe2O3/Pt nanoparticles for methanol oxidation, Journal of Power Sources, 243 (2013) 622-629.[35] S. Suzuki, T. Onodera, J. Kawaji, T. Mizukami, K. Yamaga, Effect of support materials on platinum lattice strain and its oxygen reduction activity, Appl. Catal. A-Gen., 427 (2012) 92-97.[36] S.E. Temmel, E. Fabbri, D. Pergolesi, T. Lippert, T.J. Schmidt, Investigating the Role of Strain toward the Oxygen Reduction Activity on Model Thin Film Pt Catalysts, ACS Catal., 6 (2016) 7566-7576.[37] V.M. Dhavale, S. Kurungot, Tuning the Performance of Low-Pt Polymer Electrolyte Membrane Fuel Cell Electrodes Derived from Fe2O3@Pt/C Core-Shell Catalyst Prepared by an in Situ Anchoring Strategy, Journal of Physical Chemistry C, 116 (2012) 7318-7326.[38] M. Batzill, U. Diebold, The surface and materials science of tin oxide, Progress in Surface Science, 79 (2005) 47-154.[39] S. Gan, Y. Liang, D.R. Baer, A.W. Grant, Effects of titania surface structure on the nucleation and growth of Pt nanoclusters on rutile TiO2(110), Surf. Sci., 475 (2001) 159-170.[40] F. Rieboldt, S. Helveg, R. Bechstein, L. Lammich, F. Besenbacher, J.V. Lauritsen, S. Wendt, Formation and sintering of Pt nanopartictes on vicinal rutile TiO2 surfaces, Physical Chemistry Chemical Physics, 16 (2014) 21289-21299.[41] S. Guerin, B.E. Hayden, Physical Vapor Deposition Method for the High-Throughput Synthesis of Solid-State Material Libraries, Journal of Combinatorial Chemistry, 8 (2006) 66-73.[42] S. Guerin, B.E. Hayden, C.E. Lee, C. Mormiche, A.E. Russell, High-Throughput Synthesis and Screening of Ternary Metal Alloys for Electrocatalysis, The Journal of Physical Chemistry B, 110 (2006) 14355-14362.[43] H.A. Gasteiger, J. Garche, - Fuel Cells, (2008).[44] International Centre for Diffraction Data Database no. 41-1445, 2010.

17