Novel catalyst materials for the cathode side of MEAs suitable for transportation applications G.A. 303492 17.03.2016 Deliverable D4.6 WP4 Electrochemical characterization of catalyst materials and selected catalyst/support combinations and study of chemical/morphological changes after the electrochemical workload Deliverable: Evaluation of the Nanoplasmonic sensing method Description of Deliverable: A detailed evaluation of the Nanoplasmonic sensing method as an in situ, real time method to follow electrode processes in fuel cells PM for D4.6: 1 Nature: R Dissemination Level: PU Delivery Date: 31.12.2015 Lead Beneficiary: CUT Objectives of Work Development of methods for fabrication of samples for combined electrochemical and nanoplasmonic measurements. Development of setup and methods for performing long-term corrosion tests using combined electrochemical and optical measurements and analyzing the obtained information. Evaluation of the benefits gained using combined electrochemical and nanoplasmonic sensing methods to follow processes in catalyst nanoparticles relevant for fuel cells. Relevant Sections of the Description of Work A member of this consortium recently developed a new optical method, indirect nanoplasmonic sensing (INPS), enabling to follow catalytic reactions on NPs in real time and in situ [43, 44]. The method is based on localized surface plasmon resonance sensing using an array of nano-fabricated Au-disks covered by a thin catalyst support [43, 44]. The method shall be applied in this project to the study of the ORR at selected catalyst nanoparticles in situ. These studies will supplement the studies on model compounds. Furthermore, INPS can be used to study composition/structural changes (dealloying, sintering etc.) of synthesized bimetallic NPs. A slightly modified plasmonic sensing approach can be used for similar studies on nanofabricated model systems. Task 4.3: Electrochemical characterization of Pt 5Gd and Pt 3Y catalysts (month 9-30) (TUM, UniPd, DTU-CINF, CUT) A further step ahead will be done applying the tools tested in task 4.2 to NPs of Pt 5Gd and Pt 3Y catalysts. This is an intermediate step before the new catalysts suggested by the WP2 will be available (see task 4.5). The Pt 5Gd and Pt 3Y will first be tested on standard carbon supports. In addition on selected samples the INPS method mentioned above will be applied in order to determine directly the catalysts state and other properties like catalyst dispersion during the electrochemical reaction. 1
Description of Work Done / Results Details of the work at CUT Introduction To understand and improve electrocatalysts, electrochemistry combined with other characterization methods are essential. Here, we present a versatile combined fiber-optical and electrochemical setup for in situ studies of catalyst surface state, corrosion and surface restructuring. We have shown that it is possible to form thin films of Pt-alloys using single-target co-sputtering. Clips of foil of the alloying material were fixated on the surface of the Pt-target and when sputtering, both Pt and the alloy material from the clips will be co-deposited. The number and area of the clips (i.e. total area fraction of alloying material) control the composition of the alloy. Within Cathcat, the deposition methods of sputtering and evaporation are used for preparation of model electrodes. The model electrodes are thin films, to be studied using electrochemistry in combination with optical spectroscopy. In parallel, the thin films are patterned into arrays of nanodisks. Arrays of nanodisks have interesting optical properties. The interaction between light and the confined electrons in the nanodisk gives rise to localized surface plasmon resonances (LSPR), which leads to a peak in the optical extinction spectrum of the nanodisk. The position of this plasmonic peak depends on the material properties of the nanodisk and its close surroundings. A physical change in, on or near the nanoparticle will shift the position of the plasmon-frequency and can be detected in-situ and in real-time [1]. Here arrays of nanodisks from different materials are fabricated and their use for plasmonic sensing is evaluated. The samples are studied not only with optical methods but also using electrochemistry in combination with STM-imaging. They are also physically characterized using SEM, AFM, XPS, etc. Fabrication of arrays of nanodisks Hole-Mask Colloidal Lithography (HCL) has been our workhorse for nanofabrication since 2007 [2]. It is based on electrostatic self-assembly when forming the mask and allows for efficient fabrication of quasi-random arrays of nanoparticles such as disks, ellipses or pairs on a support material of choice and covering large areas (cm 2 ) homogeneously. The process of HCL is depicted in figure 1a. Typically, particle sizes from 20 nm and up to several 100 nm can be achieved, and any metal that can be deposited by thermal evaporation can be used. Here we will show that Pt and its alloys deposited by sputtering can also be used in HCL fabrication. 2
Figure 1. a. Depiction of the process steps in HCL-fabrication of arrays of nanodisks. b and c. SEM images of arrays of Pt nanodisks with 120 nm on (b) oxidized silicon and (c) FTO. Figure 1b shows fabricated model catalysts of arrays of 120 nm Pt nanodisk arrays made by HCL on fused silica, which completes Milestone 12. Nanodisks on fused silica, figure 1b and 2a, are useful for studying optical properties of the materials. To combine optical characterization with electrochemistry the nanodisks need to be fabricated on a transparent and conductive substrate. Since the substrate also needs to be compatible with the harsh conditions during measurements of catalytic activity, the available materials are very limited. However, fluorinated tin-oxide (FTO) fulfills the requirements. Hence, nanodisks have been fabricated on glass coated with FTO, figure 1c and 2b. The extinction of transmitted light (i.e. one minus transmittance of substrate with nanodisks divided by transmittance of a similar substrate without nanodisks) shows the peaks of extinction for nanodisks, figure 2 c and d. The positions of the peaks correspond to the frequency of plasmon resonance of the nanodisks. If the nanodisks are made of Pt or its alloys, they can act as both the sensor and the material to be studied. This is called direct plasmonic sensing. However, the plasmonic peak of Pt nanodisks on fused silica is not very strong, due to damping of plasmons in Pt, as seen as blue line in figure 2c. To obtain sufficient conductivity for electrochemical characterization of nanodisks a relatively thick layer of FTO on glass is needed, giving rise to interference fringes. The plasmon peak of the Pt nanodisks then becomes hidden in the extinction spectra, red line in figure 2c. Hence, direct plasmonic sensing using Pt nanodisks is difficult on these substrates. An alternative approach is to use Au nanodisks, which have a stronger plasmon peak. The peak is clearly visible also on substrates with FTO, as seen in figure 2d. Here it is also seen that the position of the plasmon peak is sensitive to the surroundings of the nanodisks, on fused silica the peak is around 750 nm and on FTO it is around 900 nm since FTO has a higher refractive index. 3
Figure 2. a. Schematic illustration of nanodisks on fused silica. b. Schematic illustration of nanodisks on glass with FTO. c. Transmission of 120 nm diameter nanodisks of platinum on fused silica and glass substrate with FTO. d. Transmission of 120 nm diameter nanodisks of gold on fused silica and glass substrate with FTO. The sensitivity of the nanodisks to their vicinity makes it possible to use them to study physical changes of adjacent catalytic nanoparticles, e.g. sintering and corrosion. This is called in-direct nanoplasmonic sensing (INPS). The method can be used to study most materials, in particular nanoparticles of materials at sizes relevant for fuel cells. The studies of commercial catalysts using combined electrochemial and INPS-characterization will be described later in this deliverable. Nanodisk fabrication using HCL is well suited for materials deposited by evaporation, where the deposited material only arrives at the substrate at perpendicular angle. In sputtering the material is deposited at a wide distribution of angles, making patterning using lift-off (as is done in HCL) more difficult. Also HCL is not optimal for carbon-substrates, since the process involves removal of carbon species by oxygen plasma. The original process, shown in figure 1a, was improved for use on carbonsubstrates by using a thinner resist layer, hence a shorter oxygen-plasma (giving less damage to the HOPG) was needed. When sputtering the material of the nanodisks it is expected that lift-off (removal of resist mask) after sputtering will be less facile since the sidewall coverage is larger, depicted in right path of figure 3a. Still, after lift-off nanodisks of sputtered Pt remained, as seen in figure 3b. However, due to the increased sidewall coverage in sputtering, nanodisks were larger in diameter and "fences" could be found on many of the discs. 4
Figure 3. a. Depiction of HCL with left path using evaporation and right path using sputtering. b. STM image of arrays of Pt nanodisks from modified HCL using sputtering on HOPG with 41 nm particles. Imaging in AFM confirmed successful fabrication of sputtered nanodisks. However, residues from the fabrication process could also be found. The residues were believed to be due to incomplete removal the resist mask, which was confirmed by their removal in warm acetone. After cleaning only Pt nanodisks remained on the HOPG and imaging in STM was possible, as seen in figure 3b. The possibility for fabrication of sputtered nanodisks on HOPG and imaging of the nanodisks in STM has been confirmed. Arrays of nanodisks from sputtered PtY thin films on HOPG have been fabricated. The samples have been characterized by EC-STM at TUM. Setup for combined electrochemical and optical characterization In order to simultaneously measure optical nanoplasmonic and electrochemical signals from the model catalysts, an electrochemical window cell has been designed, manufactured and assambled. Optical measurements can be conducted in either transmission mode (i.e. collecting the light that passes through the sample) or reflection mode. Transmission mode requires sufficiently transparent substrates. Pt thin films on fused silica can typically not be thicker than 60 nm. Reflection mode can be used on films of all thicknesses and also on bulk material. The developed optical setup, shown in figure 4, has been designed to provide sufficiently stable optical signals over long periods of time. In addition, to guarantee that the changes in the optical signal from the sample really stems from the relevant processes occurring on it s surface and not from, e.g. lamp fluctuations, a reference signal is useful. To be able to measure a reference signal with the same spectrometer and light source as used for the sample signal, a beam-combiner in combination with a double-beam configuration was used. In this way, the signal from the sample and the reference signal can be simultaneously recorded and subtracted from each other to account for drift in the setup. This is important to be able to perform reliable long-term measurements for studying e.g. corrosion where small signal changes over long time periods are expected. 5
Figure 4. a. Depiction of the setup for combining electrochemical and optical characterization. b. Photograph of parts of the setup for combining electrochemical and optical characterization. Using the beam-combiner, measurements of transmission and reflection can be made at the same time from the same sample, if desired. Alternatively, reflection from both front and backsides can be measured simultaneously. This is for example relevant if a 3D process proceeding through the film should be analyzed. In addition, even in such an arrangement, a reference signal can be obtained using a second spectrometer to account for light-source fluctuations. In summary, the expanded possibilities of our setup achieved by integrating the beam-combiner enable collection of complementary optical characterization data in the same experiment, which is a step beyond the state-of-the-art. The setup has also been used in a study of hydride formation in Pd, which has been published. [3] Pt thin film corrosion example To validate the setup for combing electrochemistry with in-situ optical characterization, we measured the electrochemical response of a 20 nm thin Pt film (figure 5a) together with changes in optical reflectance during potential cycling between 0 and 1.4 V (figure 5b). Very similar response to oxygen adsorption and oxide formation on the surface is obtained from the optical transmittance and reflectance change signals. The measured current (black line) also correlates well with the optical signals, as clearly seen, for example, for the onset of the reduction of the oxide, at around 0.75 V. The CV is characteristic for polycrystalline Pt, as expected for our evaporated Pt thin film model electrode. Specifically, the characteristic features corresponding to adsorption of OH and O upon increasing the applied potential appear around 0.8 V in the measured current. At the same time, in the optical signals a decrease of the ΔR signals is observed. Lowering the applied potential leads to close to zero current and constant ΔR down to around 1.0 V, since the formed oxide is stable in this region. This lead to hysteresis in the extinction and reflection signals when plotted against applied potential. Further reduction of the voltage leads to a large reduction current around 0.7 V where the Pt oxide is reduced, which is accompanied by a large change in ΔR, back to the baseline level at 0.4V in the double layer region. Further reduction to of the voltage leads to hydrogen adsorption, giving rise to a slight decrease of ΔR. After reaching 0 V and increasing the voltage back to the double-layer region (0.4 V) desorption of hydrogen reversibly raises the optical signals back to the starting value. This brief analysis of simultaneously measured 6
electrochemical and optical signals also highlights their sensitivity/detection limit, which corresponds to fractions of a monolayer of adsorbed species on the surface. Figure 5. a. Depiction of Pt thin film on fused silica. Change in extinction, ΔExt, for transmittance (blue) and (a) Normalized ΔExt and ΔR signals together with the corresponding measured current (black) for the 200 th potential cycle. b. Reflectance, ΔR, for front side reflection (blue) off and current (black) from a 20 nm thin Pt film during potential cycling from 0 to 1.4 V at 50 mv s -1 in 0.5 M H 2SO 4. To study the long-term corrosion of the catalyst, we use the arrangement of our setup which allows the simultaneous measurement of transmittance and a reference signal that has not interacted with the sample (shown schematically in figure 4a), and apply it to a 20 nm Pt film in a 4000 potential cycles (between 0 and 1.6 V) long experiment (figure 6). This experiment has two purposes, to track the sample changes over longer time scales and to make sure that the long term changes observed in the optical signals really are related to a process occurring on the sample surface and not caused by, for example, a long term drift in light source intensity. Figure 6. Optical measurements on a 20 nm Pt thin film during 4000 potential cycles from 0 to 1.6 V at 50 mv s -1 in 0.5 M H 2SO 4. Change in extinction, ΔExt, for transmitted light through the sample (blue) and a reference measurement where the light does not interact with the sample (red). 7
From figure 6 it becomes clear that the ΔExt signal from the reference channel (red curve) is quite constant throughout the whole experiment and at all times lower than 0.01. In contrast, the ΔExt from the sample channel (blue curve) is initially increasing by 0.08 (up to 22 hours) and then decreased by 0.06 units during the remaining 50 hours. This clearly indicates that the obtained optical signal indeed is related to the sample itself. For the investigated Pt thin film model systems potential cycling up to 1.6 V initially (first ca. 22 hours) leads to a decrease in transmitted and an increase in reflected light. The process that gives rise to the in situ measured changes in the optical signals is believed to be increased roughening of the surface during potential cycling, as observed previously. [4] In the optical signal a second regime with a decrease in transmitted light was observed and attributed to dissolution of the Pt, in agreement with previous studies have showing the corrosion of Pt at our conditions. [5] In conclusion, combined electrochemical and optical measurements on Pt thin films show that the methods give very similar response in many respects, e.g. for individual cycles. In addition, the different information obtained during long-term corrosion measurements highlights the benefits of the simultaneous measurements using the complementary techniques. This shows that combined electrochemical and optical characterization is a promising method for in-situ characterization of electrocatalysts, in particular for degradation studies. INPS experiment on Vulcan XC-72 commercial catalyst To enable similar combined electrochamical measurements with in-direct plasmonic sensing, the developed INPS-substrates (arrays of gold nanodisks on FTO-substrates described above) were spincoated with benchmark material of 50 % PtNP on Vulcan XC-72 from Tanaka. Figure 7. a. Schematic illustration of INPS-substrates coated with PtNP on carbon black. b. Shifts in peak position (blue) and corresponding CVs (black) for 3 subsequent potential cycles for INPSsubstrate spin-coated with 50% platinum nanoparticles on Vulcan XC-72 from Tanaka. Potential cycling from 0 to 1.4 V was conducted in 0.5 M H 2SO 4 and the sweep rate was 50 mv/s. Combined in-direct plasmonic sensing (INPS) and electrochemical characterization experiments were done. For individual potential cycles, the similarity between INPS of Pt nanoparticles on carbon (figure 7b) and spectroscopy of Pt thin films (figure 5b) is apparent. This suggests that the knowledge obtained previously from combined electrochemical and optical measurements can be useful also in INPS-measurements, in particular the possibility to monitor long-term corrosion of the catalyst material. 8
Figure 8. a. Shift in peak position during long-term corrosion tests for INPS-substrates with 80 % PtNPs on C. 1500 potential cycles from 0 to 1.4 V were applied in 0.5 M H 2SO 4, and the sweep rate was 50 mv/s. b. Shift in peak position for selected potential cycles. Each curve is an average of 10 subsequent cycles. The long-term corrosion for an INPS-substrate with 80 % PtNPs on C during 1500 potential cycles between 0 and 1.4 V is seen in figure 8. There was an overall blue-shift of the peak position of ca. 25 nm during the experiment (figure 8a) and the span (the difference in peak position in the oxidized state at 1.4 V compared to the metallic state at 0.4 V within each single cycle) of the curve, was reduced by ca. 75 % during the experiment (figure 8b). Investigations of the samples in SEM, after the electrochemical workload, show severe corrosion of the PtNPs on C. The bright spots being PtNPs are no longer only situated on their carbon support, but also found on the FTO substrate and the Au nanodisks. The Pt on the FTO is also not 3 nm (being the initial size of the nanoparticles), but instead tens of nanometers in diameter. Many of the Au nanodisks seemed to be decorated by electrodeposited coatings, probably being Pt. The change in morphology of the PtNPs observed by SEM is probably due to dissolution of Pt from the nanoparticles on the carbon support and redeposition of Pt on the substrate. Possibly redeposition on the nanodisks is favored by the Au providing anchoring sites for Pt electrodeposition. Conclusions The results reported here indicate that combined electrochemical and optical measurements can provide valuable insights about the corrosion of fuel cell catalysts. In many cases electrochemical and optical results are very similar, e.g. the response within single potential cycles upon oxidation and reduction of Pt, confirming their correlation. But also differences between the two methods have been observed, e.g. the dissolution of Pt changing the optical signal while not seen in the electrochemical signal, confirming their complementarity. In particular, the benefits of using in-direct nanoplasmonic sensing to follow long-term corrosion using optical measurements, with the possibility do so for catalyst nanoparticles relevant for fuel cells, have been demonstrated. The combined electrochemical and INPS studies have shown the ability to follow the effects of potential cycling of Pt nanoparticles, for example the long-term corrosion due to Pt dissolution and redeposition. 9
Publications Björn Wickman, Mattias Fredriksson, Ligang Feng, Niklas Lindahl, Johan Hagberg, Christoph Langhammer, Depth probing of the hydride formation process in thin Pd films by combined electrochemistry and fiber optics-based in situ UV/vis spectroscopy, Physical Chemistry Chemical Physics, 17 (2015) 18953 Björn Wickman, Tomasz J. Antosiewicz, Johan Hagberg, Jun Yan, Anders Hellman, Christoph Langhammer, Resolving Electrochemical Processes at the Atomic Scale by Combined in situ Optical Spectroscopy and First Principles Multiscale Modeling of the Optical Response, Submitted to Advanced Optical Materials Niklas Lindahl, Ligang Feng, Henrik Grönbeck, Björn Wickman, Ligang Feng Christoph Langhammer, A fiber-optic spectroelectrochemical setup for simultaneous long-term optical transmittance, reflectance and cyclovoltammetry measurements, Submitted to Review of Scientific Instruments References 1. E. Larsson, C. Langhammer, I. Zorić, B. Kasemo, Nanoplasmonic Probes of Catalytic Reactions, Science, 326 (2009) 1091-1094. 2. H. Fredriksson, Y. Alaverdyan, A. Dmitriev, C. Langhammer, D.S. Sutherland, M. Zäch, B. Kasemo, Hole Mask Colloidal Lithography, Advanced Materials, 19 (2007) 4297-4302. 3. B. Wickman, M. Fredriksson, L. Feng, N. Lindahl, J. Hagberg, C. Langhammer, Depth probing of the hydride formation process in thin Pd films by combined electrochemistry and fiber opticsbased in situ UV/vis spectroscopy, Physical Chemistry Chemical Physics, 17 (2015) 18953-18960 4. X. Wei, A. Reiner, E. Müller, A. Wokaun, G. G. Scherer, L. Zhang, K. Y. Shou and B. J. Nelson, Electrochemical surface reshaping of polycrystalline platinum: Morphology and crystallography, Electrochim. Acta 53 (2008) 4051-4058. 5. A. A. Topalov, I. Katsounaros, M. Auinger, S. Cherevko, J. C. Meier, S. O. Klemm and K. J. Mayrhofer, Dissolution of platinum: limits for the deployment of electrochemical energy conversion, Angewandte Chemie 51 (2012) 12613-12615. 10