Platinum sonoelectrodeposition on glassy carbon and gas diffusion layer electrodes more

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 Available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/he Platinum sonoelectrodeposition on glassy carbon and gas diffusion layer electrodes Bruno G. Pollet*, Emmanuel F. Valzer, Oliver J. Curnick PEM Fuel Cell Research Group, Centre for Hydrogen and Fuel Cell Research, The University of Birmingham, Edgbaston Road, Birmingham B15 2TT, UK article info Article history: Received 27 October 2010 Received in revised form 17 January 2011 Accepted 26 January 2011 Available online 29 March 2011 Keywords: Platinum electrodeposition Fuel cells GC GDL Power ultrasound Sonoelectrochemistry abstract The electrodeposition of Pt on glassy carbon (GC) and gas diffusion layer (GDL) surfaces in dilute chloroplatinic acid solutions (10 mM PtCl2À in 0.5 M NaCl) was performed poten4 tiodynamically in the absence and presence of ultrasound (20 kHz) at various ultrasonic powers (up to 6 W) respectively and at (313 Æ 2) K. In our conditions, it was found that platinum electrodeposition is an irreversible process which requires a substantial overpotential to drive the formation of Pt nuclei on the GC and GDL surfaces; however, under sonication Pt electrodeposition becomes more facile due to lower concentration and nucleation overpotentials and overall currents are significantly increased compared to silent conditions. It was also observed that the specific electrochemical surface area (SECSA) was significantly affected for Pt/GC and Pt/GDL electrodes prepared in the presence of rotation (GC only) and under sonication compared to those prepared under silent conditions. This finding was explained to be due to both larger and agglomerated platinum nanoparticles formed on the GC and GDL surface caused by forced convection. It was also found that ultrasound produced larger Pt nanoparticles on GC electrodes than those on GDL electrodes. Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. 1. Introduction The main scope for the development of Polymer Electrolyte Membrane Fuel Cells (PEMFC), Direct Methanol Fuel Cells (DMFC) and Alkaline Fuel Cells (AFC) is to reduce the platinum catalyst loading of the electrodes (both anode and cathode) and the associated cost without decreasing the fuel cell performance [1,2]. In order to achieve this challenging goal, it is necessary to increase the effective surface area of the Pt catalyst, in other words, the surface contact between the electrode catalyst layers (CL), the carbonaceous electronic conductor (gas diffusion layer, GDL), the polymer electrolyte membrane (NafionÒ, PEM) and the reactants (hydrogen or methanol and oxygen). Since the electrochemical reactions occur in this active part of the electrodes (also known as the three-phase reaction zone), the fuel cell performances depend greatly on the kinetics of interfacial phenomena [1,2]. Electrodes for both PEMFCs and DMFCs are usually constituted of carbon black powder acting as a catalyst support mixed with solid polymer electrolyte e.g. NafionÒ [1]. In this case, to increase the performance of the electrodes (i.e., the ‘true’ catalyst surface area) either (i) an increase in CL thickness, for a given Pt catalyst loading or (ii) an increase in the amount of Pt catalyst in the CL is required. However, increasing the thickness of the catalyst layer leads to a decrease in reactants’ diffusion rate towards Pt catalytic * Corresponding author. Tel.: þ44 (0) 7554116546; fax: þ44 (0) 121 414 5377. E-mail address: b.g.pollet@bham.ac.uk (B.G. Pollet). URL: http://www.polletresearch.com 0360-3199/$ e see front matter Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2011.01.137 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 6249 sites, whereas increasing the weight loading generally leads to an increase in the particle size of the catalysts, thus decreasing the fuel cell efficiency [1,2]. One of the most promising methods to increase Pt active site and effective Pt utilization of the electrocatalyst is to deposit Pt on carbonaceous substrates electrochemically from commercial plating baths. For example, electrodeposition of Pt potentiostatically and galvanostatically in continuous DC or pulse modes is a viable and attractive method for the fabrication of PEMFC electrodes [1,2]. However, in view of increasing the Pt utilization, it was shown that the rate determining step of Pt electrodeposition is controlled by mass transport [1e3]. One of the many approaches to increase mass transport in such processes is to employ efficient stirring or forced convection in the form of ultrasound. Extensive work has been carried out in which high power ultrasound (20 kHze2 MHz) was applied to various electrochemical processes leading to several industrial applications and many publications over a wide range of subject areas including electrodeposition and electroplating of metals, electrochemical dissolution and corrosion testing [4]. It has been shown that the effects of high intensity ultrasonic irradiation on electrochemical processes lead to both chemical and physical effects, for example, mass-transport enhancement, surface cleaning and radical formation. Furthermore, ultrasound has been showed to influence metal electrodeposition i.e. cause an increase in deposit hardness, deposit thickness, deposition rates and efficiencies and improved deposit adhesion [4]. It is well accepted in the field that sonication leads to a decrease in the diffusion layer thickness (d) thereby giving a substantial increase in limiting current (Ilim), which can be attributed to effects of cavitation and/or micro and macrostreaming [4]. It has also been shown that ultrasonic irradiation is more effective than traditional hydrodynamic methods e.g. rotating disc electrode (RDE) in reducing d and thus both cavitational and acoustic streaming effects contribute significantly to the increase in observed experimental currents [4]. The experimental decrease in the diffusion layer thickness is also thought to be due to asymmetrical collapse of cavitation bubbles at the electrode surface leading to the formation of high velocity jets of liquid being directed towards its surface. This jetting, together with acoustic streaming, can lead to random puncture and disruption of the mass transfer boundary layer at the electrode surface. Recently, Pollet et al. [5] showed, with the aid of mathematical models based on mass-balance equations, that a Levich-like equation relating the limiting current, the square root of ultrasonic power and the inverse square root of the electrode-horn distance, may be generated for ultrasonic frequencies of 20 and 40 kHz allowing the generation of an ‘equivalent’ flow velocity under sonication, an important and useful parameter in chemical engineering. This is also known as the Pollet equation [4,5]. In this paper, we report for the first time the use of ultrasound (20 kHz) for the electrodeposition of Pt on glassy carbon (GC) and porous carbon gas diffusion media (GDL) in dilute chloroplatinic acid solutions (10 mM PtCl2À in 0.5 M NaCl) 4 performed potentiodynamically at (313 Æ 2) K. We seek to identify whether ultrasound can modify (i) the electroplating/ deposition process by influencing mass transport via diffusion/convection and the electrode ‘surface chemistry’ via cavitational events, (ii) increase Pt active sites for hydrogen adsorption and (iii) affect the Pt particle size. 2. Experimental Potentiodynamic experiments were carried out using an Autolab PGSTAT302N/FRA2 potentiostat connected to a PC for data acquisition and control. Electrochemical experiments were performed using a specially designed jacketed cooling electrochemical cell e ‘microsonoreactor’ e developed by Hihn et al. placed in a Faraday cage [6]. The cell (Fig. 1(A)) based on a particular design consists of off-setting the ultrasonic probe out of the reaction volume (inner cell, Vic ¼ 10 cm3) in order to avoid any possible contaminations and to ensure perfect electric insulation from the ultrasonic probe made of Ti-alloy (Tie6Ale4V). Ultrasound was provided by a 20 kHz Vibra-Cell VCX750 ultrasonic probe with a tip diameter of 13 mm. Ultrasonic powers were determined calorimetrically using the Margulis’ method [7] with a Fluke 51 digital thermometer fitted to a K-type thermocouple. Here, ultrasonic powers (PT) are quoted as W or otherwise stated. The microsonoreactor was linked to a Grant thermostated bath operating at preset temperatures. The working electrode was either a rotating disc glassy carbon (GC) disc (B ¼ 3 mm; geometric surface area, Ag ¼ 0.0707 cm2 determined chronocoulometrically) or a series of gas diffusion layer electrodes (Freudenberg H2315C1, thickness ¼ 252 mm, weight area ¼ 132 g mÀ2, Ag ¼ 1 cm2). A platinum gauze and a saturated calomel electrode (SCE) were used as the counter and reference electrodes respectively. For the 20 kHz sonoelectrochemical experiments, the distance between the ultrasonic probe (20 kHz, area ¼ 1.33 cm2) and the working electrode was d ¼ (10 Æ 1) mm and the distance between the microsonoreactor and the working electrode was approximately 3 mm (d0 ) (Fig. 1(A)). All Pt electrodes were electrochemically cleaned by cycling in sulphuric acid (1.0 mol dmÀ3) for 10 min prior to the experiments. They were then washed with high quality Milli-Q water. 2.1. Preparation of GC working electrodes Before each electrodeposition experiment, the GC electrodes were polished to a mirror finish first with grinding paper (Buelher-Met, P600) and then sequentially with 25 microns down to 0.3 micron alumina oxide paste and cleaned ultrasonically (40 kHz Langford ultrasonic bath) in 1 M H2SO4 prepared with Milli-Q for 5 min at room temperature. Baseline cyclic voltammograms of the polished and cleaned GC electrodes were then recorded in N2 purged 1 M H2SO4 to ensure a smooth, reproducible surface (i.e. unchanged over five cyclic voltammograms between þ1.2 V vs. sce to À0.4 V vs. sce were recorded at 50 mV sÀ1) without any evidence of platinum or/ and surface oxide functionalities. The GC electrodes was then re-polished and cleaned ultrasonically before checking via cyclic voltammetry. At this final stage only five baseline CV scans were performed so as to avoid oxidation of the GC surface. 6250 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 Fig. 1 e (A) A specially designed jacketed cooling electrochemical cell e ‘microsonoreactor’ [20] e employed for potentiodynamic studies in the absence and presence of ultrasound. Here d0 is the distance between the working electrode and the base of the inner cell and d is between the working electrode and the ultrasonic probe. (B) Plots of the electric output power (Pe) vs. the transmitted ultrasonic power (PT) at the geometries employed. (C) Plots of the diffusion layer thickness in function of the square root of transmitted ultrasonic powers for the two geometries used. 2.2. Preparation of working electrodes from carbon gas diffusion media 1 cm  5 cm strips of unwoven carbon paper (Freudenberg H2315C1, GDL) were wrapped in Teflon tape so that areas measuring 1 cm  1 cm were exposed at one end, leaving small areas of exposed carbon at the opposite end for the electrical connection. The electrodes were immersed in the working solution so that the 1 cm  1 cm deposition areas were completely submerged. For all Pt electrodeposition performed potentiodynamically in the absence and presence of rotation (GC only) and sonication, cyclic voltammograms of N2 purged 10 mM of PtCl2À 4 (K2PtCl4) in 0.5 mol dmÀ3 NaCl were performed at (313 Æ 2) K. After all potentiodynamic Pt electrodeposition experiments, the platinised electrodes (Pt/GC and Pt/GDL) were removed from the deposition solution, rinsed in Milli-Q water and immediately placed in N2 purged 1 M H2SO4 for further investigations. A cyclic voltammogram was recorded from þ0.7 V vs. sce to À0.2 V vs. sce (one cycle). The ‘real’ surface area of the platinum catalyst or the electrochemical surface area (Ar) was measured from the electrical charge used in hydrogen adsorption (QH) from cyclic voltammograms between À0.20 and þ0.05 V vs. sce assuming that a monolayer of hydrogen o corresponds to an adsorption charge of 210 mC cmÀ2 (QH ) [3] and using Equation (1) [3]: o Ar ¼ QH =QH (1) 2 where Ar is in cm . The hydrogen adsorption charge was determined using Equation (2) [3]: QH ¼ ðQads À Qdes Þ=2 (2) where Qads and Qdes are the total adsorption and desorption charge respectively (the charge corresponding to charging of the double layer was not taken into account e see black area in Fig. 5). The Pt loading, W can be calculated as follows [3]:   W ¼ Qpt M =ðnFÞ (3) where QPt is the platinum charge during the reduction of the chloroplatinic salt (assuming that this quantity of charge is mainly due to the Faradaic reaction), M (195.1 g molÀ1) is the i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 6251 atomic weight of Pt, n is the number of electrons transferred during the reduction and F is the Faraday constant (96,500 C molÀ1). Platinum electrodeposits were also characterized by their specific electrochemical surface area (SECSA) in m2 gÀ1 [3,8] Pt Pt SECSA ¼ ðAr =WÞ Â 10000 2 (4) where Ag is the geometric area of the electrode in cm . All chemical reagents were of AnalaR grade or equivalent. Aqueous solutions of K2PtCl4 (Aldrich, AR, 10 mmol dmÀ3) in NaCl (Aldrich, AR, 0.5 mol dmÀ3, employed as background electrolyte), H2SO4 (Aldrich, AR, 1.0 mol dmÀ3), K4Fe(CN)6/K3Fe (CN)6 (Aldrich, AR, 1 and 10 mmol dmÀ3) and KCl (Aldrich, AR, 0.1 mol dmÀ3) were prepared in high quality Milli-Q water (resistivity ¼ 18 MU cmÀ1). All aqueous solutions were degassed by bubbling nitrogen for 20 min prior to the experiments. Morphological studies of the Pt electrodeposits and powders were performed on both a Scanning Electron Microscope (Philips XL-30 SEM, equipped with an Oxford Instruments EDX detector) and a Transmission Electron Microscope (TEM, JEOL 1200ex, 80 kV). For some sonoelectrochemical experiments, the ‘ablated’ Pt particles were filtered with 0.05 mm Millipore filters under vacuum. The filters were then washed with pure ethanol, dried for 48 h in a silica-gel drier and stored under vacuum. electrode in the presence of ultrasound (no rotation) at several powers. Fig. 1(C) shows the diffusion layer thickness in function of the square root of transmitted ultrasonic powers for the two geometries used following the relationship showed by Pollet et al. [5]. It is evident from the figure that lower diffusion layer thicknesses (>1 mm) are obtained for the ‘direct’ face-on geometry, in the transmitted ultrasonic power range of [4e20 W] compared with the ‘indirect’ face-on geometry cell where a maximum thinning of the diffusion layer thickness of 4 mm was found in the power range of [1e6 W]. For completeness, we also studied how the distance between the working electrode and the microsonoreactor base [d0 , Fig. 1(A)] affects the RDE hydrodynamics [not shown here], and found that the diffusion layer thickness remains unchanged in the range of d’ [3 mm; 20 mm]. 3.2. Potentiodynamic studies of the electrodeposition of Pt on glassy carbon in the absence, under rotation and in the presence of ultrasound (20 kHz) 3.2.1. Silent conditions 3. 3.1. Results and discussion Microsonoreactor characterisation The microsonoreactor used in this study has been extensively studied by Hihn et al. [6] who showed that the ultrasonic power transmitted to the volume within the inner cell is low due to the energy absorbed by the inner cell. However, in order to quantify this energy loss in our conditions, calorimetric experiments were performed where the thermocouple was placed inside (‘indirect’ face-on geometry) and outside the inner cell (‘direct’ face-on geometry) at a distance d from the ultrasonic probe (Fig. 1(A)). Fig. 1(B) shows that as the electric output power (Pe) delivered by the ultrasonic generator increases, the transmitted ultrasonic power (PT) also increases for both geometries, however, at similar electric output powers, a 4-fold increase in transmitted ultrasonic power is observed for the experiments performed outside the inner cell. In other words, an average energy loss of ca. 75% caused by the inner cell was observed. We also investigated how the diffusion layer thickness (d) is affected by the transmitted ultrasonic power in the two geometries employed as shown in Fig. 1(C). Here d is defined as: d ¼ nFAg DCà =Ilim (5) Initially the study focussed on the voltammetric response associated with depositing platinum from chloroplatinic solution onto the glassy carbon electrode. Electrochemical electrode pre-treatment was required and the GC electrode was held at þ1.5 V for 5 min, followed by reduction for 5 min at þ0.5 V before each scan in views of removing any polishing residue and porous surface oxide layers. A cyclic voltammogram of a freshly polished and cleaned GC electrode in a solution of 10 mM PtCl2À in 4 0.5 M NaCl in the potential range of [Ei ¼ þ1.0 V vs. sce; e0.8 V vs. sce] at (313 Æ 1) K and at a scan rate of 10 mV sÀ1 in the absence of forced convection is shown in Fig. 2. On a freshly polished GC electrode, the reduction of PtCl2À occurs in the whole potential 4 region during the negative scan and the resulting cathodic current appears to commence at around þ100 mV vs. sce and increases continuously to reach its maximum at e328 mV vs. sce (Epc(I)). It was observed (not show here) that at potentials more negative than e0.95 V vs. sce, the cathodic current was mainly due to the evolution of hydrogen occurring on the platinum clusters already formed. It is well accepted in the literature, that this cathodic peak (I) corresponds to the reduction of Pt(II) to Pt according to Equation (6) [9]: PtCl4 þ 2eÀ /PtðsÞ þ 4Cl 2À À (6) where n is the number of electrons transferred, F is the Faraday constant, Ag is the geometric electrode surface area, D is the diffusion coefficient of the electroactive species, C* is the electroactive species bulk concentration and Ilim is the limiting current [4]. For this study, we performed linear sweep voltammograms (LSV) of 10 mM K4Fe(CN)6/K3Fe(CN)6 in 0.1 M KCl on GC Eo ¼ 0.758 V vs. she [10,11] In the reverse scan, i.e. the anodic scan, Pt(II) ions are further reduced to a maximum of ca. e400 mV vs. sce [peak (II)] and continues up to þ0.1 V vs. sce suggesting that platinum is further deposited on the nuclei already formed. Furthermore, the presence of peak (II) implies that at lower potentials the surface is blocked towards further Pt reduction. It has been shown by Gregory et al. [11] that sharp cathodic peaks on both forward and reverse scans imply a reversible inhibition of Pt(II) reduction at potentials negative of the peaks caused by H atom adsorption on the Pt surface blocking further Pt deposition. It is well-known that H-atoms adsorb on Pt surfaces reversibly in the potential range positive of H2 evolution [11]. At more positive potentials, a third peak [(III)] is observed at þ751 mV vs. sce (Epc(III)) corresponding possibly to 6252 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 Fig. 2 e Cyclic voltammogram of 10 mM PtCl2L in 0.1 mol dmL3 NaCl on a freshly polished and cleaned glassy carbon (GC) 4 electrode at 10 mV sL1 and at (313 ± 1) K in the potential range [D1.0 V vs. sce; L0.8 V vs. sce] in the absence of ultrasound. Inset figure e Cyclic voltammogram of 0.1 mol dmL3 NaCl on a freshly polished and cleaned glassy carbon (GC) electrode at 10 mV sL1 and at (313 ± 1) K in the potential range [D1.2 V vs. sce; L1.0 V vs. sce] in the absence of ultrasound. the oxidation of Pt(II) to Pt(IV). Waibel et al. [12] studied the electrodeposition of Pt on gold in a solution of 0.1 mM K2PtCl4 in 0.1 M H2SO4 and showed that the anodic peak corresponded to the oxidation of Pt2þ to Pt4þ catalyzed by the deposited Pt, according to Equation (7): PtCl4 þ 2Cl /PtCl6 þ 2eÀ 2À 2À À (7) Eo ¼ 0.726 V vs. she [9e12] Furthermore, the anodic peak (III) in Fig. 2 may possibly correspond to the oxidation of Pt(II) to form primary oxides (PteOH or OH*) as observed by Conway et al. [13,14]. It is possible that the primary oxide ‘oxidizes’ Pt2þ to Pt4þ and thereby the cathodic counterpart of the anodic peak (III) is missing, since the whole amount becomes spent in the course of such a further oxidation. These findings suggest that (i) a substantial overpotential is necessary to drive the formation of Pt nuclei on the glassy carbon surface, (ii) a surface control phenomenon exists for the Pt electrodeposition, (iii) Pt electrodeposition is irreversible, and (iv) are in excellent agreement with literature [9e11]. 3.2.2. Hydrodynamic conditions In this study, cyclic voltammograms (CV) of chloroplatinic solutions on GC electrode were analysed in the absence of ultrasound, under rotation (no ultrasound) and the presence of ultrasound (no rotation) in the objective to examine whether ultrasound affects platinum electrodeposition. [Ei ¼ þ1.2 V vs. sce; e0.5 V vs. sce] at a scan rate of 10 mV sÀ1 in the absence of ultrasound, at several rotation speeds [800 rpm (a), 1200 rpm (b) and 2000 rpm (c)] and at (313 Æ 1) K. On the forward scan i.e. towards negative potentials, the figure shows that as rotation speed increases, deposition peak (I) increases, for example, at maximum rotation speed of 2000 rpm (c), peak current density increases 4-fold compared to stationary conditions (Fig. 2) due to an increase in PtCl2À ions being 4 depleted at the surface. Furthermore, a plot of cathodic peak (I) current densities (ipc(I)) vs. square root of rotation speeds (u0.5) was constructed (not shown here). This plot did not go through zero and confirmed that, in our conditions, platinum deposition is not only diffusion control and surface reaction control takes place. On the negative scan, as rotation speed increases, peak (II) also increases with a slight shift in peak (II) potentials (Epc(II)) to more positive potentials. It is known that once platinum nuclei have been electrodeposited, further platinum deposition takes place onto the existing platinum deposit since this facilitates reduction at a lesser overpotential in comparison with ‘bare’ GC electrodes [9e12]. It is interesting to note that the magnitude of all peaks (II) are similar to that of peaks (I) indicating that efficient stirring of the solution helps in bringing about the PtCl2- from the bulk solution to the 4 electrode surface. At more positive potentials, peak (III) increases as rotation speed increases with peak (III) potentials of ca. þ1.0 V vs. sce (Epc(III)). This finding suggests that any unreduced Pt(II) ions are further oxidised to Pt(IV) ions on Pt/ GC by forced convection. 3.2.2.1. Rotating disc electrode (RDE). Rotating disc electrode experiments were performed to examine whether rotation affects platinum electrodeposition. Fig. 3 shows CVs of freshly polished and cleaned rotating GC disc electrodes in a solution of 10 mM PtCl2À in 0.5 M NaCl in the potential range of 4 3.2.2.2. Power ultrasound (20 kHz). The investigation then turned to the effect of ultrasound on the electrodeposition of platinum at various ultrasonic powers. Fig. 4 shows a series of cyclic voltammograms of freshly polished and cleaned glassy carbon (GC) electrodes in i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 6253 Fig. 3 e Cyclic voltammograms of 10 mM PtCl2L in 0.5 mol dmL3 NaCl on a freshly polished and cleaned rotating glassy 4 carbon disc electrode at 10 mV sL1 and at (313 ± 1) K in the potential range [D1.2 V vs. sce; L0.5 V vs. sce] at various rotation speeds: (a) 800 rpm; (b) 1200 rpm and (c) 2000 rpm. a solution of 10 mM PtCl2À in 0.5 M NaCl in the potential range 4 of [Ei ¼ þ1.0 V vs. sce; e0.8 V vs. sce] at 10 mV sÀ1 in the presence of ultrasound (20 kHz) at various ultrasonic powers [silent (a), 2.3 W (b), 3.2 W (c) and 4.4 W (d)] and at (313 Æ 2) K. The figure clearly shows that ultrasound affects the Pt(II) redox process in the potential range studied. In the anodic and cathodic regions, as the ultrasonic power increases, the CVs become ‘noisier’, particularly when ultrasound is applied above the cavitational threshold (>2 W) for the overall system. At the three ultrasonic powers employed and in the negative potential region, no sharp peaks are observed as shown in Fig. 4 but instead, pseudo plateaux are obtained. The inset figure shows the CVs for the forward scans only [þ1.0 V vs. sce / e0.8 V vs. sce] and shows that an average 8-fold increase in pseudo limiting current densities (ilim) [and equally in amount of Pt charge, see later] is observed compared to silent conditions (Table 1). These observations suggest that two processes occur simultaneously on the GC electrode Fig. 4 e Cyclic voltammograms of 10 mM PtCl2L in 0.5 mol dmL3 NaCl on a freshly polished and cleaned glassy carbon disc 4 electrode at 20 mV sL1 and at (313 ± 2) K in the potential range [D1.0 V vs. sce; L0.8 V vs. sce] at various ultrasonic powers: (a) silent (0 rpm); (b) 2.3 W; (c) 3.2 W and (d) 4.4 W. The inset figure shows linear sweep voltammograms of 10 mM PtCl2L in 4 0.5 mol dmL3 NaCl on GC at 10 mV sL1, at (313 ± 2) K in the potential range [D0.4 V vs. sce; L0.8 V vs. sce] and at several ultrasonic powers: (a) silent; (b) 2.3 W; (c) 3.2 W and (d) 4.4 W. 6254 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 Table 1 e Pseudo limiting current densities and potentials at foot of waves in silent and ultrasonic conditions. Values taken from insert figure in Fig. 4. Ultrasonic power/W 0 2.3 3.2 4.4 a average values. Pseudo limiting current density/A mÀ2 À38.5 À54.3a À85.2a À310a Potential at foot of wave/mV vs. sce À269 À230 À201 À145 surface in the presence of ultrasound, with possibly: (i) partial removal of the layer of H-atoms adsorbed on the Pt surface inhibiting further Pt(II) reduction as a result of cavitation bubbles imploding violently at the electrode surface. Walton et al. [15] studied the effect of sonication (38 kHz) of 1 M H2SO4 on platinised platinum electrodes and showed that ultrasound increases hydrogen evolution partly caused by the removal H-atoms adsorbed on the electrode surface; and (ii) a combination of material ablation from the surface and rapid changes in the surface ion concentration due to cavitation. In fact, during sonication at 4.4 W, large Pt particles (>100 nm) were present in the solution and, SEM and TEM pictures revealed that some Pt was present on the GC electrodes. Below 3.2 W, filters from solutions were also analysed by SEM and TEM and showed small amount of Pt nanoparticles (ca. 20 nm), possibly caused by ‘mild’ erosion induced by cavitation at low ultrasonic powers. Interestingly, in the oxidation region of the CVs, pseudo anodic limiting currents of similar magnitudes (ilim z 50 A mÀ2) are obtained at the three ultrasonic powers used. However, these currents are 50% smaller than those obtained at maximum rotation speed [2000 rpm (c); Fig. 3]. These findings suggest that under sonication, the electrode surface is no longer a complete Pt/GC but a surface containing both Pt and GC sites, where possibly the oxidation of Pt(II) to Pt(IV) still occur but only on a fewer Pt sites [Pt/Pt e platinised platinum]. Furthermore, under sonication, the CVs in the negative potentials region show that the potential required for Pt deposition becomes more positive (or less negative) as shown in Table 1. For example, a positive shift in the potential (at the foot of each wave on the forward scans) of ca. DE ¼ þ125 mV was found at maximum ultrasonic power compared to silent conditions. This observation in potential shift in metal electrodeposition under sonication is in good agreement with previous findings, for example, Pollet et al. [16e18] and Hyde et al. [19] showed that low concentrations of silver, copper, Co65Fe35, cobalt, lead, zinc and mercury electrodeposition in the presence of ultrasound started at less negative potentials. It is possible to explain this shift in potential as follows: it is known that when ultrasound is transmitted through a liquid, efficient stirring and cavitation occur in the bulk solution and near the electrode surface [4]. This leads to an increase in the movement of ions across the diffusion layer and their subsequent discharge and hence a decrease in concentration overpotential. This also leads to a decrease in nucleation overpotential. It has been shown that ultrasound affects the surface morphology of many metal electrodeposits due to highly efficient stirring caused by acoustic streaming, and also that this decrease in overpotential is due to formation of nucleation sites at the electrode surface caused by the implosion of cavitation bubbles [4,15e18]. 3.3. Evaluation of the Pt deposit characteristics from electrochemical measurements In order to examine whether rotation and ultrasound affects platinum coverage on glassy carbon electrode by influencing (i) mass transport via diffusion and convection and (ii) cavitation, the study then turned into evaluating the Pt deposit characteristics. From Figs. 3 and 4, it was observed that the reduction peaks (I) increases, thus the charge increased with increased agitation in the form of rotation and sonication. In our conditions, the charge was calculated from the area under each peak for the forward scans only in Figs. 3 and 4, assuming that this quantity of charge in mainly due to the Faradaic reaction (Equation (6)). Table 2 shows the quantity of Pt charge, QPt, increases with increasing rotation speeds and ultrasonic powers. For example, at maximum rotation speeds and ultrasonic powers the charge increases by approximately 20fold and 40-fold respectively compared to silent conditions (this is to be expected as QPt is directly proportional to PT). As QPt is available from CVs, it is also possible to calculate the platinum loading, W, at various conditions by using Equation (4) with n ¼ 2. Again, a trend is observed where the platinum loading increases with increasing rotation and ultrasonic power. It was observed that the maximum platinum loading was obtained at maximum ultrasonic power assuming that platinum deposit is not ablated during the duration of the experiment (see later). Table 2 e Pt charge (QPt), Pt loading (W ), hydrogen adsorption charge (QH), electrochemical surface area (Ar) and the specific electrochemical surface area (SECSA) at various rotation speeds and ultrasonic (20 kHz) powers. Values taken from Figs. 3e5. Rotation speed/rpm 0 QPt/mC W/mg cmÀ2 QH/mC Ar/cm2 SESCA/m2 Pt gÀ1 1.36 0.019 291.6 1.39 101 Ultrasonic power/W 2000 26.08 0.373 822.9 3.92 15 800 15.16 0.217 415.6 1.98 13 1200 17.34 0.247 621.5 2.96 17 2.3 5.75 0.082 449.6 2.14 37 3.2 9.03 0.129 e e e 4.4 49.32 0.705 e e e i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 6255 In order to demonstrate whether platinum electrodeposited on GC electrodes under rotation and ultrasound affects the electrochemical surface area (Ar) and the specific electrochemical surface area (SECSA), the investigation then turns into analysing voltammetric responses of the Pt/GC electrodes in N2 purged 1 M H2SO4. Here, the Pt/GC electrodes prepared under silent, rotating and ultrasonic conditions were removed from the microsonoreactor at the end of the forward scan (Ef ¼ e0.8 V vs. sce), rinsed in Milli-Q water and immediately placed in N2 purged sulphuric acid solution. Fig. 5 shows the CVs of Pt/GC electrodes prepared in the absence (a e 0 rpm), presence of rotation (b e 800 rpm); (d e 2000 rpm) and ultrasound (c e 2.3 W). Also, CVs of the Pt/GC electrodes above 2.3 W (not shown here) showed no evident hydrogen adsorption and desorption peaks, indicating that little Pt electrodeposits remained on the GC electrode under sonication which is in good agreement with our early observations. It is also interesting to note that for Pt/GC electrodes prepared at increasing rotation speeds and ultrasonic powers, the hydrogen adsorption and desorption peaks and QH and thus the electrochemical surface area increases, indicating that more platinum sites are available for electrochemical reaction, however, this is not the case for the SECSA, where it decreases with increasing rotation and in the presence of ultrasound compared to silent conditions (the value of SECSA found under silent conditions is in very good agreement with previous findings [20]). Also, as Pt loading increases, SECSA decreases which is in agreement with observations by WattSmith et al. [20]. They found that the specific electrochemical area of low-loaded Pt electrodes is higher compared to those with high Pt loadings due to Pt distribution on the electrode surface. In our conditions, the fraction of the catalyst that is available to participate in the electrode reaction i.e. Pt utilisation is lower for Pt/GC electrodes prepared under forced convection than for those under no hydrodynamic conditions which could be possibly caused by Pt agglomeration. For example, Maillard et al. [21] showed that the electrodeposition of Pt on GC substrate under potentiodynamic conditions gives rise to Pt agglomerates hence leading to a decrease in catalyst utilisation. To confirm our findings, morphologies of deposit experiments were conducted. SEM images of Pt/GC electrodes prepared under silent conditions (a), under and in the presence of ultrasound (b) and (c) conditions are presented in Fig. 6. Fig. 6(a) shows individual Pt nanoparticles (100e250 nm in size) deposit on glassy carbon. It is interesting to note that the Pt nanoparticles tend to be pulled together in views of forming nano-aggregates. In Fig. 6(b) and (c), no individual Pt nanoparticles are formed but instead agglomerated and larger Pt nanoparticles for both forced convection conditions were observed. By analogy with the sol-gel method (used to produce high and uniform nanodispersion of metal electrocatalysts), this observation could be possibly due to interactive supported electrocatalysts when hypo-d-oxide supports, by the interactive bonding and by the ratio in their number relative to Pt nanoparticles, define their ‘grafting’ distribution and uniform size. In this case, ultrasound should further contribute and increase the whole effect in such a sense [22,23]. Furthermore, it was found by several workers [1,2] that increasing the weight loading, W, generally leads to an increase in catalyst particle size, thus decreasing the fuel cell performance. Finally, this finding differs significantly from our previous work [9] where it was found that ultrasound increases fuel cell performance. In our previous studies [9], the platinum was electrodeposited by galvanostatic pulse electrodeposition in the presence of pulsed power ultrasound (20 kHz). This observation therefore suggests that continuous deposition under prolonged ultrasound is not favourable for obtaining good fuel cell performance as Pt nuclei on the electrode surface either (i) grow and tend to agglomerate with neighbour nuclei or (ii) are ablated from the surface or (iii) both. Fig. 5 e Cyclic voltammograms of Pt/GC electrodes prepared in the absence (a e silent and 0 rpm), presence of rotation (b e 800 rpm); (d e 2000 rpm) and ultrasound (c e 2.3 W) in N2 purged 1 M H2SO4 in the range [D0.7 V vs. sce; L0.2 V vs. sce] (one cycle) at 5 mV sL1 and at 298 K. The area in black represents the double-layer capacitance charge. 6256 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 Fig. 6 e SEM images of Pt/GC electrodes prepared: (a) under silent conditions (no rotation), (b) under maximum rotation (2000 rpm) and (c) in the presence of ultrasound (2.3 W). i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 6257 Fig. 7 e (A) Cyclic voltammograms of 10 mM PtCl2L in 0.1 mol dmL3 NaCl on a cleaned gas diffusion layer (GDL) electrode, at 4 (313 ± 2) K in the potential range [D1.0 V vs. sce; L1.0 V vs. sce] and at several ultrasonic powers: (a) silent; (b) 2.3 W; (c) 3.2 W and (d) 4.4 W. (B) Specific electrochemical surface area (SECSA) in function of ultrasonic (20 kHz) powers. (C) Particle size distribution curves: (a) no ultrasound, (b) 2.3 W, (c) 3.2 W and (d) 4.4 W. (D) SEM pictures of Pt electrodeposited on GDL (a) under silent and (b) ultrasonic (4.4 W) conditions. 3.4. Sono-electrodeposition on carbon gas diffusion media The investigation then turned to the effect of ultrasound on the electrodeposition of platinum on GDL samples at various ultrasonic powers. Fig. 7(A) shows a series of cyclic voltammograms of cleaned GDL electrodes in a solution of 10 mM PtCl2À in 0.5 M NaCl in 4 the potential range of [Ei ¼ þ1.0 V vs. sce; e1.0 V vs. sce] at 10 mV sÀ1 in the presence of ultrasound (20 kHz) at various ultrasonic powers [silent (a), 2.3 W (b), 3.2 W (c) and 4.4 W (d)] and at (313 Æ 2) K. As observed previously, the figure clearly shows that ultrasound affects the Pt(II) redox process in the potential range studied. Under silent conditions, the CV is very similar to that obtained with a GC electrode (Fig. 2), however, compared to our previous observations (Fig. 4), no pseudo plateaux are obtained in the cathodic region and no anodic peaks are observed in the anodic region but instead large currents are obtained in both the oxidation and reduction regions for all ultrasonic powers used. Furthermore, the absence of noise can also be attributed to the relatively large surface area of the GDL electrode compared with the GC electrode as ‘less’ cavitation event is ‘seen’ by the GDL electrode. The absence of peaks in the deposition region may also be a result of the porous 3-D structure of the GDL electrode. Interestingly, these unusual CV shapes were observed by Reisse et al. [24] who used the tip of the ultrasonic probe as the working electrode (sonoelectrode) for the electrodeposition of Cu. In their work, Cu deposits were continually ablated from the ‘vibrating’ electrode tip, creating a clean electrode surface on which new Cu nuclei could form. They attributed their observations to huge rates of mass transport and cavitational events corresponding to diffusion layers which are submicron in size. In our conditions, it is possible that the GDL is ‘vibrating’ thus an increase in mass transport within the 3-D porous media is observed. For experiments involving carbon gas diffusion media, the morphologies of deposits formed under silent conditions differed markedly from those of deposits formed under sonication. Fig. 7(D) shows SEM pictures of Pt electrodeposited on GDLs under silent (a) and ultrasonic (4.4 W) (b) conditions. It is interesting to note that ultrasound provides better Pt coverage along the GDL fibres compared to silent conditions. Furthermore, smaller Pt nanoparticles are formed under sonication (4.4 W) than under silent conditions as shown in Fig. 7(D-a). Interestingly, the average Pt nanoparticle size decreases with increasing ultrasonic power by ca. 5-fold [3.2 W, Fig. 7(C)] and SECSA values (Fig. 7(B)) seems to be following approximately the same trend as size distributions (here, the SECSA values are 7-fold higher at 3.2 W than under silent conditions). These observations could be due to the implosion of cavitation bubbles at the GDL surface enabling the ‘deagglomeration’ of Pt nanoparticles or/and activating nucleation sites for Pt together with high mass transport [25]. 6258 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 6 2 4 8 e6 2 5 8 4. Conclusions It was shown that platinum electrodeposition on glassy carbon and gas diffusion layer is an irreversible system even under forced agitation such as rotation and sonication. It was found that currents (GC only) are increased by 4-fold and 8-fold at maximum rotation speed and ultrasonic power respectively compared to silent conditions. Positive shifts in potentials were observed under sonication conditions and this finding was possibly attributed to a decrease in concentration and nucleation overpotentials. It was also found that the Pt deposit characteristics on GC under silent conditions differ greatly from those under rotation and ultrasonic conditions. Pt utilisations were found to decrease under forced agitation and were attributed to larger or/and agglomeration of catalyst nanoparticles. Furthermore, voltammetric analyses of Pt/GC in acid prepared in the presence of ultrasound above 2.3 W showed no hydrogen adsorption and desorption peaks. However, it was found that electrodeposited Pt on GDL samples under ultrasound led to smaller Pt particle sizes (<200 nm) compared to silent conditions (ca. 0.9e1 mm). These observations were attributed to the implosion of cavitation bubbles at the GC and GDL surfaces enabling the ‘deagglomeration’ of Pt nanoparticles or/and activating nucleation sites for Pt. Acknowledgement The authors would like to thank Advantage West Midlands (AWM) under the Birmingham Science City Initiative, EPSRC (Contract No: EP/E034888/1). references [1] Litster S, McLean G. PEM fuel cell electrodes. J Power Sources 2004;130:61e76. [2] Wee J-H, Lee K-Y, Kim SH. Fabrication methods for low-Ptloading electrocatalysts in proton exchange membrane fuel cell systems. J Power Sources 2007;165:667e77. [3] Habibi B, Pournaghi-Azar MH, Razmi H, AbdolmohammadZadeha H. Electrochemical preparation of a novel, effective and low cast catalytic surface for hydrogen evolution reaction. Int J Hydrogen Energy 2008;13(11):2668e78. [4] Pollet BG, Hihn J-Y. Sonoelectrochemistry: from theory to applications, Handbook on Applications of Ultrasound: Sonochemistry for Sustainability. In: Chen D, Sharma SK, Mudhoo A. Taylor & Francis, CRC press, in press. [5] Pollet BG, Hihn JY, Doche ML, Lorimer JP, Mandroyan A, Mason TJ. Transport limited currents close to an ultrasonic horn: equivalent flow velocity determination. J Electrochem Soc 2007;154(10):E131eR138. [6] Costa C, Hihn J-Y, Rebetez M, Doche M-L, Bisel I, Moisy P. Transport-limited current and microsonoreactor characterization at 3 low frequencies in the presence of water, acetonitrile and imidazolium-based ionic liquids. Phys Chem Chem Phys 2008;10(16):2149e58. [7] Margulis MA, Margulis IM. Calorimetric method for measurement of acoustic power absorbed in a volume of a liquid. Ultrason Sonochem 2003;10:343e5. [8] Curnick OJ, Mendes PM, Pollet BG. Enhanced durability of a Pt/C electrocatalyst derived from NafionÒ-stabilised colloidal platinum nanoparticles. Electrochem Commun 2010;12(8):1017e20. [9] Pollet BG. A novel method for preparing PEMFC electrodes by the ultrasonic and sonoelectrochemical techniques. Electrochem Comm 2009;11:1445e8. [10] Lin-Cai J, Pletcher D. A substrate effect on the catalytic activity of electrodeposited platinum layers. J Electroanal Chem 1983;149(1e2):237e47. [11] Gregory AJ, Levason W, Noftle RE, Le Penven R, Pletcher D. Studies of platinum electroplating baths Part III. The electrochemistry of Pt(NH3)4Àx(H2O)2þ and PtCl4Àx(H2O)(2Àx)-. x x J Electroanal Chem 1995;399(1e2):105e13. [12] Waibel H-F, Kleinert M, Kibler LA, Kolb DM. Initial stages of Pt deposition on Au(111) and Au(100). Electrochim Acta 2002;47: 1461e7. [13] Angerstein-Kozlowska H, Conway BE, Hamelin A, Stoicoviciu L. Elementary steps of electrochemical oxidation of single-crystal planes of AudI. Chemical basis of processes involving geometry of anions and the electrode surfaces. Electrochim Acta 1986;31:1051e61. [14] Angerstein-Kozlowska H, Conway BE, Hamelin A, Stoicoviciu L. Elementary steps of electrochemical oxidation of single-crystal planes of Au Part II. A chemical and structural basis of oxidation of the (111) plane. J Electroanal Chem 1987;228:429e53. [15] Walton DJ, Burke LD, Murphy MM. Sonoelectrochemistry: chlorine, hydrogen and oxygen evolution at platinised platinum. Electrochim Acta 1996;41(17):2747e51. [16] Pollet BG, Lorimer JP, Hihn JY, Phull SS, Mason TJ, Walton DJ. The effect of ultrasound upon the oxidation of thiosulphate on stainless steel and platinum electrodes. Ultrason Sonochem 2002;9(5):267e74. [17] Dabala’ M, Pollet BG, Zin V, Campadello E, Mason TJ. Sonoelectrochemical (20 kHz) production of Co65Fe35 alloy nanoparticles from Aotani solutions. J Appl Electrochem 2008;38:395e402. [18] Pollet BG, Hihn J-Y, Mason TJ. Sono-electrodeposition (20 and 850 kHz) of copper in aqueous and deep eutectic solvents. Electrochim Acta 2008;53(12):4248e56. [19] Hyde ME, Compton RG. How ultrasound influences the electrodeposition of metals. J Electroanal Chem 2002;531(1): 19e24. [20] Watt-Smith MJ, Friedrich JM, Rigby SP, Ralph TR, Walsh FC. Determination of the electrochemically active surface area of Pt/C PEM fuel cell electrodes using different adsorbates. J Phys D Appl Phys 2008;41(17):174004. [21] Maillard F, Schreier S, Hanzlik M, Savinova ER, Weinkauf S, Stimming U. Influence of particle agglomeration on the catalytic activity of carbon-supported Pt nanoparticles in CO monolayer oxidation. Phys Chem Chem Phys 2005;7:385e93. [22] Jaksic Jelena M, Krstajic Nedeljko V, Vracar Ljiljana M, Neophytides Stelios G, Labou Diamantoula, Falaras Polycarpos, et al. Spillover of primary oxides as a dynamic catalytic effect of interactive hypo-d-oxide supports. Electrochim Acta 2007;53(2):349e61. [23] Jaksic Jelena M, Labou Diamantoula, Papakonstantinou Georgos D, Siokou Angeliki, Jaksic Milan M. Novel spillover interrelating reversible electrocatalysts for oxygen and hydrogen electrode reactions. J Phys Chem C 2010;114(43):18298e312. [24] Reisse J, Francois H, Vandercammen J, Fabre O, Kirsch-de Mesmaeker A, Maerschalk C, et al. Sonoelectrochemistry in aqueous electrolyte: a new type of sonoelectroreactor. Electrochim Acta 1994;39(1):37e9. [25] Pollet BG. Review: the use of ultrasound for the fabrication of fuel cell materials. Int J Hydrogen Energy 2010;35(21):11986e2004.