In situ electrochemical STM study of platinum nanodot arrays on highly oriented pyrolythic graphite prepared by electron beam lithography

Surface Science 606 (2012) 1922–1933

Contents lists available at SciVerse ScienceDirect

Surface Science journal homepage: www.elsevier.com/locate/susc

In situ electrochemical STM study of platinum nanodot arrays on highly oriented pyrolythic graphite prepared by electron beam lithography A. Foelske-Schmitz ⁎, A. Peitz, V.A. Guzenko, D. Weingarth, G.G. Scherer, A. Wokaun, R. Kötz Paul Scherrer Institut, 5232 Villigen PSI, Switzerland

a r t i c l e

i n f o

Article history: Received 23 March 2012 Accepted 31 July 2012 Available online 4 August 2012 Keywords: E-beam lithography Pt HOPG Model electrodes Electrochemical oxidation In situ EC-STM

a b s t r a c t Model electrodes consisting of platinum dots with a mean diameter of (30 ± 5) nm and heights of 3–5 nm upon highly oriented pyrolytic graphite (HOPG) were prepared by electron beam lithography and subsequent sputtering. The Pt nanodot arrays were stable during scanning tunnelling microscopy (STM) measurements in air and in sulphuric acid electrolyte, indicating the presence of “anchors”, immobilising the dots on the HOPG surface. Electrochemical STM was used to visualise potential induced Pt, carbon and Pt-influenced carbon corrosion in situ in 0.5 M sulphuric acid under ambient conditions. Potentiostatic hold experiments show that the Pt dots start to disappear at electrode potentials of E > 1.4 V vs. SHE. With increasing time and potential a hole pattern congruent to the original dot pattern appears on the HOPG basal planes. Corrosion and peeling of the HOPG substrate could also be followed in situ. Dissolution of Pt dots appears to be accelerated for potential cycling experiments compared to the potential hold statistics. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Carbon-supported platinum is an electrocatalyst widely used in polymer electrolyte fuel cells (PEFC). Despite high corrosion resistance of Pt as well as carbon, numerous studies have confirmed that lifetime and stability of PEFC is greatly limited by corrosion and degradation processes occurring on the surface of the catalyst resulting for example in a reduction of catalytically active surface area (see for instance [1–8]). There is general agreement that the loss of Pt active area can be assigned to three fundamental degradation processes such as (i) Pt dissolution and redeposition (Ostwald ripening), (ii) migration and sintering of Pt atoms on the carbon support and (iii) Pt detachment triggered by corrosion of the carbon support [1,8]. Most of this knowledge is based on post mortem analysis of different catalysts degraded under different working conditions in different PEFC setups. As a consequence, a detailed assignment of the degradation mechanisms to defined reaction conditions, such as time, potential or temperature is missing. Fundamental research on well defined model electrodes consisting of carbon supported Pt is rarely portrayed. Such model systems could be investigated via e.g. in situ scanning probe methods, thus enabling records of changes in morphology of the electrode materials during degradation. Some of the requirements for an “ideal” model catalyst were listed by Zoval et al. in 1998 [9]: (i) the particles should be of monodisperse size and shape, (ii) should have an ⁎ Corresponding author at: Electrochemistry Laboratory, OLGA 115, Paul Scherrer Institut, 5232 Villigen PSI, Switzerland. Tel.: +41 563104193; fax: +41 310 4415. E-mail address: [email protected] (A. Foelske-Schmitz). 0039-6028/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.susc.2012.07.040

electrical connection to a (catalytically inert) support, which facilitates the characterisation of the particles, (iii) should be well-separated from each other, (iv) their structure and the structure of the support should be accessible to certain measurement techniques before, during and after the involved catalytic process and, finally, (v) the particles should be stable for a long time. A method to prepare monodisperse, well-separated Pt particles on a flat carbon substrate was described e.g. by Lindström et al. [10], who studied oxidation of carbon monoxide and formaldehyde on nanostructured Pt/glassy carbon (GC) electrodes, prepared by hole mask colloidal lithography [10]. Pt nanoparticles with a well-defined diameter and a controlled arrangement on GC can also be prepared by extreme ultraviolet inference lithography (EUV-IL) or electron beam lithography (EBL), followed by Pt deposition via sputtering, as reported recently [11,12]. In case of EBL, an area of one square centimetre on GC could be homogeneously patterned by the Pt dots after modifying several parameters relevant for the EBL process (for details refer to [12]). Recently, we communicated that platinum nanodot arrays on HOPG can likewise be prepared by EBL and subsequent platinum deposition via sputtering [13]. It was shown that the Pt dot pattern was stable during scanning tunneling microscopy (STM) measurements in air and in situ in 0.5 M sulphuric acid electrolyte at a potential of 0.7 V vs. SHE, indicating immobilisation of the Pt dots, which are otherwise reported to be mobile on the untreated HOPG surface [14,15]. Stepwise increase of the electrode potential led to a Pt loss, which was indicated by a hole pattern appearing on the HOPG surface congruent with the formerly existing dot pattern. The exact origin of the holes observed after oxidation, however,

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

required clarification. They may be either defects created on the HOPG surface during the e-beam exposure or the subsequent sputter process. Pt induced carbon corrosion could also explain the formation of holes on the HOPG surface [13]. In this paper, the possible reasons for the strong Pt adhesion to HOPG and the origin of holes are discussed in more detail. Further, in situ STM investigations of Pt-nanodot arrays on HOPG in 0.5 M sulphuric acid providing more statistics are presented. The oxidation of Pt/HOPG model electrodes is performed by means of stepwise potential increase and for comparison by potential cycling with increasing upper vertex potential. In good agreement with literature, if cycled accelerated dissolution of the Pt dots occurs. Besides Pt loss and subsequent appearance of the hole pattern, carbon corrosion starting on the step edges of the HOPG terraces is observed. 2. Experimental 2.1. E-beam lithography The EBL preparation procedure was performed according to [12,13]. Pieces of HOPG (Tectra GmbH, ZYH-grade) with the dimensions 15× 15× 1 mm3 were freshly cleaved with a scotch-tape. Two-layers of a poly(methyl-methacrylic acid) (PMMA)/ethyl lactate solution, a 2% solution of PMMA with the molecular weight (MW) of 50 k and a 1% solution of PMMA with the molecular weight of 950 k, were used as photoresist. The HOPG samples were first spin-coated with a 35 nm PMMA layer of the lower molecular weight. Afterwards, the samples were soft-baked on a hot plate at 175 °C for 5 min. Then, a second PMMA layer (MW 950 k) of approximately 30 nm was spincoated upon the first layer. After this step the sample was soft-baked again on a heated plate at 175 °C for 5 min to remove the solvent. Therefore, it was possible to obtain an undercut profile during the lithography process promoting the lift-off process after metal deposition. A Vistec EBPG5000Plus electron beam lithography tool, operated at 100 keV, was used to pattern the photoresist. This e-beam writer is equipped with a pattern generator, which can be operated at beam stepping frequencies up to 50 MHz, and a deflection system, capable of deflecting the beam to large angles so that a writing field as large as 512 × 512 μm 2 can be exposed without any stage movements. The Gaussian shaped beam was focused to approximately 10 nm in diameter at a beam current of 10 nA. Rectangular dot arrays on an area of one square centimetre were generated by setting the beam step size equal to the dot pitch of 70, 100 or 200 nm, so that the exposure of each dot was performed within a single “beam shot” with a dose of 105, 75 and 14 μC cm −2, respectively. The exposed samples were developed in an isopropanol (IPA) : methyl-isobutylketone (MIBK) 3:1 mixture for 45 s, rinsed in IPA for 30 s and dried by centrifugation at 3000 rpm. Deposition of Pt on the developed samples was performed with a DC magnetron-sputtering device TIPSI with an Ar pressure of 10 −3 mbar and a power of 30 W. Alternatively, vapour deposition of Pt was performed in a Balzers BAK 600 (Oerlikon Balzers, Liechtenstein) evaporator. After Pt deposition the lift-off process was accomplished by soaking the samples overnight in acetone and rinsing them with acetone and isopropanol the following day. 2.2. Characterisation methods and instrumentation The surface of the obtained samples was investigated by scanning electron microscopy (SEM) (ZEISS SUPRA 55 VP). The detection of back scattered electrons was carried out by an In-Lens detector in vacuum at acceleration voltages between 2 and 4 kV with a working distance between 2 and 7 mm. STM measurements were conducted with a 10 μm STM scanner in constant current mode with the set point current of 1 nA using an

1923

Agilent PicoLe™ microscope equipped with a bipotentiostat (PicoStat, Molecular Instruments). The in situ EC STM measurements were conducted in an electrochemical STM cell with a three electrode arrangement where the sample acted as working electrode (WE) and two annealed high purity Pt wires with a diameter of 0.5 mm (FG 999.5, Carl Schaefer AG) as quasi-reference (RE) and counter (CE) electrodes, respectively. An area of 0.8 cm 2 of the working electrode was exposed to the electrolyte (Silicon sealing, inner diameter 10 mm). The 0.5 M H2SO4 electrolyte was diluted from sulphuric acid 95–97% (pa, Merck) using UHQ (ultra high quality, Millipore) water (> 18 MΩ cm). For tip preparation a gold wire with a diameter of 0.25 mm (FG 999.9, Carl Schaefer AG) was etched in hydrochloric acid (30% HCl, Baker Analyzed™ Reagent) by applying a DC voltage of 1.7 V between the gold wire and the surrounding Pt ring. After etching, the tips were immersed into hot UHQ water (Millipore, > 18 MΩ cm and partially coated with apiezon wax (B7276, Plano GmbH, Germany). The STM images were processed using WSxM 5.0 Develop 3.1. software [16]. The mean diameters of Pt dots on the support were determined from SEM and STM images using image software Image J 1.43 (W. Rasband, NIH, USA). 2.3. Electrochemistry The oxidation of Pt/HOPG model electrodes was performed in air saturated 0.5 M sulphuric acid by the means of stepwise potential increase (potentiostatic hold) or potential cycling at room temperature. All potentials in the paper have been converted to the standard hydrogen electrode (SHE) scale by adding 0.89 V to the values measured with the Pt quasi-reference electrode which was calibrated against an Hg/Hg2SO4 reference electrode. For the potential hold measurements, the STM tip was approached under potential control at Estart =0.7 V. An STM image was recorded within 8–10 minutes (scan rate: 0.8–1.0 lines/s). Then, the potential was increased by 0.1 V at a scanrate of 5 mV s −1 and an STM image acquired at 0.8 V, the time for the image acquisition was again 8–10 min. The procedure was repeated in 0.1 V steps up to the potential of 1.9 V. After reaching 1.9 V, the potential was scanned back to Eend = 0.7 V with a scanrate of 5 mV s−1 and again images were recorded at Eend = 0.7 V. In case of potential cycling, the scan-rate was set to 20 mV s −1 and the sample cycled for 10 cycles between 0.6 V and the upper vertex potential, which was stepwise increased from 1.0 V to 1.9 V. All images were recorded at the lower vertex potential of 0.6 V, avoiding gas evolution on the electrode and bias-related artefacts possibly occurring during the potential hold experiments, as the STM tip was set to a constant potential of 0.4 V. The time for the image acquisition was again 8–10 minutes. 3. Results 3.1. Preparation and characterisation of Pt/HOPG model electrodes Typical SEM images of model electrodes after the lift-off procedure are shown in Fig. 1a, b. Fig. 1a represents an image of an area with a complete pattern, while Fig. 1b displays an example of an only partially successful exposure/lift-off. Some areas on HOPG were “under” exposed, bearing no dots at all (a1 in Fig. 1b) while other areas were still covered by Pt covered photoresist after lift-off (a2 in Fig. 1b). Finally, on some areas both the exposure and the lift-off process went well exhibiting an intact pattern (a3 in Fig. 1b). However, on most of the structured samples, areas of (50 × 50) μm 2 showing almost intact pattern of Pt dots could be observed with SEM. Fig. 1c, d show the particle size distribution on two different samples by means of two histograms. The vertical lines indicate the respective mean diameter as obtained by this analysis. Table 1 lists

1924

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

a

b

c

d

Fig. 1. SEM images of areas with a) a complete and b) an incomplete lift-off on a Pt/HOPG EBL sample (pitch 100 nm) indicating an “under”exposed area with missing dots (a1), Pt/photoresist residues (a2) and an area with intact dot pattern (a3); c) and d) diameter distribution by means of two histograms on two different samples (CI and AI, see Table 1) at different magnification.

the mean, min and max diameters, and the standard deviation of the analysis as obtained from SEM images of four different samples (A, B, C, D) at different locations (I, II) and magnifications (see table). From this statistics the diameters of the Pt dots range from 22 to 45 nm, the mean diameter of the different samples varies between 29 and 35 nm. The STM measurements in air confirmed the SEM observations. Well lifted areas showing an intact pattern could be observed on most of the EBL structured samples. The Pt dot pattern remained stable in air for longer than 10 days, as shown for a Pt/HOPG model electrode in Fig. 2a. “Under” exposed and Pt/PMMA covered areas could also be observed (not shown here). Finally, areas in which the pattern was not complete, as some of the Pt dots were missing, could be

Table 1 Mean, min and max diameters, and the standard deviation of the prepared Pt dots as obtained from SEM images of four different samples (A, B, C, D) at different locations (I, II) and magnifications. Sample

Magnification

No. of particles

dmean, nm

dmin, nm

dmax, nm

Std. Dev., nm

AI AI AII AII BI CI CII DI DII

200 500 200 250 150 100 150 100 250

45 6 25 18 88 110 314 150 29

30.2 29.6 32.2 32.2 34.4 34.6 34.4 32.6 32.3

23.5 27.3 23.4 23.3 27.7 26.4 24.6 22.8 27.2

35.1 31.5 42.6 39.8 44.1 45.6 43.8 41.0 36.9

2.5 1.5 5.3 4.0 3.6 4.2 4.6 4.3 2.6

k k k k k k k k k

observed (Fig. 2b). The STM images of locations showing incomplete pattern revealed that the otherwise intact HOPG surface showed a slight corrugation at the missing dot locations (indicated by the dashed black circles in Fig. 2b). Fig. 2c, d show the particle size distribution on two different samples by means of two histograms. The vertical lines indicate the respective mean diameter as obtained by this analysis. Table 2 lists the mean, min and max diameters, and the standard deviation of the analysis as obtained from STM images of four different samples (A, B, C, D) at different locations (I, II, III). The statistics for these samples indicate diameters between 19 and 43 nm, the mean diameter is in the range of 25 to 34 nm. Considering the standard deviation of 7– 13%, the mean diameters as determined from the STM data are in good agreement with the ones obtained from SEM images. The Pt dots could be well-resolved with STM (Fig. 3a). High resolution STM imaging revealed that a dot of 30–40 nm in diameter and 3–5 nm in height can be described as agglomeration of smaller particles of up to 5 nm in size (line scan, Fig. 3a). The missing dot locations (Fig. 3b) appear as undefined spots, having a diameter of approximately 20 nm and a roughness well below 1 nm on otherwise well-defined HOPG basal planes (line scan, Fig. 3b). This observation as well as the observed stability of the Pt dots suggests that the immobilisation of the dots was most probably due to defects at the Pt/HOPG boundary on the HOPG surface, created by e-beam damage or sputtering. The e-beam damage is not unlikely, as it is reported that radiation can cause severe damage to carbon substrates [17]. To investigate whether the EBL process created the defects, a freshly cleaved HOPG was spin-coated, patterned by the EBL, lifted in acetone and, subsequently, scanning probe microscopies

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

a

b

c

d

1925

Fig. 2. a) STM image of Pt/HOPG EBL model electrode (pitch 200 nm) in air after lift-off and subsequent storage under ambient conditions for 10 days; The tunnelling current is 1 nA, the bias is −0.2 V; the Δz-range is 8 nm; b) STM image of Pt/HOPG EBL model electrode in air, the black dotted circles indicate defects instead of dots. The tunnelling current is 1 nA, the bias is −0.2 V; the Δz-range is 3 nm; c) and d) diameter distribution and by means of two histograms on two different samples (CI and DI, see Table 2) at different magnification.

were employed to look for the defects. However, neither SEM nor STM (air) revealed any presence of defects on the HOPG. Omitting the spin-coating step and structuring a freshly cleaved HOPG surface by EBL also revealed the absence of defects, as measured by SEM and STM (air). Sputtering may be another cause for defect creation, as the sputtered Pt atoms hit the substrate with a speed of 2 km s−1 (4 eV). So-called implantation, slight metal cluster embedding into the surface, and local damage of the surface is known to occur at high impact energies (>1 eV) [18].

Table 2 Mean, min and max diameters, and the standard deviation of the prepared Pt dots as obtained from STM images in air of four different samples (A, B, C, D) at different locations (I, II, III). Sample

No. of particles

dmean, nm

dmin, nm

dmax, nm

Std. Dev., nm

AI AII BI BI BII BIII CI DI DII

41 18 39 25 3 5 55 17 5

27.5 32.6 28.6 31.2 34.1 26.9 25.0 28.6 27.0

23.3 19.0 17.4 26.3 29.1 25.6 21.8 24.0 26.7

35.4 37.8 42.9 36.4 36.6 28.3 31.8 32.2 27.4

3.3 4.5 4.5 3.0 4.3 1.1 2.0 2.3 0.4

In order to figure out whether the defects are due to sputtering damage, the sputtering step was replaced by vapour deposition. After the lift-off step, the SEM images of the surface showed a high amount of Pt dots, which arrangement, however, was not according to the pattern written by the e-beam (Fig. 4a). Some of the dots were still part of a square pattern with a pitch of 200 nm (see white circles, Fig. 4b), but most of them were randomly distributed on the HOPG surface or clustered into long strings (Fig. 4a), probably decorating the steps and the edges of the HOPG substrate, thus assuming the energetically more favourable position [14]. All the above-mentioned data indicates that the Pt dots of 30–40 nm in diameter and 3–5 nm in height are “anchored” to defects present at the Pt/HOPG boundary and created by the sputter process. The observed defects probably indicate locations at which this “anchoring” boundary was not sufficient to adhere the Pt particles during the lift-off procedure. 3.2. In situ STM investigations of Pt/HOPG model electrodes The in situ STM measurements were performed in 0.5 M sulphuric acid in laboratory atmosphere. Electrochemical oxidation of Pt/HOPG model electrodes was done by means of stepwise potential increase followed by potential hold during the STM image acquisition or potential cycling with increasing upper vertex potential and constant lower vertex potential of E = 0.6 V, which was also chosen as imaging potential.

1926

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

Fig. 3. STM image of Pt/HOPG EBL model electrodes in air. a) image (left) and corresponding height profile (right) of a Pt dot (Δz-range: 8 nm); b) image (left) and corresponding height profile (right) of a defect (Δz-range: 2 nm). The tunnelling current is 1 nA, the bias −0.2 V.

3.2.1. Stepwise potential increase A cyclic voltammogram (CV) recorded during an in situ STM measurement is shown in Fig. 5 for a sample with a pitch of 200 nm. The current measured upon immersion at 0.7 V was typically between − 0.5 and − 2.0 μA, which may be ascribed to oxygen reduction as the measurements were performed under ambient conditions. Upon potential increase the current increased and zero crossing always occurred at a potential of 0.89 ± 0.03 V. The oxidation current further increased with further potential increase due to Pt oxide formation at E > 0.9 V and oxygen evolution at E > 1.4 V [19,20]. The strong oxidation currents observed at E > 1.7 V may be attributed to mixed contributions of oxygen evolution and carbon oxidation. The oxygen, the oxidised Pt and C may be reduced during the backscan to Eend = 0.7 V (between 1.0 V and 0.7 V).

a

As already described in ref. [13], in situ STM measurements show that this electrochemical treatment leads to a disappearance of the Pt dots and to the appearance of a hole pattern, congruent to the formerly existing Pt pattern. This observation could be reproduced for all investigated samples, as exemplarily shown in the STM images of Fig. 6a, showing the Pt dots at 0.8 V and the holes of the same location after the in situ STM experiment at Eend = 0.7 V. For two samples showing a complete Pt dot pattern (pitch 200 nm) and one sample showing an area with incomplete lift-off (pitch 100 nm), the quality of the STM data was statistically relevant (more than 40 dots in the scanned area, no significant drift or noise during the acquisition time of the in situ experiment) to plot the disappearance of the Pt dots as a function of the applied potential (Fig. 6a). The plot shows the change of the relative number of dots

b

Fig. 4. SEM images of a Pt/HOPG model electrode, upon which Pt has been deposited via vapour deposition. The circles in (b) trace remains of the EBL pattern.

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

Fig. 5. CV of a Pt/HOPG model electrode recorded during an in situ STM experiment (pitch 200 nm) in air saturated 0.5 M H2SO4 electrolyte.

(N Pt dots/%) with respect to the number of dots appearing in the STM image recorded at Estart = 0.7 V (100%) against the oxidation potential of the subsequent STM scans of the same area. A dot is no longer counted if its apparent height is smaller than 1.5 nm in the STM image. This statistic indicates that the Pt dots are stable up to a potential of 1.4 V. Further potential increase leads to a disappearance of the dots; at 1.8 V the dots are completely removed from the HOPG surface. In order to follow changes in the time domain, the potential of sample 200 nm B was held at 0.9 V and at 1.6 V for > 30 min, while recording STM data (Fig. 6b). No Pt particle loss is observed within this time range at a potential of 0.9 V (black triangles in Fig. 6b), indicating that tip induced Pt loss may be excluded not only for environmental air, but also for scanning in the sulphuric acid electrolyte. The potential hold images at 1.6 V show that this potential is sufficient to remove all dots from the surface and to form holes. The depth of the holes is determined to be 1.5 nm ± 0.3 nm in case of sample 200 nm B after 32 minutes at 1.6 V. Direct comparison of the depth of individual holes observed after 32 min (inset Fig. 6b) with the values obtained for the same holes after potential increase to 1.9 V and scanning back the potential to 0.7 V shows no significant changes. Comparison of dimensions of individual dots recorded at 0.8 V with the corresponding holes after potential increase to 1.9 V and scanning back the potential to 0.7 V indicates that the holes are of same or smaller size than the formerly present dots. Fig. 7 shows some selected in situ STM images of a recorded sequence, displaying an area with a complete Pt pattern and two step edges of the HOPG substrate (sample 200 nm A in Fig. 6). The height

a

1927

of step 1 is determined to be 0.7 nm, the one of step 2 determined to be 0.4 nm and the Pt particles show an average height of (3.0 ± 0.5) nm and a mean diameter of 24 nm at Estart = 0.7 V (Fig. 7a). At a potential of E ≥ 1.0 V the Pt dots appear fuzzy, complicating the exact determination of an average particle diameter and height. However, more than 90% of the dots, showing a minimum height of 2.5 nm, are present on the surface at a potential of 1.4 V (Fig. 7b). The shape of steps 1 and 2 does not show any significant change, if compared to the image recorded at 0.7 V. During the STM scan at 1.5 V the number of Pt particles starts to decrease significantly (Fig. 7a, 200 nm A), whereas the steps keep their shape. At 1.6 V most of the Pt particles are disappeared, defects, indicated by spots having a roughness below 1 nm, appear instead and carbon corrosion starting from step edges of the upper terrace is observed (see markers in Fig. 7c). Further potential increase leads to corrosion of large parts of the upper basal plane. First holes appear in the STM image at E = 1.9 V (not shown here). Scanning back the potential to Eend = 0.7 V shows the surface damages resulting from the applied oxidation procedure (Fig. 7d). The depth of the holes is determined to be 0.4–0.7 nm with a mean diameter of about 18 nm. The dotted circles in Fig. 7d indicate locations at which Pt dots were observed on the upper terrace at the beginning of the experiment. For these locations no holes are observed in the STM image. Fig. 8 shows a sequence of in situ STM images of an area with an incomplete pattern (pitch 100 nm). Well resolved Pt dots, the defects described in chapter 3.1 (black, dotted circles, inset Fig. 8a) as well as locations with completely unmodified HOPG areas are present on the surface at E = 1.0 V. The average dot height is 4.0 nm ± 0.5 nm. Potential increase in 0.1 V intervals and subsequent STM measurements show that the sample surface remains unchanged up to 1.4 V (compare Fig. 8a, b). Upon application of a potential of 1.5 V, some Pt dots start to disappear, as highlighted by black circles in Fig. 8c and d, the number of missing dots further increases for potentials of 1.6 V (Fig. 8d, yellow circles) and 1.7 V (Fig. 8e). Pronounced corrosion of an HOPG basal plane can be observed at 1.7 V (see white arrow, Fig. 8e). At a potential of 1.8 V all initially present dots or defects are replaced by holes and the corrosion of the upper basal plane proceeds further (Fig. 8f). A more detailed analysis of the images indicates that the holes formed at locations having Pt dots on top prior to the oxidation appear deeper (1.4 nm ± 0.3 nm) in comparison with the deepenings measured at locations showing defects on top of the HOPG prior to the oxidation (b 1.0 nm) (compare dotted circles and former dot locations in the insets in Fig. 8). The mean diameter of the Pt dots is determined to be between 24 and 26 nm in the potential range of 0.7 V ≤ E ≤ 1.5 V and 22 nm for the holes recorded at Eend = 0.7 V.

b

Fig. 6. a) relative number of Pt dots (in %) present on the sample in dependence on the applied potential; b) relative number of Pt dots (in %) present on the sample in dependence on the potential hold time at E = 0.9 V (black triangles) and E = 1.6 V (white triangles); pitch as indicated.

1928

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

lower vertex potential of 0.6 V and increasing anodic vertex potentials (Erev) of 1.x V (0≤x ≤9). The images were recorded at 0.6 V, thus avoiding any possible bias-related artefacts in the STM imaging, which might have occurred during the potential hold experiments. Some selected CVs recorded on a Pt/HOPG model electrode with a pitch of 70 nm during an in situ STM experiment are shown in (Fig. 9a). Considering the reduction current recorded at 0.6 V as an indicator for Pt oxide, carbon oxide, and oxygen reduction one may conclude that the electrode shows highest activity at the vertex potentials of 1.4 V and 1.5 V. Further increase of the upper potential leads to a decrease of the reduction current. The oxidation currents occurring in a potential range of 1.0 V b E b 1.5 V also show a decrease. A more detailed analysis of the recorded CVs is given in Fig. 9b. Here the reduction currents measured at the potential of E = 0.6 V are plotted in dependence of the vertex potential, Erev. The values for the first (black dots) and the last (white dots) cycles are given, respectively. One can qualitatively distinguish between an “activation” region (E ≤ 1.4 V) and a “deactivation” region (E ≥ 1.5 V). The deactivation may be attributed to a surface area loss of Pt caused by corrosion or ripening processes. Fig. 10 shows some selected STM images (left) recorded during the cycling experiment described above. The corresponding line scans are shown on the right. The series starts with an almost intact Pt dot pattern on a model electrode with a nominal pitch of 70 nm recorded after cycling the sample between 0.6 and 1.3 V. The pattern stays almost intact up to Erev = 1.4 V. The line scan does not reveal any changes in apparent height of the Pt dots if compared to 1.3 V (approximately 6 nm), the diameter of the dots, however, seems to decrease. This observation could also be an artefact, during upwards and downwards scan the pattern of the Pt dots does not show the same orientation, and, therefore, the dimensions of the dots may not be compared directly. However, after increase of the vertex potential to 1.5 V the Pt dots disappear. The terrace appears mainly flat in the image and a few tiny holes become visible in the lower part of the image (see also line scan). The holes become more and more pronounced for increasing potentials and very well-resolved holes showing a depth of 0.5 - 0.9 nm are recorded upon the potential back-scan to 0.6 V after cycling to the final vertex potential of 1.9 V (Fig. 10, bottom). The diameter of the holes is 20 nm ± 5 nm, which is significantly smaller if compared with the apparent diameter of the Pt dots of > 30 nm recorded after cycling to 1.3 V (compare line scans Fig. 10 top and bottom). The line scans (Fig. 10d) show no indication of possible decoration of the step edges of the holes (see also inset image in Fig. 11). A summary of the experiments performed by potential cycling is displayed in Fig. 11. Herein, the potential values are plotted against the abundance of dots on the surface. The plots show that the first cycles up to 1.3 V do not lead to any significant change of the number of the dots. In all cases the dots disappear completely at an upper vertex potential of 1.6 V. 4. Discussion

Fig. 7. EC-STM images of a Pt/HOPG model electrode (pitch 200 nm) in 0.5 M sulphuric acid at a) 0.7 V, b) 1.4 V, c) 1.6 V, d) 0.7 V. Tip potential is 0.4 V. The Δz-range is 5 nm, potential and scan direction as indicated, arrows indicate carbon corrosion, dotted circles mark former dot locations.

3.2.2. Potential cycling For some model electrodes potential cycling experiments with stepwise increased anodic vertex potentials were performed. The scan-rate was set to 20 mV s −1 and the sample cycled for 10 cycles between the

In contrast to previous results obtained on GC [12] the prepared model electrodes used for this study are not perfectly patterned over the whole structured area of 1 cm 2. This is supposed to mainly refer to the surface properties of the HOPG. The surface of this substrate has a wavy nature on a macroscopic level, which leads — as opposed to the mirror-like GC — to uneven spreading of the photoresist during the spin-coating step. During the e-beam scan, the areas with different photoresist thicknesses are exposed to the same radiation dose, leading to under-exposure of some areas as shown in Fig. 1b. Therefore, the current responses recorded in the electrochemical experiments can only be considered qualitatively as the amount of un-lifted Pt contributing to oxidation and reduction currents is not accessible. However, on the length scale relevant for the STM investigations (scan

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

1929

Fig. 8. EC-STM image of a Pt/HOPG model electrode (pitch 100 nm) in 0.5 M sulphuric acid at a) 1.0 V, b) 1.4 V, c) 1.5 V Black circles indicate locations of dots disappearing first, d) 1.6 V yellow circles mark locations of dots disappeared at 1.6 V, e) 1.7 V, the arrow indicates carbon corrosion, f) 1.8 V. The insets magnify a location showing dots and defects (dotted circles). Tip potential is 0.4 V. The Δz-range is 7.5 nm for a–e and 5 nm for f. Potential and scan direction as indicated.

a

b

Fig. 9. a) CVs recorded for a Pt/HOPG EBL electrode (pitch 70 nm) in 0.5 M sulphuric acid during an in situ STM experiment; b) evolution of reduction current at 0.6 V during cycling in dependence of the upper vertex potential (Erev); black circles mark the values of the first, white circles of the tenth cycle up to the respective Erev.

1930

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

Fig. 10. EC-STM images of a Pt/HOPG model electrode (pitch 70 nm) in 0.5 M sulphuric acid after cycling to the upper limit of a) 1.3 V; z-range: 10 nm, b) 1.4 V; z-range: 10 nm, c) 1.5 V; z-range: 4 nm and f) after cycling to 1.9 V upon the end of the cycling experiment; z‐range: 2 nm. Tip potential is 0.4 V, bias: −0.3 V. Scan direction as indicated.

window smaller than 10 μm × 10 μm) well lifted areas showing an intact pattern could be observed on most of the EBL structured samples and reproducible observations were made allowing for some conclusive statements and further discussion.

The fact that the lithographically prepared Pt/HOPG model electrodes could be investigated via in situ EC STM is very interesting as it is known that Pt dots of this size are rather mobile on a smooth un-treated HOPG surface in air [14,15].

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

Fig. 11. Relative number of Pt dots (in %) present on the sample after cycling at a given upper vertex potential (Erev), the inset image shows a high magnification image of the holes at the end of an experiment.

As elaborated in Section 3.1, this is most probably the result of the generation of a defective interface between the HOPG and the Pt dots caused by the sputter procedure applied for the metal deposition step. The introduced defects are supposed to act as anchors for the Pt dots and show a corrugation of b1 nm. The defects observed in air and the holes generated after electrochemical removal of the Pt dots appear with similar or smaller dimensions than the formerly present dots. This may be explained by the undercut profile generated during the E-Beam procedure by using the two layer photoresist. Probably, the sputtered Pt atoms diffuse to the area shadowed by the upper PMMA layer during the sputter process where they coalesce with Pt anchored to the substrate (schematically shown in Fig. 12). High resolution STM images (Fig. 3a) show that the Pt dots consist of agglomerates of Pt particles of up to 5 nm in size. This grainy structure may also be a consequence of coalescence of sputtered Pt atoms. Crystallinity, orientation of the surface atoms or surface roughness of the Pt is, however, unknown. Three different locations on the model electrodes are considered in the following discussion: i) the Pt dots anchored on the HOPG; ii) the defects without Pt on top and iii) the step edges of the HOPG. 4.1. Stability of the Pt dots on HOPG The collected data of the potential hold experiments indicate that the Pt dots are stable on the HOPG at E = 0.9 V for > 30 min and up to an electrode potential of E = 1.4 V for the time of image acquisition at the respective potentials, i.e. 8–10 min. This is in good agreement with data published for polycrystalline Pt as well as Pt single crystal electrodes [19,21–25]. Inzelt et al. [23] as well as Jerkiewicz et al. [22]

1931

have shown by electrochemical quartz nanobalance (EQCN) measurements on bulk Pt electrodes that Pt oxide formation occurs in the potential range of 0.85≤ E ≤ 1.4 V. The exact mechanism is still under discussion, but there is common consent that it proceeds via O/OH adsorption followed by “place exchange” reactions occurring at E >1.2 V leading to the development of a Pt surface oxide with a bilayer structure at higher potentials [26]. Very recently, Pt oxide formation was followed via in situ STM in 10 mM HF solution on a Pt(111) surface, showing that the O/OH adsorption occurs together with a formation of tiny spots of 0.5 nm diameter and a height increase of 0.08 nm, which were attributed to adsorbed oxygen on the single crystal surface [25]. An in situ AFM study performed on cubic, iodine capped Pt nanoparticles supported on either a platinum or a carbon substrate in air saturated 0.1 M NaClO4 electrolyte indicated a height change of up to 1 nm in the potential region of Pt oxide formation [27,28]. In our in situ STM studies the Pt dots have a consistently fuzzy appearance, therefore no height changes caused by O or OH adsorption or Pt oxide formation could be resolved in detail. However, we can conclude that in agreement with measurements performed on bulk platinum as well as on cubic Pt nanoparticles supported on a carbon substrate, the Pt dots anchored on the HOPG substrate are stable during the Pt oxide formation process (Fig. 13b).

4.2. Dissolution of the Pt dots on HOPG At potentials above 1.4 V the Pt dots start to disappear (Fig. 13c). The dissolution proceeds with time and potential (Fig. 6). As evident from the recorded in situ STM sequences several different surface processes may be considered. At some locations deepenings indicating the final hole pattern on the HOPG are observed directly after disappearance of the Pt dots (Figs. 8c, d, 13e). At other locations the dissolution appears to proceed via the defects induced by the sputtering, which is indicated by the almost flat appearance of the basal plane where neither dots nor holes are observed (deduced from the image of Fig. 7c and schematically shown in Fig. 13d). Some dots accidentally located at step edges of the HOPG disappear at 1.6 V, other dots at comparable locations stay on the surface up to 1.7 V, similar to those located on the basal planes (Fig. 8d, e). Therefore, detailed determination of the mechanisms behind the Pt loss remains complex, although strong efforts were made to prepare well defined model electrodes and study these mechanisms using this imaging method. However, it is evident that the Pt oxide film formed on the model electrodes does not prevent the Pt from being dissolved at E > 1.4 V. At first glance, this observation does not seem to be in good agreement with literature, as it is generally assumed that the Pt oxide layer formed at higher potentials protects the platinum from anodic dissolution [29,30]. Wang et al. [29] observed that the equilibrium concentration of dissolved Pt increases monotonically from 0.65 to 1.1 V and decreases at potentials >1.1 V vs SHE in 0.57 M perchloric acid electrolyte. However, these results were obtained from bulk platinum electrodes after holding the WE at the respective potential for 76 h. In our study Pt loss occurs much faster, i.e. the dots disappear within 32 min at 1.6 V. This strongly suggests that one has to consider the nature of the Pt/C interface. Due to the defect generation during sputtering this interface is supposed to be very prone to carbon corrosion. According to [31] the oxidation of carbon to CO2 (Eq. (1)), which is catalysed by the presence of platinum, proceeds at an appreciable rate at the electrode potentials greater than approximately 1.4 V. þ

−

C þ 2H2 O→CO2 þ 4H þ 4e Fig. 12. Schematic drawing of the Pt/HOPG interface induced by the sputter procedure applied for metal deposition on the E-Beam structured photoresist.

ð1Þ

Chaparro et al. [20] showed by means of “membrane inlet mass spectrometry” (MIMS) in 0.5 M H2SO4 at room temperature that CO2

1932

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

can be formed at this interface under potentiostatic conditions at E > 1.4 V. However, for determination of the exact mechanisms behind the Pt loss during the potential hold procedure, more experiments would be required. Especially the time scale and resolution of the imaging should be improved. The potential cycling experiments show that at an upper vertex potential of 1.6 V all Pt dots vanish within 10 cycles (Fig. 11) corresponding to approx. 200 s at an electrode potential of >1.4 V. This observation indicates that dissolution of the Pt dots is accelerated if compared to the potential hold statistics which is in good agreement with literature. The effect of Pt dissolution accelerated by potential cycling on platinum is well known ([2] and references therein) and recently attracted new attention (see for instance [23,26,32]). Kim and Meyers [26] published a combined EQCN, rotating ring disc electrode and inductively coupled plasma mass spectroscopy study on Pt dissolution using a commercial Pt black catalyst (E-Tek). The EQCN measurements showed that during cycling in 0.5 M H2SO4 electrolyte, the dissolution rate is highest if the lower potential limit is in the range of 0.2–0.6 V and the upper potential limit set to 1.6 V. They concluded that the protecting PtO2 film formed during oxidation at E > 1.4 V might form soluble Pt 2+ species during reduction according to reaction 2. 4.3. Carbon corrosion

Fig. 13. Schematic drawing of a Pt nanodot anchored to HOPG at low potentials(a), indicating Pt oxide layer formation(b), dissolution(c) and carbon corrosion (d, e) at given electrode potentials.

is detected at 1.5 V on an uncatalysed, Vulcan covered carbon cloth. For the carbon cloth covered with a film of Pt/C catalyst these authors observe CO2 generation already at 0.35 V and show a pronounced increase of CO2 at E > 1.0 V. Roen et al. [4] attributed the feature occurring at low potentials to electro-oxidation of CO adsorbed on the catalyst surface. The fact, that we do not observe significant carbon corrosion at lower potentials may refer to the different substrate morphologies of HOPG if compared to carbon blacks. However, the onset of reaction (1) may comprise several consequences detrimental to the stability of the Pt/C boundary explaining our observations. Firstly, the anchors assumed to be responsible for the good adhesion of the Pt dots would be destroyed. Secondly, the oxidised carbon atoms may leave electrons at the catalyst surface, which might lead to reduction of presumably formed Pt 4+ to Pt 2+ thus forming the soluble platinum species according to: þ



2þ

PtO2 þ 4H þ 2e →Pt

þ 2H2 O

ð2Þ

The third point is that the pH is locally decreased by the release of protons, and therefore, might also enhance the probability of chemical dissolution of PtO (Eq. (3)) as for instance accounted in ref [30]: þ

2þ

PtO þ 2H →Pt

þ H2 O

ð3Þ

In summary, our results indicate that the Pt attached to the defects may catalyse the corrosion of defective HOPG locations and, therefore, the anchoring sites are corroded and/or no stable PtO2 layer

Carbon corrosion has been visualised at E > 1.5 V on two different locations on the prepared model electrodes, these are i) the upper terraces of the basal planes, and ii) the locations anchoring the Pt dots forming the hole pattern after the electrochemical oxidation (Figs. 8 and 9). The holes seem to grow in depth with time and potential. The final depth of the holes varied from sample to sample and was determined to be in the range of 0.4–1.5 nm, corresponding to 1–5 layers of the HOPG substrate. Probably, this rather broad range refers to locally different surface damage introduced by the Pt sputtering process. However, no obvious correlation of the final depth of the holes and the stability potential of the Pt dots on the HOPG is observed. As evaluated in Section 3.2, the holes generated after electrochemical removal of the Pt dots appear with smaller or similar size if compared to the Pt dots formerly present on the HOPG. Once formed, no significant growth in diameter of the holes with time or potential could be observed. Pronounced corrosion of basal planes starting from a hole has likewise never been observed in our study. This observation might indicate that carbon corrosion is restricted to the defective area underneath Pt (Fig. 13e). The remaining sites bordering the holes seem to be stable under the conditions applied here. However, one might also have to consider that convolution with the actual tip shape can let protrusions appear broader than depressions. Well pronounced corrosion of carbon always starts at step edges of the basal planes. The images shown in Fig. 7c may indicate that defects at the step edges are most prone to corrosion. As carbon corrosion occurs parallel to Pt dissolution, this process might also be catalysed by Pt which probably diffuses to step edges of the HOPG during the dissolution process. However, as already stated above, more work is required to study these processes in more detail, especially, speed and resolution of the imaging should be improved. 5. Conclusion The STM measurements on Pt/HOPG EBL model electrodes have shown that the samples are stable in air and in 0.5 M sulfuric acid electrolyte. The immobilisation of the Pt dots seems to be a consequence of the sputter-induced surface defects anchoring the Pt dots on the HOPG.

A. Foelske-Schmitz et al. / Surface Science 606 (2012) 1922–1933

The Pt dots are stable at E = 0.9 V for > 30 min and during the Pt oxide formation process up to E = 1.4 V for at least 8–10 min at room temperature. Gradual disappearance of the dots and appearance of a hole pattern congruent to the original dot pattern occurs at E > 1.4 V. Hole formation refers to carbon corrosion and is found to be restricted on the defective areas underneath the Pt dots. The Pt dissolution process is accelerated by potential cycling. Corrosion of larger parts of the basal planes of the HOPG starts at the step edges of the terraces. As this process occurs parallel to the Pt dissolution it might be catalysed by Pt. Acknowledgements The authors would like to thank Michael Horisberger (LDM, PSI) for Pt sputtering and Swiss National Science Foundation (SNSF) for financing the project 200021_121719/1. References [1] P.J. Ferreira, G.J. la O’, Y. Shao-Horn, D. Morgan, R. Makharia, S. Kocha, H.A. Gasteiger, J. Electrochem. Soc. 152 (2005) A2256. [2] R. Borup, J. Meyers, B. Pivovar, Y.S. Kim, R. Mukundan, N. Garland, D. Myers, M. Wilson, F. Garzon, D. Wood, P. Zelenay, K. More, K. Stroh, T. Zawodzinski, J. Boncella, J.E. McGrath, M. Inaba, K. Miyatake, M. Hori, K. Ota, Z. Ogumi, S. Miyata, A. Nishikata, Z. Siroma, Y. Uchimoto, K. Yasuda, K. Kimijima, N. Iwashita, Chem. Rev. 107 (2007) 3904. [3] M.S. Wilson, F.H. Garzon, K.E. Sickafus, S. Gottesfeld, J. Electrochem. Soc. 140 (1993) 2872. [4] L.M. Roen, C.H. Paik, T.D. Jarvi, Electrochem. Solid-State Lett. 7 (2004) A19. [5] D.A. Stevens, J.R. Dahn, Carbon 43 (2005) 179. [6] S. Maass, F. Finsterwalder, G. Frank, R. Hartmann, C.J. Merten, J. Power. Sources 176 (2008) 444. [7] N. Linse, L. Gubler, G.G. Scherer, A. Wokaun, Electrochim. Acta 56 (2011) 7541.

1933

[8] Y.Y. Shao, G.P. Yin, Y.Z. Gao, J. Power. Sources 171 (2007) 558. [9] J.V. Zoval, J. Lee, S. Gorer, R.M. Penner, J. Phys. Chem. B 102 (1998) 1166. [10] R. Lindström, Y. Seidel, Z. Jusys, M. Gustavsson, B. Wickman, B. Kasemo, R.J. Behm, J. Electroanal. Chem. 644 (2010) 90. [11] A. Savouchkina, A. Foelske-Schmitz, R. Kötz, A. Wokaun, G.G. Scherer, C. Padeste, J. Ziegler, V. Auzelyte, H. Solak, ECS Trans. 25 (2010) 175. [12] V.A. Guzenko, J. Ziegler, A. Savouchkina, C. Padeste, C. David, Microelectron. Eng. 88 (2011) 1972. [13] A. Savouchkina, A. Foelske-Schmitz, V.A. Guzenko, D. Weingarth, G.G. Scherer, A. Wokaun, R. Kötz, Electrochem. Commun. 13 (2011) 484. [14] F. Atamny, A. Baiker, Surf. Interface Anal. 27 (1999) 512. [15] J. Chen, K.-Y. Chan, Mol. Simul. 31 (2005) 527. [16] I. Horcas, R. Fernandez, J.M. Gomez-Rodriguez, J. Colchero, J. Gomez-Herrero, A.M. Baro, Rev. Sci. Instrum. 78 (2007) 013705. [17] F. Banhart, Rep. Prog. Phys. 62 (1999) 1181. [18] V.N. Popok, I. Barke, E.E. Campbell, K.-H. Meiwes-Broer, Surf. Sci. Rep. 66 (2011) 347. [19] A. Sun, J. Franc, D.D. Macdonald, J. Electrochem. Soc. 153 (2006) B260. [20] A.M. Chaparro, N. Mueller, C. Atienza, L. Daza, J. Electroanal. Chem. 591 (2006) 69. [21] Z. Nagy, H. You, Electrochim. Acta 47 (2002) 3037. [22] G. Jerkiewicz, G. Vatankhah, J. Lessard, M.P. Soriaga, Y.-S. Park, Electrochim. Acta 49 (2004) 1451. [23] G. Inzelt, B. Berkes, A. Kriston, Electrochim. Acta 55 (2010) 4742. [24] B.E. Conway, Prog. Surf. Sci. 49 (1995) 331. [25] M. Wakisaka, S. Asizawa, H. Uchida, M. Watanabe, Phys. Chem. Chem. Phys. 12 (2010) 4184. [26] S. Kim, J.P. Meyers, Electrochim. Acta 56 (2011) 8387. [27] H. Kikuchi, W. Ouchida, M. Nakamura, C. Goto, M. Yamada, N. Hoshi, Electrochem. Commun. 12 (2010) 544. [28] N. Hoshi, M. Nakamura, C. Goto, H. Kikuchi, J. Electroanal. Chem. 667 (2012) 7. [29] X.P. Wang, R. Kumar, D.J. Myers, Electrochem. Solid-State Lett. 9 (2006) A225. [30] R.M. Darling, J.P. Meyers, J. Electrochem. Soc. 150 (2003) A1523. [31] S.D. Knights, K.M. Colbow, J. St-Pierre, D.P. Wilkinson, J. Power. Sources 127 (2004) 127. [32] S.G. Rinaldo, W. Lee, J. Stumper, M. Eikerling, Electrochem. Solid-State Lett. 14 (2011) B47.