In Situ Surface-Enhanced Raman Spectroscopic Studies of Nafion

Nov 21, 2011 - In this study, adsorption of Nafion ionomer on Au and Pt surfaces was ... The SER spectra suggest the direct interaction of sulfonate g...
0 downloads 0 Views 1017KB Size
ARTICLE pubs.acs.org/Langmuir

In Situ Surface-Enhanced Raman Spectroscopic Studies of Nafion Adsorption on Au and Pt Electrodes Jianbo Zeng,† Deok-im Jean,† Chunxin Ji,‡ and Shouzhong Zou*,† † ‡

Department of Chemistry & Biochemistry, Miami University, Oxford, Ohio 45056, United States Electrochemical Energy Research Lab, General Motors LLC, Honeoye Falls, New York 14472, United States

bS Supporting Information ABSTRACT: Understanding interactions between Nafion (perfluorosulfonic acid) and Pt catalysts is important for the development and deployment of proton exchange membrane fuel cells. However, study of such interactions is challenging and Nafion/Pt interfacial structure remains elusive. In this study, adsorption of Nafion ionomer on Au and Pt surfaces was investigated for the first time by in situ surface-enhanced Raman spectroscopy. The study is made possible by the use of uniform SiO2@Au core shell particle arrays which provides very strong enhancement of Raman scattering. The high surface sensitivity offered by this approach yields insightful information on interfacial Nafion structure. Through spectral comparison of several model compounds, vibration assignments of SERS bands were made. The SER spectra suggest the direct interaction of sulfonate group with the metal surfaces, in accord with cyclic voltammetric results. Comparison of present SERS results with previous IR spectra was briefly made.

In this work, we demonstrate for the first time that in situ surface-enhanced Raman spectroscopy (SERS) can be used to probe Nafion structure on Au and Pt electrode surfaces. This study is enabled by Au nanoshells overgrown on SiO2 nanoparticles which provide a very strong Raman enhancement and therefore render superior surface sensitivity. In contrast to the previous infrared (IR) spectroscopic studies where Nafion films casted on Pt were used,5,9,10,12,13,17 the present SERS study was conducted in a diluted Nafion aqueous suspension. Because the spectral contribution from the bulk material is insignificant and SERS is most sensitive to the first layer of adsorbate, the background reduction techniques used in the previous IR spectroscopic studies of Nafion adsorption on Pt are not needed; thereby, the uncertainty caused by the IR spectral subtraction can be avoided. To aid vibration band assignments, vibration spectra of several model compounds were obtained and compared. The SER spectra show a significantly downshifted sulfonate symmetric stretch band, suggesting that Nafion adsorbs with sulfonic groups directly on the metal surfaces. This band is largely absent in the previous IR studies. Based on the spectroscopic results, a tentative model depicting the Nafion metal interface structure is proposed.

1. INTRODUCTION Proton exchange membrane fuel cells (PEMFCs) are attractive future energy sources for transportation and stationary electric powers.1,2 A central component of PEMFCs is the membrane electrode assembly (MEA), which consists of a cathode and an anode separated by a solid polymer electrolyte (membrane) that permits proton conduction while preventing electron conduction. In addition to electrocatalysts, e.g., Pt or Pt alloy nanoparticles supported on a high surface area carbon black, an important component in PEM fuel cell electrodes is the proton conducting polymer. It was not until the early 1990s that proton-conducting ionomers were mixed with catalysts to prepare inks for the electrodes, originally developed by Wilson and Gottesfeld.3 Nafion, a polymer with a tetrafluoroethylene backbone and a perfluorosulfonic acid side chain (Scheme 1), has been extensively used as an effective ionomer for PEMFCs.4 Since fuel cell reactions occur on the catalyst surfaces, the ionomer plays a critical role to shuttle protons within the electrodes. Understanding Nafion-catalyst interfacial structure is of paramount importance for improving PEMFC performance and durability, which are overarching challenges in fuel cell development and deployment.1,2 There are many studies devoted to uncovering the structure of bulk Nafion structure under different conditions,4 17 but few have focused on the Nafion-catalyst interfacial structure by using in situ infrared spectroscopy.5,10,12,13,17 Differences in the measured electrochemically active surface area, oxygen reduction activity, and carbon monoxide (CO) adsorption on Pt electrodes or Pt nanoparticles with and without Nafion coating have been reported by many research groups.10,16,18 20 On the basis of cyclic voltammetric and CO displacement results, Subbaraman et al. concluded that Nafion adsorbs specifically on Pt electrodes through its sulfonate groups.21,22 r 2011 American Chemical Society

2. EXPERIMENTAL SECTION 2.1. Chemicals and Materials. 5 wt % Nafion suspension (equivalent weight 1100, in 50% water and 50% alcohols) was purchased from Alfa Aesar, and 20 wt % Nafion suspension (equivalent weight Received: September 9, 2011 Revised: November 19, 2011 Published: November 21, 2011 957

dx.doi.org/10.1021/la2035455 | Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

Scheme 1. Structure of Nafion and Model Compound PFMDS

Briefly, an atomic layer of Cu was first deposited by underpotential deposition (UPD) on Au nanoshell arrays in 0.1 M H2SO4 + 5 mM CuSO4 solution at 0.05 V for 10 min. Then, the Cu layer-coated Au nanoshell arrays was immersed in a deaerated 0.1 M HClO4 + 5 mM K2PtCl6 for 20 min to replace the atomic layer of Cu . After repeating this procedure four times, the atomic layer of Cu was deposited at 0.1 V for 10 min and replaced by K2PtCl6. To obtain pinhole-free Pt-coated Au nanoshell arrays, this procedure was repeated four more times. 2.4. Spectroscopic Measurements. Surface-enhanced Raman measurements were performed in a two-compartment, three-electrode glass cell with an optically flat glass disk as the window at the bottom. The SiO2@Au array on a glassy carbon electrode was the working electrode, a Pt wire was used as the counter electrode, and a Ag/AgCl electrode with a saturated KCl aqueous solution was the reference electrode. For SERS measurements of model compounds, a Au electrode was used as the working electrode after it was electrochemically roughened to obtain SERS activity following the approach developed by Gao et al.27 The electrode potential was controlled by a voltammograph (CV27, BAS) in potential-dependent experiments. Raman spectra were collected with a portable micro-Raman probe system as described elsewhere.28 The spectra were plotted with the intensity converted to electron counts per second (cps). The IR spectra were recorded using a Perkin-Elmer Spectrum One spectrometer with an ATR accessory. All of the measurements were conducted at room temperature (22 ( 1 °C).

1100, in 42.5 wt % water and 57.5 wt % alcohols) was from DuPont (DE2021). 10 wt % Nafion suspension in water was obtained from Fuel Cell Store (DE1021). The following chemicals are from Sigma Aldrich: sodium 1-hexanesulfonate (HeS, 99.0+%), sodium 1-dodecanesulfonate (DoS, 99.0+%), potassium hexachloroplatinate(IV) (K2PtCl6, 99.99%), tetrachloroauric (III) acid (HAuCl4 3 3H2O, 99%), tetrakis(hydroxymethyl) phosphonium chloride (THPC, 80% solution in H2O), sodium hydroxide pellets (99.998%), ammonium hydroxide (28 30%), 3-aminopropyltriethoxysilane (APTES, 99%), tetraethyl orthosilicate (TEOS, 99+%), and L-ascorbic acid (AA, 99+%). Toluene (99.97%) and ethanol (200 proof) are from Pharmco. Potassium carbonate (K2CO3, ACS certified) and potassium iodide (KI, ACS certified) are from Fisher Scientific. Perchloric acid (HClO4, double distilled, 70%) is from GFS Chemicals. Potassium perfluoro(4-methyl-3,6-dioxaoctane) sulfonate (PFMDS, 99%) and potassium nonafluorobutanesulfonate (NFBS, 99%) is from SynQuest. All of the chemicals were used without further purification. Nafion solution used in the SERS study was diluted to 0.5 wt % from 5 wt % suspension by using 0.1 M HClO4 electrolyte. Glassy carbon electrodes were obtained from CH Instruments. All of the aqueous solutions were prepared by ultrapure water (18 MΩ, TOC < 5 ppb) from a Milli-Q system. 2.2. Preparation of SiO2@Au Nanoshell Arrays. SiO2@Au core shell nanoparticles were synthesized by a modified approach based on that developed by Halas and co-workers.23 Briefly, a solution of 4.0 mL of ammonium hydroxide was mixed with 50 mL of ethanol. 1.5 mL of TEOS was then added drop by drop, and the solution was stirred about 10 h to obtain SiO2 nanoparticles with a diameter of ca. 220 nm. SiO2 nanoparticle surface was then modified with APTES to obtain NH2 groups for anchoring Au nanoparticles. Aqueous solutions of small Au nanoparticles (2 3 nm in diameter) were prepared by the reduction of tetrachloroauric acid with THPC.23 500 μL APTES-modified SiO2 nanoparticles was added into 50 mL 2 3 nm Au solution and stirred for 24 h to attach small Au nanoparticles onto the SiO2 surface. Then, SiO2@Au seeds were washed by centrifugation at 3800 rpm about 8 times to get rid of additional Au seeds in solution. SiO2@Au core shell nanoparticles were formed by adding 1 mM ascorbic acid (AA) and K-gold solution (1.8 mM K2CO3 + 0.375 mM HAuCl4) with a given volume ratio of AA/K-gold/SiO2@Au seeds (e.g., 2.64 mL/40 mL/4 mL). The growth process can be repeated to obtain a thicker Au shell. SiO2@Au core shell particle arrays were formed by a water/oil interfacial entrapment method.24 Details of particle synthesis and array formation will be given elsewhere. The nanoshell arrays were transferred onto desired solid substrates, such as Si wafer, glass, and glassy carbon electrodes for characterizations. The residual organic molecules were removed by using iodide adsorption and oxidation.25 The particles were characterized by a field emission scanning electron microscope (Zeiss Supra 35 VP). 2.3. Preparation of Pt-Coated Nanoshell Arrays. To prepare Pt-coated Au surface, a galvanic redox replacement method was used.26

3. RESULTS AND DISCUSSION The SiO2@Au core shell nanoparticles were synthesized following the approach developed by Halas’ group with some modifications.23 The SiO2 core is about 220 nm in diameter and the Au shell thickness can be varied from 20 to 80 nm by increasing the number of growth cycles. However, the Au shell thickness used in the present study is typically about 50 to 60 nm, which yields optimal SERS intensity with a 785 nm laser excitation. These SiO2@Au core shell particles were assembled onto a solid support by an interfacial entrapment method24 to form ordered arrays of particles that yield uniform SERS signals across a large area (on the scale of millimeters). Figure 1A displays a scanning electron microscopic (SEM) image of SiO2@Au particle arrays formed on a Si wafer. The uniformity of particle size and distribution is clearly evident. Similar arrays can also be obtained on other supports, such as indium tin oxide coated glass slides and glassy carbon electrodes, which are commonly used for electrochemical studies. To obtain SERS on Pt surfaces, an overlayer SERS strategy was used. In this approach,29 ultrathin (nominally about 2 monolayers) Pt films were deposited on SiO2@Au core shell particle arrays supported on glassy carbon electrodes by a redox replacement approach.26 Figure 1B shows an SEM image of the Pt-coated SiO2@Au particle arrays. The Pt film was formed by repetitive redox replacement cycles until the Au surface was completely covered by Pt, as confirmed by (1) the absence of Au oxide reduction peak in cyclic voltammograms (CVs) and (2) the disappearance of C O stretching band from CO adsorbed on residual Au sites at ca. 2100 cm 1 and the appearance of a band at ca. 2060 cm 1 characteristic of CO adsorbed on Pt.29 As evident in the image, the uniformity of the arrays is largely retained. These particle arrays are stable against electrode potential scanning in acidic solutions as suggested by stable CVs obtained in deaerated 0.1 M HClO4. This important property allows their use in in situ electrochemical studies. Details of the particle array preparation and characterization will be given elsewhere. Figure 2 displays SER spectra obtained on SiO2@Au and Pt-coated SiO2@Au particle arrays supported on a glassy carbon 958

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

Figure 1. SEM images of (a) SiO2@Au nanoshell arrays on a Si wafer and (b) Pt-coated SiO2@Au nanoshell arrays on a glassy carbon electrode.

Figure 2. (a) Normal Raman spectrum recorded from Nafion gels; (b,c) SER spectra from SiO2@Au nanoshell arrays in 0.5 wt % Nafion suspension in 0.1 M HClO4 at (b) +0.60 V and (c) 0.20 V; (d,e) SER spectra from Pt-coated SiO2@Au nanoshell arrays in the same solution at (d) +0.60 V and (e) 0.20 V. Spectra are baseline corrected by using the GRAMS/AI.

electrode in a 0.5 wt % Nafion suspension at two different applied potentials (+0.60 V and 0.20 V vs Ag/AgCl). For comparison, a normal Raman spectrum of Nafion gel in water is also included. The Nafion gel sample was prepared by first evaporating the solvent of a 20 wt % Nafion in water/alcohols (mainly n-propanol) suspension. After the solvent evaporation, a given amount of water was added to make up a solution with a nominal concentration of 5 wt %. However, the majority of the Nafion exists as gels in water even after sonication. Figure 2a is a typical spectrum obtained by focusing the laser light on the gels. The spectrum resembles those obtained by others on Nafion 117 membranes.8,11,30 The most intense band in the spectrum is at 733 cm 1, which has been assigned to the CF2 symmetric stretch.8,11 In a spectroscopic study of octafluoropropane and hexafluoroacetone, Pace et al. assigned the strongest Raman band at 780 cm 1 to skeletal C C C symmetric stretch.31 Our DFT calculations (B3LYP, 6311+g(d), Gaussian 0332) of n-C4F10 also show that the strongest Raman band at 740 cm 1 is from C C C skeletal stretch coupled to the C F stretch. Therefore, we assign the 733 cm 1 band to CF2 symmetric stretch coupled to C C C skeletal stretch. The symmetric stretch of the sulfonate group (SO3 ) appears at 1060 cm 1. The C O C symmetric stretch of the side chain is located at 972 cm 1. In recent DFT calculations,13,33 it was suggested that the 1060 cm 1 band is mainly from the C O C asymmetric stretch of the ether groups on the Nafion side chain. However, a band at a similar position was observed on molecules with a sulfonate group but without any ether groups (Figure 3). This observation suggests that this band is mainly from the sulfonate symmetric stretch. The assignments of other peaks are summarized in Table 1. Note that the 860 and 892 cm 1

peaks are from residual n-propanol solvent, as they disappeared after prolonged drying and were also observed in the Raman spectrum of the pure solvent. Compared to the normal Raman spectrum, the SER spectra of Nafion on Au and Pt surfaces show many more peaks. This observation is common in SERS studies and has been attributed to either lowered molecular symmetric upon adsorption,34 or the field gradient effect,34,35 though the importance of each contribution depends on the molecule metal system and is still under debate.36,37 Due to the low molecular symmetry of Nafion, the field gradient effect is more likely operative in the present case. In addition to the richer spectral features, the interaction of molecules with the surface results in shifts of vibration frequency of several bands involving the interacting functional groups. These factors make SERS powerful for probing adsorbate surface interaction, but also make the band assignment challenging. To aid band assignments, we studied normal Raman and SER spectra of several model compounds, including hexanesulfonate (HeS), dodecanesulfonate (DoS), and perfluoro(4-methyl-3,6dioxaoctane) sulfonate (PFMDS) (Figure 3). The latter has a structure similar to that of the Nafion side chain (Scheme 1), while the former two linear alkanesulfonates do not contain C F bonds or ether groups, which have bands overlapping with those from the sulfonic groups. Comparison of spectra from these model compounds will therefore help to identify overlapping bands. 959

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

Table 1. Vibration Assignments of Normal Raman and SER Spectra of Nafion normal Raman

SERS on Au

SERS on

(cm 1)

(cm 1)

Pt (cm 1)

483

δ(CF2)8,11 ν(Pt-CO)29

683

w(CF2)8,11

391

δ(CF2)

510 678

assignment

δ(SO3 )

715 733

755

755

νs(CF2)+νs(C C C)

805

814

829

ν(C S)8,11

972

971

968

δ(C O C)8,11

1060

1001 1030

1001 1065

νs(SO3 ) νs(SO3 )8,11,30

1183

1178

νas(SO3 )8

1233

1233

νs(CF2) or νs(CF3)8,11

νs(CF2)

1150

1269

1270

νas(CF3)

1296

1301

1302

νs(C C)8,11

1382

1379 1506

1384

1544

1544

1602

1610

1342 ν (C C)8,11

δ(H2O)46,47

dissolved in water as compared to the solid state. Comparison of Raman and SER spectra of PFMDS also reveals that, upon adsorption, the SO3 symmetric and asymmetric stretch frequencies red-shifted from 1070 and 1179 cm 1 to 1000 and 1128 cm 1, respectively. We now return to discussion of Nafion spectra obtained on Au (Figure 2b and c). A common feature in the SER spectra of the model compounds and Nafion on Au is a strong band at around 1000 cm 1. This observation gives us confidence to assign this band to the symmetric stretch of the sulfonate group. On the Au surface, this band red-shifted significantly from 1060 cm 1 in the bulk, suggesting that Nafion sulfonate groups interact with the Au surface. This 1000 cm 1 band is largely absent in the previous IR studies.5,10,12,13,17 We will return to this point later. The symmetric SO3 stretching frequency does not change significantly with the applied potential (Figure 2b and c). In SERS and IR study of bis(3-sulfopropyl)disulfide (SPS) adsorption on Cu surfaces, Schultz et al. reported a significant downshift of SO3 symmetric stretching mode upon adsorption and its frequency is also independent of the applied potential, though the molecule is anchored on the surface with its thiolate group.43 Unlike the normal Raman spectrum of Nafion gels, the SER spectra do not show a strong peak at 733 cm 1 from the coupled symmetric stretching mode of the CF2 groups in the backbone, suggesting that Nafion ionomers adsorb on Au through the side chain with the CF2 backbone away from the surface, as the Raman enhancement effect decays with the distance from the Au surface.34 The next strong band is located at 1269 cm 1, whose vibration assignment is more challenging. The C F stretches of CF3 and CF2 groups, and the SO3 asymmetric stretching bands all appear between 1100 and 1300 cm 1.7,8,11,15,30 For free-standing Nafion membranes or thin films coated on Pt, two bands in the IR spectra at 1300 and 1250 cm 1 were assigned to the degenerated asymmetric SO3 stretches, respectively.14,17

Figure 3. Normal Raman (a,c,e) and SER spectra (b,d,f) of model compound HeS (a,b), DoS (c,d), and PFMDS (e,f). Raman spectra were recorded with sulfonate salts, and SER spectra were recorded at +0.6 V in 1 mM PFMDS or 10 mM alkanesulfonate +0.1 M HClO4. Spectra are baseline corrected by using the GRAMS/AI.

In the normal Raman spectra of alkanesulfonate (Figure 3a and c), a strong peak appears at 1068 cm 1 that can be assigned to the symmetric stretch of the sulfonic group.38 The band at about 1180 cm 1 is from the asymmetric stretch of SO3 ,38 and the other degenerated asymmetric SO3 stretch is expected at 1200 cm 1 based on the IR spectra (Supporting Information Figure S1).39 The other band in this region (1119 cm 1 for HeS and 1131 cm 1 for DoS) can be assigned to the antisymmetric C C stretch.40 The C S stretch appears at around 800 cm 1. These spectral features in the normal Raman spectra are similar to those reported by others.38 Upon adsorption on Au, the 1068 cm 1 downshifted significantly to 1008 and 1003 cm 1, respectively, for HeS and DoS. The asymmetric SO3 stretch bands also red-shifted to about 1150 cm 1. Meanwhile, the C S stretch blueshifted significantly to 860 cm 1 for HeS and 846 cm 1 for DoS. The different C S stretching frequencies of HeS and DoS may come from the conformational difference.38 These spectral transitions upon adsorption strongly suggest that the alkanesulfonates adsorbed on Au via their sulfonic groups. The redshift of the SO3 symmetric stretch can be attributed to either the softening of the S O bond or the change of vibration coupling, or both. A similar, albeit smaller, redshift of the SO3 symmetric stretching frequency was also observed from Sn(CF3SO3)2 in acetonitrile,6 alkanesulfonate intercalated in Mg2Al layer double hydroxide,41 and long chain copper alkylsulfonates.42 The SO3 stretches also shifted to lower frequencies upon being 960

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

Figure 4. Raman (A) and IR (B) spectra of NFBS.

The C C stretch bands at 1296 and 1382 cm 1 in the normal Raman spectrum are nearly unchanged and show up in the SER spectra at 1300 and 1380 cm 1, respectively. This leads us to believe that the C C bonds on the Nafion side chains or the backbone do not interact directly with the Au surface. The origins of the 1447 and 1544 cm 1 bands are not clear at present. Peak positions and tentative vibration assignments of the bands in SER spectra are summarized in Table 1, together with those of the bands in the normal Raman spectrum. When the bands are severely overlapped, their positions were obtained through peak fitting using the GRAMS program. It should be emphasized that the assignment of a Raman band to a single vibration does not mean that this vibration is the sole contributor to this band, rather it only reflects the dominant contributor to the band, as some of the vibrations are strongly coupled.13,30,33 Figure 2d and e displays SER spectra recorded on Pt-coated Au nanoshell arrays in the 0.5 wt % Nafion suspension prepared as described above. The center position of most of the bands is within 10 cm 1 of the corresponding peaks in Figure 2b and c. The assignment of these peaks follows the foregoing arguments and is listed in Table 1. Of special interest is again the observation of a relatively strong band at about 1000 cm 1 assigned to the symmetric SO3 stretch. The significant redshift of this band suggests that the sulfonate group is directly interacting with the metal surface. This is in agreement with IR and electrochemical studies of Nafion Pt interfaces which show that the presence of Nafion on the surface partially blocks CO adsorption on Pt(111)10 and hydrogen adsorption,10,16,18 20 and lowers the ORR activity,20,21 suggesting the strong interaction between Nafion sulfonic group and Pt surfaces.21,22 Our cyclic voltammetric results on particle array electrodes are consistent with these observations. Figure 5 displays comparisons of cyclic voltammograms obtained in 0.1 M HClO4 on SiO2@Au arrays and Pt-coated SiO2@Au arrays before and after the electrodes were coated with about 5 μm of Nafion. To avoid the interference from the residual alcohols in the film, we used 10 wt % Nafion dispersed in water diluted two times by water in these experiments. The films were formed by carefully putting a drop (about 14 μL) of the suspension onto particle arrays supported on a GC electrode and drying in a stream of N2 for 2 h. The film thickness was estimated from the volume and density (2.05 g/mL)48 of the Nafion dispersion. Figure 5A shows that the presence of Nafion film on the SiO2@Au nanoparticle arrays pushed the Au surface oxidation

However, comparison of Raman spectra of a Nafion membrane and a polytetrafluoroethylene film suggest that one of these bands is more likely located at ca. 1180 cm 1, though the authors did not assign this band to the asymmetric SO3 stretch.11 Kujawski assigned an IR band at 1206 cm 1 to the lower frequency asymmetric SO3 stretch of Nafion.15 Miles et al. also assigned a Raman band at 1188 cm 1 from 4 M sodium trifluoromethanesulfonate to SO3 asymmetric stretch.44 Our normal Raman spectra data of the model compounds show that one of the asymmetric SO3 stretches is at about 1180 cm 1, regardless of the molecular structure (Figure 3). The other asymmetric stretch mode is too weak to be detected in the Raman spectra, but is strong at about 1200 cm 1 in the corresponding IR spectra (Supporting Information Figure S1). Furthermore, IR and Raman spectra of potassium nonafluorobutanesulfonate (NFBS) show a pair of peaks between 1170 and 1200 cm 1 that can be assigned to the degenerated SO3 asymmetric stretches (Figure 4). On the basis of these observations, we postulate that the SO3 asymmetric stretches of Nafion ionomer more likely fall in this frequency region. In a recent IR and DFT study of perfluoro(2-ethoxyethane) sulfonic acid (PES), Warren and McQuillan assigned two strong bands at 1192 and 1220 cm 1 to CF2 and CF3 stretches, respectively.30 These two bands are likely from degenerated asymmetric SO3 stretches. On the basis of the above discussion, the band at 1183 cm 1 in the SER spectra (Figure 2b and c) is attributed to the asymmetric stretch of SO3 of adsorbed Nafion ionomer. Several pieces of evidence suggest the 1269 cm 1 band is from asymmetric stretch of the CF3 on the Nafion side chain. Miles et al. assigned a pair of bands at 1230 and 1285 cm 1 to the CF3 symmetric and asymmetric stretches of trifluoromethanesulfonate, respectively.44 Albinsson and Michl assigned an IR band at 1250 cm 1 in the matrix isolation IR spectra of n-perfluorobutane to asymmetric CF3 stretch.45 Our Raman and IR spectra of NFBS also show a pair of peaks at ca. 1250 and 1280 cm 1 that can be assigned to the CF3 symmetric and asymmetric stretches. Note that the CF2 symmetric stretches are much weaker than the CF3 stretches in the Raman spectra (Figure 4). The 1233 cm 1 band in the SER spectra is therefore likely from either the symmetric CF3 stretch or the asymmetric stretch of CF2. In a combined PM-IRRAS and DFT calculation study, Kendrick et al. assigned a band at 1260 cm 1 to the asymmetric stretch of adsorbed CF3 of the Nafion side chain.13 961

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

Figure 5. CVs of (A) SiO2@Au (B) Pt-coated SiO2@Au particle arrays supported on a GC electrode in deaerated 0.1 M HClO4 with (solid line) and without (dashed line) Nafion thin films. Scan rate: 0.1 V/s.

and reduction peaks to more positive potentials, and the associated charges decreased by about 10%. The smaller reduction peak at about 0.35 V on the bare SiO2@Au particles is likely from the reduction of some impurities from particle synthesis. The pair of peaks at about 0.30 V is caused by the presence of Nafion film and is similar to that observed with PFMDS (Supporting Information Figure S2) and alkanesulfonate (not shown) on Au electrodes. In the case of alkanesulfonate, the pair of current peaks was observed even after the Au electrode was transferred to 0.1 M HClO4 without alkanesulfonates, suggesting the adsorption is irreversible. On Pt-coated SiO2@Au particle arrays, both the hydrogen adsorption/desorption and the surface oxidation/ reduction peaks are smaller in the presence of Nafion thin film. Similar observations were also obtained with Pt-coated SiO2@Au in 0.1 M HClO4 with 1 mM PFMDS (Supporting Information Figure S2). The decrease of current peaks is not due to the detachment of particles from the GC electrode, as similar surface blocking was observed on a 2 mm polycrystalline Pt disk electrode. The CO oxidation peak on Pt-coated SiO2@Au was shifted to positive potentials by about 20 mV in the presence of Nafion thin films (Supporting Information Figure S3). These observations on Pt-coated SiO2@Au agree with those reported by others on Pt electrodes.10,16,18,19 Taken together with SERS data presented above, we believe there is strong evidence supporting sulfonate specific adsorption on Au and Pt surfaces, in contrast to the previous view of its weak adsorption property. Compared to Figure 2b and c, there is a new band at 483 and 492 cm 1 in Figure 2d and e, respectively. This strong band is from adsorbed CO formed by the dissociation of the residual npropanol from the Nafion dispersion. This assignment was further confirmed by the observation of a band at around 2030 cm 1 (not shown), which can be attributed to the C O stretch. The 492 cm 1 band intensity decreases at +0.6 V due to CO oxidation, and its position redshifts to 483 cm 1. These observations agree with those of extensively studied CO adsorption on Pt.49 The two bands at 860 and 897 cm 1, also observed on Au, are from propanol vibrations. The 934 cm 1 band is from symmetric stretch of ClO4 . Its intensity decreased when the potential moved to more negative values as a result of repulsive columbic interaction of the negatively charged surface and the anion, consistent with recent electrochemical and in situ infrared reflection absorption studies.10,22 The spectra in Figure 2d and e show several other noticeable differences when compared to Figure 2b and c. First, the overall

spectral intensity is lower. This lower intensity on Pt-modified surface is expected as a result of the adsorbates being further away from the Au shell which supports the Raman enhancement. In addition, the CO adsorption from alcohol dissociation reduces the number of sites available for sulfonate group adsorption. The CO coverage estimated from comparison of peak intensity with a full monolayer of CO adsorption is about 0.5. This condition mimics the anode environment of the direct methanol fuel cell where CO adsorption is significant. Second, the relative band intensities change significantly. For example, the SO3 asymmetric stretch band at 1178 cm 1 on Pt is nearly as intense as the symmetric stretch at 1001 cm 1. This change may arise from the molecular conformational or orientation difference of Nafion side chains on Au and Pt surfaces. It has been shown that the Raman spectra of alkanesulfonates are very sensitive to their molecular conformation.38 A third spectral difference between Nafion on Au and Pt surfaces is a strong band at about 1610 cm 1 on Pt. This band is tentatively assigned to the water bending mode on the basis of its frequency.46,47 In the SER spectrum of DoS on Au, a band at 1646 cm 1 was also observed (Figure 3d). Our preliminary study showed that this band in the DoS spectrum vanished when the spectra were recorded in D2O, which provides indirect support of the vibration assignment of the 1610 cm 1 band. A weaker but discernible band at 1602 cm 1 was also observed in the SER spectra from Nafion adsorbed on Au. The observed water bending band is red-shifted significantly from that of the bulk water, suggesting that it is from water molecules interacting with the metal surfaces through the oxygen atoms.47 The presence of the water bending band suggests either that sulfonate groups coadsorb with water molecules, or that some of the sulfonate groups are still partly solvated and adsorb through the water molecules in their solvation shells or simply present near the surface (Figure 6). A piece of evidence supporting this latter argument can be found from the appearance of a SO3 symmetric vibration band at 1065 cm 1 (1030 cm 1 on Au) which is close to the SO3 symmetric stretch of Nafion in the solution. Such water sulfonate interactions may play an important role in proton transport from the electrochemically active sites to Nafion membrane. A recent neutron reflectometry study of Nafion/Pt and Nafion/Au interfaces also suggests the presence of water molecules presumably solvating the sulfonate groups.9 On the basis of the above discussions, the adsorption of Nafion on Au and Pt surfaces is illustrated in Figure 6. 962

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

In addition, subtraction of IR spectra from a reference spectrum taken at a different potential for removing the bulk signal may weaken and even completely remove the desired information when the peak position and intensity are not sensitive to the applied potential. Nevertheless, the intrinsic sensitivity of IR spectroscopy provides valuable information complementary to the SER spectra. From this point of view, it will be interesting to compare IR and SER spectra recorded under identical conditions. A caveat in the present SERS study of Nafion adsorption on Pt is the presence of significant amount of coadsorbed CO, which only mimics the methanol fuel cell operation condition. It will be of interest to further explore Nafion catalyst interactions by using SERS under conditions closer to H2 O2 fuel cell operation, e.g., at higher applied potentials as well as elevated temperatures, and without the influence of coadsorbed CO by removing alcohol cosolvents. Efforts in this direction are proceeding in our laboratories.

Figure 6. Schematic illustration of Nafion adsorption on Au and Pt surfaces. Polytetrafluoroethylene backbone, perchlorate, cation, coadsorbed CO, and water molecules not interacting with sulfonate groups are omitted. Blue: fluorine; Red: oxygen; Yellow: sulfur; Dark gray: carbon; Light gray: hydrogen.

This tentative model depicts the adsorption of Nafion on the metal electrode surfaces through the sulfonate group. Our model is similar to that proposed by Kendrick et al. based on their IR studies,13 except that we are not proposing the CF3 group is interacting with the surfaces. We emphasize that the schematic presented in Figure 6 only depicts the Nafion side chain that interacts with the surface. It is conceivable that there are sulfonate groups that are pointing away from the surface, given that the ionomers have a micellar structure.4 Finally, we compare our results with those obtained by in situ IR studies.5,10,12,13,17 Most of the bands observed in the present study are similar to those in the IR spectra, albeit the relative intensity may differ due to different selection rules. The most apparent difference between the SER spectra and the IR spectra is the appearance in the former of a strong peak at about 1000 cm 1 assigned to the SO3 symmetric stretch from the adsorbed sulfonate group. The confidence of this assignment is ensured by the appearance of this band in several model compounds containing a sulfonic group (Figure 3). This demarcation may be due to the interferences from the IR absorption of the bulk species and the overwhelming IR absorption of other molecular groups (e.g., CF2 and C O C vibrations). In IR studies, the Nafion ionomer was coated on Pt surfaces as a thin layer,5,10,12,13,17 while in the present study, a diluted ionomer dispersion was used. Different methods including internal reflection and polarization modulation were applied to minimize the IR absorption contribution from the bulk species to the spectra. Given that the penetration depth of radiation in the mid-IR region is several hundred nanometers to a few micrometers, the IR spectra are likely tainted with signals from ionomers that are away from the interfacial region, even when an ATR configuration is used. PMIRRAS exploits the insensitivity of s-polarized IR light to the adsorbed species to remove the interference from the bulk species by subtracting spectra taken with the s-polarized light from those recorded with the p-polarized light, which contain both bulk and interfacial signals.50,51 This is a very useful technique for reducing the interference of bulk signal. However, as has been pointed out, the PM-IRRAS is also sensitive to the species present within a quarter of wavelength from the metal electrode surface,50 and the background subtraction can be ineffective due to the reflectivity difference of s- and p-polarized radiation.10,51 The large bulk-solution bands can obscure the observation of weaker vibration features from interfacial species.

4. CONCLUSIONS In summary, SER spectra of adsorbed Nafion on SiO2@Au and Pt-coated SiO2@Au arrays were successfully obtained. By comparing with the Raman and IR spectra of several model compounds, vibration assignments of bands in SER spectra of Nafion on Au and Pt were made. The results suggest that Nafion adsorbed on Pt and Au through sulfonate groups, yielding a significantly red-shifted symmetric sulfonate stretch. The specific adsorption of Nafion on Pt and Au was also supported by the cyclic voltammetric results. The present work demonstrates for the first time that SERS is able to provide detailed information on Nafion adsorption which will help to advance our understanding of Nafion-catalyst interfacial structure. ’ ASSOCIATED CONTENT

bS

Supporting Information. IR spectra of PFMDS, DoS, and HeS; cyclic voltammetric comparisons of SiO2@Au and Ptcoated SiO2@Au with and without Nafion coating in 0.1 M HClO4 solution with PFMDS. CO stripping voltammograms on Pt-coated SiO2@Au with and without Nafion film in 0.1 M HClO4. SER spectrum obtained with SiO2@Au in 0.1 M HClO4. This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Tel: 1-513-529 8084. Fax: 1-513529 5715.

’ ACKNOWLEDGMENT This work was supported by General Motors LLC and Miami University. ’ REFERENCES (1) Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal., B 2005, 56, 9–35. (2) Wagner, F. T.; Lakshmanan, B.; Mathias, M. F. J. Phys. Chem. Lett. 2010, 1, 2204–2219. (3) Wilson, M. S.; Gottesfeld, S. J. Electrochem. Soc. 1992, 139, L28–L30. (4) Mauritz, K. A.; Moore, R. B. Chem. Rev. 2004, 104, 4535–4585. 963

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964

Langmuir

ARTICLE

(41) Xu, Z. P.; Braterman, P. S. J. Phys. Chem. C 2007, 111, 4021–4026. (42) Park, S.-H.; Lee, C. E. Chem. Mater. 2006, 18, 981–987. (43) Schultz, Z. D.; Feng, Z. V.; Biggin, M. E.; Gewirth, A. A. J. Electrochem. Soc. 2006, 153, C97–C107. (44) Miles, M. G.; Doyle, G.; Cooney, R. P.; Tobias, R. S. Spectrochim. Acta 1969, 25A, 1515–1526. (45) Albinsson, B.; Michl, J. J. Phys. Chem. 1996, 100, 3418–3429. (46) Li, J.-F.; Huang, Y.-F.; Duan, S.; Pang, R.; Wu, D.-Y.; Ren, B.; Xu, X.; Tian, Z.-Q. Phys. Chem. Chem. Phys. 2010, 12, 2493–2502. (47) Osawa, M.; Tsushima, M.; Mogami, H.; Samjeske, G.; Yamakata, A. J. Phys. Chem. C 2008, 112, 4248–4256. (48) Zook, L. A.; Leddy, J. Anal. Chem. 1996, 68, 3793–3796. (49) Zou, S. Z.; Weaver, M. J. J. Phys. Chem. 1996, 100, 4237–4242. (50) Zamlynny, V.; Lipkowski, J. In Diffraction and spectroscopic methods in electrochemistry, Alkire, R. C., Kolb, D. M., Lipkowski, J., and Ross, P. N., Eds.; Wiley-VCH: Weinheim, 2006; Vol. 9, pp 315 376. (51) Iwasita, T.; Nart, F. C. In Advances in Electrochemical Science and Engineering, Gerischer, H., Tobias, C. W., Eds.; VCH: Weinheim, 1995; Vol. 4, pp 123 216.

(5) Ayato, Y.; Kunimatsu, K.; Osawa, M.; Okada, T. J. Electrochem. Soc. 2006, 153, A203–A209. (6) Bergstrum, P.-A.; Frech, R. J. Phys. Chem. 1995, 99, 12603–12611. (7) Blanchard, R. M.; Nuzzo, R. G. J. Polym. Sci., Part B: Polym. Phys. 2000, 38, 1512–1520. (8) Bribes, J. L.; Elboukari, M.; Maillols, J. J. Raman Spectrosc. 1991, 22, 275–279. (9) Dura, J. A.; Murthi, V. S.; Hartman, M.; Satija, S. K.; Majkrzak, C. F. Macromolecules 2009, 42, 4769–4774. (10) Gomez-Marin, A. M.; Berna, A.; Feliu, J. M. J. Phys. Chem. C 2010, 114, 20130–20140. (11) Gruger, A.; Regis, A.; Schmatko, T.; Colomban, P. Vib. Spectrosc. 2001, 26, 215–225. (12) Kanamura, K.; Morikawa, H.; Umegaki, T. J. Electrochem. Soc. 2003, 150, A193–A198. (13) Kendrick, I.; Kumari, D.; Yakaboski, A.; Dimakis, N.; Smotkin, E. S. J. Am. Chem. Soc. 2010, 132, 17611–17616. (14) Korzeniewski, C.; Snow, D. E.; Basnayake, R. Appl. Spectrosc. 2006, 60, 599–604. (15) Kujawski, W.; Nguyen, Q. T.; Nee, J. J. Appl. Polym. Sci. 1992, 44, 951–958. (16) Malevich, D.; Li, J.; Chung, M. K.; McLaughlin, C.; Schlaf, M.; Lipkowski, J. J. Solid State Electrochem. 2005, 9, 267–276. (17) Malevich, D.; Zamlynny, V.; Sun, S.-G.; Lipkowski, J. Z. Phys. Chem. 2003, 217, 513–525. (18) Gottesfeld, S.; Raistrick, I. D.; Srinivasan, S. J. Electrochem. Soc. 1987, 134, 1455–1462. (19) Zecevic, S. K.; Wainright, J. S.; Litt, M. H.; Gojkovic, S. L.; Savinell, R. F. J. Electrochem. Soc. 1997, 144, 2973–2982. (20) Yano, H.; Higuchi, E.; Uchida, H.; Watanabe, M. J. Phys. Chem. B 2006, 110, 16544–16549. (21) Subbaraman, R.; Strmcnik, D.; Paulikas, A. P.; Stamenkovic, V. R.; Markovic, N. M. ChemPhysChem 2010, 11, 2825–2833. (22) Subbaraman, R.; Strmcnik, D.; Stamenkovic, V.; Markovic, N. M. J. Phys. Chem. C 2010, 114, 8414–8422. (23) Wang, H.; Kundu, J.; Halas, N. J. Angew. Chem.; Int. Ed. 2007, 46, 9040–9044. (24) Reincke, F.; Hickey, S. G.; Kegel, W. K.; Vanmaekelbergh, D. Angew. Chem.; Int. Ed. 2004, 43, 458–462. (25) Li, M. D.; Cui, Y.; Gao, M. X.; Luo, J.; Ren, B.; Tian, Z. Q. Anal. Chem. 2008, 80, 5118–5125. (26) Brankovic, S. R.; Wang, J. X.; Adzic, R. R. Surf. Sci. 2001, 474, L173–L179. (27) Gao, P.; Gosztola, D.; Leung, L. W. H.; Weaver, M. J. J. Electroanal. Chem. 1987, 233, 211–222. (28) Gruenbaum, S. M.; Henney, M. H.; Kumar, S.; Zou, S. Z. J. Phys. Chem. B 2006, 110, 4782–4792. (29) Mrozek, M. F.; Xie, Y.; Weaver, M. J. Anal. Chem. 2001, 73, 5953–5960. (30) Warren, D. S.; McQuillan, A. J. J. Phys. Chem. B 2008, 112, 10535–10543. (31) Pace, E. L.; Plaush, A. C.; Samuelson, H. V. Spectrochim. Acta 1966, 22, 993–1006. (32) Frisch, M. J. et al. Gaussian 03, revision B04; Gaussian Inc.: Pittsburgh, PA, 2003. (33) Webber, M. D., N.; Kumari, D.; Fuccillo, M.; Smotkin, E. S. Macromolecules 2010, 43, 5500–5502. (34) Moskovits, M. Rev. Mod. Phys. 1985, 57, 783. (35) Ayars, E. J.; Hallen, H. D.; Jahncke, C. L. Phys. Rev. Lett. 2000, 85, 4180–4183. (36) Hulteen, J. C.; Young, M. A.; Van Duyne, R. P. Langmuir 2006, 22, 10354–10364. (37) Lombardi, J. R.; Birke, E. L. J. Phys. Chem. C 2008, 112, 5605– 5617. (38) Ohno, K.; Naganobu, T.; Matsuura, H.; Tanaka, H. J. Phys. Chem. 1995, 99, 8477–8484. (39) Fujimori, K. Bull. Chem. Soc. Jpn. 1959, 32, 850–851. (40) Boerio, F. J.; Koenig, J. L. J. Chem. Phys. 1970, 52, 3425–3431. 964

dx.doi.org/10.1021/la2035455 |Langmuir 2012, 28, 957–964