Article pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Probing Molecular-Scale Catalytic Interactions between Oxygen and Cobalt Phthalocyanine Using Tip-Enhanced Raman Spectroscopy Duc Nguyen,‡ Gyeongwon Kang,‡ Naihao Chiang,† Xu Chen,† Tamar Seideman,‡,† Mark C. Hersam,‡,†,§ George C. Schatz,‡ and Richard P. Van Duyne*,‡,† ‡
Department of Chemistry, †Applied Physics Graduate Program, and §Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208, United States S Supporting Information *
ABSTRACT: Ultrahigh vacuum tip-enhanced Raman spectroscopy (UHVTERS) is used to investigate adsorption of molecular oxygen (O2) on cobalt(II) phthalocyanine (CoPc) supported on Ag(111) single crystal surfaces, which is the initial step for the oxygen reduction reaction (ORR) using metal Pc catalysts. Two adsorption configurations are primarily observed, assigned as O2/CoPc/Ag(111) and O/CoPc/Ag(111) based on scanning tunneling microscopy (STM) imaging, TERS, isotopologue substitution, and density functional theory (DFT) calculations. Distinct vibrational features are observed for different adsorption configurations such as the 18O−18O stretching frequency at 1151 cm−1 for O2/CoPc/Ag(111), and Co−16O and Co−18O vibrational frequencies at 661 and 623 cm−1, respectively, for O/CoPc/Ag(111). DFT calculations show vibrational mode coupling of O−O and Co−O vibrations to the Pc ring, resulting in different symmetries of oxygen-related normal modes. This study establishes UHVTERS as a chemically sensitive tool for probing catalytic systems at the molecular scale.
■
INTRODUCTION The oxygen reduction reaction (ORR) is one of the most widely studied chemical reactions with a broad range of applications spanning biology, catalysis, and fuel cells.1−3 Efficient ORR catalysts contain a large fraction of precious metals such as Pt, which are challenging to scale up.1−3 Consequently, extensive research has been devoted to the search for alternative precious metal-free ORR catalysts.1−7 In this search, metal phthalocyanines (MPcs, M = Co, Fe) have emerged as particularly promising because of their high catalytic activity, chemical stability, and abundance.8−12 To further improve ORR catalytic performance of MPc catalysts, a fundamental understanding of adsorption and interaction of MPc with molecular O2 is necessary. In heterogeneous MPc catalyst systems, ORR activity and mechanism strongly depend on how molecular O2 adsorbs and interacts with MPc molecules.11 Different adsorption configurations of O2 on metal surface-supported MPc have been observed using scanning tunneling microscopy (STM) imaging at the single molecule level.8,13,14 However, STM imaging does not provide direct chemical information, which leads to ambiguity in determining the structures of oxygen-bound adsorbed species. On the other hand, surface-enhanced Raman spectroscopy (SERS) provides the rich chemical information that is necessary to unambiguously determine structures of adsorbates.15 However, the spatial resolution of SERS is limited to ∼10 nm,16 which conceals the details of site-specific molecule−substrate interactions and heterogeneity related to © XXXX American Chemical Society
different reaction pathways. Thus, it is highly desirable to combine SERS with STM to investigate ORR of MPc catalysts at the molecular scale. Ultrahigh vacuum tip-enhanced Raman spectroscopy (UHVTERS) is a powerful method combining the rich chemical information on SERS with the ultrahigh spatial resolution of STM.17−23 UHV-TERS provides chemical information at subnm spatial resolution under controlled environment.19,24,25 UHV-TERS has previously been used to investigate the adsorption of single porphyrins and related molecules on metal surfaces.19,25 UHV-TERS also provides information about adsorption geometry even when STM imaging is not reliable.26 Although previous studies have attempted to use TERS to study chemical reactions and molecular adsorption,27−29 these studies were not performed in highly controlled environments, which complicated data analysis and assignment of reaction mechanisms. Herein, we use UHV-TERS to investigate the chemical interaction and adsorption of molecular O2 with cobalt(II) phthalocyanine (CoPc) supported on Ag(111) surfaces under highly controlled environments, which is the initial step of ORR using MPc catalysts. Different adsorption configurations are observed from STM images of CoPc/Ag(111) after exposure to O2 including O2/CoPc/Ag(111) and O/CoPc/Ag(111). Characteristic features in both STM images and TERS are Received: January 30, 2018 Published: April 23, 2018 A
DOI: 10.1021/jacs.8b01154 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
sites of the Ag slab with two different molecular orientations as possible adsorption configurations (Figure S3). The adsorbate and the top Ag layer were then relaxed. The energy difference for oxygen adsorption on different CoPc adsorption configurations is negligible ( 97%, SigmaAldrich) was then sublimed at ∼650 K (P < 5 × 10−11 Torr) onto the atomically clean Ag(111) surface. Specifically, approximately 1 monolayer (ML) of CoPc was deposited with a deposition rate of ∼0.3 ML/min. Surface cleanliness and CoPc coverage were confirmed by STM imaging. CoPc on Ag(111) was then exposed to ∼1800 L (1 × 10−6 Torr, 30 min) of 16O2 (99.999%, Airgas) or 18O2 (99.93%, Sigma-Aldrich) in a sample preparation chamber (base pressure ∼ 3 × 10−11 Torr), and then transferred to the STM chamber (P ∼ 2 × 10−11 Torr) for STM and TERS characterization. UHV-STM and TERS. STM imaging was conducted at room temperature in a home-built UHV-STM system (P ∼ 2 × 10−11 Torr).18 Both electrochemically etched tungsten (W) and silver (Ag) tips were used for imaging. TERS measurements were performed using plasmonically active Ag tips. The instrument was equilibrated for at least ∼1−2 h after the laser was illuminated to minimize thermal drift. A drift rate of ∼0.05 nm/min was obtained (Figure S1). Single molecules or clusters of a few molecules of interest were identified by STM imaging. The STM tip was then parked on top of the preidentified molecules for TERS measurements, while keeping the feedback loop on to maintain the tip−sample distance. The excitation wavelength was 632.8 nm (HeNe laser, Research Electro-optics). The excitation beam was p-polarized with an incidence angle of ∼15° with respect to the surface plane. The beam was focused to a ∼10 μm × 10 μm spot18 with a power of 0.4−0.8 mW. Raman scattered light was collected using a spectrograph (Princeton Instrument SCT320) equipped with a charged-coupled device (CCD) detector (Princeton Instrument PIXIS 400BR). A 1200 grooves/mm grating was used, which gives a spectral resolution of ∼4 cm−1.21 The sample bias voltage and tunneling current were varied to optimize the TER signal. The acquisition time was typically 6 min per TER spectrum. All TER spectra are presented as difference of tip-engaged and tip-retracted spectra. Typical tip-engaged, tip-retracted, and difference spectra are given in Figure S2. The same setup was used to perform UHV-SERS measurements of CoPc on Ag film-overnanospheres (AgFON) substrates (200 nm Ag thermally deposited onto a close-packed 540 nm SiO2 nanospheres on Si) optimized for 632.8 nm excitation30 with the same doses of O2. DFT Calculations. STM images and vibrational modes of CoPc− oxygen complexes on Ag(111) were simulated based on density functional theory with the long-range dispersion correction approach by Grimme (DFT-D2)31 using the Vienna ab initio simulation package (VASP).32 Long-range dispersion correction was included to consider the substantial interaction between an adsorbate and surface. The electron−ion interactions were calculated using the projector augmented wave (PAW) method33 with a plane wave up to an energy cutoff of 400 eV. The Perdew−Burke−Ernzerhof (PBE) exchange-correlation functional34 was used. Because of the large size of the oxygen/CoPc/Ag(111) system, spin unpolarized calculations were employed. The silver fcc unit cell was optimized without long-range dispersion correction. The calculated lattice constant was 0.416 nm which agrees well with the experimental value of 0.408 nm.35 A Ag(111) slab structure was built by repeating three layers of the p(5/ √2 × 9/√6) Ag unit cell followed by adding 4 nm of vacuum space along the z direction. Each slab layer contains 30 Ag atoms. For the slab geometry optimization, only the top layer Ag atoms were relaxed while other layers were frozen. The Brillouin zone of slab structures was sampled on a (2 × 2 × 1) Monkhorst−Pack k-point grid. The cobalt atom in the CoPc molecule was placed on the atop and hollow
■
RESULTS AND DISCUSSION Figure 1A shows a typical STM image of CoPc on a Ag(111) surface at ∼1 ML coverage. CoPc molecules appear as a fourlobed structure, which corresponds with the four arms of the Pc ring, with a brighter center attributed to the Co atom. The average height corrugation of the CoPc molecules is ∼0.05 nm, as seen in the height profile in Figure 1A. The experimental STM image of CoPc/Ag(111) agrees well with the calculated STM image shown in Figure 1C. After exposing CoPc/Ag(111) to ∼1800 L O2, distinct adsorption configurations are observed. As shown in Figure 1B, there are rounded bright spots of two different height levels after O2 exposure, in comparison with pristine CoPc/Ag(111) molecules (one highlighted by a green dot). The brightest spots are ∼0.14 nm higher than pristine CoPc/Ag(111) molecules, as shown in the black line profile in Figure 1B. This difference is comparable to the height difference of FePc/Au(111) before and after O2 adsorption.14 We assign these brightest features as molecular O2 adsorbed on CoPc/Ag(111) in the end-on configuration, or O2/CoPc/Ag(111). This assignment is consistent with the DFT calculation shown in Figure 1D and TERS measurements discussed later in Figure 2. Higher electron density is observed on adsorbed O2 making the center brighter than pristine CoPc/Ag(111). Note that in the calculated STM image, the bright center is asymmetric due to the tilted angle of the adsorbed O2. At room temperature, adsorbed O2 has enough thermal energy to freely rotate, and thus appears as a rounded bright spot in the experimental STM images. The dimmer spots in Figure 1B (blue line profile) are ∼0.07 nm higher than pristine CoPc/Ag(111). We attribute these features to the dissociated O atom adsorbed on CoPc/Ag(111) (i.e., O/CoPc/Ag(111)), again consistent with DFT calculations (Figure 1E). This adsorption configuration has not been observed previously for MPc, but has been observed for porphyrins such as manganese 5,10,15,20-tetraphenyl porphyrin (MnTPP) where MnTPP/Ag(111) dissociates molecular O2 to form O/MnTPP/Ag(111) species.37 The O2/CoPc/Ag(111) and O/CoPc/Ag(111) adsorption configurations concurrently coexist on the Ag(111) surface and are randomly distributed even in single domains (Figure 1B and S4). This observation is supported by DFT calculations indicating that different adsorption configurations can be sampled due to the low free energy for oxygen binding and high O2 dosing condition (Table 1). We further characterize the two adsorption configurations by measuring TER spectra at different spots on the surface. Figure 2A shows a set of TER spectra before and after O2 dosing measured on the O2/CoPc/Ag(111) molecules. Figure 2B is a zoom-in of the spectra in Figure 2A with Lorentzian fits to the peaks highlighting changes after O2 dosing. Both 16O2 and 18O2 are used to untangle CoPc vibrational bands from oxygenB
DOI: 10.1021/jacs.8b01154 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 3 shows calculated normal modes for the O−O stretching and related modes in the O2/CoPc/Ag(111) adsorption configuration. The experimental 18O−18O stretching frequency at 1151 cm−1 is in good agreement with the calculated value at 1177 cm−1 (Figure 3 and Table 2). The 18 O−18O stretching mode couples with a CoPc/Ag(111) mode at 1180 cm−1 and a weak Pc ring motion with the same symmetry as the 1180 cm−1 CoPc/Ag(111) mode is observed (Movie S1). The isotope-insensitive 1180 cm−1 CoPc/Ag(111) mode that is slightly red-shifted to 1178 cm−1 after 16O2 and 18 O2 adsorption (Figure 3), is not observed experimentally. Because of its close energy to the 18O−18O stretching, an appreciable 18O−18O vibration is observed after 18O adsorption (Figure 3 and Movie S1). Isotope switching from 18O2 to 16O2 blue shifts the O−O stretching frequency by ∼70 cm−1 to 1247 cm−1. DFT calculations also show a lack of CoPc/Ag(111) modes near the 16O−16O stretching frequency at 1247 cm−1 (at least separated by ∼30 cm−1), thus preventing efficient coupling of 16O−16O stretching to Pc ring vibrations. As a result, the 16 O−16O stretching mode is localized and isolated from Pc ring motion (Figure 3 and Movie S1). Our observation of the 18O−18O stretching frequency at 1151 cm−1 is in between the 18O−18O stretching frequency of 18 O2/CoPc at ∼1225 cm−1, and 18O2/CoPc(pyridine) at ∼1090 cm−1.39,40 In the case of O2/CoPc, the O−O bond is strong and the Co−O interaction is weak due to weak charge transfer from the Co atom. By adding a pyridine molecule to the axial position as the fifth ligand, charge is transferred from pyridine to the Co atom, making the Co−O interaction stronger and consequently the O−O bond weaker. Our observed 18O−18O stretching is in between these two cases because the Ag(111) surface acts as a fifth axial ligand in the case of CoPc/Ag(111). It is known from prior literature that charge is transferred from the Ag(111) surface to adsorbed CoPc molecules,41,42 thereby promoting O2 adsorption. This result is verified by charge analysis of DFT optimized geometries (Table 1) and calculated O−O stretching frequencies for O2/CoPc (Figure S5) and O2/ CoPc/Ag(111) (Table 2). For free CoPc molecules, the Co atom receives 0.77e of charge from the Pc ring. When CoPc is supported on Ag(111), the Co atom receives more charge from the Pc ring and Ag surface (0.98e of charge), thus making the Co−O interaction stronger and O−O stretching weaker. In contrast to UHV-TERS measurements, UHV-SER spectra acquired on AgFON substrates with the same doses of O2 show no changes after dosing, as shown in Figure 2C, even when a lower coverage of CoPc is used (∼0.5 ML). Wider range UHVSER spectra are shown in Figure S6. The UHV-SERS measurements are explained by the nanostructured nature of AgFONs,30 where the hot spots are at the crevice sites and/or between Ag nanostructures. Because preferential adsorption is likely to happen at the hot spots, CoPc molecules contributing to most of the intensity are buried underneath other CoPc molecules and are inaccessible to O2. In addition, SERS measurements are spatially averaged over different geometries of adsorbed and unadsorbed molecules, which smears out the weak oxygen-related signal. The experimental observation of the 18O−18O stretching band in TERS measurements thus highlights the advantage of TERS over SERS for this study of ORR. Figure 4A shows a set of TER spectra that highlights changes in the lower wavenumber region (
