14702
2008, 112, 14702–14705 Published on Web 08/27/2008
Site-Dependent
13C
Chemical Shifts of CO Adsorbed on Pt Electrocatalysts
Patrick McGrath, Aurora Marie Fojas, Jeffrey A. Reimer,* and Elton J. Cairns Department of Chemical Engineering and Lawrence Berkeley National Laboratory, UniVersity of California, Berkeley, California 94720 ReceiVed: July 09, 2008; ReVised Manuscript ReceiVed: August 14, 2008
13C
NMR and hydrogen-region cyclic voltammetry are used to parse the distribution of adsorbed CO on Pt electrocatalysts into two different types of sites. Trends in the NMR shift data show that 13CO adsorbed on so-called weakly bound H sites show larger Knight shifts as compared to 13CO adsorbed onto strongly bound H sites, and thus experience greater back-bonding from the Pt conduction band. These results are discussed in the context of local electron densities of states and the varying oxidation reactivities associated with these sites on the Pt surface.
I. Introduction The behavior of CO adsorbates on platinum surfaces has far reaching implications in electrocatalysis. Electrocatalysts are subject to poisoning by small amounts of CO present in fuel streams, or by CO adsorbates generated in the direct oxidation of carbonaceous fuels such as alcohols. For this reason it is of interest to study the behavior of CO on platinum surfaces in an electrochemical environment. Understanding the nature of the metal-adsorbate interaction under realistic operating conditions could provide valuable information about the behavior of these adsorbates on electrocatalyst surfaces. A powerful tool to investigate the surfaces of these materials has been nuclear magnetic resonance spectroscopy (NMR).1 The advantages afforded by NMR spectroscopy of 13COads make it an attractive tool for electrocatalysis: it is quantitative and can be applied to any catalyst that will adsorb CO (or CH3OH).2,3 The importance of conduction electrons in Pt-CO, predicted by the Blyholder model of 1964,4 was definitively demonstrated by the observation of a Knight shift in the 13C NMR of 13CO adsorbed on Pt nanoparticles in 1985.5 The presence of unpaired conduction electrons at the 13C nucleus, as determined by the so-called Knight shift, proved conclusively that the bonding of CO to a Pt surface involves the mixing of molecular orbitals from CO with conduction electrons from Pt. Similar observations of a Knight shift for 13COads have followed for the electrochemical adsorption of 13COads and 13CH OH on Pt and Pt alloys.1,6-10 3 Electroanalytical techniques such as cyclic voltammetry have proven to be a powerful complement to NMR studies of electrochemical adsorption.6,8 In the present work, we relate 13C NMR of 13COads to cyclic voltammetry studies11,12 that parse adsorbing surface sites into cubic-packed Pt sites associated with strongly bound hydrogen (SB sites) and close-packed step sites associated with weakly bound hydrogen (WB sites). We present a spectroscopic link between the adsorption site of COads and the 13C NMR shift of the adsorbate. Our observation of a larger 13C shift for CO ads at the WB sites suggests an increased surface electron density vis-a`-vis SB sites. This result portends a link * To whom correspondence should be addressed.
10.1021/jp806068t CCC: $40.75
between reactivity and surface electronic structure (noting that the WB sites are more active toward oxidation of COads), and suggests that 13C NMR may be useful in parsing site occupancies in alloy nanoparticles that cannot be characterized quantitatively via voltammetry in the hydrogen region. II. Experimental Section This Letter compiles NMR results from studies of carbonsupported and unsupported Pt-nanoparticles. The NMR experiments for 13CH3OH on platinum presented herein were performed concurrently with the voltammetry as described in references.11,12 Pt/C and Pt-black electrode stacks (from gold film electrodes as described in ref 12) were assembled and cleaned by potential cycling between 50 and 980 mV vs RHE using a PAR 263A potentiostat. Our flow cell uses a platinum mesh counterelectrode, and a Hg/HgSO4 reference electrode (Koslow Scientific). The clean platinum surface was recorded using cyclic voltammetry (CV), and sweeping the low potential hydrogen region monitored surface coverages. For the studies of supported Pt/C electrodes, preparations of 13CO ads were tailored to follow ref 8. Partial adsorptions were performed by varying the length of exposure (10 s, 1 min, 1 h, saturated (12 h)) to 50 mM 13CH3OH in 1.50 M H2SO4. Partial oxidations for the Pt/C electrode followed the protocol from ref 8. The studies of unsupported platinum black electrodes were performed according to the protocol of ref 12 (which is similar to other studies of methanol on Pt electrodes). Partial adsorptions were performed by 2 h adsorptions of 10 mM, 50 mM, 300 mM, and 1.50 M 13CH3OH in 0.50 M H2SO4 at 250 mV vs RHE. Partial coverages were also obtained from the oxidation of the saturated surface for varying times (5, 20, 60, and 120 min) at 450 mV vs RHE. Certified ACS Plus grade sulfuric acid (Fisher Scientific), deionized (DI) water (Millipore Organ-X system at 18 MΩ), and 13CH3OH (13C, 99%, Cambridge Isotopes) were used for all experiments. Pt/C and Pt-black electrode stacks were rinsed for 2 h with DI H2O following the preparation of the 13CO adlayer and the working electrode compartment was then closed and subjected to NMR. All NMR experiments were performed 2008 American Chemical Society
Letters
J. Phys. Chem. C, Vol. 112, No. 38, 2008 14703
Figure 1. Representative 13C NMR spectra from electrode stacks after various surface preparations using 13CH3OH. Plots to the left show spectra for a Pt/C electrode stack after partial adsorptions for (a) 10s, (b) 1 min, (c) 1 h, and (d) 12 h with 50 mM 13CH3OH. Plots to the right show spectra for a Pt/C electrode stack after partial oxidations of the 13CO-saturated surface, created through potential sweeps to (f) 550 mV, (g) 600 mV, and (h) 650 mV vs RHE. (e) 13C spectrum of the saturated surface preparation before partial oxidation.
in a 300 MHz magnet with a home-built probe. A standard Hahn echo experiment, with a 4.5 µs 90° pulse, a 40 µs delay, and a 13 µs 180° pulse, was used for all NMR spectra presented herein. Signal averaging of 50000-100000 transients was performed with a 1.2 s delay between acquisitions. A detailed description of the data processing scheme is described in ref 11. Tabulated values of shifts and line widths derive from leastsquares fits of the transfer-function corrected 13C NMR spectra to Lorentzian line shapes. Background subtraction occasionally leaves artifacts at 13C shifts associated with either Teflon or carbon cloth. III. Results and Discussion Figure 1 shows the 13C NMR spectra of 13CO-adlayers on the Pt/C electrode produced by partial oxidations of the
adsorbate after saturation by 50 mM 13CH3OH. Table 1 gives the 13C NMR analyses (shifts, widths) along with the derived results from the hydrogen region voltammogram for the COads occupation of WB (below 200 mV) and SB (above 200 mV) sites. The reported shifts for 13COads on the saturated surfaces are consistent with previously reported results, as is the disparity in shifts for 13COads between the supported and unsupported Pt (which is attributed to a support-metal interaction).7-10 A changing 13C NMR shift for 13COads on Pt nanoparticles has been reported previously8 for partial oxidations of an adlayer derived from 13CO gas, but an explanation for the changing shift was not forthcoming at the time. From the present study, it is clear that the 13C NMR shift reflects the relative occupations of 13COads on WB and SB sites. This is shown in Figure 2, which plots the NMR shifts from Table 1 against the ratio of
14704 J. Phys. Chem. C, Vol. 112, No. 38, 2008
Letters
TABLE 1: Combined NMR and Voltammetry Results sample Pt/C Pt/C Pt/C Pt/C Pt/C Pt/C Pt/C Pt black Pt black Pt black Pt black Pt black Pt black Pt black Pt black
preparation 13
10s ads. 50 mM CH3OH 1m ads. 50 mM 13CH3OH 1 h ads. 50 mM 13CH3OH Sat. 50 mM 13CH3OH pot. sweep to 550 mV pot. sweep to 600 mV pot. sweep to 650 mV 2 h ads 10 mM 13CH3OH 2 h ads 50 mM 13CH3OH 2 h ads 300 mM 13CH3OH 2 h ads 1.5 M 13CH3OH 5 m oxidation at 450 mV 20 m oxidation at 450 mV 60 m oxidation at 450 mV 2 h oxidation at 450 mV
chem shift (ppm, TMS)
fwhm (ppm)
WBa occ. (µC/cm2)
SBa occ. (µC/cm2)
WB/SB
304 ( 3 319 ( 1 326 ( 1 348 ( 1 334 ( 1 323 ( 1 324 ( 1 328 ( 3 335 ( 4 334 ( 5 342 ( 4 337 ( 4 338 ( 4 332 ( 6 322 ( 3
35 ( 3 74 ( 1 82 ( 2 81 ( 1 77 ( 1 64 ( 1 56 ( 3 68 ( 2 74 ( 5 100 ( 7 81 ( 5 83 ( 5 92 ( 5 62 ( 10 55 ( 2
1 23 50 85 61 27 14 55 94 107 112 105 95 76 48
17 45 68 84 74 50 18 52 74 81 83 80 72 65 53
0.054 0.50 0.74 1.02 0.82 0.53 0.73 1.05 1.28 1.32 1.36 1.31 1.32 1.18 0.91
a The ‘”WB occ.” and “SB occ.” columns show the voltammetric H-charge that was displaced by COads on the WB and SB sites, respectively. This is a difference mode value, obtained by the subtraction of the total anodic H-charge for the COads-treated electrode from the H-charge of the clean electrode (with corrections for double-layer charging). This is described in ref 12.
Figure 2. NMR shift (obtained from line shape fits) vs ratio of COads occupation on WB and SB sites (obtained by cyclic voltammetry). Red diamonds represent data from Pt/C catalyst; black circles represent data from unsupported Pt black catalyst.
WB/SB occupation by 13COads obtained by cyclic voltammetry. It is clear that the 13C NMR shift increases as the prevalence of 13CO 13CO ads on WB sites increases, and thus the shift for ads on a WB site is larger than on a SB site. It is worth noting that the NMR line for 13COads on Pt appears as a single broad peak for all preparations on both Pt/C and Pt black. A change in the occupation ratio of 13COads on WB/SB sites causes the NMR peak to shift its apparent center-of-mass, but it does not appear to alter the shape significantly (the line width does change for different preparations, but the overall shape remains a single peak). This behavior is characteristic of systems undergoing motional averaging: if each 13COads is not fixed to one environment but samples all available environments on a fast enough time scale (i.e., faster than the spread of NMR frequencies across those environments), the NMR line should appear as a single, averaged peak.3 It is well established the 13CO ads on Pt is a highly dynamic system, and thus evidence of motional averaging is not surprising.15-18 Line shape analyses of NMR spectra taken with various excitation schemes and using specific dynamic models may more clearly elucidate the effects of CO dynamics.17,18 It was noted above that the Knight shift is related to the Fermi-level local electronic density of states (Ef -LDOS) at the nucleus, and thus changes in NMR shift for 13COads on Pt nanoparticles are typically interpreted in terms of Ef -LDOS.
In the Blyholder model for CO adsorption on Pt, conduction electrons from Pt are back-donated to the 2π* molecular orbital of CO, weakening the CO bond. A clear link has been established between the IR CO stretching frequency and the Ef -LDOS at the Pt surface: as the electronic density at the Pt surface increases, the IR frequency decreases.14 It has also been shown that the CO IR stretching frequency decreases as the 13C NMR Knight shift for 13CO-Pt increases.9 In that study (of 13CH3OH-derived 13COads on different Pt/C electrocatalysts), it was suggested that larger Knight shifts in 13CO-Pt systems are caused primarily by an increase in 2π* Ef -LDOS at the 13C nucleus, indicative of a greater degree of back-bonding from the Pt conduction band to the CO molecular orbitals. In this light, the trend depicted in Figure 2 shows that 13COads on WB sites, which show larger Knight shifts as compared to 13COads on SB sites, experience greater back-bonding from the Pt conduction band: the Ef-LDOS at 13COads on WB Pt sites is larger than on SB sites. It was shown previously12 that COads is preferentially oxidized from WB sites of Pt, and thus the chemical shift, Ef-LDOS, and reactivity are all related to one another via the chemical properties of the adsorbing site. The implications of this observation could be far reaching. 13C NMR can distinguish between different adsorption sites of 13CO ads on Pt nanoparticles, showing that site-specific information can be obtained from commercially available electrocatalysts. Further, the trend in Figure 2 establishes a spectroscopic link between the active site in 13COads oxidation and 13C NMR shift (and, by extension, Ef-LDOS). It should be noted that this increase in surface electronic density is not necessarily the direct cause of increased activity, as it is possible that the more active WB sites can better accommodate OHads, thereby facilitating faster oxidation of COads. This appears to be the case for PtRu, as 13COads shows decreased Ef -LDOS on the Pt sites, but increased oxidation rates due to the oxophilicity of Ru.16 Nevertheless, 13C NMR is able to identify the active WB sites through their increased Knight shift. Although adsorption sites and active sites have long been observed (and, to some degree, controlled) in single crystal surface science studies, this kind of information has been difficult to ascertain in real electrocatalytic systems. 13C NMR should prove to be a powerful tool for electrocatalysis because of the availability of site-specific information on surface electronic structure. Acknowledgment. This material is based upon work supported by the U.S. Army Research Laboratory and the U.S.
Letters Army Research Office under contract/grant number 48713CH. JAR acknowledges the Deutsche Forschungsgemeinschaft (DFG) for his Mercator Professorship at RWTH Aachen University. References and Notes (1) Wu, J.; Day, J.; Franaszczuk, K.; Montez, B.; Oldfield, E.; Wieckowski, A.; Vuissoz, P.; Ansermet, J. J. Chem. Soc., Faraday Trans. 1997, 93, 1017. (2) Slichter, C. Principles of Magnetic Resonance; Springer, Berlin, 1990. (3) Levitt, M. Spin Dynamics; John Wiley & Sons: West Sussex, U.K., 2001. (4) Blyholder, G. J. Phys. Chem. 1964, 68, 2772. (5) Rudaz, S.; Ansermet, J.; Wang, P.; Slichter, C.; Sinfelt, J. Phys. ReV. Lett. 1985, 54, 71. (6) Yahnke, M.; Rush, B.; Reimer, J.; Cairns, E. J. Am. Chem. Soc. 1996, 118, 12250. (7) Day, J.; Vuissoz, P.; Oldfield, E.; Wieckowski, A.; Ansermet, J. J. Am. Chem. Soc. 1996, 118, 13046. (8) Rush, B.; Reimer, J.; Cairns, E. JES 2001, 148, A137.
J. Phys. Chem. C, Vol. 112, No. 38, 2008 14705 (9) Tong, Y.; Rice, C.; Godbout, N.; Wieckowski, A.; Oldfield, E. J. Am. Chem. Soc. 1999, 121, 2996. (10) Tong, Y.; Wieckowski, A.; Oldfield, E. J. Phys. Chem. B 2002, 106, 2434. (11) McGrath, P. Ph.D. thesis, University of California, Berkeley, 2007. (12) McGrath, P.; Fojas, A. M.; Rush, B.; Reimer, J.; Cairns, E. Electron. Acta 2007, 53, 1365. (13) Van Bramer, S.; Glatfelter, A.; Bai, S.; Dybowski, C.; Neue, G. Conc. Magn. Reson. 2002, 14, 365–387. (14) Tong, Y.; Billy, J.; Renouprez, A.; van der Klink, J. J. Am. Chem. Soc. 1997, 119, 3929. (15) Wu, J.; Day, J.; Franaszczuk, K.; Montez, B.; Oldfield, E.; Wieckowski, A.; Vuissoz, P.; Ansermet, J. J. Chem. Soc., Faraday Trans. 1997, 93, 1017. (16) Babu, P.; Kim, H.; Oldfield, E.; Wieckowski, A. J. Phys. Chem. 2003, 107, 7595. (17) Kobayashi, T.; Babu, P.; Gancs, L.; Chung, J.; Oldfield, E.; Wieckowski, A. J. Am. Chem. Soc. 2005, 127, 14164. (18) Kobayashi, T.; Babu, P.; Chung, J.; Oldfield, E.; Wieckowski, A. J. Phys. Chem. C 2007, 111, 7078.
JP806068T
