Article pubs.acs.org/ac
Step-Scan IR Spectroelectrochemistry with Ultramicroelectrodes: Nonsurface Enhanced Detection of Near Femtomole Quantities Using Synchrotron Radiation Scott M. Rosendahl,†,‡ Ferenc Borondics,‡ Tim E. May,‡ and Ian J. Burgess*,† †
Department of Chemistry, University of Saskatchewan, Saskatoon, Saskatchewan, S7N 5C9 Canada Canadian Light Source, Saskatoon, Saskatchewan, S7N 0X4 Canada
‡
S Supporting Information *
ABSTRACT: The result of interfacing step-scan spectroelectrochemistry with an IR microscope and synchrotron infrared (SIR) radiation is provided here. An external reflectance cell containing a 25 μm gold ultramicroelectrode is employed to achieve an electrochemical time constant less than one microsecond. The use of a prototypical electrochemical system, i.e., the mass-transport controlled reduction of ferricyanide, allows for a proof of principle evaluation of the viability of SIR for step-scan spectroelectrochemistry. An analysis of the importance of accounting for synchrotron source variation over the prolonged duration of a step-scan experiment is provided. Modeling of the material flux in the restricted diffusion space afforded by the external reflectance cell allows the quantitative IR results to be compared to theoretical predictions. The results indicate that only at very short times does linear diffusion within the cavity dominate the electrode response and the majority of the transient signal operates under conditions of quasi-hemispherical diffusion. The analytical information provided by the IR signal is found to be considerably less than that derived from the current response due the latter’s pronounced edge effects. The results provide a detection limit of 36 fmol for step-scan SIR measurements of ferrocyanide. Implications for future IR spectroelectrochemical studies in the microsecond domain are discussed.
A
resolution in IR spectroelectrochemical studies. First, the use of internal reflection geometry and surface enhanced infrared absorption spectroscopy (SEIRAS) eliminates the highly resistive electrolyte layers found in external reflection methods such as PM-IRRAS. There are several reports where SEIRAS has permitted the study of kinetic processes on the ∼1−50 ms time scale.12−14 In fact, picosecond time-resolved SEIRAS on electrode surfaces can be achieved through laser-induced temperature jump experiments,15,16 although this method is not readily amenable to precise control of the electrode’s potential. Alternatively, the time constant can be reduced below ca. a microsecond if an ultramicroelectrode (ume) is used. Working with a ume requires the use of reflectance-mode microspectroscopy which can be disadvantageous as the inherently more complex optics of a microscope leads to higher relative throughput losses compared to internal reflectance SEIRAS. Sun and co-workers succeeded in coupling FT-IR microscopy, a conventional thermal source, and large ultramicroelectrodes (radius ca. 200 μm) while reporting studies of processes occurring on electrode surfaces on the time scale of tens to hundreds of microseconds.10,17,18 Although Sun et al.’s systems exhibited surface enhancement via anomalous IR
lthough not intuitive to the uninitiated, the utilization of mid and far-infrared radiation from synchrotron sources is now well-established, as evidenced by several recent reviews.1−3 The relative advantages of synchrotron infrared radiation (SIR) compared to thermal sources (e.g., globars) have been previously addressed in detail, particularly with respect to SIR’s enhanced spatial resolution in microspectroscopy.4 This group has recently reported a proof of principle experiment illustrating an example of this advantage by using SIR to perform spectroelectrochemistry on ultramicroelectrodes.5,6 Further demonstrating how SIR can be employed for electroanalytical studies is part of this group’s ongoing interests. Although IR spectroelectrochemistry is now a mature field with a host of well-developed external7 and internal8 reflection techniques, most studies to date have typically employed large electrodes to facilitate high photon throughput which prevents fast kinetic studies from being properly performed. This is a consequence of the fact that the establishment of the interfacial potential is determined by the cell’s time constant which is the product of the electrolyte resistance, R, and the electrode’s capacitance, C. As the time constant increases with increasing working electrode area, only very slow processes (on the order of seconds) can be meaningfully studied with large electrodes. Acquisition of spectroscopic data with temporal resolution smaller than five times the cell constant (5xRC) results in the convolution of a time dependent thermodynamic parameter (electrode potential) and the kinetically controlled response.9−11 Two approaches exist to reach faster time © XXXX American Chemical Society
Received: June 7, 2013 Accepted: August 10, 2013
A
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
enhanced spectroscopy (AIRES),19 it is likely that further improvements in their temporal resolution were prohibited by insufficient signal-to-noise levels when equivalent experiments using smaller Pt umes were attempted. Signal to noise (S/N) will be increasingly limited by the instrument’s ability to focus large photon intensity onto very small areas providing an impetus to move to nonconventional sources of IR radiation such as SIR. Herein, we report the use of synchrotron IR as a source for fast, step-scan spectroelectrochemical measurements. The high brilliance of SIR should, in principle, offer lower noise levels compared to thermal sources when using small electrodes ( 0)
(1)
Fe(CN)64−
a mass-transport controlled generation of will result. A detailed description of the diffusion equation for diffusion to an embedded disk electrode under conditions of restricted geometry has been provided in our earlier work.5 It is entirely analogous to the operative conditions found in negative feedback scanning electrochemical microscopy (SECM) studies where finite element (FE) simulations have been successfully employed.21−23 We performed FE simulations using commercial software (FlexPDE) for both the thick and thin cavities. Figure 2 shows semilogarithmic plots of the simulated and measured current transients for a 25 μm radius ume and 10 mM Fe(CN)63−. Comparison of simulation and experiment for the larger cavity allows for evaluation of the microelectrode response and the appropriateness of the simulation routine as Shoup and Szabo provided the following equation for the
■
EXPERIMENTAL SECTION Potassium hexacyanoferrate(III) (K3Fe(CN)6, ≥99.0% trace metals basis), potassium hexacyanoferrate(II) trihydrate (K4Fe(CN)6·3H2O, ≥99.0% trace metals basis), and sodium fluoride (99.998%) were purchased from Sigma Aldrich and were used as received. All aqueous solutions were prepared from Milli-Q water (>18.2 MΩ cm−1). The three electrode flow-through in situ spectroelectrochemical (SEC) cell was similar to previous SEC cells.6 Full experimental details concerning the electrochemical and step-scan measurements are provided in the Supporting Information.
■
RESULTS AND DISCUSSION Electrochemistry in Confined Geometry. The thin layer geometry of the spectroelectrochemical cell creates a restricted diffusion space where semi-infinite conditions exist radially from the electrode’s circumference but a finite diffusion volume exits perpendicular to the electrode’s surface. To emphasize the effect the cavity layer’s thickness has on the electrochemistry, Figure 1 provides cyclic voltammograms (CVs) for a 25 μm radius, re, gold ume in 10 mM Fe(CN)63− plus 0.5 M NaF in the spectroelectrochemical cell with both a 14 μm cavity (the cavity thickness used in the IR measurements, vide infra) and a 5 mm cavity. In the thicker cavity, enhanced mass transfer, arising from the fact that quasi-hemispherical diffusion to the electrode is operative, results in nearly ideal steady-state behavior.20 Much smaller currents are observed in the 14 μm cavity, and the CV exhibits current peaks rather than the sigmoidal shape of a steady-state voltammogram. It is most convenient to define the spatial distribution of the Fe(CN)6 3− concentration in the cell using cylindrical coordinates, C(r,z), where r and z, respectively, run parallel
Figure 2. Current transients for an embedded 25 μm radius Au ultramicroelectrode in 10 mM Fe(CN)63− plus 0.5 M NaF. The larger magnitude curves are for a thick (5 mm) cavity, and the smaller currents are for a 14 μm thick cavity. Open data points are experimental data; blue lines are simulations using finite difference methods, and the solid red line is calculated from eq 2a. B
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
current arising from semi-infinite diffusion to a disk ume24 (equivalent to the large cavity cell) i(τ ) = 4nFDC*ref (τ )
element bounded by the CaF2 window and the ume’s surface. To illustrate the substantial difference in the IR and electrochemical signals, the simulated diffusion results in the restricted cell geometry were used to determine the number of moles of Fe(CN)64− produced during the potential step. This was achieved in the case of the electrochemical signal by integrating the current transient and applying Faraday’s law. To simulate the signal expected from the IR measurements, contour maps of the Fe(CN)63− concentration at different times were integrated in cylindrical coordinates within the region bounded by the electrode surface and the IR window. Previous experiments with much larger electrode surfaces showed excellent agreement between experimental results and the calculated response for finite linear diffusion (i.e., radial diffusion could be considered negligible). Figure 4 compares
(2a)
where f (τ ) = 0.7854 + .8862τ −1/2 + 0.2146e−0.782τ
−1/2
(2b)
and
τ = 4Dt /re 2
(2c)
As shown in Figure 2, the simulated current response and the calculated current transient are almost superimposable and both are very close to the experimentally measured curve. Mauzeroll et al. reported similar consistencies when modeling SECM approach curves.25 Figure 2 also compares finite difference simulations and experimentally measured i(t) curves for the 14 μm cavity cell. The observed agreement between simulation and experiment is also very good, and the difference in the total charge passed (integral of the transients) for the two plots is less than 2.5%. Comparisons between the signal assessed by electrochemical detection and that observable through infrared spectroscopy can be made using the simulated results. These analytical signals will be significantly different as the former will be dominated by semi-infinite diffusion to the perimeter of the electrode whereas the IR is indifferent to this source of Fe(CN)63− flux.5 Figure 3 illustrates the edge effect by
Figure 4. Comparison of the predicted amount of Fe(CN)64− produced during the current transient in the thin cavity spectroelectrochemical cell. The red line is the result of integrating the current transient; the blue line is the calculated response assuming finite linear diffusion, and the black line is the simulated IR response.
these three possible responses by plotting the expected number of moles of Fe(CN)64− produced as a function of time. As expected, the signal from the current transient greatly exceeds the other curves and, as shown in the inset of Figure 4, increases with time. In contrast, the finite linear-diffusion treatment results in a plateau of 2.7 × 10−13 moles which is reached approximately 0.5 s after the potential perturbation. The finite differences solution of this diffusion problem can also simulate the change in the concentration of ferricyanide species in the thin-volume cavity which would be in the beam path of the IR radiation. This simulated IR response reveals similar behavior to that of the finite linear diffusion albeit with a systematically smaller extent of conversion. In the case of finite linear diffusion, the depletion zone rapidly extends across the entire cell cavity and exhaustively consumes the ferricyanide. However, if the radius of the electrode is comparable to the thickness of the electrolyte cavity, the depletion zone of Fe(CN)63− extends radially away from the electrode center and the cavity volume above the electrode is not completely electrolyzed within the time frame of this experiment. The simulated IR response provides a significantly smaller (∼13%) maximum analytical signal compared to that predicted assuming finite linear diffusion. IR Step-Scan Spectroelectrochemistry with Synchrotron Radiation in Decay Mode. The ferri/ferrocyanide redox couple is well suited for IR spectroelectrochemistry owing to the clear spectral shift in the cyanide stretches of the reduced and oxidized forms which are both frequency resolved
Figure 3. Contour maps of the simulated Fe(CN)63− normalized concentration profiles within the thin (14 μm) spectroelectrochemical cell, 1 s after a potential step. The red dotted box defines the crosssection of the cavity volume that is sampled by the incident IR radiation.
providing a contour map of the finite elements simulated Fe(CN)63− concentration in the electrode vicinity one second after the potential step. An animation of the evolving diffusion space over the duration of the transient is available (Figure S-1 in the Supporting Information). In the snapshot provided in Figure 3, one can observe a pronounced concentration gradient extending radially along the electrode’s mantle (z = 0, r/re > 1) which gives rise to a large flux of material to the electrode circumference. The red box in Figure 3 delineates the space that is sampled by the incident IR radiation. This volume does not contain large gradients but instead shows a much more uniformly distributed Fe(CN)63− concentration. Species electrogenerated at the edges will not be sampled by the infrared radiation as they remain excluded from the volume C
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
from water absorption bands.26−28 The spectra obtained using step-scan interferometry exhibit a strong peak at 2038 cm−1 which is assigned to ferrocyanide (Figure 5). A much weaker
box in Figure S-3, Supporting Information, encloses a complete set of mirror positions representing a single block of 100 μs time-resolved inteferograms. A single block of data requires ∼10 min of measurement and is conducted n = 256 times for a total experiment time of nearly 48 h. Two data processing approaches were employed to explore the impact of source intensity variation. In method A, each block was treated individually by Fourier transforming the interferogram for every time slice producing a family of energy curves, Sn(t,υ̃), and then subtractively normalizing using the energy curves for t < 0 (i.e., prior to the potential step) as a reference. Each block was processed in this fashion on an individual basis and then coadded as expressed mathematically below ΔS(t , υ)̃ S
= A
Sn(t , υ)̃ −1 Sn(t = 0, υ)̃
(3)
In method B, the 256 measurements for every time slice at each mirror position were first coadded to produce ⟨Pδ(t)⟩ (reddotted boxes in Figure S-3, Supporting Information). A single Fourier transform was applied to these averaged signals to generate an average energy curve for each time slice, ⟨S(t,υ̃)⟩, and then subtractively normalized as follows
Figure 5. Step-scan spectra for the ferro/ferricyanide system in the electrochemical cell at t = 1 s using an embedded 25 μm radius Au ultramicroelectrode in a 14 μm thick cavity. A comparison is made between the coaddition of 35 SIR interferograms (black line) and an equivalent number of interferograms using a conventional thermal source (red line).
ΔS(t , υ)̃ S
band at 2115 cm−1 is attributable to ferricyanide and becomes more pronounced with higher resolution (Figure S-2, Supporting Information). Figure 5 also demonstrates the advantage of increased brilliance by overlaying the SIR spectrum with an equivalent measurement made with the microscope’s thermal source. Any analytical signal in the latter spectrum is masked by significant noise levels. While it is clear from Figure 5 that step-scan interferometry using SIR is suitable to follow the ferro/ferricyanide conversion, it is important to also consider data acquisition and noise limitations whose general relevance to step-scan interferometry is described elsewhere.29−31It is prudent to discuss two aspects that are particularly relevant to this study using SIR. First, the successful utilization of step-scan spectroscopy requires a system that exhibits long-term stability and a repeatable response to a triggered perturbation. Herein, a double potential step (E+0.4 V − E−0.4 V ; E−0.4 V − E+0.4 V) was applied to the microelectrode at each of the 170 individual mirror positions that constitute a single interferogram. In the process of acquiring 256 interferograms, the potential was stepped approximately 90 000 times across the ferri/ferrocyanide formal potential. To assess the stability of the system, the current transients for each potential step were recorded and integration revealed that the total charge passed only decreased by ∼15% over the course of the two-day experiment indicating very good system stability. Second, unlike a conventional thermal source, some (but not all) synchrotrons such as the CLS operate in socalled “normal mode” whereby the source intensity decays exponentially with time. At the CLS, storage ring refills occur every eight hours and the beam current decays by about 25% over this period. It should be emphasized that some synchrotrons such as the Advanced Light Source and SOLEIL operate in “top-up” mode whereby a constant beam current is maintained. Figure S-3, Supporting Information, illustrates how beam decay can affect data processing through a schematic representation of the position of the moveable mirror as a function of time. Each rectangular point corresponds to the triggered acquisition of t = 1.5 s worth of data. The black dotted
= B
⟨S(t , υ)̃ ⟩ −1 ⟨S(t = 0, υ)̃ ⟩
(4)
Thus, the two methods can be considered to be (A) an average of normalized spectra and (B) a normalization of averaged spectra. The former method is more demanding computationally, due to the much greater number of FFTs needed to be performed, but provides an internal normalization of source variation over a much shorter time period (∼10 min) compared to the latter method (∼2 days). To illustrate the differences, the ferricyanide spectroelectrochemical data was analyzed using the two data processing methodologies. The noise level, sN, of each method was determined by considering the standard deviation of the noise in a 50 ms window of the subtractively normalized signal at 2040 cm−1 prior to the application of the potential step (i.e., t = −0.25 s ± 25 ms). Figure 6 provides plots of the inverse of sN as a function of the square root of the number of coadded spectra, n. After coadding all 256 spectra, the noise level from method B is
Figure 6. Inverse of the standard deviation of the noise plotted as a function of the square root of the number of coadded spectra. Black open circles are sequential coadditions using eq 3, and the red open triangles are determined using eq 4. The blue open squares are the result of randomly coadding the 256 normalized spectra using eq 3. D
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
the analytical signal giving rise to the LOD is not reached until 10 ms after the potential step. The thickness of the depletion layer, zd, after 10 ms can be estimated, using zd = (4Dt)1/2, to be approximately 5 μm. Clearly, the current experimental configuration provides no surface enhancement, and yet, detection limits on the order of tens of femtomoles are possible. Optimization of conditions, particularly by using different redox couples that have larger intrinsic absorptivities and peak absorbances at frequencies removed from beamline interferences (e.g., a diamond window at the 01B1 mid-IR beamline at the CLS attenuates the throughput at 2000−2200 cm−1 by approximately 50%) could lower the LOD by an estimated further order of magnitude. While impressive, this number must be placed in context with the overarching goal of measuring electrochemical processes generated at an ultramicroelectrode with a temporal resolution on scale with the electrode’s time constant. The present study demonstrates that it is not plausible for completely diffusion controlled processes to provide sufficient electrochemical turnover on microsecond time scales. However, surface confined reactions such as CO poisoning of Pt during the oxidation of methanol or other fuels on electrocatalytic surfaces are obvious targets for future stepscan SIR spectroelectrochemical studies. To reach adequate detection levels for submonolayer coverages, SIR for spectroelectrochemical applications will require a means to achieve surface enhancement through the implementation of grazing angle objectives and/or the design of plasmonic structures. For example, individual gold nanorods with lengths of a few micrometers provide 5-fold enhancement factors,32,33 whereas arrays of such nanorods have recently been reported to give between ∼103 and 105 surface enhancement factors.34−36
found to be nearly double that of method A, clearly emphasizing the advantage of a data processing method that inherently accounts for source intensity variation. Interestingly, the order in which the spectra were coadded affects the shape of the ensemble averaging results for method A. If the 256 subtractively normalized spectra are sequentially coadded in the order of which they were measured, portions of the plot are linear (with varying slope) but in other regions additional coadditions do not improve the noise level. On the other hand, if the 256 spectra produced from method A are randomly coadded, a nearly perfect linear dependence of sN−1 on √n results. This demonstrates the degree of noise variation that occurs during the course of a prolonged SIR experiment. Noise can originate from source beam motion and along the optical path as mechanical vibration and thermal drifts affect the mirrors and their mounting chambers. The large number of mirrors in the beam path means that isolating the primary contribution that leads to vacillating noise levels is not simple nor is it immediately obvious when an anomalously noisy interferogram has been recorded. However, Figure 6 does indicate that the overall quality of the step-scan SIR data is degraded by a relatively small number of compromised interferograms. Further improvements in this regard will improve the benefits of ensemble averaging. The unfiltered IR transient showing the formation of eletrogenerated ferrocyanide within the thin volume cavity is shown in Figure 7. The subtractively normalized peak height at
■
SUMMARY AND CONCLUSIONS This work has improved upon previous spectroelectrochemical studies using SIR radiation by implementing step-scan interferometry and a true ultramicroelectrode. The masstransport limited electrochemical signal has been successfully modeled using a finite-differences approach, and the diffusion behavior within a finite cavity volume has been fully described. Whereas purely electrochemical responses in the thin-cavity cell include contributions from both linear and radial diffusion, it has been shown that the IR spectroscopic signal arises only from species originating in the thin-cavity between the electrode surface and the IR window resulting in a maximum analytical signal of 235 fmol. Time-resolved detection of this electrogenerated ferrocyanide has been accomplished by coupling step-scan experiments with SIR. Spectroelectrochemical measurements with time resolution approaching the time constant of the cell (ca. 1 μs) can be readily made with the configuration described. This work has also provided a description of data processing methods that illustrate a possible deleterious effect caused by synchrotron-derived source variation. Careful consideration of data processing strategies is needed to account for beam current decay. By correctly doing so, noise signals behave in a random fashion and ensemble averaging provides the expected √n dependence. The estimated limit of detection for ferrocyanide detection approaches the femtomole level. However, it is evident that further advancement in microsecond-resolved spectroelectrochemical measurements of electrochemical processes will require surface enhancement strategies.
Figure 7. Transient response of the 2040 cm−1 band. The inset shows the filtered transient (purple line) and is compared to the calculated results for finite, linear diffusion (blue line) and the simulated results using finite differences (black line). The limit of detection (LOD) is shown as the horizontal black line in the main body of the figure.
the position of the ferrocyanide band maximum (υ̃max = 2040 cm−1) was converted first to an absorbance and finally to the number of moles of ferrocyanide produced using Beer’s law. Molar extinction coefficients for ferrocyanide as a function of spectrometer resolution were determined in separate experiments and are shown in Figure S-4 and Table S-1, Supporting Information. The data presented in the main body of Figure 7 was subjected to a 50 Hz low pass filter to assess the agreement between the experimentally measured transient and that expected from the finite difference simulations. Figure 7’s inset reveals that the simulation of diffusion within the thin cavity is in very good agreement with the measured result. Figure 7 also provides some insight on the maximum achievable sensitivity using synchrotron IR microspectroscopy and ultramicroelectrodes. The limit of detection (LOD = three times the standard deviation of the noise level) is 36 fmol, and E
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
■
Article
(14) Wisitruangsakul, N.; Zebger, I.; Ly, K. H.; Murgida, D. H.; Ekgasit, S.; Hildebrandt, P. Phys. Chem. Chem. Phys. 2008, 10 (34), 5276−5286. (15) Yamakata, A.; Uchida, T.; Kubota, J.; Osawa, M. J. Phys. Chem. B 2006, 110 (13), 6423−6427. (16) Yamakata, A.; Osawa, M. J. Phys. Chem. C 2008, 112 (30), 11427−11432. (17) Zhou, Z. Y.; Tian, N.; Chen, Y. J.; Chen, S. P.; Sun, S. G. J. Electroanal. Chem. 2004, 573 (1), 111−119. (18) Zhou, Z. Y.; Sun, S. G. Electrochim. Acta 2005, 50 (25−26), 5163−5171. (19) Lu, G. Q.; Sun, S. G.; Cai, L. R.; Chen, S. P.; Tian, Z. W.; Shiu, K. K. Langmuir 1999, 16 (2), 778−786. (20) Heinze, J. Angew. Chem. 1993, 105 (9), 1327−1349. (21) Bard, A. J.; Denuault, G.; Friesner, R. A.; Dornblaser, B. C.; Tuckerman, L. S. Anal. Chem. 1991, 63 (13), 1282−1288. (22) Kwak, J.; Bard, A. J. Anal. Chem. 1989, 61 (11), 1221−1227. (23) Mirkin, M. V. Theory. In Scanning Electrochemical Microscopy; Bard, A. J., Mirkin, M. V., Eds.; Marcel Dekker: New York, 2001; pp 145−200. (24) Shoup, D.; Szabo, A. J. Electroanal. Chem. Interfacial Electrochem. 1982, 140 (2), 237−245. (25) Mauzeroll, J.; Hueske, E. A.; Bard, A. J. Anal. Chem. 2003, 75 (15), 3880−3889. (26) Korzeniewski, C.; Severson, M. W.; Schmidt, P. P.; Pons, S.; Fleischmann, M. J. Phys. Chem. 1987, 91 (22), 5568−5573. (27) Pharr, C. M.; Griffiths, P. R. Anal. Chem. 1997, 69 (22), 4673− 4679. (28) Pons, S. J. Electroanal. Chem. 1984, 160 (1−2), 369−376. (29) Andrews, S. S.; Boxer, S. G. Appl. Spectrosc. 2001, 55 (9), 1161− 1165. (30) Rodig, C.; Siebert, F. Instrumental Aspects of Time-Resolved Spectra Generated Using Step-Scan Interferometers. In Handbook of Vibrational Spectroscopy; John Wiley & Sons, Ltd: New York, 2006. (31) YaoWu, X.; Mark, B. Step-Scan Time-Resolved FT-IR Spectroscopy of Biopolymers. In Vibrational Spectroscopy of Biological and Polymeric Materials, CRC Press: Boca Raton, FL, 2005; pp 353− 418. (32) Neubrech, F.; Kolb, T.; Lovrincic, R.; Fahsold, G.; Pucci, A.; Aizpurua, J.; Cornelius, T. W.; Toimil-Molares, M. E.; Neumann, R.; Karim, S. Appl. Phys. Lett. 2006, 89 (25), 253104−/1−253104/3. (33) Neubrech, F.; Weber, D.; Lovrincic, R.; Pucci, A.; Lopes, M.; Toury, T.; de, L. C. Appl. Phys. Lett. 2008, 93 (16), 163105−1− 163105/3. (34) Adato, R.; Yanik, A. A.; Amsden, J. J.; Kaplan, D. L.; Omenetto, F. G.; Hong, M. K.; Erramilli, S.; Altug, H. Proc. Natl. Acad. Sci. 2009, 106 (46), 19227−19232. (35) Brown, L. V.; Zhao, K.; King, N.; Sobhani, H.; Nordlander, P.; Halas, N. J. J. Am. Chem. Soc. 2013, 135 (9), 3688−3695. (36) Pucci, A.; Neubrech, F.; Weber, D.; Hong, S.; Toury, T.; de la Chapelle, M. L. Phys. Status Solidi B 2010, 247 (8), 2071−2074.
ASSOCIATED CONTENT
S Supporting Information *
Experimental details, an animation of the concentration contour map of the thin-cavity cell during the electrolysis, a subtractively normalized IR spectrum of ferri/ferrocyanide at different resolution, a schematic of the time course of a prolonged step-scan experiment, a plot of absorptivity as a function of ferricyanide concentration, and a table of molar extinction coefficients as a function of resolution. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC). S.M.R. acknowledges NSERC for graduate funding through a post graduate scholarship (PGS-D). The authors are grateful to Andrzej Baranski for assistance in the preparation and characterization of gold ultramicroelectrodes. Research described in this paper was performed at the Mid-IR beamline of the Canadian Light Source, which is supported by the Natural Sciences and Engineering Research Council of Canada, the National Research Council Canada, the Canadian Institutes of Health Research, the Province of Saskatchewan, Western Economic Diversification Canada, and the University of Saskatchewan.
■
REFERENCES
(1) Dumas, P.; Miller, L. Vib. Spectrosc. 2003, 32 (1), 3−21. (2) Dumas, P.; Sockalingum, G. D.; Sule-Suso, J. Trends Biotechnol. 2007, 25 (1), 40−44. (3) Miller, L. M.; Dumas, P. Biochim. Biophys. Acta, Biomembr. 2006, 1758 (7), 846−857. (4) Miller, L. M.; Smith, R. J. Vib. Spectrosc. 2005, 38 (1−2), 237− 240. (5) Rosendahl, S. M.; Borondics, F.; May, T. E.; Pedersen, T. M.; Burgess, I. J. Anal. Chem. 2011, 83 (10), 3632−3639. (6) Rosendahl, S. M.; Borondics, F.; May, T. E.; Pedersen, T. M.; Burgess, I. J. Rev. Sci. Instrum. 2011, 82 (8), 083105. (7) Zamlynny, V.; Lipkowski, J. Quantitative SNIFTIRS and PM IRRAS of Organic Molecules at Electrode Surfaces. In Diffraction and Spectroscopic Methods in Electrochemistry; Alkire, R. C., Kolb, D. M., Lipkowski, J., Ross, P. N., Eds.; Wiley-VCH: Weinheim, 2006; Vol. 9, pp 315−376. (8) Osawa, M. In-Situ Surface-Enhanced Infrared Spectroscopy of the Electrode/Solution Interface. In Diffraction and Spectroscopic Methods in Electrochemistry; Alkire, R. C., Kolb, D., Ross, P. N., Lipkowski, J., Eds.; Wiley-VCH: Weinheim, 2006; pp 269−314. (9) Nowak, C.; Luening, C.; Schach, D.; Baurecht, D.; Knoll, W.; Naumann, R. L. C. J. Phys. Chem. C 2009, 113 (6), 2256−2262. (10) Zhou, Z. Y.; Lin, S. C.; Chen, S. P.; Sun, S. G. Electrochem. Commun. 2005, 7 (5), 490−495. (11) Ataka, K.; Hara, Y.; Osawa, M. J. Electroanal. Chem. 1999, 473 (1,2), 34−42. (12) Osawa, M.; Yoshii, K.; Ataka, K.; Yotsuyanagi, T. Langmuir 1994, 10 (3), 640−642. (13) Pronkin, S.; Wandlowski, T. J. Electroanal. Chem. 2003, 550− 551, 131−147. F
dx.doi.org/10.1021/ac401799z | Anal. Chem. XXXX, XXX, XXX−XXX
