Detection of Low-Concentration Contaminants in Solution by

Aug 18, 2014 - ... School of Mathematics and Physics, Queen's University Belfast, BT7 ... Serstech AB, Ideon Science Park, Scheelev 15, SE-223 70, Lun...
9 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Detection of Low-Concentration Contaminants in Solution by Exploiting Chemical Derivatization in Surface-Enhanced Raman Spectroscopy Mike Hardy,† Matthew D. Doherty,† Igor Krstev,‡ Konrad Maier,‡ Torgny Möller,⊥ Gerhard Müller,‡ and Paul Dawson*,† †

Centre for Nanostructured Media, School of Mathematics and Physics, Queen’s University Belfast, BT7 1NN, Belfast, United Kingdom ‡ Airbus Group Innovations, D-81663, Munich, Germany ⊥ Serstech AB, Ideon Science Park, Scheelev 15, SE-223 70, Lund, Sweden S Supporting Information *

ABSTRACT: A simple derivatization methodology is shown to extend the application of surface-enhanced Raman spectroscopy (SERS) to the detection of trace concentration of contaminants in liquid form. Normally in SERS the target analyte species is already present in the molecular form in which it is to be detected and is extracted from solution to occupy sites of enhanced electromagnetic field on the substrate by means of chemisorption or drop-casting and subsequent evaporation of the solvent. However, these methods are very ineffective for the detection of low concentrations of contaminant in liquid form because the target (ionic) species (a) exhibits extremely low occupancy of enhancing surface sites in the bulk liquid environment and (b) coevaporates with the solvent. In this study, the target analyte species (acid) is detected via its solid derivative (salt) offering very significant enhancement of the SERS signal because of preferential deposition of the salt at the enhancing surface but without loss of chemical discrimination. The detection of nitric acid and sulfuric acid is demonstrated down to 100 ppb via reaction with ammonium hydroxide to produce the corresponding ammonium salt. This yields an improvement of ∼4 orders of magnitude in the low-concentration detection limit compared with liquid phase detection. fingerprinting; the fluorescent- or marker-free nature of the technique is also of immense utility in biosensing applications. It has also found use in a wide range of applied fields, including, for example, medicine8 and the detection of drugs and explosives.9−12 In SERS, the Raman cross-section of molecules adsorbed on a surface supporting LSPRs is massively increased, first due to the enhanced local fields exciting the Raman dipole and, then, as a result of optical reciprocity because of the enhancement of the instantaneous re-emission of radiation by the dipole.6 The SERS enhancement factor (EF) is generally defined as the increase of Raman scattering from a molecule at a specific point over the scattering for the same molecule under normal conditions13 and is related to the local electric field enhancement near a surface by the well-known “E4” formula.6 However, the more experimentally relevant figure is the average EF across a given nanostructured surface. As a result of the localization of electromagnetic energy to very small volumes (or “hot-spots”) by LSPRs on structured surfaces such substrates can routinely exhibit average EFs of 104−106, with

N

oble metal nanoparticles and nanostructured surfaces have attracted widespread interest for well over a decade now, forming the backbone of the burgeoning field of plasmonics as a result of their ability to support localized surface plasmon-polariton resonances (LSPRs).1,2 These excitations are of broad interest and application because of two fundamental properties: strong wavelength (and often polarization) selective optical resonances in the far-field and highly enhanced and localized concentration of electromagnetic radiation in the near-field.3,4 The subdiffraction limit focusing of electromagnetic fields upon optical excitation of LSPRs is responsible for the huge signal increases observed in surfaceenhanced spectroscopies, in particular surface-enhanced Raman spectroscopy (SERS).5−7 The extraordinary signal enhancing capabilities of SERS have obvious utility in low-concentration chemical detection, and in this work, we present a simple methodology to extend the use of SERS to the unexploited area of detection of trace chemicals in the liquid phase. Raman spectroscopy is a technique for observing vibrational modes via the inelastic scattering of light from a molecule or other system. It is commonly used in chemistry, since vibrational information is specific to chemical bonds and the symmetry of molecules, and hence can be used for molecular © 2014 American Chemical Society

Received: April 17, 2014 Accepted: August 18, 2014 Published: August 18, 2014 9006

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

Article

Figure 1. (a) Sketch illustrating liquid analyte deposition for Raman spectroscopy on a SERS substrate; magnified 3D schematic of the SERS surface illustrates the very low percentage of analyte molecules in close proximity to the surface. (b) Sets of Raman spectra of 5% nitric acid solutions on nanostructured gold Klarite SERS substrate (top) and a flat gold surface (bottom) illustrating, first, the very weak SERS effect experienced by analytes in the liquid phase and, second, that the Raman signal decays upon drying since the analyte species evaporates congruently from the surface with the solvent molecules. (c) Scanning electron micrograph of the Klarite substrate. (Details of experimental conditions and procedure are given below.)

maximum EFs exceeding 1011 under ideal conditions.13,14 Signal enhancements of this size have obvious potential for extending the application of conventional Raman spectroscopy to systems where signal level presents a problem, such as trace chemical detection. However, the effective utilization of SERS depends on the reliable fabrication of nanostructured, enhancing substrates and on placing the species to be detected in intimate proximity to the substrate, specifically and crucially, within hot-spot regions of extreme field localization. The importance of such molecular placement is underscored in the work of Fang et al.15 who demonstrated that >50% of the SERS signal can originate from less than 1% of the substrate area. Typically comprising an intermetallic gap of a few nanometers in dimension16 such hotspots will often determine the (near-field) SERS response while having very little or no far-field optical signature.16,17 In previous work, using Au nanorod substrates, we have quantified a gap dimension of ∼7−8 nm below which the SERS response of the substrate is hot-spot dominated.18 Experiment and modeling in our own16−18 and other studies19,20 show a significant falloff in near-field intensity within a few nanometers of the enhancing surface outside such regions. Two of the most commonly used procedures for functionalizing a SERS substrate with low concentrations of

analyte molecules are drop-casting and chemisorption. Both of these techniques can lead to the adsorption of analyte molecules in the requisite nanometer-scale proximity to the SERS surface to yield a significant EF. In chemisorption the SERS surface is exposed to the analyte solution which then bonds with the metal to form a layer over the surface;21,22 the surface will typically then be rinsed to remove excess molecules. This technique is obviously only applicable to molecules which form a strong bond with noble metal surfaces (typically thiols or similar), severely limiting applicability. In drop-casting a low concentration of a solid analyte is dissolved in solvent (typically water or ethanol), a small drop of which is cast onto the SERS surface and then allowed to evaporate, precipitating out a layer of analyte molecules over the surface.17,18 Both of these techniques can lead to the adsorption of analyte molecules in the requisite nm-scale proximity to the SERS surface to yield a significant EF. However, a severe disadvantage of both techniques is that they are limited to analytes which are able to adsorb on SERS substrates much more tightly than the solvent ones. This latter condition is only fulfilled for a very limited set of analytes, such as thiols, which severely limits the more general applicability of the SERS technique,23,24 despite its 40 years in existence.25 9007

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

Article

Figure 2. (a) Sketch illustrating the derivatization drop-casting technique where contaminant in solution forms salt and is then allowed to air-dry; magnified 3D schematic of SERS surface illustrates higher concentration of derivative molecules very close to the surface after drying. Panels b and c show SERS spectra from nitric acid and sulfuric acid solutions, respectively: 5% (black), 1% (blue) concentration liquid solutions on SERS surface before evaporation, and 100 ppm (red) solution derivatized using ammonium hydroxide and allowed to dry as described in the text. Each set of spectra is plotted on same Raman intensity scale with individual spectra offset for clarity.

substrates. In the present paper we show that this class of analytes can be successfully SERS-detected by applying a chemical derivatization technique to the adsorbed liquid solutions. This derivatization process transforms the dissolved analytes into solid precipitates and brings them into close contact with the hot-spot regions of the SERS substrates, while at the same time it allows the remaining solvent to evaporate. We show that in this way, enhancements with EF ≈ 4 orders of magnitude relative to those on the liquid adsorbate layers can be obtained. With this method minimum detectable nitrate ion concentrations lower than 100 ppb in water solute layers can be obtained. It is argued that these results open up perspectives for NO2 and SO2 gas sensors which are comparable with the sensitivities of state of the art solid state gas sensors providing, at the same time, a much higher degree of gas selectivity. This latter fact is demonstrated by comparative measurements on nitrate and sulfate ion-containing solutions.

An important class of analytes which do not conform to the above requirements are acid-forming gases, such as NO2 and SO2, which are toxic in low ppm concentrations in the ambient air and which need to be monitored for a variety of reasons. NO2 and SO2 have a strong affinity toward water vapor. When present in the atmosphere, these tend to form very dilute acid adsorbate layers on SERS substrates. SERS measurements on the liquid adsorbates, however, do not succeed as most of the target analytes in such dilute solutions are far from the hotspots of the SERS substrates. This problem is illustrated schematically in Figure 1a. It is underscored emphatically and quantitatively in Figure 1b where the Raman spectrum from a drop of solution of nitric acid deposited on a nanostructured gold SERS substrate (Klarite) is enhanced by less than 3× relative to the signal retrieved from a drop of nitric acid solution deposited on a smooth Au film (red curves in each set of spectra); this very modest enhancement factor is further reduced toward unity if the exposed area is taken into account. (Normally, as already noted, if the SERS substrate were effective the EF should be ≥104 approximately.) Further, Figure 1b illustrates the compounding problem that drying the sample is ineffective, indeed quite counterproductive. Drying is not successful because the acid solutions evaporate congruently and thus do not leave any measurable amount of analyte on the

■

EXPERIMENTAL SECTION To experimentally demonstrate the methodology we have chosen acid analyte and base reagent molecules. The first example is the acid−base reaction between nitric acid and ammonium hydroxide 9008

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

Article

camera attached to a Jobin-Yvon HR640 spectrometer operating with a 300 lines mm−1 grating. The samples were mounted on an X−Y translation stage to facilitate positioning relative to the input laser.

HNO3(aq) + NH4OH(aq) → NH4NO3(aq) + H 2O → NH4NO3(s)

(1)

■

producing ammonium nitrate, a solid which then precipitates out of solution and can be detected at low concentration using SERS. Sulfuric acid follows a similar reaction scheme, producing ammonium sulfate

RESULTS AND DISCUSSION Figure 2b and c shows Raman spectra obtained on Klarite substrates from nitric and sulfuric acid in solution and from their ammonium salt derivatives prepared using the technique above. The EF obtained for the liquid samples when using the SERS substrate is only ∼3 relative to the unenhanced signal (see Figure 1b), whereas for the derivatized, then adsorbed chemicals a concentration in the original solution 2 orders of magnitude or more lower (100 ppm vs 1% or 5%) clearly produces a much more intense signal. These data thus unambiguously illustrate how the technique facilitates exploitation of the massive enhancement offered by SERS when interrogating low concentration liquid contaminants in solution. No peaks in the Raman signal corresponding to those of Figure 2b and c are observed when 1% NH4OH solution is added to deionized water, giving assurance that the obtained spectra are indeed associated with the respective ammonium salts, and thus indicative of presence of the original acid and not impurities in the NH4OH reagent. To illustrate more quantitatively how this technique benefits the observed Raman signal, and ultimately the detection limit, measurements were taken using nitric acid over a wide concentration range. Figure 3 shows the observed signal

H 2SO4 (aq) + 2NH4OH(aq) → (NH4)2 SO4 (aq) + 2H 2O ⇌ (NH4)2 SO4 (s)

(2)

We comment later on the use of the equilibrium symbol in this case. Stock solutions of acid were prepared by standard dilution procedures, using a graduated pipet to dispense a small volume of concentrated (70%) nitric acid or concentrated (98%) sulfuric acid into 100 mL volumetric flasks with the balance being made up with freshly prepared deionized water. Sequential dilutions yielded solutions of acid with concentrations down to 100 ppb. These volumetric ratios relate to other concentration measures as follows. For HNO3 solutions, a 1% or 1 in 100 concentration (used only for liquid case reference purposes in this study) corresponds to 1 molecule of HNO3 per 230 molecules of H2O and a molality of ∼0.240 mol kg−1; for H2SO4, this translates to 1 molecule of H2SO4 per 295 molecules of H2O and a molality of ∼0.188 mol kg−1. The substrates used in this work are of two types, planar Au surfaces where Au was deposited onto oxidized silicon wafers and commercial SERS substrates, Klarite (Renishaw Diagnostics Ltd.). The latter are comprised of a highly granular gold coating in pyramidal pits etched into silicon substrates (Figure 1c). They offer relatively uniform and reproducible SERS EF (± 10%) and were used as the enhancing substrate throughout this study. The planar Au surfaces thus offer a control-case SERS substrate of smooth Au. Two microliter drops of the various acid solutions were applied to both the planar Au surfaces and the Klarite substrates using a micropipette, and Raman measurements were taken immediately after drop-casting. Apart from the spectra of Figure 1b Raman measurements were not taken during or after air drying since the analyte signal decays significantly, tending to zero, as evaporation proceeds (Figure 1b). To assess the effect of chemical derivatization, the same experiment as above was repeated, but this time a second 2 μL drop of excess reagent (1% concentration NH4OH) was added to the analyte droplet to react any target acid present to yield its corresponding ammonium salt. This droplet of reacted acid and excess reagent was then allowed to air-dry, leaving only the adsorbed derivative molecules on the SERS substrate; Raman spectra were then recorded. This process is illustrated in Figure 2a, the principle being that the target anionic species (now incorporated in a solid derivative) is in contact or much closer proximity to the substrate surface, where field enhancement is greatest (and in greater density that the liquid case) leading to a significant SERS effect. This Raman spectra were acquired with a lab-built Raman spectroscopy system in the backscattering configuration. The laser excitation wavelength was 780 nm with an input power of ∼30 mW focused to a spot size of ∼100 μm diameter. Using a narrow band-pass filter on the input to the sample and a longpass edge filter on the output from the sample Raman spectra were acquired using an air cooled Andor DU-420-OE CCD

Figure 3. Plot of the intensity of the main 1050 cm−1 Raman band against concentration in parts per million for nitric acid in solution (blue triangles), the derivatized ammonium nitrate on a planar gold surface (red diamonds), and ammonium nitrate on a Klarite SERS substrate (black squares). Concentration axis refers to that of original nitric acid solution, before derivatization using ammonium hydroxide. All Raman intensities plotted on the same scale in arbitrary units.

intensity at 1050 cm−1 of the NO3− band of nitric acid in solution and of ammonium nitrate derivative on both a planar gold surface and on Klarite, as the concentration of the original solution is varied from 5% down to 100 ppb. For all three situations investigated it is clear that as the concentration of the original acid falls so does the signal, however the signal intensities and the detection limits vary dramatically. The detection limit for nitric acid in solution is above 1000 ppm. When the ammonium nitrate is dried onto a planar gold surface 9009

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

Article

(where there is very little enhancement) the signal intensity increases and the detection limit falls into the 100−1000 ppm range. The increased signal is due to an increase in the concentration of molecules sampled; as illustrated in Figure 2a when molecules precipitate out of solution they become concentrated on the surface, and hence the number in the sampling volume of the laser focus increases. However, by far the largest increase in signal, and corresponding decrease in detection limit, is observed when the derivative molecule is adsorbed on the SERS substrate. In this case there is still an observable signal at a concentration of 0.1 ppm, an improvement of ∼4 orders of magnitude over the original acid solution. This result clearly and quantitatively demonstrates that the derivative molecules take advantage of the substrate SERS enhancement, whereas the liquid does not. It can also be seen that the Raman intensity does not fall off linearly with concentration, a feature that is likely linked to the distribution of molecule−substrate separations where those molecules adsorbed closest to the surface experience the largest EF and hence generate the most signal.15 However, further careful experimentation would be required to yield a reliable calibration of the SERS signal referenced to concentration. It is important to emphasize that this increase in sensitivity does not come at the cost of chemical discrimination: the spectra of different derivatives can still be clearly distinguished at low concentration. The Raman spectra of nitric acid and the ammonium nitrate salt both display a prominent Raman band at ∼1050 cm−1. This isolated spectral feature, characteristic of both dilute nitric acid solutions and of many nitrate salts, is due to a vibrational mode of the f ully dissociated nitrate anion NO3−.26,27 In contrast, the sulfuric acid solution displays two peaks, one at 1046 cm−1 and one at 980 cm−1. H2SO4 undergoes a 2-stage dissociation process where, in the first stage, there is complete dissociation to HSO4− and H+ over a broad range of concentrations while the second stage, producing SO42− and a further H+, is temperature and concentration dependent.28 These anionic moieties generate vibrational peaks at 1046 and 980 cm−1 respectively (Figure 2c) as identified in refs 29 and 30, where the presence of the lower wavenumber peak yields clear and instantly identifiable spectral contrast with the case of nitric acid and ammonium nitrate (Figure 2b). In addition, since the ionic molal scattering coefficients are almost identical (and temperature independent) for the two ions, HSO4− and SO42−28,31 the integrated intensities of the Raman peaks give a direct measure of the relative concentration of each species. Interestingly, however, both peaks can persist after the derivative salt has been formed, indicating the presence of both species. This is an incongruous result for the case of (NH4)2SO4 where the Raman spectrum should be dominated by peaks associated with only SO42− in the range 600−1200 cm−1, principally the stretching vibration at 980 cm−1. The persistence of the HSO4− peak may be associated with the hygroscopic nature of the resultant salt and an incomplete drying process. In the final part of this study we aim to confirm the link between the HSO4− peak and residual water on the sample and to illustrate, through the use of mixed acid solutions, that the increased chemical sensitivity, established in Figure 3, does not come at the cost of chemical discrimination, that is, that the spectra of different derivatives are still clearly distinguished at low concentration. Figure 4a shows a spectrum from a typical H2SO4 sample, derivatized with excess NH4OH and dried in the ambient laboratory environment (relative humidity is

Figure 4. (a) SERS spectra of 1 μL of 1000 ppm H2SO4 solution with excess NH4OH taken by drop-casting in ambient conditions and in a dry environment showing dominant 980 cm−1 SO42− peak. Note the presence of additional HSO4−, 1046 cm−1 peak in ambient spectrum (b) SERS spectra from mixed solution of 1 μL of 100 ppm H2SO4 and 1 μL of 100 ppm HNO3 with excess NH4OH under continuous laser exposure at 780 nm taken over a 20 min period.

typically >50%) and a spectrum from a sample under dry nitrogen flow conditions: note in this case that the 1046 cm−1 HSO4− peak is attenuated into the background. The reaction to form (NH4)2SO4 is thus not brought to completion under ambient laboratory conditions, the 1046 cm−1 peak indicating the presence of the HSO4− anion which we take to be associated with water molecules that have not departed the surface. The intensity of the HSO4− peak actually varies under ambient conditions, presumably depending on ambient humidity in the lab at the time and indeed, to some extent, on location within a particular sample. A similar attenuation effect on this peak can also be induced by continuous exposure to the stimulating laser beam over a period on the order of 20 min. In Figure 4b the SERS spectra arising from the respective ammonium salts derived from 2 μL of a premixed solution of 100 ppm of H2SO4 and 100 ppm of HNO3 drop-cast on the surface with 2 μL of NH4OH are shown. The three spectra, of 50 s acquisition time each, were taken over a 20 min interval during which the same part of the sample was continuously exposed to the 780 nm laser; the first (uppermost) spectrum was taken at the start of the sequence, the middle spectrum part way through while the lowermost spectrum was taken at the end of the sequence. The peak at 980 cm−1 yields unambiguous 9010

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

Article

evidence of the SO42− anion originating from the presence of trace H2SO4. The peak at ∼1050 cm−1 is interpreted as being primarily due to the NO3− anion in the form of NH4NO3 originating from the trace HNO3 solution (peaks at 1046 and 1050 cm−1 are not resolved in our system). In the main part of the figure the spectra are arbitrarily displaced for clarity of individual presentation while in the inset the raw spectra are simply overlaid, showing that while the SO42− peak remains almost constant in intensity there is a small decrease in the peak at 1050 cm−1 with increasing, integrated laser exposure time. This peak stabilizes after ∼20 min with no further noticeable decrease. We therefore suggest that a minor component of this peak arises from a HSO4−-water molecule complex which diminishes through drying by exposure to the laser beam. In the process there appears to be no net benefit to the SO42− peak, suggesting on the one hand that there is no further NH4OH to react with at this stage and also that the HSO4−-containing entity evaporates from the surface. To complete the present discussion of the SERS derivatization method we consider issues of applicability. First, in the Supporting Information, this technique is compared to other established techniques for the detection of trace ions in solution with reference to a range of criteria including sensitivity, minimum sample volume and sampling time, but also particularly portability and cost. The second issue is how broadly applicable the technique might be for a range of salt derivatives, especially those that may have a very high solubility product. NH4NO3 and (NH4)2SO4 have solubility products of 219 and 611 mol L−1 respectively, that is, are relatively soluble in water, but nonetheless precipitate readily onto the nanostructured substrates used here, favoring detection by SERS. Indeed, since the water evaporates away, a solid precipitate should form on the SERS substrate irrespective of the solubility of the parent ions in water (and, of course, notwithstanding that the solubility of the forming nanoscale precipitate product will have a size-dependent solubility, arising from a large molar surface area, that is greater than the thermodynamic bulk values quoted above.) In the context of the detection of low concentration acid-forming gases (NO2, SO2), the more limiting factor is actually the reduced temperature solubility of the target molecules in the water film condensing on a cooled SERS substrate. This issue is well documented in previous work on low-temperature gas detection based on a variety of semiconductor gas sensor materials.32,33

mist is sprayed onto a substrate for the detection of dinitrotoluene by Raman spectroscopy to indicate the presence of explosive precursors,35,36 this general goal in fact being the context of the work presented here. The principal point of contrast is that in a diffusion-controlled, reactive absorption the drying of the substrate leads to a significant decay in the Raman signal. In large part this would appear to be due to crystallization of the NaOH interfering with the detection process, whereas in the derivatization method it is the crystallization of the (ammonium) salt that drives the improvement in the SERS detection limit. Finally, it seems rather curious that a simple derivatization methodology has not been methodically investigated or routinely exploited in the context of SERS, especially given that a bulk form of the technique is well established in the context of aerosol spectroscopy.37 To summarize, we have demonstrated an improvement of ∼104 in the Raman spectroscopy detection limit for low concentration acids by chemical derivatization on a SERS substrate over the case of detection in solution on the same substrate; this detection threshold limit corresponds to ∼100 ppb by volume. Shown only for the case of bench acids and ammonium salts here, this technique should apply to any lowconcentration liquid contaminant which has a derivatization reaction leading to a solid precipitate in intimate proximity to the surface, making this method applicable to a very broad range of systems. Chemical discrimination in the detection of nitric acid and sulfuric acid in a low-concentration, mixed solution was also clearly demonstrated. In addition to allowing for detection of very low concentrations, this technique may offer an additional benefit in that the derivatization stage can act not only to facilitate SERS enhancement, but also to further discriminate the desired analyte from other contaminants. These features, along with the relative simplicity and compatibility with small volumes and existing equipment, make this method an attractive alternative to other lowconcentration liquid detection techniques.

■

ASSOCIATED CONTENT

S Supporting Information *

A comparison of the derivatized SERS method relative to a selection of established techniques for the detection of trace ions in solution with reference to various parameters including sensitivity, minimum required sample volume, sampling time, and equipment cost and portability. This material is available free of charge via the Internet at http://pubs.acs.org/

■

CONCLUSIONS In conclusion, we draw clear distinction between the derivatization method described here and other physical and chemical means of optimally exploiting the enhancement of nanostructured SERS substrates, specifically the use of (super)hydrophobicity34 and of reactive absorption of the analyte species to the substrate surface,35,36 essentially a variant of direct chemisorption.21,22 The hydrophobic nature of certain SERS substrates can be exploited to concentrate the target species onto a very small area of the surface from a very low concentration in solution;34 however, this entails no change in the chemical form of the entity to be detected, in contrast to the present study; in fact, it would seem that the use of hydrophobicity and derivatization could be combined in a complementary fashion. The use of an intermediate chemical species to tether the target analyte to the surface has been exploited and is well exemplified by a scheme in which NaOH

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

M.H., M.D.D., I.K., and K.M. conducted the measurements on the basis of experimental concept by G.M. and experiment and apparatus design by T.M., K.M., and G.M. P.D. and M.H. performed further experimental development and analyzed the data. P.D. wrote the bulk of the manuscript with input from all others, particularly M.H. and M.D.D. Notes

The authors declare no competing financial interest. 9011

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012

Analytical Chemistry

■

Article

(28) Knopf, D. A.; Luo, B. P.; Krieger, U. K.; Koop, T. J. Phys. Chem. A 2003, 107, 4322. (29) Querry, M. R.; Waring, R. C.; Holland, W. E.; Earls, L. M.; Herman, L. M.; Nijm, W. P.; Hale, G. M. J. Opt. Soc. Am. 1974, 64, 39. (30) Cox, R. A.; Haldna, U. L.; Idler, K. L.; Yates, K. Can. J. Chem. 1981, 59, 2591. (31) Dawson, B. S. W.; Irish, D. E.; Toogood, G. E. J. Phys. Chem. 1986, 90, 334. (32) Helwig, A.; Müller, G.; Garrido, J. A.; Eickhoff, M. Sens. Actuators, B 2008, 133, 156. (33) Helwig, A.; Müller, G.; Sberveglieri, G.; Eickhoff, M. J. Sensors 2009, 2009, 620720. (34) De Angelis, F.; Gentile, F.; Mecarini, F.; Das, G.; Moretti, M.; Candeloro, P.; Coluccio, M. L.; Cojoc, G.; Accardo, A.; Liberale C, C.; Zaccaria, R. P.; Perozziello, G.; Tirinato, L.; Toma, A.; Cuda, G.; Cingolani, R.; Di Fabriz, E. Nat. Photonics 2011, 5, 682. (35) Wang, J.; Yang, L.; Boriskina, S.; Yan, B.; Reinhard, M. Anal. Chem. 2011, 83, 2243. (36) Sylvia, J. M.; Janni, J. A.; Klein, J. D.; Spencer, K. M. Anal. Chem. 2000, 72, 5834−5840. (37) Signorell, R.; Reid, J. P. Fundamentals and Applications in Aerosol Spectroscopy; CRC Press (Taylor-Francis Group): Boca Raton, FL, 2011.

ACKNOWLEDGMENTS This work has been supported by EC Project BONAS under the 7th Framework Programme of the European Commission, Contract No. 261685. The authors thank Dan Marlow of the School of Mathematics & Physics, QUB, for his support and assistance with the necessary chemical preparations. We also thank Taifur Rahman, School of Chemistry & Chemical Engineering, and Josh Einsle, School of Mathematics & Physics, both QUB, for valued conversations on the work and literature guidance. M.H. gratefully acknowledges the provision of an EPSRC studentship from the QUB Doctoral Training Pool.

■

REFERENCES

(1) Kelly, K. L.; Coronado, E.; Zhao, L. L.; Schatz, G. C. J. Phys. Chem. B 2003, 107, 668. (2) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 8410. (3) Lal, S.; Link, S.; Halas, N. J. Nat. Photonics 2007, 1, 641. (4) Schuller, J. A.; Barnard, E. S.; Cai, Y. C.; Jun, Y. C.; White, J. S.; Brongersma, M. L. Nat. Mater. 2010, 9, 193. (5) Willets, K. A.; Van Duyne, R. P. Annu. Rev. Phys. Chem. 2007, 58, 267. (6) Le Ru, E. C.; Etchgoin, P. G. Principles of Surface-Enhanced Raman Spectroscopy; Elsevier: Amsterdam, 2009. (7) Stiles, P. L.; Dieringer, J. A.; Shah, N. C.; Van Duyne, R. P. Annu. Rev. Anal. Chem. 2008, 1, 601. (8) Choo-Smith, L.-P.; Edwards, H. G. M.; Endtz, H. P.; Kros, J. M.; Heule, F.; Barr, H.; Robinson, J. S., Jr.; Bruining, H. A.; Puppels, G. J. Biopolymers 2002, 67 (1), 1. (9) Vankeirsbilck, T.; Vercauteren, A.; Baeyens, W.; Van der Weken, G.; Verpoort, F.; Vergote, G.; Remon, J. P. Trends Anal. Chem. 2002, 21 (12), 869. (10) Fini, G. J. Raman Spectrosc. 2004, 35 (5), 335. (11) Eliasson, C.; Macleod, N. A.; Matousek, P. Anal. Chem. 2007, 79 (21), 8185. (12) Izake, E. L. Forensic Sci. Int. 2010, 202 (1), 1. (13) Le Ru, E.; Blackie, E.; Meyer, M.; Etchegoin, P. G. J. Phys. Chem. C 2007, 111, 13794. (14) Etchegoin, P. G.; Le Ru, E. C. Phys. Chem. Chem. Phys. 2008, 10, 6079. (15) Fang, Y.; Seong, N.-H.; Dlott, D. D. Science 2008, 321 (5887), 388. (16) Dawson, P.; Duenas, J. A.; Boyle, M. G.; Doherty, M. D.; Bell, S. E. J.; Kern, A. M.; Martin, O. F. J.; The, A.-S.; Teo, K. B. K.; Milne, W. I. Nano Lett. 2011, 11, 365−371. (17) Doherty, M. D.; Murphy, A.; Pollard, R. J.; Dawson, P. Phys. Rev. X 2013, 3, No. 011001. (18) Doherty, M. D.; Murphy, A.; McPhillips, J.; Pollard, R. J.; Dawson, P. J. Phys. Chem. C 2010, 114, 19913. (19) Ye, Q.; Fang, J.; Sun, L. J. Phys. Chem. B 1997, 101, 8221. (20) Camden, J. P.; Dieringer, J. A.; Zhao, J.; Van Duyne, R. P. Acc. Chem. Res. 2008, 41 (12), 1653. (21) McFarland, A. D.; Young, M. A.; Dieringer, J. A.; Van Duyne, R. P. J. Phys. Chem. B 2005, 109, 11279. (22) Abdelsalam, M. E.; Bartlett, P. N.; Baumberg, J. J.; Cintra, S.; Kelf, T. A.; Russell, A. E. Electrochem. Commun. 2005, 7 (7), 740. (23) Li, J. F.; Huang, Y. F.; Ding, Y.; Yang, Z. L.; Li, S. B.; Zhou, X. S.; Fan, F. R.; Zhang, W.; Zhou, Z. Y.; Wu, D. Y.; Ren, B.; Wang, Z. L.; Tian, Z. Q. Nature 2010, 464, 392. (24) Sharma, B.; Frontiera, R. R.; Henry, A.-I.; Ringe, E.; Van Duyne, R. P. Mater. Today 2012, 15 (1−2), 16. (25) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. Chem. Phys. Lett. 1974, 26 (2), 163. (26) Aksenenko, V. M.; Murav’ev, N. S.; Taranenko, G. S. Zh. Prikl. Spektrosk. 1986, 44 (1), 87. (27) Ruas, R.; Pochon, P.; Simonin, J.-P.; Moisy, P. Dalton Trans. 2010, 39, 10148. 9012

dx.doi.org/10.1021/ac5014095 | Anal. Chem. 2014, 86, 9006−9012