Utilizing Molecular Hyperpolarizability for Trace Analysis: A Surface-Enhanced Hyper-Raman Scattering Study of Uranyl Ion.

Surface-enhanced hyper-Raman scattering (SEHRS), the nonlinear analog of surface-enhanced Raman scattering (SERS), provides unique spectral signatures arising from the molecular hyperpolarizability. In this work, we explore the differences between SERS and SEHRS spectra obtained from surface-bound uranyl ion. Exploiting the distinctive SEHRS bands for trace detection of the uranyl ion, we obtain excellent sensitivity (limit of detection = 90 ppb) despite the extreme weakness of the hyper-Raman effect. We observe that binding the uranyl ion to the carboxylate group of 4-mercaptobenzoic acid (4-MBA) leads to significant changes in the SEHRS spectrum, whereas the surface-enhanced Raman scattering (SERS) spectrum of the same complex is little changed. The SERS and SEHRS spectra are also examined as a function of both substituent position, using 2-MBA, 3-MBA, and 4-MBA, and the carbon chain length, using 4-mercaptophenylacetic acid and 4-mercaptophenylpropionic acid. These results illustrate that the unique features of SEHRS can yield more information than SERS in certain cases and represent the first application of SEHRS for trace analysis of nonresonant molecules.

Introduction

Surface-enhanced hyper-Raman scattering (SEHRS), the nonlinear analog of surface-enhanced Raman scattering (SERS), has primarily been used to probe the physical properties of molecules. SEHRS can, for example, reveal the mode-specific enhancements from resonance with different electronic states,[1−3] the molecular orientation of molecules on metal surfaces,[4] and identify local chemical effects, which do not appear in SERS data.[5] Its usefulness as an analytical technique, however, has been limited by the weakness of the hyper-Raman scattered light relative to the Raman scattered light.[6−8] SEHRS-based studies, therefore, mostly utilize highly polarizable dye molecules, such as rhodamine derivatives[2,4] and triphenylmethane dyes,[3,9] as the combination of large hyperpolarizability and resonance enhancement yields large signal, even allowing for single-molecule detection.[9] It is worth noting that although hyper-Raman scattering is weaker than Raman scattering, surface enhancement can significantly shrink the gap between the two because the SEHRS enhancement can be several orders of magnitude greater than the SERS enhancement.[6,7]

Several recent studies have explored applications of SEHRS, such as imaging[10,11] and measurement of biologically relevant small molecules.[12,13] Aqueous pH measurement[14] and pH measurement of cellular environments[15] using scattering signal from the protonated and deprotonated forms of carboxylate-functionalized nanoparticles have also been reported. The SEHRS signal, excitingly, shows sensitivity to a larger pH range than the similarly constructed SERS probes. The measurement of pH, however, relies on signal arising from a large number of species on the nanoparticle surface, and does not require a trace detection scheme.

Extending SEHRS to trace analyte detection (sub-ppm) would add to the already existing advantages, such as large surface enhancement, use of longer wavelengths for deeper sample penetration while still yielding signal at easily detectable wavelengths,[16] and complementary information available with respect to SERS. Herein, we exploit the unique and complementary aspects of SEHRS spectra to demonstrate trace (90 ppb) detection of uranyl (UO22+) in aqueous solutions. Furthermore, the high symmetry of the uranyl ion yields fundamental insights into the differences between the SERS and SEHRS spectra of identical analytes.

Detection of uranium is also interesting in its own right, due to the importance of maintaining safe drinking water and to identify potentially trafficked radioactive material.[17] Recently, concerns of increased mobility of the uranyl ion, due to liberal application of nitrate rich fertilizers,[18] have increased fears of uranium-contaminated drinking water. A variety of SERS-based schemes have been developed that address these concerns.[19−24] SEHRS-based methods, however, have not yet been explored. In this work, we show that SEHRS-based detection schemes are viable and further use this model system to explore the differences between SERS and SEHRS.

Results and Discussion

Figure shows the SEHRS and SERS spectra that result upon addition of uranyl nitrate solution to 4-mercaptobenzoic acid (4-MBA)-functionalized silver particles. The presence of uranyl nitrate dramatically changes the relative intensities of many vibrational bands in the SEHRS spectrum, whereas the SERS spectra derived from the exact same sample demonstrate almost no change. Interestingly, three new bands appear in the SEHRS spectra between 800 and 900 cm–1: 830, 855, and 900 cm–1, which can be seen in an expanded view in Figure S1. The 830 and 900 cm–1 bands are attributed to the symmetric (νs) and asymmetric (νas) stretching of the uranyl ion, respectively. The uranyl modes in the SEHRS (νas, 900 cm–1) and SERS (νs, 830 cm–1) spectra are shifted lower by approximately 30 cm–1 with respect to the Raman and IR positions obtained from solid uranyl nitrate hexahydrate,[25] indicating the binding of the uranyl ion by the 4-MBA ligand.[23]

Figure 1

Left: SEHRS spectra of 4-MBA-functionalized silver nanoparticles before (red) and after (black) addition of uranyl nitrate solution. Right: SERS spectra of 4-MBA-functionalized silver nanoparticles before (red) and after (black) addition of uranyl nitrate solution. All spectra were background subtracted and normalized by dividing by the maximum intensity. The SERS spectrum of the particles has minor changes in the presence of uranyl, whereas the SEHRS spectrum changes dramatically after addition of uranyl.

For molecules with a center of inversion, e.g., linear UO22+, the activities of its vibrational modes are mutually exclusive between Raman and hyper-Raman scattering.[26] If the uranyl ion retains D∞ symmetry, the symmetric stretch is Raman active and hyper-Raman inactive, whereas the asymmetric stretch is Raman inactive and hyper-Raman active. The symmetric stretch (830 cm–1), however, appears in both the SERS and SEHRS spectra, indicating that the symmetry of the uranyl ion must be altered upon binding.

The third band (855 cm–1) must either correspond to a shifted nitrate or 4-MBA carboxylate mode. SEHRS spectra of 4-MBA with added uranyl acetate were therefore measured to determine the 855 cm–1 signal origin. Appearance of the same modes in the SEHRS spectrum of 4-MBA particles with uranyl acetate and additional rinsing steps ensures that the 855 cm–1 is indeed a shifted 4-MBA carboxylate mode (see Supporting Information and Figure S2 for more detail).[27] The shift of the carboxylate mode, from 840 to 855 cm–1, is also observed in the SERS spectrum in the presence of uranyl (Figure , right). It is interesting that the 4-MBA mode at 855 cm–1 becomes SEHRS active upon uranyl binding, whereas there is little change in activity in the SERS spectrum (Figure ). The four 4-MBA bands between 1100 and 1500 cm–1 in the SEHRS spectrum have previously been observed in low pH solutions of 4-MBA-coated nanoparticles.[15] The presence of these bands in low pH solution indicates the protonated form of the carboxylate. It is not surprising that the same bands appear in the presence of uranyl, where a single carboxylate oxygen is likely bound to a uranyl ion instead of a proton, similar to the reported crystal structure of uranyl bound to a single carboxylate oxygen of isonicotinic acid.[28] As a control (Figure ), SEHRS spectra of 4-MBA-functionalized particles were recorded at low pH (pH = 1) and no signal was obtained between 800 and 900 cm–1 where our analysis is done for the presence of the uranyl ion. Furthermore, no SEHRS signal is observed between 800 and 900 cm–1 when the 4-MBA particles are exposed to uranyl, indicating that at low pH the competition between the proton and the uranyl favors the protonated form of the carboxylate (Figure ). As expected, the basic (pH = 12) and neutral pH 4-MBA SEHRS spectra are identical to the neutral pH spectrum, as both neutral and basic pHs are higher than the pKa of the carboxylate group (Figure ). The formation of more complex uranyl species at higher pH inhibits uranyl binding resulting in the black spectrum in Figure and is consistent with previous reports.[29] Integration under the curves in the 800–900 and 1050–1100 cm–1 regions are tabulated as a function of uranyl concentration, and a calibration curve is constructed (Figure ). This peak-area method was chosen as the ratio of the signal to a reference band accounts for fluctuations in enhancement, commonly seen in SERS detection schemes.[30,31] These results indicate that SEHRS can deliver quantitative detection of uranyl with a limit of detection (LOD) of 90 ppb. Here, we calculate the LOD from 3σ/m, where σ is the standard deviation obtained from three replicate measurements of five independently prepared samples each and m is the slope of the calibration curve.

Figure 2

Left: SERS spectra of 4-mercaptobenzoic acid (4-MBA)-functionalized silver nanoparticles at acidic (top) and basic (bottom) pH, both with (blue) and without (red) added uranyl. Right: SEHRS spectra of 4-MBA-functionalized particles in acidic (top) and basic (bottom) pH and with (black) and without (green) added uranyl ion. At both low and high pH values (pH 1, 12 respectively), presence of uranyl ion in solution does not change the resulting spectra.

Figure 3

Left: representative SEHRS spectra of 4-MBA (red) and 4-MBA in the presence of uranyl nitrate at a concentration of 400 ppb (black) and 800 ppb (green). Right: integration of the region from 800 to 900 cm–1 relative to the integrated area of the 1100 cm–1. 4-MBA mode as a function of concentration yields a calibration curve (red trace) with a limit of detection of 90 ppb.

Determining what characteristics of ligands lead to the largest change in the SEHRS spectrum could lead to the design of high-performance ligands for SEHRS-based detection. We specifically studied the effect of the benzene ring on the enhanced 4-MBA carboxylate band (855 cm–1) by changing the number of carbons separating the carboxylate from the phenyl ring. Adding uranyl to a suspension of 4-mercaptophenylacetic acid (4-MPAA)-functionalized particles resulted in SEHRS spectrum similar to that observed with 4-MBA (Figure ) although the intensity of the signal is decreased by 40% relative to that of 4-MBA. Identical experiments were done using 4-mercaptophenylpropionic acid (4-MPPA); the resulting spectra surprisingly show no response in the 800–900 cm–1 in the presence of uranyl ion (Figure ). These experiments demonstrate that separating the carboxylate from the phenyl ring by a single carbon greatly decreases the observed signal.

Figure 4

Surface-enhanced hyper-Raman spectra of 4-mercaptobenzoic acid (4-MBA) particles (top), 4-mercaptophenylacetic acid (4-MPAA) particles (middle), and 4-mercaptophenylpropionic acid (4-MPPA) particles (bottom), before and after addition of uranyl nitrate solution (red and black traces, respectively). It is clear that at constant concentration (800 ppb), the signals from the uranyl ion and carboxylate group (800–900 cm–1) dramatically decrease with additional carbons separating the carboxylate from the benzene ring.

The location of the carboxylate on the phenyl ring may also be a factor in the enhancement of the 4-MBA band. Therefore, additional experiments were performed using 2-mercaptobenzoic acid (2-MBA) and 3-mercaptobenzoic acid (3-MBA) in the place of the 4-MBA. Interestingly, a comparison of the SERS spectra of 3-MBA and 4-MBA, both without and with added uranyl nitrate, shows that uranyl can be detected with a greater spectral change using 3-MBA (Figure ). This is consistent with our previous work suggesting that changes in the 3 position of the benzene ring yield more dramatic changes in the SERS spectra than changes in the 4 position.[32] The differences between the 3-MBA and 4-MBA SEHRS spectra indicate that hyper-Raman may detect changes in the 4 position with more sensitivity than changes in the 3 position (Figure ). No change is observed upon addition of uranyl to 2-MBA functionalized particles (Figure S3). This is likely due to orientation effects or less efficient packing on the surface arising from steric interaction of the carboxylate groups leading to significantly weaker signal.

Figure 5

Left: SERS spectra of 4-MBA (top) and 3-MBA (bottom) before and after addition of uranyl nitrate solution (red and black traces, respectively). Right: SEHRS spectra of 4-MBA (top) and 3-MBA (bottom) before and after addition of uranyl nitrate solution (red and black traces, respectively). These spectra indicate that different parameters must be optimized for maximum sensitivity in SEHRS-based schemes than those of SERS-based schemes.

Conclusions

We demonstrate the first trace detection assay based on surface-enhanced hyper-Raman scattering (SEHRS). Our model system using 4-mercaptobenzoic acid-functionalized silver nanoparticles provides a LOD of 90 ppb for uranyl and suggests future viability of SEHRS-based assays. We additionally present data demonstrating that SEHRS can be more sensitive to the local structure and chemical environment than SERS obtained with identical conditions, especially when the carboxylate is present in the para position of the benzene ring. This sensitivity combined with the benefits of nonlinear spectroscopies, such as greater sample penetration and less interference from fluorescence, suggests that SEHRS has great potential for further application in analytical sciences.

Experimental Section

Silver nitrate, sodium citrate tribasic, uranyl nitrate, uranyl acetate, sodium nitrate, hydrochloric acid, nitric acid, sodium hydroxide, sodium bromide, 4-mercaptobenzoic acid (4-MBA), 3-mercaptobenzoic acid (3-MBA), 2-mercaptobenzoic acid (2-MBA), and 4-mercaptophenylacetic acid (4-MPAA) were purchased from Sigma. 4-Mercaptophenylpropionic acid (4-MPPA) was purchased from Santa Cruz Biotechnology. All reagents were used as received without further purification. Silver colloids were prepared using the Lee and Meisel method.[33] Specifically, 91.8 mg of silver nitrate was added to 200 mL of water and brought to boiling. Tribasic sodium citrate (115 mg) was added to reduce the silver, forming spherical silver nanoparticles approximately 70 nm in diameter, verified by dynamic light scattering.

4-MBA-functionalized particles were prepared by adding 15.6 μL of 1.6 mM 4-MBA in methanol solution to 5 mL of prepared silver colloids. The resulting suspension was then stirred by vortex mixer. Sodium bromide (1 mL, 1 M) was added causing the particles to aggregate. The aggregated particles were allowed to settle, and the supernatant was removed to eliminate unwanted complexation of the uranyl by citrate in solution. 4-MBA-functionalized particles were then resuspended in water. Varying volumes of a 21 ppm uranyl nitrate stock solution were then added to the resuspended particles, resulting in final uranyl concentrations between 200 and 1000 ppb, which were then aggregated using 1 M NaBr. Samples were then ready for analysis.

SERS data were acquired using a 633 nm HeNe laser (Thorlabs) aligned into an inverted microscope system (Nikon Ti-U). The beam, typically 50 μW at the objective, was then focused onto the sample using a 20× objective (Nikon, NA = 0.5). Backscattered light was then collected through the same objective, passed through a Rayleigh rejection filter (Semrock), and analyzed using a dispersive imaging spectrometer (Princeton Instruments, Acton SP2300, 1200 g mm–1) operated using Winspec software (Princeton Instruments).

SEHRS spectra were acquired using an optical parametric oscillator light source (picoEmerald, Applied Physics & Electronics) with an idler wavelength of 1266 nm. The beam was pulsed (approximately 5 ps pulse width) at 80 MHz. The beam was then aligned into the same inverted microscope system, analyzed in a dispersive imaging spectrometer (Princeton Instruments, SP2300, 1200 g mm–1), and detected using a back-illuminated deep-depletion CCD (PIXIS, Spec-10, Princeton Instruments). SEHRS spectra were obtained with an average laser power measured at the objective of approximately 2 mW and acquisitions of 2 min. All presented spectra were background corrected and normalized to the maximum intensity.