Anal. Chem. 2006, 78, 445-451
Pushing the Limits of Mercury Sensors with Gold Nanorods Matthew Rex, Florencio E. Hernandez, and Andres D. Campiglia*
Department of Chemistry, P.O. Box 25000, University of Central Florida, Orlando, Florida 32816-2366
The method presented here provides a direct way to determine mercury in tap water samples at the parts-pertrillion level. Its outstanding selectivity and sensitivity results from the well-known amalgamation process that occurs between mercury and gold. The entire procedure takes less than 10 min. No sample separation or sample preconcentration is required. The only step prior to mercury determination consists of mixing the water sample with a gold nanorod solution in sodium borohydride. The analytical figures of merit demonstrate precise and accurate analysis at the parts-per-trillion level. The limit of detection (6.6 × 10-13 g‚L-1) shows excellent potential for monitoring ultralow levels of mercury in water samples. Mercury is a known environmental pollutant routinely released from power plants burning fossil fuels. According to the Environmental Protection Agency (EPA), coal-burning power plants are the largest human-caused source of Hg emissions to the air in the United States.1 Once released into air, Hg eventually settles into water or onto land where it is washed. Because human exposure to high Hg levels can harm the brain, heart, kidneys, lungs, and immune system of people of all ages, it is important to monitor Hg levels in the aquatic ecosystem as a potential source of contamination. Several methods exist to monitor concentration levels of Hg in water samples. Established techniques, such as atomic absorption spectroscopy,2,3 gas chromatography-inductively coupled plasma-mass spectrometry,4 atomic fluorescence spectrometry (AFS),5,6 inductively coupled plasma-atomic emission spectrometry,7,8 and reversed-phase high-performance liquid chromatography,9 provide limits of detection at the parts-per-billion level. * To whom correspondence should be addressed. Tel.: 407-823-3289. Fax: 407-231-8831. E-mail:
[email protected]. (1) http://www.epa.gov/mercury/about.htm. (2) Vil′pan, Y. A.; Grinshtein, I. L.; Akatove, A. A.; Gucer, S. J. Anal. Chem. 2005, 60, 45-51. (3) Kopysc, E.; Pyrzynska, K.; Garbos, S.; Bulska, E. Anal. Sci. 2000, 16, 13091312. (4) Karunasagar, D.; Arunachalam, J.; Gangadharan, S. J. Anal. At. Spectrom. 1998, 13, 679-682. (5) Yu, L.; Yan, X. At. Spectrom. 2004, 25, 145-153. (6) Bloxham, M. J.; Hill, S. J.; Worsfold, P. J. J. Anal. At. Spectrom. 1996, 11, 511-514. (7) Trimble, C. A.; Hoenstine, R. W.; Highley, A. B.; Donoghue, J. F.;Ragland, P. C. Mar. Georesour. Geotechnol. 1999, 17, 187-197. (8) Smigelski, T.; O′Brien, K.; Fusco, J.; Schaumloffel, J. C.; Tausta, J. 226th ACS National Meeting 2003, September 7-11. 10.1021/ac051166r CCC: $33.50 Published on Web 12/15/2005
© 2006 American Chemical Society
Their excellent performance, however, is achieved at the expenses of elaborate and time-consuming sample preparation and preconcentration procedures. The EPA method 1631 is a classical example.10 Prior to Hg detection by cold vapor AFS, Hg in the sampleswhich includes, but is not limited to, Hg(II), Hg(0), strongly organocomplexed Hg(II) compounds, adsorbed particulate Hg, and several covalently bound organomercurials such as CH3HgCl, (CH3)2Hg, and C6H5HgOOCCH3sis oxidized to Hg(II) and then reduced to volatile Hg(0). The vapor is then carried into the atomic fluorescence spectrometer via purge and trap with two high-surface area gold “traps” (usually Au-coated sand). As a tentative means of reducing analysis time and cost, onsite sensing approaches capable of providing real-time Hg determination have been actively pursued. These include optical test strips,11 remote electrochemical sensors,12 ion-selective electrodes,13 fluorescence-based sensor membranes,14,15 and piezoelectric quartz crystals.16 Although these approaches provide low detection limits and fast response times, they still lack the procedural simplicity for on-site analysis. Our approach takes advantage of the strong affinity between Au and Hg. The interaction between these two elements has been extensively studied by monitoring physical properties of Au upon Hg adsorption. Numerous gravimetric sensors exist based on the mass change when Hg adsorbs onto a gold-coated piezoelectric substrate.17-22 Au films (∼10-40-nm thickness) have also been employed to monitor Hg upon changes in resistivity,23-25 reflectivity,26-28 and surface plasmon resonance (SPR).29 We (9) Yin, X.; Xu, Q.; Xu, X. Fenxi Huaxue 1995, 23, 1168-1171. (10) http://www.epa.gov/waterscience/methods/1631.html. (11) Capitan-Vallvey, L. F.; Cano Raya, C.; Lopez Lopez, E.; FernandezRamos, M. D. Anal. Chim. Acta 2004, 524, 365-372. (12) Wang, J.; Tian, B.; Lu, J.; Luo, D.; MacDonald, D. Electroanalysis 1998, 10, 399-402. (13) Vlasov, Y. G.; Ermolenko, Y. E.; Kolodnikov, V. V.; Ipatov, A. V.; Al-Marok, S. Sens. Actuators, B 1995, 24-25, 317-319. (14) Murkovic, I.; Wolfbeis, O. S. Sens. Actuators, B 1997, 38-39, 246-251. (15) Chan, W. H.; Yang, R. H.; Wang, K. M. Anal. Chim. Acta 2001, 444, 261269. (16) Palenzuela, B.; Manganiello, L.; Riso, A.; Valcarcel, M. Anal. Chim Acta 2004, 511, 289-294. (17) Manganiello, L.; Rios, A.; Valcarcel, M. Anal. Chem. 2002, 74, 921-925. (18) Ruys, D. P.; Andrade, J. F.; Guimaraes, O. M. Anal. Chim. Acta 2000, 404, 95-100. (19) Yao, S.; Tan, S.; Nie, L. Fenxi Huaxue 1986, 14, 729-734. (20) Casilli, S.; Malitesta, C.; Conoci, S.; Petralia, S.; Sortino, S.; Valli, L. Biosens. Bioelectron. 2004, 20, 1190-1195. (21) Rogers, B.; Bauer, C. A.; Adams, J. D. Micro-electro-mech. Syst. 2003, 5, 663-666. (22) Gomes, M.; Teresa, S. R.; Morgado, E. V.; Oliveira, Joao A. B. P. Anal. Lett. 1999, 32, 2715-2723.
Analytical Chemistry, Vol. 78, No. 2, January 15, 2006 445
determine Hg upon wavelength changes in absorption spectra of Au nanorods. To the extent of our literature search, only a few reports exist correlating this optical property to Hg-Au interactions.30-32 Herein, we demonstrate the analytical potential of Au nanorods for monitoring Hg in water samples. The outstanding selectivity and sensitivity of the method provide a unique way to determine Hg in water samples without previous separation or preconcentration of the original sample. EXPERIMENTAL SECTION Chemicals. Analytical-reagent grade chemicals were used in all experiments. Acetone, tetradodecylammonium bromide, cyclohexane, hexadecyltrimethylammonium bromide (CTAB), hydrogen tetrachloroaurate(III), silver nitrate, and ammonium fluoride were obtained from Sigma-Aldrich. Mercury(II) chloride, sodium borohydride, sodium nitrate, lead(II) nitrate, barium acetate, copper sulfate, arsenic pentoxide, chromic nitrate, and sodium chloride were purchased from Fisher chemicals. All water used was obtained from a Barnstead Infinity ultrapure water system. Synthesis of Gold Nanorods. Nanorods were synthesized according to the photochemical method developed by Kim et al.33 Briefly, 3 mL of a 0.08 M CTAB and 0.42 mg/mL aqueous solution of tetradodecylammonium bromide were mixed with 0.25 mL of a 0.024 M of chloroauric acid trihydrated solution. Then 0.065 mL of acetone and 0.045 mL of cyclohexane were added to the mixture followed by 31.5 µL of a 0.01 M AgNO3 aqueous solution. The final mixture was then irradiated for 24 h with a 254-nm UV light (420 mW/cm2). The appearance of a blue solution indicated nanorod formation. Instrumentation. Absorbance measurements were carried out with a single-beam spectrophotometer (model Cary 50, Varian) equipped with a 75-W pulsed xenon lamp, 2-nm fixed band-pass, and 24 000 nm‚min-1 maximum scan rate. Instrumental performance was monitored daily with a commercial standard purchased from Photon Technology International. The standard consists of a single crystal of dysprosium-activated yttrium aluminum garnet mounted in a cuvette-sized holder with well-characterized quasiline absorption spectrum. Wavelength accuracy was monitored by comparing the position of several atomic lines to the maximum wavelengths provided by the manufacturer. Table 1 summarizes the typical precision of measurements obtained in our laboratory within 2 h of instrumental use. Our results confirm the performance of our spectrometer according to specifications. More (23) Skreblin, M.; Byrne, A. R. Vestnik Slovenskega Kemijiskega Drustva 1991, 38, 521-536. (24) Mazzolai, B.; Mattoli, V.; Raffa, V.; Tripoli, G.; Accoto, D.;Menciassi, A.; Dario, P. Sens. Microsyst. 2003, (Feb 12-14), 369-375. (25) George, M. A.; Glaunsinger, W. S. Thin Solid Films 1994, 245, 215-224. (26) Morris, T.; Szulczewski, G. Langmuir 2002, 18, 5823-5829. (27) DiMasi, E.; Tostmann, H.; Ocko, B. M.; Huber, P.; Shpyrko, O. G.;Pershan, P. S.; Deutsch, M.; Berman, L. E. Mater. Res. Soc. Symp. Proc. 2000, 590, 183-188. (28) Butler, M. A.; Ricco, A. J.; Baughman, R. J. J. Appl. Phys. 1990, 67, 43204326. (29) Chah, S.; Yi, J.; Zare, R. N. Sens. Actuators, B 2004, 99, 216-222. (30) Henglein, A.; Giersig, M. J. Phys. Chem. B 2000, 104, 5056-5060. (31) Morris, T.; Copeland, H.; McLinden E.; Wilson, S.; Szulczewiski, G. Langmuir 2002, 18, 7261-7264. (32) Morris, T.; Kloepper, K.; Wilson, S.; Szulczewiski, G. J. Colloid Interface Sci. 2002, 254, 49-55. (33) Kim, F.; Song, J. H.; Yang, P. J. Am. Chem. Soc. 2002, 124, 14316-14317.
446 Analytical Chemistry, Vol. 78, No. 2, January 15, 2006
Table 1. Instrumental Performance of Cary 50 Spectrophotometera wavelength (nm)
absorption intensity (r.u.)b
352.00 ( 0.00 366.00 ( 0.00 447.00 ( 0.00 752.00 ( 0.00
0.407 00 ( 0.001 00 0.334 00 ( 0.001 00 0.088 00 ( 0.001 00 0.101 78 ( 0.001 00
a Measurements were done with a commercial standard consisting of a single crystal of dysprosium-activated yttrium aluminum garnet mounted in a cuvette-sized holder. Reported values are the average of results extracted from 10 absorption spectra. br.u., relative units.
importantly, we demonstrate that its response does not contribute significantly to the precision of measurements of experimental results presented in this article. For transmission electron microscopy (TEM), a FEI Tecnai F30 microscope was used operating at an accelerating voltage of 300 kV. This electron microscope has a field emission electron source and a super twin objective lens to yield a point-to-point resolution of 2 Å. For compositional analysis with energydispersive X-ray spectroscopy (EDX), the sample was tilted 15° toward the detector. For TEM sample preparation, drops of solution were deposited and dried on a standard TEM copper grid with amorphous carbon film. RESULTS AND DISCUSSION The first article reporting on the absorbing properties of metal nanorods appeared a few years ago. On the basis of experimental evidence, El-Sayed and co-workers34 provided the theoretical foundation to understand the two absorption bands typically observed in the UV-visible absorption spectra of gold nanorods. According to the authors,34 the two absorption bands correspond to the transversal and longitudinal modes of SPR. The transversal mode band belongs to the SPR along the short axis of the rod and appears at a shorter absorption maximum than the longitudinal mode band. The maximum absorption wavelength of the longitudinal modeswhich corresponds to the SPR along the long axis of the rodspresents a linear correlation with the aspect ratio (length/diameter) of the nanorod. As the aspect of the nanorod increases, the longitudinal mode band shifts to longer wavelengths. The same behavior is observed as the dielectric constant of the medium increases.34 In this study, we use nanorods with an average aspect ratio of 1.6. The solid line in Figure 1A shows a typical absorption spectrum recorded from a nanorod suspension in pure water. Clearly, the absorption maximums of the transversal and longitudinal modes appear at 520 and 612 nm, respectively. Preliminary Studies. The first step of our approach consists of reducing all existing forms of oxidized Hg to Hg(0). This is accomplished by mixing the water sample with NaBH4. This strong reducing agent is also capable of reducing any oxidized form of Au into Au(0). The broken line in Figure 1A depicts the absorption profile of a nanorod suspension prepared in 1.67 × 10-3 M NaBH4. Comparison of the two spectra in Figure 1A shows a blue shift (∆λmax) in the absorption maximums of the transversal (∆λmax ) 6 nm) and the longitudinal (∆λmax ) 12.1 nm) modes. (34) Link, S.; Mohamed, M. B.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 3073-3077.
Figure 2. TEM showing the effect of 1.67 × 10-3M NaBH4 on Au nanorods. Adding reducing agent to the nanorods solution changes the aspect ratio from 1.8 to 1.6. Each picture is a representative image of the average size distribution of a nanorods solution.
Figure 3. UV-visible absorption spectra showing the spectral shift at several Hg(II) concentrations. The concentration range between 1.6 × 10-11 and 6.3 × 10-11 M shows the spectra within the linear dynamic range of the calibration curve. The remaining spectra show the overlapping between the longitudinal and transversal absorption bands at higher Hg(II) concentrations. Figure 1. (A) UV-visible absorption spectrum of gold nanorods in Nanopure water (solid line) and in 1.67 × 10-3 mol/L NaBH4 (dotted line). (B) Plot showing the wavelength shift of the longitudinal (9) and transversal (2) modes of gold nanorods with as a function of NaBH4 concentration.
Monitoring of the spectral shift as a function of NaBH4 concentration yielded the plot in Figure 1B. The wavelength shifts experienced by the two modes reach plateaus at different NaBH4 concentrations. The longitudinal mode reaches a maximum at 6.0 × 10-4 M whereas the transverse mode reaches a maximum at 1.5 × 10-3 M. In addition, the absorption maximum of the longitudinal mode band shifts ∼2× longer than the one corresponding to the transversal mode band. This is consistent with the well-known higher sensitivity the longitudinal mode band presents toward the chemical environment of the nanorod.34-36 The observed blue shift is attributed to the presence of unreacted Au3+ in the nanorod suspension prior to the addition of NaBH4. The reduction of Au3+ to Au(0), followed by its deposition on the surface of the nanorods, augments the size of the nanorods and decreases the density of CTAB molecules (35) Raether, H. Surface Plasmons on Smooth and Rough Surfaces and on Gratings; Springer-Verlag: New York, 1988. (36) Chang, J.; Wu, H.; Chen, H.; Ling, Y.; Tan, W. Chem. Commun. 2005, 8, 1092-1094.
covering their surface. Because the dielectric constant of the medium perceived by the nanorods is reduced, the maximum absorption bands of the two modes shift to shorter wavelengths. Experimental evidence of our assumption was obtained via TEM analysis (see Figure 2). The presence of NaBH4 increases the nanorods’ diameter and, therefore, reduces their aspect ratio from 1.8 to 1.6. Effect of Hg(II) in the Absorption Spectrum of Gold Nanorods. To guarantee the complete reduction of oxidized Hg eventually present in unknown water samples, all further studies were performed with an excess of reducing agent (1.67 × 10-3 M NaBH4). Monitoring of the absorption spectrum of gold nanorods over a period of 8 h showed no further wavelength shifts, providing a reliable spectrum for reference purposes. Figure 3 shows the typical spectral shifts observed in the presence of Hg(II). The water samples were prepared by adding incremental volumes (1.0, 10.0, and 20.0 µL) of an aqueous standard 10-9 M HgCl2 solution to a constant volume (600 µL) of a 1.64 × 10-5 M gold nanorod solution in 1.67 × 10-3 M NaBH4. The concentration of nanorods is estimated by assuming a 100% synthetic yield. All the spectra were recorded after 4 min of HgCl2 addition, i.e., the minimum time at which a plateau was reached, and no further spectral shifts Analytical Chemistry, Vol. 78, No. 2, January 15, 2006
447
Scheme 1. Chemical Reactions Involved in the Amalgamation of Mercury and Gold
were observed. The blue shift of the maximum absorption wavelength of the longitudinal band shows a direct correlation with Hg(II) concentration. The presence of Hg(II) at higher concentration ranges causes an overlap between the longitudinal and the transversal absorption bands of the nanorods. At [HgCl2] ) 1.57 × 10-4 M, only one absorption band is observed. This evidence suggests the conversion of Au nanorods into Au nanospheres. Scheme 1 presents the chemical reactions involved in the observed phenomenon. In the presence of liquid Hg, two probable mechanisms of action should be considered: surface covering of the nanorods by the liquid metal, amalgamation between Au and Hg, or both. Because Hg increases the effective dielectric constant of the medium perceived by the nanorods, the first possibility should result in the displacement of the longitudinal band toward longer wavelengths. In more chemical terms, the surface coating of the nanorods caused by the presence of Hg should dislocate the structure of the surrounding micelles that prevent the nanorods from aggregation and precipitation. As a result of Au dissolution, the second possibilitysi.e., amalgamationsshould cause a reduction of the effective aspect ratio of the nanorods and a blue shift of the maximum absorption wavelength of the longitudinal mode band. Our experiments strongly suggest the main mechanism of action as being through amalgamation. Based on the fact that the active sites of Au nanorods mainly belong to the tips of the nanostructures,36 amalgamation should take place more efficiently on the tips of the nanorods, and therefore, it should decrease their effective ratio. Preferential deposition at the tips of the nanorods is also favored by the presence of surfactant (CTAB) on the lateral sides of the nanorods. Their shielding effect restricts the amalgamation of Hg on the lateral walls of the nanorods. The fact that no further appreciable change is observed in the SPR of Au nanospheres beyond [HgCl2] ) 1.57 × 10-4 M is consistent with its lack of sensitivity to particle size changes between ∼3 and ∼60 nm.35 Figure 4A provides a schematic explanation to our observations. Experimental evidence of our hypothesis is shown in Figure 4B and C. TEM images show the shape and the aspect ratio of Au nanorods in the absence of Hg2+ and the presence of 1.25 × 10-5 and 1.57 × 10-4 M Hg2+ solutions. All pictures were taken in the presence of 1.67 × 10-3 M NaBH4. As the concentration of Hg2+ increases, the aspect ratio decreases to the point at which the shape of the nanoparticles becomes spherical. By comparing Figures 4B and 2 one can correlate the changes in aspect ratio and shape of Au nanorods to the changes observed in their absorption spectra. The blue shift of the longitudinal band observed in the absorption spectra in the presence of 1.25 × 10-5 M Hg2+ can be attributed to the change in aspect ratio. Similarly, the single absorption band observed in the presence of 1.57 × 10-4 M Hg2+ solution can be attributed to the spherical shape of the nanoparticles. Additional experimental evidence is provided 448
Analytical Chemistry, Vol. 78, No. 2, January 15, 2006
by EDX analysis. As expected, the Hg content in nanoparticles is inversely proportional to their aspect ratio. Spherical nanoparticles contain more Hg than nanorods. Analytical Figures of Merit. The limit of detection (LOD ) xL) of an analytical method can be calculated by xL ) xB + ksB, where xB is the main average of blank responses, sB is the standard deviation of blank measurements, and k is a numerical factor chosen in accordance to the desired confidence level.37 Therefore, the best LOD is usually obtained with the minimum blank signal value. Because our approach is based on absorption spectral shifts upon mercury interaction with gold nanorods, our LOD also depends on the reproducibility (sR) of the reference wavelength (λR), i.e., the maximum absorption wavelength of gold nanorods in the absence of Hg. In the lack of matrix interference, the main source of wavelength variation is instrumental noise. The extent to which instrumental noise deteriorates sR depends on the magnitude of the reference signal. Because the magnitude of the reference signal is directly proportional to the concentration of gold nanorods, one can adjust this parameter to optimize λR reproducibility. In doing so, one should also keep in mind that there is a direct correlation between gold nanorods and Hg concentration and that Hg traces are detected only with low nanorod concentrations. Careful investigation of these parameters led us to set a 1.64 × 10-5 M working concentration. Multiple runs (n ) 16) of this standard solution yielded an average maximum absorption peak at λR ) 599.72 ( 0.02 nm. Each λR was calculated by taking the first derivative of each absorption spectrum between 560 and 640 nm. Keeping in mind the average intensity of the reference signal (0.055 00 ( 0.002 r.u.), comparison of λR ( sR to values in Table 1 shows no significant contribution from instrumental noise. Table 2 summaries the analytical figures of merit obtained with a 1.64 × 10-5 M gold nanorod solution. The calibration curve was built by plotting the wavelength shift of the first absorption derivative as a function of Hg2+ molar concentration. The wavelength shifts plotted in the calibration graph are the averages of individual measurements taken from three aliquots of the same working solution. The linear dynamic range of the calibration curve is based on at least six mercury concentrations. The correlation coefficient and the slope of the log-log plot are close to unity, demonstrating a linear relationship between mercury concentration and wavelength shift. The relative standard deviation at medium concentrations was below 1%, which shows excellent reproducibility of measurements. It is important to note that the upper concentration of the linear dynamic range is set by the concentration of nanorods in solution. Higher concentrations expand the upper limit of the linear dynamic range but deteriorate the LOD. The LOD was calculated with the equation LOD ) 3sR/ m, where m is the slope of the calibration curve and sR ) (0.02 nm. The standard deviation was calculated based on 16 measurements of the reference signal (λR). Although a straightforward comparison with reported LOD is difficult because different instrumental setups and mathematical approaches have been used for their determination, we can safely state that our LOD (1.53 × 10-10 g‚L-1) is at least 1 and 3 orders of magnitude better than (37) Miller, J. N.; Miller, J. C. Statistics and Chemometrics for Analytical Chemistry, 4th ed.; Pearson Prentice Hall: New York, 2000.
Figure 4. (A) Schematic diagram showing the amalgamation of Hg with Au nanorods. (B) TEM and EDX analysis of Au nanorods in the absence and the presence of Hg. I ) no Hg; II ) 1.25 × 10-5 M and III ) 1.57 × 10-4 M Hg2+. All solutions were prepared in 1.67 × 10-3 mol/L NaBH4. (C) TEM analysis of multiple Au nanorods in the absence and the presence of Hg. Sol A, no Hg; Sol B, 1.25 × 10-5 M Hg; Sol C, 1.57 × 10-4 M Hg2+. All solutions were prepared in 1.67 × 10-3 mol/L NaBH4. Table 2. Analytical Figures of Merit for Mercury Determination with Gold Nanorodsa linear dynamic rangeb(g/L) limit of detectionc(g/L) correlation coefficientd relative standard deviatione(%)
1.98 × 10-12-3.11 × 10-8 6.6 × 10-13 0.9998 0.33
a Analytical figures of merit were obtained adding small aliquots of a 1 × 10-9 mol/L Hg(II) solution to a 1.64 × 10-5 M gold nanorods solution in 3.34 × 10-3 M NaBH4. Each intensity value plotted in the calibration graph is the average of six replicate measurements. b Linear dynamic range. Lower concentration limit was calculated as 3 × LOD. c Limit of detection was calculated using 3S /m, where S is the R R standard deviation of the blank (N ) 16). d Correlation coefficient of calibration curve. Each calibration curve was built on six different concentration levels, with three replicates for each level (N ) 18). e Relative standard deviation; RSD ) (100S)/X, where S is the standard deviation of a concentration in the middle of the LDR and X is the average wavelength shift at that concentration.
those previously reported with established techniques2-9 and with the most sensitive sensors,11-22 respectively. Mercury Determination in Tap Water. Based on the Safe Water Drinking Act, all community water systems must comply with a set of standards established by the EPA.38 The maximum concentration of mercury allowed in drinking water is 2 × 10-9 g‚L-1. Because this concentration is well above the LOD of our approach, we found it appropriate to investigate its potential for monitoring mercury in tap water. Several inorganic ions were first screened for potential interference (see Table 3). Our selection
Table 3. Effect of Inorganic Ions on the Maximum Absorption Wavelength of Gold Nanorodsa compoundb
concentrationc
Au abs peakd
As2O5 Ba(C2H3O2)2 Cr(NO3)3 CuSO4 NH4F NaCl NaNO3 Pb(NO3)2
5 ng‚mL-1 36 ng‚mL-1 2 ng‚mL-1 590 ng‚mL-1 1 µg‚mL-1 29 µg‚mL-1 120 ng‚mL-1 1 ng‚mL-1
599.71 ( 0.02 599.72 ( 0.02 599.72 ( 0.02 599.72 ( 0.02 599.71 ( 0.02 599.72 ( 0.02 599.70 ( 0.02 599.71 ( 0.0
a All measurements were made using a 1.64 × 10-5M gold nanorods solution in 3.34 × 10-3M NaBH4. b Compound dissolved in Nanopure water. c Final concentration of ion in the nanorods aqueous solution. dAverage wavelength and standard deviation of three replicate measurements of maximum absorption wavelength in the presence of potential interferent. These values show no statistical difference36 when compared to the reference value in the absence of inorganic ions (599.72 ( 0.02).
followed the EPA list of potential contaminants commonly found in tap water composition.38 Their concentrations were adjusted to the maximum contaminant level allowed by the EPA. Although such a contaminated sample is most likely difficult to find, the goal of our study was to test the selectivity of the method with a highly complex matrix. The reference sample consisted of a (38) http://www.epa.gov/safewater/consumer/pdf/mcl.pdf.
Analytical Chemistry, Vol. 78, No. 2, January 15, 2006
449
Figure 5. Plot showing the first derivative of the UV-visible absorption spectrum of gold nanorods in Nanopure water (black) and gold nanorods in tap water (red). First derivative was taken over the peak of the longitudinal absorption band. Average peak in Nanopure water was 599.72 ( 0.01 nm and in tap water was 599.64 ( 0.01 nm.
Nanopure water solution with 1.64 × 10-5 M Au nanorods and 1.67 × 10-3 M NaBH4. Each ion was individually tested by spiking the reference solution with a Nanopure water solution of appropriate ion concentration. None of the ions caused noticeable spectral interference. The maximum absorption wavelength of the reference signal remained the same, i.e., λR ) 599.72 ( 0.02 nm. Lack of interference was also observed in the presence of all the ions in solution. Quantitative analysis was performed via the multiple standard addition method. The reference spectrum was recorded from a mixture of 5 mL of Nanopure water with an equal volume of a stock reference solution containing 3.28 × 10-5 M Au nanorods and 3.34 × 10-3 M NaBH4. The spectrum corresponding to the zero standard addition was recorded from a mixture of 5 mL of tap water with an equal volume of stock reference solution. The standard additions were made to 5-mL aliquots of tap water. Each aliquot was spiked with microliters of a 1.5 × 10-9 M HgCl2 standard solution. Their absorption spectra were recorded after mixing them with equal volumes of the stock reference solution. The least-squares fitting of the standard addition curve provided a straight line (correlation coefficient, 0.9975) best described by the equation Y ) 2.2634 × 1011X + 1.887 54. Extrapolation to Y ) 0 provided a mercury concentration (1.67 × 10-9g‚L-1) in the tap water sample above our LOD and slightly below the maximum level allowed by the EPA. Figure 5 compares the first derivatives of the longitudinal absorption bands recorded from three aliquots of Nanopure and tap water. The maximum absorption wavelength of each aliquot was obtained at the intercept with the zeroderivative wavelength shift. For three sample aliquots, the average maximum peak in Nanopure water was 599.72 ( 0.01 nm and in tap water was 599.64 ( 0.01 nm. Within a 95% confidence interval, 450
Analytical Chemistry, Vol. 78, No. 2, January 15, 2006
these values are statistically different,37 which shows our ability to accurately determine trace levels of Hg. Even at these low concentration levels, the relative standard deviation of our method is excellent (
