Anal. Chem. 1999, 71, 596-601
Quantitative Assessment of Surface-Enhanced Resonance Raman Scattering for the Analysis of Dyes on Colloidal Silver Joanna C. Jones, Clare McLaughlin, David Littlejohn, Daran A. Sadler, Duncan Graham, and W. Ewen Smith*
Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow, G1 1XL, UK
Factors that affect quantitative analysis by surfaceenhanced resonance Raman scattering (SERRS) have been investigated using azobenzotriazol and reactive dyes. Preaggregation of the silver colloid was the most effective method to obtain repeatable and reproducible scattering. Aggregation by poly(L-lysine) or spermine provided better precision than aggregation by sodium chloride or nitric acid. Repeatable quantitative analysis was achieved with the azobenzotriazol dyes. A linear calibration graph was obtained over different concentration ranges below 10-8 M, depending on the nature of the colloid. Calculations estimate that 10-8 M is the concentration at which monolayer coverage of the dye on the silver colloid is achieved. Above 10-8 M, there was only a minor increase in the scattering intensity from the azobenzotriazol dyes. In contrast, the reactive dyes did not give a response proportional to concentration over the range studied. The different responses obtained for the two types of dye are believed to be caused by differences in the nature of the interaction of the molecules with the silver surface. The conclusion reached is that control of the colloid preparation, aggregation process, and surface chemistry are essential for successful quantitative analysis of dyes on colloidal silver by SERRS. Surface-enhanced Raman scattering (SERS)1 requires that the analyte is in close proximity to a suitably roughened surface of certain metals, such as can be created on an electrode or by aggregating a colloidal suspension. SERS is believed to be due to either or both of two general mechanisms: electromagnetic enhancement and charge-transfer (or chemical) enhancement.2-4 Only metals with a plasmon resonance at a frequency similar to that of the Raman excitation will give electromagnetic surface enhancement, and with visible excitation, silver and gold are the most commonly used SERS-active metals. The main advantage of the technique from an analytical viewpoint is that an enhancement in scattering of up to 106 is obtained over conventional (1) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. Chem. Phys. Lett. 1974, 25, 163-166. (2) Mal’shukov. G. Phys. Rep. (Rev. Sect. Phys. Lett.) 1990, 194, 343-349. (3) Otto. A.; Mrozek, I.; Grabhorn, H.; Akemann, W. J. Phys. Codens. Matter 1992, 4, 1143-1212. (4) Selected Papers on Surface Enhanced Raman Scattering; Kerker, M., Ed.; SPIE: Bellingham: WA, 1990.
596 Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
Raman scattering. However, the magnitude of the enhancement is sensitive to the physical nature of the enhancing surface, e.g., surface roughness for electrodes and size and shape of particles for colloid solutions. In addition, the orientation-sensitive nature of SERS can hinder effective discrimination of the analyte signal in the presence of a contaminant. Surface-enhanced resonance Raman scattering (SERRS) arises when the analyte displays an additional resonance enhancement providing greater sensitivity.5 It has several additional advantages over SERS for quantitative analysis. When the frequency of the laser is chosen to be at or close to that observed with molecular resonance, the spectrum obtained is related to that of molecular resonance. This gives a molecularly specific signal from the analyte on the metal surface. Together with the significant increase in scattering efficiency from the resonant species, this enables positive identification and selective detection of the analyte in the presence of nonresonant interferences. In many cases, the interaction of the incident light with the chromophore reduces the dependence of the scattering on the orientation of the molecule to the surface.6 Energy transfer can occur from excited states of the molecule to the surface, causing fluorescence quenching and enabling the use of a wide range of chromophores.14 The sensitivity achieved permits the use of low-power lasers, which in turn reduces the likelihood of analyte decomposition. Quantitative analysis by SERRS has recently been reported for both metal electrodes7 and metal colloids.5,8-11 However, only a few reports have given adequate data on the repeatability and reproducibility of intensity measurements, and in some instances, (5) Kneipp, K.; Wang, Y.; Dasari, R. D.; Feld, M. S. Appl. Spectrosc. 1995, 49, 780- 784. (6) Rodger, C.; Smith, W. E.; Dent, G.; Edmondson, M. J. Chem. Soc., Dalton Trans. 1996, 791-799. (7) Norrod, K. L.; Sudnik, L. M.; Rousell, D.; Rowlen, K. L. Appl. Spectrosc. 1997, 51, 994-1001. (8) Kellner, R.; Mizaikoff, B.; Jakusch, M.; Wanzenbock, H. D.; Weissenbacher, N. Appl. Spectrosc. 1997, 51, 495-503. (9) Hill, W.; Wehling, B.; Fallourd, V.; Klockow, D. Spectrosc. Eur. 1995 7, 20-22. (10) Laserna, J. J.; Sutherland, W. S.; Winefordner, J. D. Anal. Chim. Acta 1990, 237, 439-450. (11) Xue, G.; Lu, Y.; Gao, J. Appl. Surf. Sci. 1994, 78, 11-15. (12) Kerker, M.; Wang, D S.; Chew, H.; Siiman, O.; Bumm. L. A. In Surface Enhanced Raman Scattering; Chang, R. K., Furtak, T. E., Eds.; Plenum: New York, 1982; p 109. (13) Laserna, J. J.; Campigpla, A. D.; Winefordner, J. D. Anal. Chem. 1989, 61, 1697-1701. (14) Munro, C. H.; Smith, W. E.; White, P. C. Analyst 1995, 120, 993-1003. 10.1021/ac980386k CCC: $18.00
© 1999 American Chemical Society Published on Web 12/29/1998
the signals obtained have been shown to be irreproducible and nonquantitative. Several factors, including differences in the degree of aggregation, the colloid particle size,12 variations in adsorption energies, and the distribution and orientation of molecules at various sites on the metal surface, may have contributed to the poor reproducibilities observed.13 This paper reports the results of a study to determine the factors that need to be controlled to achieve quantitative analysis by SERRS. The active surface was a silver colloid prepared according to a modified Lee and Miesel method.14 The colloid is nearly monodisperse with a mean particle diameter of 27 nm15 and is stable for several months. As the plasmon resonance wavelength is approximately 402-404 nm, aggregation of the colloid is required to obtain optimum surface enhancement when visible excitation is used. The degree of aggregation and control of this process are critical in obtaining repeatable SERRS signals. In this study, the choice of aggregating agent and the influence of the aggregation procedure on precision have been investigated for a wide range of concentrations of two sets of dyes, namely, azobenzotriazol compounds16 and commercial reactive dyes. The azobenzotriazol dyes were synthesized to have properties that would optimize surface adsorption and so enhance the possibility of achieving a quantitative relationship between the concentration of the dye and the measured Raman signal. EXPERIMENTAL SECTION Raman scattering was collected using a Renishaw 2000 microprobe spectrometer (Renishaw Ltd.), with the excitation at 514.5 nm provided by an OmNichrome (Lambda Photometrics) argon ion laser (25 mW). The Renishaw instrument has a notch filter, a single grating, and a cooled charge-coupled device detector. A microscope is used to focus the laser radiation and to collect the 180° Raman scattering. Samples were analyzed in a cuvette placed in a macrosampler attached to a microscope. The quartz glass cuvettes (1 × 1 × 4 cm) were cleaned overnight in concentrated sulfuric acid and then washed with distilled water, followed by repeated washings with methanol and distilled water. Resonance Raman (RR) spectra were recorded by placing several crystals of the compounds on a microscope slide positioned under the microscope. All scans were of 10 s with the grating centered at 1400 cm-1 giving a measured spectral range of 824-1724 cm-1. Net peak height intensities were obtained by subtracting the baseline estimated from close by background points. Five separate aggregates for each dye concentration were prepared and analyzed, unless otherwise stated. The spectra shown and the relative standard deviation (RSD) values quoted are based on the mean intensities of five scans. The silver colloid was prepared according to a modified Lee and Miesel procedure,14 which gives a concentration of 2.6 × 1010 colloidal particles mL-1 (see Appendix I for calculation). Two colloidal solutions were used: (A) with λmax ) 402 nm and pH ) 5.8; (B) with λmax ) 410 nm and pH ) 6.4. Four separate aggregating agents were used: nitric acid (1% v/v), sodium chloride (1% m/v), poly(L-lysine) hydrobromide, average Mr of (15) Munro, C H.; Smith, W. E.; Garner, J.; Clarkson, J.; White, P. C. Langmuir 1995, 11, 3712-3720. (16) Graham, D.; McLaughlin, C.; McAnally, G.; Smith, W. E.; Jones, J. C.; White, P. C. Chem. Commun. 1998, 11, 1187.
15 000-30 000 (0.01% m/v), and spermine (8 × 10-4 M). Two methods of preparing the aggregated colloid were used. Method 1. Colloid (1 mL) and water (1 mL) were placed in a cuvette; the aggregating agent (50-200 µL) was added followed by immediate addition of the analyte solution (150 µL). The mixture was shaken for a few seconds and left for 10 min prior to analysis. Method 2. Colloid (10 mL) and water (10 mL) were preaggregated with one of the four aggregating agents (1500 µL) and placed in a cuvette. The resultant aggregated solution was left to stabilize for 10 min. An aliquot of 2.15 mL was removed from the batch of preaggregated colloid, the dye solution (150 µL) added, and the mixture left for 10 min. All concentrations quoted, for either method, are the final concentrations of the dyes in the aggregated colloid solution and are in the range 4 × 10-4-4 × 10-12 M. Poly(L-lysine), Mr 4 000-15 000 (Sigma), spermine H2N(CH2)3NH(CH2)4NH(CH2)3NH2 (Aldrich), reactive blue 4, reactive blue 74, acid blue 25, and 1,4-diaminoanthraquinone (Aldrich) were of analytical grade. 4-(5′-Azobenzotriazol)-aminonaphthalene, 3-methoxy-4-(5′-azobenzotriazol)phenylamine, and 3,5-dimethoxy-4-(5′azobenzotriazol)phenylamine were synthesized at the University of Strathclyde and checked for purity using thin-layer chromatography, NMR spectrometry, and C, H and N analysis.16 Water used in the preparation of solutions was HPLC grade (BDH). The structure of the dyes are given in Figure 1. RESULTS AND DISCUSSION Azobenzotriazol Dyes. Examples of the SERRS spectra of the azobenzotriazol dyes are shown in Figure 2; none of the spectra showed evidence of fluorescence. When no reagents were added to enhance aggregation, SERRS signals for 4-(5′-azobenzotriazol)aminonaphthalene were obtained over a concentration range of 10-3-10-9 M. Use of nitric acid or sodium chloride as an aggregating agent (method 1) gave greater SERRS enhancement but adversely affected the RSD values, as indicated in Table 1, for 10-7 M, at 1424 cm-1. The effects of different volumes of each of the aggregating reagents (method 1) on the magnitude and stability of the SERRS signals of 4-(5′-azobenzotriazol)aminonaphthalene, are illustrated in Figure 3, for measurements at 1424 cm-1. Although the largest enhancement in scattering intensities is observed when sodium chloride or nitric acid are added (Figure 3a and b, respectively), 50-150 µL of spermine or poly(L-lysine) gave a better signal stability over 60 min. At a volume of 200 µL of poly(L-lysine) or spermine, poorer signal stability was observed. Consequently, 150 µL of poly(L-lysine) or spermine was normally used, to maximize the enhancement effect and achieve the best signal stability. Initially, small quantities of the colloid were pre-aggregated (method 1) and used for quantitative analysis; however, it was observed that the reproducibility of five separate aggregations was poor, as shown in Table 1. To improve the precision, method 2 was devised that involved preaggregation of a larger batch (e.g., 20 mL) of the colloidal suspension. Using colloid preaggregated with poly(L-lysine), RSD values of 3-5% were obtained for five scans of five separate solutions to which the 10-7 M dye was added separately in each case (Table 1). Similar results were obtained Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
597
Figure 1. Structures of azobenzotriazole and reactive dyes: (a) 4-(5′-azobenzotriazol)aminonaphthalene, (b) 3,5-dimethoxy-4-(5′-azobenzotriazol)phenylamine, (c) 3-methoxy-4-(5′-azobenzotriazol)phenylamine, (d) reactive blue 74, (e) reactive blue 4, (f) acid blue 25, and (g) 1,4diaminoanthraquinone.
Figure 2. SERRS spectra of (a) 3,5-dimethoxy-4-(5′-azobenzotriazol)phenylamine, (b) 3-methoxy-4-(5′-azobenzotriazol)phenylamine, and (c) 4-(5′-azobenzotriazol)aminonaphthalene, all at 10-7 M; aggregation method 2. Spectra are overlaid for clarity.
with spermine. Hence, method 2 was applied in all subsequent analyses. Each of the azobenzotriazol dyes gave a consistent spectrum in the range 850-1750 cm-1, for concentrations from 10-3 to 10-13 M, as indicated by typical spectra of 10-5, 10-8, and 10-10 M 4-(5′azobenzotriazol)aminonaphthalene, shown in Figure 4. There were no significant peaks in other parts of the spectra between 200 and 4000 cm -1. Quantitative measurements were made for each of the azobenzotriazol dyes, but as the results and trends obtained were similar for all the compounds, only the data for 4-(5′azobenzotriazol)-aminonaphthalene are given in detail. A log-log graph of intensity versus the concentration of 4-(5′azobenzotriazol)aminonaphthalene is shown in Figure 5a, for measurements at 1424 cm-1 using colloid solution A. A linear relationship exists between 10-8 and 10-12 M, the gradient of which is 1 for a log-log graph. However, when the exact same procedure was followed using colloid B, a similar shape of graph was obtained, but the linearity of response was over a smaller concentration range (Figure 5b). For both colloids, several calibration graphs were recorded at different times to prove the reproducibility of the features illustrated in Figure 5a and b. 598
Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
To examine whether the variable calibration response was a consequence of changes in the laser intensity, three different neutral density filters with optical densities of 0, 0.3, and 0.6 were used to reduce the laser intensity at the sample. The filters provide relative transmissions of 100, 50, and 25%, respectively. However, the results in Figure 5c show that no significant effect on the extent of linearity was observed for measurements at 1424 cm-1 using colloid A. Therefore, the extent of the linearity is independent of the incident laser intensity. The intensity-concentration plots in Figure 5a-c have several common features: (i) at lower concentrations there is an increase in intensity, which is not always proportional to the change in concentration; (ii) there is a concentration (10-8 M) at which no further increase in intensity is observed; and (iii) after this turnover point, the intensity is constant or decreases at higher concentrations, probably as a consequence of self-absorption. Calculations (see Appendix I) estimate that monolayer coverage of the dyes on the silver particles occurs at approximately 10-8 M. This is the concentration at which there is the change in gradient for each of the intensity-concentration plots. Once monolayer coverage is reached, any further dye molecules do not contribute significantly to the SERRS signal. A linear response is observed using both colloid solutions (A and B). However, the extent of linearity is different between the two colloids and may be dependent upon the characteristics of the colloid, such as pH, surface charge, and average particle size distribution. Further studies are necessary to identify the cause of these differences and their effect upon the extent of linearity. Reactive Dyes. SERRS spectra of the reactive dyes (Figure 1) were recorded for a wide range of concentrations. Examples of the spectra recorded for reactive blue 74 are given in Figure 6. At concentrations above and below 10-8 M, the spectra are different, unlike the situation for the azobenzotriazol dyes, which gave a consistent spectrum at all concentrations. Furthermore, a proportional increase in intensity was not obtained for any concentration range of reactive blue 74 or the other reactive dyes. The spectrum of reactive blue 74 obtained at concentrations greater than 10-8 M is similar to that of the RR spectrum of a
Table 1. RSD of SERRS Measurements of 4-(5′-Azobenzotriazol)aminonaphthalene at 10-8 M, Obtained at 1424 cm-1 aggregating agent (150 µL)/ aggregation method 1 or 2
RSD of 1 solution (1 × 5 scans)/%
RSD of 5 solutions (5 × 5 scans)/%
self-aggregation nitric acid (1% v/v) (1) sodium chloride (1% m/v) (1) poly(L-lysine) (0.01% m/v) (1) spermine (8 × 10-4 M) (1) dye + colloid preaggregated with poly(L-lysine) (2)
1-4 12-25 14-18 1-4 1-4 1-3
7-8 31-41 28-35 17-21 17-18 3-5
Figure 3. Changes in SERRS intensity of 10-7 M 4-(5′-azobenzotriazol)aminonaphthalene at 1424 cm-1, for aggregation with (a) 1% (m/v) sodium chloride, (b) 1% (v/v) nitric acid, (c) 0.01% (m/v) poly(L-lysine), and (d) 8 × 10-4 M spermine. Spectra were recorded every minute for a total of 60 min; 0 (a), 50 (b), 100 (c), 150 (d), and 200 (e) µL of each aggregating reagent was used (method 1).
solid sample of this compound (Figure 7, spectra c and d). Compared with the RR spectrum of the dye in solution (Figure 7, spectrum b), the fluorescence background of the SERRS spectrum for 4 × 10-5 M is much reduced. This suggests that at concentrations of >10-8 M, the dye is adsorbed onto the silver in some way that allows efficient fluorescent quenching and enhancement. Both these effects could be achieved by the formation, on the colloidal surface, of small crystallites or aggregates, or the building up of multilayers of dye molecules. At concentrations below 10-8 M, the change in SERRS spectrum indicates that the interaction of the molecules of reactive blue 74 on the silver surface is different. As the spectrum at all the concentrations examined below 10-8 M is identical, the interaction is consistent. As there was not a proportional change in intensity with dye concentration, at either concentration range, it seems that the processes causing the dye and colloid interactions are not controlled. Consequently, quantitative analysis of reactive blue 74 by SERRS is not currently feasible. Similar observations were made for reactive blue 4, acid blue 25, and 1,4-diaminoanthraquinone (Figure 1).
Figure 4. Overlaid SERRS spectra of 4-(5′-azobenzotriazol)aminonaphthalene, at (a) 6.7 × 10-5, (b) 6.7 × 10-8, and (c) 6.7 × 10-10 M; aggregation with poly(L-lysine) (150 µL), method 2.
When the results for both sets of dyes are compared, it is clear that quantitative SERRS measurements can be obtained with the azobenzotriazol dyes. However, for the reactive dyes, this could not be achieved under the conditions described. The main Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
599
Figure 6. Overlaid SERRS spectra of reactive blue 74, at (a) 4 × 10-10, (b) 4 × 10-8 (c) 4 × 10-7, and (d) 4 × 10-5 M; aggregation with poly(L-lysine) (150 µL), method 2.
Figure 7. Spectra of reactive blue 74: (a) SERRS spectrum at 4 × 10-11 M, (b) RR spectrum at 4 × 10-5 M, (c) SERRS spectrum at 4 × 10-5 M, and (d) RR spectrum of a solid sample.
Figure 5. Calibration graphs for 4-(5′-azobenzotriazol)aminonaphthalene based on measurements at 1424 cm-1; aggregation with poly(L-lysine) (150 µL), method 2, (a) and (b) using different batches of colloid and (c) using filters permitting 100, 50, and 25% transmission.
difference is believed to be in the nature of the interaction of the dyes on the silver surface. The stability of the silver colloid is attributed to a citrate layer on the surface of the particles.15 The azobenzotriazol dyes are believed to displace the citrate layer and bond strongly to the colloid particle surfaces through the triazol group, whereas the reactive dyes are expected to bond by polar attachment to the citrate layer without displacement. This type of polar interaction is likely to be more susceptible to changes in the chemical environment of the silver colloid, which is a possible cause of the differences observed between the two types of dye. 600 Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
CONCLUSIONS It has been shown that, for SERRS, preparation of a stable aggregate of silver colloidal particles can be achieved with poly(L-lysine) or spermine. Preaggregation of the colloid is preferable to obtain repeatable and reproducible spectra. A proportional change in intensity occurs with azobenzotriazol dye concentration, although the sensitivity of response is reduced abruptly above monolayer concentrations. A colloid solution with λmax ) 402 nm and pH ) 5.8 gave a linear response over a larger concentration range than a colloid with λmax ) 410 nm and pH ) 6.4. In contrast, the interaction of reactive blue 74 and similar dyes is more variable. Two different SERRS spectra can be obtained depending on the dye concentration, and there is a lack of proportionality in the change in intensity with concentration. It is apparent that the nature of a molecule influences the surface interactions with the silver colloid. Without adequate control of the process, a proportional and repeatable increase in intensity with analyte concentration is not achieved. This indicates that, for some compounds, quantitative analysis by SERRS may not be possible, unless measures are taken to optimize the mechanism of interaction with the colloid. ACKNOWLEDGMENT We acknowledge support provided by the Engineering and Physical Sciences Research Council (EPSRC) under the analytical
initiative including support for C.M. J.C.J. acknowledges CASE support from EPSRC and Zeneca. D.A.S. acknowledges the Royal Society for the receipt of a University Research Fellowship. APPENDIX I Estimation for Monolayer Coverage and Number of Colloidal Particles per Milliliter of Colloid. The azobenzotriazol dye was assumed to lie flat (0°) to the colloid sphere surface. The area occupied by one molecule of azobenzotriazol dye was estimated at 1.505 nm2. Assuming spherical colloid particles of mean diameter 27 nm, obtained from TEM15 studies, the surface area of one colloidal particle is 2290 nm2. To calculate the total surface area available to the analyte, the number of colloidal particles in 1 mL must be estimated. In 500 mL of water, 90 mg of AgNO3 is used, which is equivalent to 1.06 × 10-3 mol of AgNO3. Therefore, the number of moles of Ag in 1 mL is 2.12 × 10-6 and the number of atoms of Ag is 1.28 × 1018 atoms mL-1. The number of colloidal particles in 1 mL is the total number of Ag atoms in 1 mL divided by the number of Ag atoms in 1 colloidal particle. The number of Ag atoms in 1 colloidal particle may be estimated from the metal radius of 0.144 nm, assuming
Ag to be a 2-coordinate face-centered cubic (fcc) formation with 14 atoms per unit. The volume of such a cube is therefore equal to 2.99 × 10-3 nm3. The volume of a sphere with diameter 27 nm is 10 306 nm3. Assuming the distance between individual fcc Ag is ignored, the number of such structures which would occupy 1 colloidal particle are 3.45 × 106. Therefore, the number of Ag atoms in 1 colloidal particle are 4.83 × 107. The number of colloidal particles in 1 mL is the total number of Ag atoms in 1 mL divided by the number of Ag atoms in 1 particle, which is 2.6 × 1010. The available surface area is therefore 6.05 × 1013 nm2. The number of azobenzotriazol molecules occupying this area is 4.02 × 1013. The monolayer coverage of the azobenzotriazol dye is estimated at 6.7 × 10-11 mol in the total solution volume of 2.3 mL. Therefore, monolayer coverage of the dye on the colloid is estimated to occur at a concentration of about 10-8 M.
Received for review April 7, 1998. Accepted October 20, 1998. AC980386K
Analytical Chemistry, Vol. 71, No. 3, February 1, 1999
601
