Anal. Chem. 2007, 79, 5862-5868
Determination of Chromium(VI) by Surface Plasmon Field-Enhanced Resonance Light Scattering Zhiqiang Han,† Li Qi, Gangyi Shen,† Wei Liu,† and Yi Chen*
Beijing National Laboratory of Molecular Science, Laboratory of Analytical Chemistry for Life Science, Institute of Chemistry, Chinese Academy of Sciences, Beijing, 100080, People’s Republic of China
A surface plasmon-enhanced resonance light scattering method has been developed. The method features strong light scattering but very weak background, and after incorporating with selective sample extraction and ionassociation complexation using rhodamine B and KI as reactants, it could selectively determine Cr(VI) in both of standard and real samples, reaching a limit of detection down to 20 nM which is about 40-fold as sensitive as flame atomic absorption spectrometry and 140-fold as sensitive as fluorescent spectroscopy. Its linear working range was found in between 40 and 320 nM, with a relative standard deviation of peak height at 10). An aliquot of 25 mL of filtrate was thoroughly mixed, in a 100 mL separatory funnel, with 7 mL of 1 M H2SO4, 1 mL of saturated Na2SO4 solution, 10 mL of saturated NaCl solution, and 7 mL of water. The mixed solution was shaken with 5 mL of 8%, (v/v) TOA in MiBK for 3 min and placed at room temperature until phase separation was completed. The organic phase was (27) Yu, F.; Persson, B.; Lofas, S.; Knoll, W. J. Am. Chem. Soc. 2004, 126, 89028903. (28) Futamata, M. J. Phys. Chem. 1995, 99, 11901-11908. (29) Perkins, E. A.; Squirrell, D. J. Biosens. Bioelectron. 2000, 14, 853-859. (30) Jory, M. J.; Cann, P. S.; Sambles, J. R. Appl. Phys. Lett. 2003, 83, 30063008. (31) Wang, Z.; Zheng, Q. Y.; Chen, Y. Anal. Lett. 2001, 34, 2609-2619. (32) Wang, Z.; Chen, Y. Carbohydr. Res. 2001, 332, 209-213. (33) Chen, Y.; Huang, H. W.; Yu, X.; Qi, L. Carbohydr. Res. 2005, 340, 20242029. (34) Jordan, C. E.; Frutos, A. G.; Thiel, A. J.; Corn, R. M. Anal. Chem. 1997, 69, 4939. (35) Patel, N.; Davies, M. C.; Hartshorne, M.; Roberts, C. J.; Tendler, S. J. B.; Williams, P. M. Langmuir 1997, 13, 6485-6490. (36) Gestwicki, J. E.; Cairo, C. W.; Mann, D. A.; Owen, R. M. Anal. Biochem. 2002, 305, 149-155. (37) Liu, S. P.; Liu, Q.; Liu, Z. F.; Li, M.; Huang, C. Z. Anal. Chim. Acta 1999, 379, 53-61.
Figure 1. Setup of the laboratory-built SP-RLS system. 1, He-Ne laser; 2, polarizer; 3, goniometer; 4, prism; 5, sensor film; 6, flow cell; 7, lens; 8, filter; 9, PMT; 10, data treating part; 11, inlet; 12, outlet.
transferred into a 25 mL separatory funnel and back-extracted with 5 mL of 0.3 M NaOH. The left organic phase was discarded, and an aliquot of 0.84 mL of the water phase was pipetted into a flask. (Caution: Because TOA and MiBK have moderate toxicity, all the extraction procedures should be performed in a fume hood and the operator should wear safety goggles and protective gloves.) After being adjusted to pH 7.0 with 4 M HCl, the aqueous solution was mixed with 1.8 mL of 0.005% RhB, 0.2 mL of 10% KI, 0.3 mL of 4 M HCl, and 5.26 mL of water and allowed to react for 10 min prior to measurements. Ammonium nitrate solution (60 mM) was adjusted to pH 9.3 by NH3 to work as the HPLC mobile phase. The mobile phase was prepared daily and filtered through a 0.45 µm membrane before use. SP-RLS Setup and Measurement. Figure 1 shows the schematic setup of a laboratory-built SP-RLS system. A light beam from an 8 mW He-Ne laser (1, model 3227H-PC, Hughes aircraft, U.S.A) was polarized through a polarizer (2), and its perpendicular component or p-polarized light was directed into an isosceles BK7 glass prism (4, n ) 1.515) from one side and projected on a sensor film (5) attached to the bottom of the prism via an underlay of refractive index matching oil (n ) 1.516). The scattering signal was detected from the bottom side of the prism with a PMT (9, model R121, Hamamatsu, Japan) through a lens coupler (7) and an interference filter (8, λ ) 632.8 nm, ∆λ ) 10 nm) to block the potential fluorescence and other lights. The prism and scattering light detection parts were mounted on a rotating goniometer (3, angular scanning). The sensor film was assembled under a glass cover and sealed using Teflon and silicone gaskets to form a 110 µL flow cell (6). Flow injection analysis (FIA) was used for easy and convenient delivery of various solutions into the detection cell. A highly simplified FIA unit was coupled to the flow cell. The injection block of the FIA was made of an injection valve (model 7725, Rheodyne, U.S.A) connected to a micropiston pump (model SB-II, Beijing Oriental Science and Technology Company, Beijing, China) and the flow cell (17 cm from valve to detection point) with PEEK tubes. Prior to measurement, all the channels of the system were flushed with 0.1 M NaOH and carrier solution, for 5 min each, at a flow rate of 0.1 mL/min. The carrier flow rate was then adjusted to 1.5 mL/min (to speed up analysis), and samples (100 µL) were sequentially injected via the FIA and brought into the carrier stream to generate peak series. The scattering signals were Analytical Chemistry, Vol. 79, No. 15, August 1, 2007
5863
Figure 2. Comparison between evanescent wave- and orthogonal light-excited light scatterings. (A) Schematic setup of light scattering excited by an either evanescent field or orthogonal light. 1, He-Ne laser; 2, block and reflection mirror inserted only when evanescent wave excitation is required; 3, prism; 4, mirror to direct an orthogonal excitation light to the flow cell; 5, 110 µL flow cell; 6, signal collecting lens; 7, PMT. (B) Rayleigh scattering signals of water produced by evanescent wave (lower) and orthogonal light (upper).
recorded and treated through a chromatographic working station SC-1100 (Computer Department, Beijing University of Chemical Technology). All the experiments were performed at an ambient temperature of 20 ( 1 °C. The light incident angle was adjusted at around 72.3° to achieve the maximum resonance and the most sensitive detection. Preparation of Gold Sensing Film. Gold sensing films were deposited on glass slides and modified to get rid of potential sample absorption problems. Briefly, onto polished BK7 glass slides was deposited a layer of 50 nm of gold film over a 2 nm chromium undercoat by a high-vacuum evaporator model HUS5GB (Hitachi, Japan). The resulted gold-coated slides were cut into required sections, washed with ethanol, and dried under a stream of nitrogen (high purity). These freshly cleaned and dried gold films were immersed into 5 mM MUA (in ethanol) for at least 5 h to form an MUA monolayer on the gold surface. The assembling process was monitored by surface plasmon resonance (SPR) using a homemade multiwavelength SPR system reported in previous papers.31-33 A variation of resonance wavelength shift (∆λ) of the slides would be observed as the modification reaction progresses. Once ∆λ reached 8.31 nm and kept unchanged, which indicated the surfaces had reached the maximum coverage,34 the assembling was stopped and the gold films were rinsed with ethanol and triply distilled water to thoroughly remove the unbound MUA. After being immersed in an aqueous solution of 75 mM EDC and 15 mM NHS for 30 min,35 they were soaked in aqueous EDA (250 mM, pH 8.5) for 4 h to introduce primary amines onto the gold surface and thoroughly rinsed with water and ethanol before use.36 (Caution: EDA has moderate toxicity. All the procedures with EDA should be performed in a fume hood or in an airtight system, and the operator should wear safety goggles and protective gloves.) LC-ICPMS Experiment. LC-ICPMS determination was performed for comparison. A sample was injected, via an autoinjector with a 100 µL sample loop, onto a Hamilton PRP-X100 anionexchange column (250 mm × 4.1 mm, 10 µm, Hamilton, Reno, NV), eluted by an Elite P230 pump (Elite, Dalian, China) at a flow 5864
Analytical Chemistry, Vol. 79, No. 15, August 1, 2007
rate of 1 mL/min and finally transferred into an ELAN6000 ICPMS (Perkin-Elmer, Sciex, Norwalk, CT). The main conditions of ICPMS were as follows: plasma flow 18 L/min, auxiliary flow 1.65 L/min, nebulizer flow 0.24 L/min, sheath flow 0.98 L/min, plasma rf power 1.40 kW, monitored ion 52Cr, and dwell time 0.5 s.41 RESULTS AND DISCUSSION Effectiveness of SP-RLS and Mechanism. As expected, the newly developed SP-RLS method was demonstrated to be very sensitive in the determination of Cr(VI), reaching a limit of detection (LOD) down to 20 nM Cr(VI) (S/N ) 6). This is improved by about 6-fold in comparison with the common orthogonal RLS detection format20 that has an LOD of 128 nM Cr(VI) obtained in our laboratory. This improvement was first attributed to the effective reduction of background by using the evanescent field to excite the scattering. In theory, this is a better type of dark (or even black) background excitation than the orthogonal excitation format: There is no condition to produce reflection and surface scattering from the walls of the flow cell because the source light cannot penetrate the glass-solution interface beyond hundreds of nanometers when its incident angle is greater than the critical for total reflection. By measuring the background using a setup shown in Figure 2, it was found that the new excitation method was able to reduce the background by 61% and to improve the signal-tonoise by 81%. It should be mentioned that the evanescent field was also unable to remove the insuperable Rayleigh and Raman scatterings and potential fluorescence from a solution and the flow cell wall. Complete black background has thus not yet been achieved. The sensitivity improvement was second or more importantly ascribed to the surface plasmon enhancing mechanism. It is (38) Wang, L. Y.; Wang, L.; Xia, T. T. Anal. Sci. 2004, 20, 1013-1017. (39) Ronot-Trioli, C.; Trouillet, A.; Veillas, C.; Gagnaire, H. Sens. Actuators, A 1996, 54, 589-593. (40) KotasA ˆ , J.; Stasicka, Z. Environ. Pollut. 2000, 107, 263-283. (41) Byrdy, F.; Olson, L.; Vela, N.; Caruso, J. J. Chromatogr., A 1995, 712, 311320.
Figure 3. Peak series of Cr(VI) measured at room temperature on a bare BK7 glass slide (A) or amino-modified gold film (B) at a flow rate of 1.5 mL/min. The sample was prepared by a 10 min reaction of 320 nM Cr(VI) with 0.005% (w/w) RhB, 10% (w/w) KI, 4 M HCl, and H2O at a volume ratio of 9:1:1.5:31.5.
theoretically clear that a gold film will produce a surface plasmon wave due to the fluctuation of its highly free electrons.26 Once the resonance condition is achieved, the evanescent field for excitation will be strengthened and in turn the scattering intensity increases. Figure 3, parts A and B, compares the experimental difference between the bare glass surface and the modified gold surface, and the latter made the peaks rise for about 1 order of magnitude than the former. It should be noted that the SP also increased the background scattering intensity according to the baselines of the two figures. This is theoretically reasonable based on the above-mentioned mechanism. The high sensitivity of SP-RLS obtained in this investigation was third resulted from the use of a second resonance mechanism, that is, by introducing the ion-association complex of [RhB+][I3-] into the measuring system. This ion-association complex is formed by using RhB and KI as reactants where I3- will be generated once an oxidative solute is added:20,42
Cr(VI) + I- f I3- + Cr(III) I3- + RhB+ f [RhB+][I3-] The reactions are suitable for the measurement of all samples able to produce or destroy iodine, and selective sample preparation is thus required as discussed in the section Interference and Sample Preparation. This is a typical RLS system, easily incorporating with a He-Ne laser (632.8 nm) to produce strong resonance scattering at 625-640 nm but weak background.20 In addition, the reaction to form the ion-association complex should last for at least 10 min at a temperature between 20 and 30 °C to achieve a sensitive detection. However, the measurement should also be finished within 1 h after the reaction started because oxygen in air and water reacts with I-, causing a gradual increase of detection background. Blank reaction (without the addition of analytes) is hence suggested for control. The novel method features thus at least weak background and strong resonance scattering of Cr(VI) caused by two resonance (42) Liu, S. P.; Liu, Z. F.; Luo, H. Q. Anal. Chim. Acta 2000, 407, 255-260.
mechanisms, much superior to the common orthogonal detection formats and SP-free evanescent field excitation approaches. Key Factors. In spite that SP-RLS has intrinsic advantages, optimization of some critical factors remains important to achieve sensitive detection. After a systematic screening and inspection, gold surface modification, light incident angle, and sample preparation were shown to be essential. Incident Angle. Similar to SPR,39 the incident angle of light is a control factor of sensitivity for SP-RLS and was systematically studied. Figure 4 shows that the scattering intensity is a peakshaped function of the incident angle, with its maximum at 72.3° measured from a solution of 160 nM Cr(VI), implying that the measured scattering is not directly excited by the incident light but by the SP evanescent field.30 It was thus anticipated that SP-RLS would have angular “detuning” phenomenon: The maximum incident angle is dependent on sample concentration because the dielectric constant and hence the evanescent field are also a function of sample concentration, similar to the SP fluorescence.26 This anticipation has been demonstrated by experiment as shown in Figure 6, where the measured scattering intensity bends negatively from the linear curve as the sample concentrates. The bending becomes very serious at the concentration far above 320 nM Cr(VI). This angular detuning problem is not preferred for quantitation but can be overcome either by optimizing the incident angle per concentration or by diluting the samples to a linear responding range. The latter way is convenient in use and was adopted in this study, that is, keeping the incident angle at 72.3° and Cr(VI) in the range 40 to 320 nM for quantitative determination. Interference and Sample Preparation. In principle the new method is applicable to the determination of all solutes able to produce or destroy I2 in solutions. It is thus not a selective sensing approach but somewhat universal. However, it could be applied to the selective determination of Cr(VI) after incorporating with a selective sample preparation procedure,37 that is, its aqueous samples were selectively extracted with TOA and the extraction efficiency is 88.9%. The potential interferences of coexisting substances in such wastewater samples have been inspected Analytical Chemistry, Vol. 79, No. 15, August 1, 2007
5865
and Bi(III). Hg(II) and Fe(III) have also fairly low LOT (around micromolar). Fortunately, these substances are commonly present at a very low concentration in real Cr(VI) samples,40 and their influences were negligible. The interference of NO2- was successfully removed with urea.37 In addition, MnO4- had also strong interference but could be removed by sequentially adding drops of dilute sodium nitrite and urea solution into the sample solutions.37 MnO4- was reduced by NO2- to produce Mn2+ and NO3- which could be tolerated at quite a high concentration (Table 1), whereas the excessive NO2- was finally decomposed by urea:
2MnO4- + 5NO2- + 6H+ f 2Mn2+ + 5NO3- + 3H2O 2NO2- + 2H+ + CO(NH2) f CO2 + 3H2O + 2N2v Figure 4. Plot of scattering intensity against light incident angle measured from 160 nM Cr(VI) on an amino-modified gold film. Other conditions are as in Figure 3.
Figure 5. Peak series of Cr(VI) produced by SP-RLS at an increasing concentration sequence on a bare gold film. Other conditions are as in Figure 3. Peaks: 1 ) background or [RhB+][I-], 2-7 ) Cr(VI) at 80, 160, 320, 400, 640, and 800 nM, respectively.
Figure 6. Peak series of Cr(VI) produced by SP-RLS at an increasing concentration sequence on an amino-modified gold film. The inset shows the relationship of relative scattering intensity with the concentration of Cr(VI) on the amino-modified gold film, where ∆I is the relative intensity of scattered light. Other conditions are as in Figure 3. Peaks: 1 ) background or [RhB+][I-], 2-9 ) Cr(VI) at 20, 40, 80, 160, 320, 400, 640, and 800 nM, respectively.
carefully, and their limit of tolerance (LOT) was collected in Table 1. By LOT was meant the maximum concentration of a nontarget composition allowed to coexist in a sample solution, causing a signal variation less than (5% for a solution of 160 nM Cr(VI). As shown in Table 1, the nonoxidative ions have high LOT, whereas the oxidative and I2 (or I-)-combining substances have quite a low LOT. The most harmful substances are NO2-, Cd(II), Se(IV), 5866
Analytical Chemistry, Vol. 79, No. 15, August 1, 2007
Modification of Gold Sensing Film. Since SP-enhanced evanescent field has its maximum at the interface of gold and solution, decaying exponentially away from the surface, it can be expected that dielectric substances imposing in the way will disturb the evanescent field. This disturbance will become significant once solutes randomly adsorb on the gold surface. Such adsorption may lead to a great reduction of sensitivity, signal drifting (Figure 5), or even irreproducible measurement and should be suppressed. In this study, RhB, especially its ionic form RhB+, was observed to easily absorb onto the gold sensing surface. To reduce the adsorption, several techniques were considered, but chemical modification of the gold surface with amino groups was shown to be the most effective, without imposing obvious influence on determination. The amino groups are positively charged under the experiment conditions, repelling the attachment of RhB+ and other cations onto the gold surface. Figure 6 clearly demonstrates the effectiveness of the modified surface as compared with the unmodified one (Figure 5) which produces evident baseline fluctuation. The modification also improved the peak reproducibility, reducing the relative standard deviation of peak height from 3.8% to less than 2% (n ) 5). Carrier. Either buffer (pH 3-8) or water was inspected as a carrier solution. No evident difference has been observed yet, and water was thus adopted for its “green” feature and low cost. The flow rate of the water carrier was optimized at 1.5 mL/min. The peaks were observed to broaden in a slow carrier stream or to decrease in their height in a very fast carrier flow due to the delayed response of electronic circuits. Quantitation. SP-RLS of standard samples which was conducted at an increasing concentration sequence (Figures 5 and 6) generated linear curves between the relative peak height (∆I, subtracted by the averaged peak height of background) of Cr(VI) and its concentration (c, /nM) as shown in Figure 6 (the inset) and Table 2. Commonly, the modified gold film gave the widest linear range, lowest LOD, and highest precision (Table 2, line 3) compared with the unmodified (Table 2, line 2) and bare glass (Table 2, line 1). Strikingly, SP-RLS (LOD ) 20 nM) was much more sensitive than FLAA (LOD ) 807 nM9) and fluorescent spectroscopy (LOD ) 2.8 µM38). Three wastewater samples and three synthetic “real” samples (prepared under the guidance of interference experimental data,
Table 1. Tolerance of Coexisting Substances Measured from 160 nM Cr(VI)a
a
coexisting ion
amount tolerated/µM
coexisting ion
amount tolerated/µM
coexisting ion
amount tolerated/µM
K(I) Na(I) Mg(II) Al(III) Pb(II) Ca(II) ClO4Mo(VI)
1800 1300 1200 260 26 1600 40 69
Fe(III) NH4+ Mn(II) Ag(I) NO2Ni(II) Cd(II) Bi(III)
4 1500 300 8 0.1 550 0.3 0.4
As(III) NO3Se(IV) SO42Cu(II) ClHg(II) PO43-
240 900 0.1 1700 6 240 2 250
By tolerance was meant the maximum concentration of coexisting substances causing a 5% signal increment.
Table 2. Quantitation-Related Data
a
film
detection limit/nM
linear range/nM
linear regression function/nM
r2
RSD/%, n)5
1a 2c 3d
90 40 20
300-940 90-320 40-320
∆I ) 0.288 + 3.2 × 10-3cb ∆I ) 0.803 + 1.83 × 10-2c ∆I ) 0.641 + 3.38 × 10-2c
0.994 0.987 0.996
2.59 3.79 1.99
Bare BK7 glass slide. b ∆I ) relative peak height. c Bare gold film. d Modified gold film.
Table 3. Recovery of Cr(VI) Measured from Various Synthetic Samples added/nM
coexisting ions/µM
found/nM
recovery/%, n ) 5a
RSD/%, n)5
60
Fe(III) 2, Pb(II) 5, Cu(II) 3, Zn(II) 3, Ni(II) 2, Ca(II) 30, As(I) 0.2 Fe(III) 3, Al(III) 20, Ca(II) 30, Cu(II) 2, Zn(II) 4, Cr(III) 1 Fe(III) 2, Ca(II) 40, As(III) 1, Mg(II) 40, Cd(II) 0.07, Bi(III) 0.1
62
96.3-103.4
2.85
58
97.1-104.9
2.98
98
98.2-103.6
1.89
60 100 a
The recoveries were obtained by adding 0.40 mL of sample solution and 0.40 mL of 80 nM standard solution of Cr(VI).
Table 4. Results for the Determination of Content of Cr(VI) in Real Samples LC-ICPMSa
this method
a
sample
Cr(VI), /nM
RSD/%, n ) 5
electroplating wastewater river water tap water
124
