Ratiometric Fluorescence Detection of Mercury Ions in Water by

Jan 26, 2012 - We present dye-doped polymer nanoparticles that are able to detect mercury in aqueous solution at parts per billion levels via fluoresc...
0 downloads 12 Views 2MB Size
Download PDF
Letter pubs.acs.org/ac

Ratiometric Fluorescence Detection of Mercury Ions in Water by Conjugated Polymer Nanoparticles Elizabeth S. Childress, Courtney A. Roberts, Desmarie Y. Sherwood, Clare L.M. LeGuyader, and Elizabeth J. Harbron* Department of Chemistry, The College of William and Mary, Williamsburg, Virginia 23187-8795, United States S Supporting Information *

ABSTRACT: We present dye-doped polymer nanoparticles that are able to detect mercury in aqueous solution at parts per billion levels via fluorescence resonance energy transfer (FRET). The nanoparticles are prepared by reprecipitation of highly fluorescent conjugated polymers in water and are stable in aqueous suspension. They are doped with rhodamine spirolactam dyes that are nonfluorescent until they encounter mercury ions, which promote an irreversible reaction that converts the dyes to fluorescent rhodamines. The rhodamine dyes act as FRET acceptors for the fluorescent nanoparticles, and the ratio of nanoparticle-to-rhodamine fluorescence intensities functions as a ratiometric fluorescence chemodosimeter for mercury. The light harvesting capability of the conjugated polymer nanoparticles enhances the fluorescence intensity of the rhodamine dyes by a factor of 10, enabling sensitive detection of mercury ions in water at levels as low as 0.7 parts per billion.

M

cence via fluorescence resonance energy transfer (FRET) from

ercury contamination in the environment and in living organisms continues to be a critical issue of concern on a global scale. With the toxic effects of mercury on ecosystems and health well-established,1−3 the need to detect mercury at extremely low concentrations persists. Detection schemes based on fluorescence spectroscopy are prevalent, due in part to the sensitivity of fluorescence compared to other optical techniques.4,5 Many fluorescent probes for mercury function as “turn-on” probes that are nonfluorescent until they encounter a mercury ion, which induces a change in the probe that renders it fluorescent. Such probes have achieved low detection limits but face the challenge of quantifying mercury concentration based solely on the fluorescence intensity of the probe, especially at very low mercury concentrations. Fluorescence intensity readings are also subject to misleading intensity fluctuations due to instrumental or environmental factors. This limitation can be addressed by ratiometric fluorescence measurements, which involve intensity readings at two wavelengths such that intensity fluctuations due to spurious factors are canceled out. However, development of ratiometric systems for mercury detection that operate in aqueous environments has lagged behind that of aqueous-based single wavelength probes.5 We present herein a fluorescence detection system for mercury ions that is sensitive, aqueous-based, and ratiometric and thus addresses the limitations of many existing systems. Conjugated polymer nanoparticles (CPNs) doped with a mercury-responsive rhodamine derivative enable aqueous detection of Hg2+ with enhanced probe intensity and twocolor, ratiometric measurements. The CPNs fluoresce greenyellow until the rhodamine derivatives on the particle surface encounter mercury and begin to exhibit orange-red fluores© 2012 American Chemical Society

the CPNs to the rhodamine dyes (Scheme 1). The FRETexcited dyes on the CPNs feature enhanced intensity compared Scheme 1

Received: January 3, 2012 Accepted: January 26, 2012 Published: January 26, 2012 1235

dx.doi.org/10.1021/ac300022y | Anal. Chem. 2012, 84, 1235−1239

Analytical Chemistry

Letter

to directly excited dyes due to the light-harvesting capabilities of the donor conjugated polymer chains. Able to detect mercury ions at concentrations below the U.S. EPA’s 2 parts per billion upper limit for drinking water,6 these doped CPNs function as a sensitive chemodosimeter for mercury in water and build on other recent examples of CPNs in aqueous sensing applications.7−10 The mercury-responsive CPNs are composed of the conjugated polymer poly[9,9-dioctylfluorenyl-2,7-diyl)-co-1,4benzo-{2,1′-3}-thiadiazole)] (PFBT) and a nonfluorescent rhodamine B spirolactam derivative (RB-SL) shown in Scheme 1. Spherical, surfactant-free, and stable in aqueous suspension, CPNs also possess brightness and photostability levels that far exceed those of typical small molecule fluorophores.11 Their outstanding photophysical properties are derived from the parent conjugated polymers, which are multichromophoric chains with high absorption cross sections and radiative rates.11 Here, CPNs are used as FRET donors to a fluorescent rhodamine B derivative (RB-Hg) that is formed when nonfluorescent RB-SL encounters a mercury ion. RB-SL and RB-Hg are analogues of dyes originated by Tae and co-workers, who determined that thiosemicarbazide-functionalized rhodamine spirolactams undergo an irreversible, mercury-promoted reaction to form 1,3,4-oxadiazole-functionalized rhodamine derivatives that are highly fluorescent.12,13 RB-SL is classified as a chemodosimeter rather than a chemosensor due to the irreversibility of this reaction.14 Analogous dyes have also been used by others as part of a fluorescent dyad for ratiometric mercury detection in solvent/water mixtures15,16 and for twophoton excitation of the rhodamine derivative in an acetonitrile/buffer mixture.17 These mercury-responsive dyes are exemplars of the rhodamine spirolactam family of dyes, which were pioneered for use in metal ion detection by Czarnik18 and are prized for their turn-on fluorescence in sensing applications.19,20 PFBT and RB-Hg were selected for use here because they possess strong spectral overlap, a prerequisite for efficient FRET. RB-SL does not absorb in the visible region of the spectrum and thus does not overlap the CPN fluorescence and cannot act as a FRET acceptor. In contrast, the absorbance of the acceptor RB-Hg (λmax = 568 nm) substantially overlaps the fluorescence of the donor PFBT CPNs (λmax = 537 nm, Figure 1). In addition to spectral overlap, the PFBT/RB-Hg donor−

R 0 = 0.211[κ 2n−4Q D

∫0

∞

FD(λ)εA (λ)λ4 dλ]1/6

(1)

where κ2 is an orientational factor generally assumed to be 2/3, n is the refractive index of the medium, QD is the quantum yield of the donor in the absence of an acceptor, FD is the fluorescence intensity of the donor, and εA is the extinction coefficient of the acceptor. As calculated according to eq 1, R0 for the PFBT/RB-Hg FRET pair is 4.8 nm. The doped CPNs are prepared by a reprecipitation method in which a dilute tetrahydrofuran (THF) solution of all CPN components is injected into sonicating water.23 After vacuum removal of the THF, a stable aqueous suspension of CPNs is obtained. Clafton et al. determined that the CPNs’ excellent stability is due to the formation of negatively charged, oxygencontaining chemical defects on the particle surface during CPN preparation.24 Previous work by others23 and us25 demonstrated that hydrophobic dyes that are included in the THF precursor solution are incorporated into the CPNs and do not partition into the water. Our work with photochromic dye dopants has also revealed that dyes on the CPN surface are able to undergo structural transformations in response to an external stimulus.25,26 Building on that work, we here used a THF precursor solution of PFBT, RB-SL, and poly(vinyl butyral-covinyl alcohol-co-vinyl acetate) (PVB-VA-VA, Scheme 1) to prepare the CPNs. CPNs prepared without PVB-VA-VA were mercury responsive but exhibited unexpected fluctuations in the quantum yield of the conjugated polymer that disrupted the ratiometric measurement. We attributed these fluctuations to the binding of mercury ions to the negatively charged, oxygencontaining groups on the CPN surface and found that they disappeared when optically transparent PVB-VA-VA was used as an additional stabilizer. PVB-VA-VA’s balance of hydrophobic and hydrophilic properties presumably allows it to be localized primarily on the CPN surface, blocking metal ion binding to the conjugated polymer. Chiu and co-workers previously demonstrated improved CPN properties with the inclusion of various polystyrene derivatives.8,21,27 Their work includes CPNs coated with a carboxyl-functionalized polystyrene that binds copper and iron ions, inducing CPN aggregation and concomitant fluorescence quenching.9 The size distribution of our CPNs was measured by dynamic light scattering, and 100% of the number-weighted distribution was 10.8 nm or smaller for both RB-SL-doped CPNs and control CPNs containing only PFBT and PVB-VA-VA. The measured CPN size corresponds to 3 PFBT polymer chains per particle based either on typical organic density28 or with reference to the hydrodynamic diameters of polystyrene standards.29,30 Figure 2 reveals how the absorbance (A) and fluorescence (B) spectra of the doped CPNs change as mercury ions are introduced to the aqueous system, first at nM concentrations and then with incremental increases up to a final concentration of 0.57 μM. Prior to mercury addition, the spectra are dominated by the PFBT CPNs, with λmax,abs = 460 nm and λmax,fl = 537 nm in agreement with literature values.21 Both absorbance and fluorescence spectra show a small contribution from RB-SL dyes that have been opened to a rhodamine form by impurities, a result consistent with the literature.13 The rhodamine B-impurity adducts can be distinguished from RBHg by their slightly blue-shifted absorbance and fluorescence spectra (λmax,abs = 565 vs 568 nm, λmax,fl = 586 vs 590 nm). Addition of small aliquots of mercury ions to the doped CPNs does not alter the PFBT absorbance but induces an increase in

Figure 1. Absorption (blue) and fluorescence (green) spectra of PFBT CPNs in aqueous suspension and absorption (red) and fluorescence (black) spectra of RB-Hg in water.

acceptor pair possesses other photophysical properties associated with efficient FRET including a moderately high quantum yield donor (0.3)21 and an acceptor with a large extinction coefficient (9.8 × 104 M−1cm−1).16 The Förster radius (R0), the donor−acceptor distance at which the energy transfer is 50% efficient, is given by eq 122 1236

dx.doi.org/10.1021/ac300022y | Anal. Chem. 2012, 84, 1235−1239

Analytical Chemistry

Letter

Figure 2. Absorption (A) and fluorescence (B) spectra of RB-SL/ PFBT CPNs in aqueous suspension as mercury ion concentration is increased from 0 to 0.57 μM by incremental addition of 1 × 10−5 M mercury(II) acetate. The inset shows an expansion of the RB-Hg absorbance.

Figure 3. (A) Ratio of RB-Hg to PFBT fluorescence intensities (I590/ I537) as a function of mercury ion concentration. Inset: Stern−Volmer plot of PFBT fluorescence quenching (λem = 537 nm) versus RB-Hg concentration expressed as number of quencher dyes per polymer chain. (B) Fluorescence intensity of RB-Hg at 590 nm when excited directly (triangles, λexc = 550 nm) and via FRET from PFBT (squares, λexc = 450 nm). The fluorescence intensities were corrected for emission from PFBT and RB-impurities at 590 nm, the difference in fluorimeter excitation intensities at 450 and 550 nm, and the small fraction of emission upon excitation at 450 nm that is due to direct excitation of RB-Hg.

the absorbance at 568 nm, consistent with the formation of RBHg. The concentration of RB-Hg can be calculated from the absorbance data: at the highest mercury ion concentration shown in Figure 2A, 0.57 μM, there are approximately 2 RB-Hg dyes per polymer chain or 6 per CPN. This small number of dyes has a tremendous impact on the fluorescence of the doped CPNs, as shown in Figure 2B. As mercury ions are added and RB-SL is converted to RB-Hg, the PFBT donor fluorescence is quenched while the RB-Hg acceptor fluorescence concomitantly increases in intensity. This change in fluorescence occurs with an isoemissive point at 562 nm, indicating that the PFBT CPNs and RB-Hg dyes are the only emissive species present. The sensitivity of the doped CPNs to parts per billion (ppb) levels of mercury is demonstrated in Figure 3A. The ratio of RB-Hg to PFBT fluorescence intensities (I590/I537) jumps noticeably upon the first addition of mercury, which corresponds to a concentration of 0.7 ppb. The U.S. EPA standard for mercury concentration in drinking water in 2 ppb,6 and the doped CPNs are clearly able to detect mercury both below and above this level. The intensity ratio continues to increase steeply up to nearly 15 ppb of mercury and then more gradually up to 85 ppb, where it levels off. The nonlinear ratiometric plot appears to exhibit saturation behavior due to a limited number of accessible RB-SL dyes that are able to react with mercury. Indeed, at a mercury concentration of 14.3 ppb, 67% of dyes that will ultimately react to form RB-Hg have already done so according to absorbance data. The more gradual change in the fluorescence ratio at mercury concentrations >15 ppb thus reflects the fact that the population of new RB-Hg dyes being formed is much smaller than the existing population. Absorbance data reveal that generation of new RB-Hg dyes essentially ceases at mercury concentrations at and above 85 ppb (data not shown). It thus appears likely that the population of mercury-accessible RB-SL dyes has been exhausted by this point and that the upper end of the concentration range over which the doped CPNs are responsive is limited by the population of accessible dyes on the CPNs. In an effort to increase this population, we experimented with higher RB-SL doping levels but found that the more heavily doped CPNs tended to aggregate. The CPNs described herein showed no evidence of aggregation over the mercury

concentration range studied, 0−0.57 μm. All experiments were conducted on as-prepared suspensions, and it might be possible to improve the detection limit by concentrating the CPNs, provided that they remain free from aggregation. The doped CPNs’ ability to detect less than 1 ppb of mercury compares very favorably to other ratiometric fluorescence systems that utilize a rhodamine spirolactam derivative as the mercury-responsive unit. Dyads with a covalently linked FRET donor and acceptor generally must be studied in organic solvent/water mixtures for solubility reasons and are unable to detect mercury concentrations much below 10 ppb.15,16,31 Like the doped CPNs presented here, macromolecular systems based on block copolymer micelles,32,33 polymeric nanoparticles,34 and inorganic nanoparticles35,36 are able to detect mercury in aqueous solution. With one exception,35 these systems have detection limits above 10 ppb. The fluorescence quenching of the PFBT CPN fluorescence by RB-Hg dyes can be evaluated by the Stern−Volmer equation (eq 2)

I0 = 1 + KSV [Q ] I

(2)

where I0 and I are the unquenched and quenched fluorescence intensities of the donor CPNs, KSV is the Stern−Volmer quenching constant, and [Q] is the concentration of RB-Hg quencher dyes as it increases throughout the mercury titration experiment.22 The linear Stern−Volmer plot (Figure 3A inset) shows I0/I versus [Q], which is expressed as the number of RBHg quencher dyes per polymer chain as calculated from absorbance data and the known polymer concentration. Analysis of the graph’s slope yields a KSV value of 0.36, which indicates that each RB-Hg dye quenches the emission of just over a third of a polymer chain. Given that the polymer has a 1237

dx.doi.org/10.1021/ac300022y | Anal. Chem. 2012, 84, 1235−1239

Analytical Chemistry

Letter

molecular weight of 99 000, each chain contains numerous chromophores. Therefore, each RB-Hg dye quenches multiple chromophores. That the KSV value is lower than in some other dye-doped CPNs reflects the fact that the quencher dyes in this case are found only on the CPN surface. While RB-SL dyes are distributed randomly throughout each CPN, only those dyes on the surface are exposed to mercury ions in aqueous solution and can undergo the transformation to RB-Hg. Quenching of chromophores deep inside the CPN by surface dyes is predicted to be far from efficient by FRET theory. Given that the Förster radius (R0) is 4.8 nm, the FRET efficiency (E, eq 322) is predicted to be 33% when the donor-acceptor distance (r) is 5.4 nm, which is the radius of the CPNs as measured by dynamic light scattering. Thus, the fact that CPN emission is never completely quenched by RB-Hg is consistent with particle size and donor−acceptor properties.

E= 1+

1

Figure 4. Ratio of RB-Hg to PFBT fluorescence intensities (I590/I537) in the presence of 1 μM Hg2+, Ag+, Co2+, Mn2+, Cr2+, Na+, Pb2+, Cu2+, Ba2+, Mg2+, or Ni+ ions. Each intensity ratio was divided by the initial ratio for that sample in the absence of metal ions to normalize for slight sample-to-sample fluctuations in the initial ratio. Inset: Fluorescence spectra of undoped control CPNs in the presence (dotted) and absence (solid) of 1 μM Hg2+ ions.

⎛ r ⎞6 ⎜ ⎟ ⎝ R0 ⎠

fluctuations. Collectively, these features demonstrate the continued promise of CPNs for chemical sensing applications.

■

(3)

The quenching of multiple PFBT chromophores by each RBHg dye yields significant amplification of the RB-Hg fluorescence intensity. Indeed, it is this amplification that enables the very sensitive detection of mercury ions in this system. To quantify this effect, the RB-Hg fluorescence intensity was measured twice at each concentration of mercury ions studied: once upon direct excitation of the dye (λexc = 550 nm) and again upon FRET via excitation of the PFBT CPNs (λexc = 450 nm). As shown in Figure 3B, the fluorescence intensity of the RB-Hg dyes when excited by FRET from the PFBT donor fluorophores is dramatically higher than when excited directly. The RB-Hg intensity is enhanced by a factor of 10 at low mercury concentrations (