Multicolor Quantum Dot Encoding for Polymeric Particle-Based Optical

Apr 17, 2007 - exchange. Polymeric microspheres are generated under mild, nonreactive conditions with a particle caster that breaks down a polymer str...
0 downloads 0 Views 248KB Size
Anal. Chem. 2007, 79, 3716-3723

Multicolor Quantum Dot Encoding for Polymeric Particle-Based Optical Ion Sensors Chao Xu and Eric Bakker*

Department of Chemistry, Purdue University, 560 Oval Drive, West Lafayette, Indiana 47907

Multicolor quantum dot-encoded polymeric microspheres are prepared with controllable and uniform doping levels that function as chemical sensors on the basis of bulk optode theory. TOP/TOPO-capped CdSe quantum dots and CdTe quantum dots capped with CdS (λem ) 610 and 700 nm, λex ) 510 nm) are blended with a THF solution of poly(methyl methacrylate-co-decyl methacrylate), poly(n-butylacrylate), or poly(vinyl chloride) plasticized with bis(2-ethylhexyl) sebacate without a need for ligand exchange. Polymeric microspheres are generated under mild, nonreactive conditions with a particle caster that breaks down a polymer stream containing the quantum dots into fine droplets by the vibration of a piezocrystal. The resulting microspheres exhibit uniform size and fluorescence emission intensities. Fluorescent bar codes are obtained by subsequent doping of two quantum dots with different colors and mass ratios into the microspheres. A linear relationship is found between the readout fluorescence ratio of the two types of nanocrystals and the mixing ratio. Quantum dot-encoded ion sensing optode microspheres are prepared by simultaneous doping of sodium ionophore X, chromoionophore II, a lipophilic tetraphenylborate cation exchanger, and TOPOcapped CdSe/CdS quantum dot as the fluorescent label. A net positive charge of the quantum dots is found to induce an anion-exchange effect on the sensor function, and therefore, an increased concentration of the lipophilic cation exchanger is required to achieve proper ion sensing properties. The modified quantum dot-labeled sodium sensing microspheres show satisfactory sodium response between 10-4 and 0.1 M at pH 4.8, with excellent selectivity toward common interferences. The amount of the carried positive charges of the CdSe quantum dots is estimated as 2.8 µmol/g of quantum dots used in this study. As sensor miniaturization continues to thrive, micrometer-sized beads carrying spectroscopic signatures have gained growing interest for clinical diagnosis, drug screening, and multiplexed bioassays, because they provide a highly flexible platform on the micrometer scale that can be simultaneously functionalized by many types of chemistry in a cost-effective manner.1 Reagents as well as the analyte may be deposited more evenly on the surface * To whom correspondence should be addressed. E-mail: [email protected]. (1) Epstein, J. R.; Lee, M.; Walt, D. R. Anal. Chem. 2002, 74, 1836-1840.

3716 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

of the spherical sensors than on larger substrates, which improves reliability and reproducibility. The required incubation time is shortened, and the detection limit of the analyte may be improved. Rapid readout can be achieved with a microarray or a flow system. Porous materials such as silica gels, and polymeric materials such as polyacrylates and plasticized PVC are often used to prepare such microspheres.2-5 For multiplexed assays, it is advantageous to label the microbeads for effective signal classification and further improve the sensor accuracy. However, this is difficult to achieve with conventional dyes because of the relatively narrow excitation windows, broad emission spectra, and risk of photobleaching. Semiconductor quantum dots (QDs)6 possess size-dependent absorption and emission properties and thus can be tuned for a specific wavelength of interest. QDs have narrow fluorescence emission spectra that are independent of the excitation wavelength, making it convenient to incorporate several kinds of QDs for multiple colors. Moreover, QDs are highly stable against photobleaching, and different QDs can be excited with a single light source, making quantum dots an ideal fluorescent label. Two research groups7,8 in the United States independently reported QD-conjugated proteins for cell staining in 1998, leading to many papers published in biological applications of water-soluble quantum dots. Incorporating quantum dots into the microsphere moiety can yield tremendous advantages of miniaturized sensors with highly stable fluorescence labels. Quantum dots that are capped with organic ligands such as trioctylphosphine (TOP) have a hydrophobic nature and are therefore compatible with lipophilic polymers. However, the fluorescence of TOP/TOPO-capped QDs is easily quenched by free radicals that are commonly used in polymerization.9 Therefore, ligand exchange or cross-linking are often needed to protect the QDs during the polymerization. For instance, Alivisatos’ group reported a composite with a poly(2) Albert, K. J.; Walt, D. R.; Gill, D. S.; Pearce, T. C. Anal. Chem. 2001, 73, 2501-2508. (3) Buck, S. M.; Xu, H.; Brasuel, M.; Philbert, M. A.; Kopelman, R. Talanta 2004, 63, 41-59. (4) Peper, S.; Ceresa, A.; Qin, Y.; Bakker, E. Anal. Chim. Acta 2003, 500, 127136. (5) Xu, C.; Wygladacz, K.; Qin, Y.; Retter, R.; Bell, M.; Bakker, E. Anal. Chim. Acta 2005, 537, 135-143. (6) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226-13239. (7) Chan, W. C.; Nie, S. Science 1998, 281, 2016-2018. (8) Bruchez, M.; Jr.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Science 1998, 281, 2013-2016. (9) Sheng, W.; Kim, S.; Lee, J.; Kim, S.-W.; Jensen, K.; Bawendi, M. G. Langmuir 2006, 22, 3782-3790. 10.1021/ac0701233 CCC: $37.00

© 2007 American Chemical Society Published on Web 04/17/2007

thiophene conductive polymer and CdSe nanorods and fabricated a dye-sensitized photovoltaic device.10 Lee et al. prepared a crosslinked QD-polymer composite for a solid light-emitting diode by radical polymerization of a mixture containing CdSe/ZnS quantum dots with TOP and the lauryl methacrylate monomer.11 Emrick et al. reported on solvent cast films containing CdSe QDs by modifying the QD surface with a p-vinylbenzylDOPO, a phosphine derivative.12 In term of microbead fabrication, polymer swelling and suspension polymerization as well as emulsion polymerization have been reported. The group of Nie published QD-encoded polystyrene microbeads by swelling micrometer-sized polystyrene beads in a mixture of chloroform and propanol. The microbead surface was consequently functionalized with streptavidin and conjugated with oligonucleotide probes.13 The same group also doped multicolor QDs in mesoporous silica beads and performed two-color flow cytometry for fast fluorescence readout.14,15 In a work reported by O’Brien et al., hexyadecylamine-capped CdSe QDs went through a ligand exchange to be capped with several organophosphine derivatives, and QD-polystyrene beads were prepared by suspension polymerization in the presence of additional ligands.16 Li et al. reported on polystyrene beads encoded with CdS quantum dots by suspension polymerization.17 The group of Bawendi modified the QD surface with an oligomeric ligand and prepared polystyrene-QD beads in situ under different loading conditions.9 Joumaa et al. studied the synthesis of submicrometer latex particles via miniemulsion polymerization with either TOPO-coated or vinyl-functionalized QDs and investigated the effect of QD concentrations as well as surfactants.18 The QD-labeled microspheres prepared by in situ polymerization suffer to a large extent from nonuniform distribution of quantum dots, accompanied by poor reproducibility in size and fluorescence signals. In situ polymerization also requires long preparation time and sometimes additional coating steps.19 The QD-encoded microspheres prepared by swelling the mesoporous materials such as latex or silica usually yield more uniform beads,15 but the reproducibility is dictated by the manufacturing of the mesoporous beads (standard deviation 5-15%), and multiple steps were involved prior to doping. For multiplexed bioassays, one would prefer to achieve a simultaneous loading of the QD labels with other sensing components. The QDs are used here as fluorescent labels in polymer-based sensors with fluorescence transduction in view of multiplexed sensing. Ion-selective bulk optodes are a unique class of chemical sensors for measuring the activity of ionic species.20-23 Sharing a (10) Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295, 24252427. (11) Lee, J.; Sundar, V. C.; Heine, J. R.; Bawendi, M. G.; Jensen, K. F. Adv. Mater. 2000, 12, 1102-1105. (12) Skaff, H.; Ilker, M. F.; Coughlin, E. B.; Emrick, T. J. Am. Chem. Soc. 2002, 124, 5729-5733. (13) Han, M.; Gao, X.; Su, J. Z.; Nie, S. Nat. Biotechnol. 2001, 19, 631-635. (14) Gao, X.; Nie, S. J. Phys. Chem. B 2003, 107, 11575-11578. (15) Gao, X.; Cui, Y.; Levenson, R. M.; Chung, L. W. K.; Nie, S. Nat. Biotechnol. 2004, 22, 969-976. (16) O’Brien, P.; Cummins, S. S.; Darcy, D.; Dearden, A.; Masala, O.; Pickett, N. L.; Ryley, S.; Sutherland, A. J. Chem. Commun. 2003, 2532-2533. (17) Li, Y.; Liu, E. C. Y.; Pickett, N.; Skabara, P. J.; Cummins, S. S.; Ryley, S.; Sutherland, A. J.; O’Brien, P. J. Mater. Chem. 2005, 15, 1238-1243. (18) Joumaa, N.; Lansalot, M.; Theretz, A.; Elaissari, A.; Sukhanova, A.; Artemyev, M.; Nabiev, I.; Cohen, J. H. M. Langmuir 2006, 22, 1810-1816. (19) Yang, X.; Zhang, Y. Langmuir 2004, 20, 6071-6073. (20) Bakker, E.; Buehlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083-3132.

number of similarities with ion-selective electrodes in terms of materials and response principles, bulk optodes are based on a true equilibration with the bulk sample.20 In addition to the ionophore and the ion exchanger in the polymeric matrix, a bulk optode also involves a reference ionophore (often a lipophilic pH indicator, or the so-called H+-chromoionophore or H+-fluoroionophore). The equilibrium of the competitive or collaborative extraction between the primary analyte ion and protons into the polymeric phase is characterized by the degree of protonation of the chromoionophore. The measuring range of the bulk optode sensors can be tuned by choosing an appropriate sensor composition and the basicity of the H+-chromoionophore. Recently, miniaturized ion-selective optodes have been prepared in our group in the form of micrometer-scale beads, with improvement in the response time and lower detection limit, showing great promise for applications in trace level analysis and clinical diagnostics.24,25 These sensing beads have also been explored in suspension array technologies such as analytical flow cytometry,26 where multicolor quantum dot encoding would be highly beneficial. In this paper, we report on a simple method to prepare QDencoded microspheres with controllable and uniform doping, which is compatible with microsphere ion sensor materials. We apply TOP/TOPO-capped QDs directly without ligand exchange and use a sonic particle casting method for formation of polymeric microspheres. This recently developed technique adopts microfluidics and piezoeffects to break a thin polymer stream surrounded by an aqueous sheath flow into uniform droplets.27 Our group has developed several types of fluorescent microspheres with this type of sonic particle caster for the sensing of ions and neutral species.24-26 Since no chemical reaction is involved during the casting process, the risk of deteriorating the optical properties of the doped components is minimized. EXPERIMENTAL SECTION Materials. Selenium (99.99%), tellurium (99.9%), TOP (90%), tri-n-octylphosphine oxide (TOPO, 90%), and n-tetradecylphosphonic acid (99%) were purchased from Alfa Aesar (Ward Hill, MA). Hexadecylamine (95%) and cadmium acetate were obtained from Sigma-Aldrich (St. Louis, MO). The chromoionophore 9-dimethylamino-5-[4-(16-butyl-2,14-dioxo-3,15-dioxaeicosyl)phenylimino]benzo[a]phenoxazine (ETH 2439), the lipophilic cation exchanger tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (NaTFPB), the plasticizer bis(2-ethylhexyl) sebacate (DOS), poly(vinyl chloride) (PVC), poly(urethane) (Tecoflex), dimethyl phthalate (DMP), and tetrahydrofuran (THF) were Selectophore quality or the highest grade available from Fluka (Milwaukee, WI). The reference dye 1,1′′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiIC18) was obtained from Molecular Probes (Eugene, OR). Dichloromethane, chloroform, anhydrous metha(21) Bakker, E.; Simon, W. Anal. Chem. 1992, 64, 1805-1812. (22) Shortreed, M.; Bakker, E.; Kopelman, R. Anal. Chem. 1996, 68, 26562662. (23) Badr, I. H. A.; Meyerhoff, M. E. Anal. Chem. 2005, 77, 6719-6728. (24) Wygladacz, K.; Radu, A.; Xu, C.; Qin, Y.; Bakker, E. Anal. Chem. 2005, 77, 4706-4712. (25) Telting-Diaz, M.; Bakker, E. Anal. Chem. 2002, 74, 5251-5256. (26) Retter, R.; Peper, S.; Bell, M.; Tsagkatakis, I.; Bakker, E. Anal. Chem. 2002, 74, 5420-5425. (27) Tsagkatakis, I.; Peper, S.; Retter, R.; Bell, M.; Bakker, E. Anal. Chem. 2001, 73, 6083-6087.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3717

nol, and polystyrene were purchased from Fisher Scientific (Fair Lawn, NJ). Cyclohexanone (99.8%) was purchased from SigmaAldrich, while poly(n-butyl acrylate) and poly(ethylene glycol) (PEG) were from Polysciences, Inc. (Warrington, PA). The copolymerized methyl methacrylate-co-decyl methacrylate (MMADMA) was synthesized in-house according to the literature with slight modification.28 Sodium chloride, calcium chloride, potassium chloride, magnesium chloride, magnesium acetate, and 2-morpholinoethanesulfonic acid were puriss grade from Fluka. All aqueous samples were prepared with Nanopure water (18.2 MΩ cm-1). Synthesis of TOPO-Capped Core-Shell Quantum Dots. CdSe (QD1)and CdTe (QD2) quantum dots with CdS shells, capped with trioctylphosphine, were synthesized according to the literature.29,30 The reactions were carried out under nitrogen atmosphere, and the purified quantum dots were redispersed in chloroform after the last washing step in methanol and were left for evaporation in nitrogen atmosphere for 48 h. The resulting dry CdSe and CdTe quantum dots with CdS shell exhibited an orange and dark reddish brown color, respectively. Polymeric Film Preparation. In a glass vial, 0.5 mg of QD1 was mixed with 1 mL of the tested solvent to choose the suitable solvent for film preparation. QD1 of 0.5 mg was then dissolved together with a polymer matrix to be tested (45 mg) in 1 mL of solvent for 20 min. A 50-µL aliquot of the mixture was carefully deposited onto a 2.2-cm-wide microscope cover slide. The films were covered for 30 min for the solvent to fully evaporate. Microsphere Preparation. Polymeric microspheres were generated by an in-house particle-caster previously described27 with modification of some casting parameters. First, all components were dissolved in THF and diluted with 50 mL of dichloromethane and 0.5 mL of xylenes to prepare the casting cocktail, which was filtered with a 0.45-µm syringe filter before casting. The cocktail was then delivered to a ceramic tip (46 or 63 µm) by a syringe pump (model 100, Stoelting, Wood Dale, IL). A piezoelectric crystal was mounted above the tip orifice, whose vibration was controlled by a frequency generator (BK Precision, Placentia, CA). The polymer stream was broken into tiny droplets under the influence of the piezocrystal and was surrounded by a deionized water sheath upon reaching the mixing chamber. A solution of 3% PEG was added to the 25-mL receiving vial in order to avoid excessive clumping of the microspheres. For all castings involving QDs, the microspheres settled to the bottom of the vials within 1 min. The suspension was quickly decanted with a glass pipet, and fresh deionized water was added to replace the mixed solvent. The syringe pump was operated at 0.3 mL/min. A flow rate of the water stream of 50 mL/min was applied for rapid removal of the solvent from the polymer and to stabilize the formed microspheres. The frequency generator was operated at 10-15 kHz. For the preparation of QD-bar code microspheres containing only the quantum dots and the polymer, THF solutions of two individual types of quantum dots (QD1 and QD2) were mixed at different ratios and were subsequently blended with the polymer matrix to make a total solute mass of 90 mg in the casting cocktail. (28) Qin, Y.; Peper, S.; Bakker, E. Electroanalysis 2002, 14, 1375-1381. (29) Peng, Z. A.; Peng, X. J. Am. Chem. Soc. 2001, 123, 183-184. (30) Mekis, I.; Talapin, D. V.; Kornowski, A.; Haase, M.; Weller, H. J. Phys. Chem. B 2003, 107, 7454-7462.

3718

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

For the QD-encoded sodium ion-sensing microspheres, several cocktail compositions varying in the amount of ion exchanger were prepared to estimate the apparent charge for the CdSe/CdS quantum dots. They contained 90 mg of total solute mass of Na ionophore X (19.4 mmol kg-1), NaTFPB (8.8, 14.1, 18 mmol kg-1, respectively), ETH 2439 (5.2 mmol kg-1), CdSe/CdS 1.9 wt %, and PVC/DOS (1:2 mass ratio). The non-QD sodium ion-sensing microspheres were prepared for comparison, which contained Na ionophore X (19.2 mmol kg-1), NaTFPB (8.9 mmol kg-1), ETH 2439 (5.4 mmol kg-1), and the reference dye DiIC18 (∼0.1 mmol kg-1), with the rest being PVC/DOS (1:2 mass ratio). In order to minimize the photobleaching of the fluoroionophore in this study, the cocktail was covered with aluminum foil during casting, and the microspheres were kept in the dark before fluorescent measurement. Sample Preparation. For the QD-encoded sodium-selective microspheres, sodium response curves for the ion-sensing microspheres were recorded in 100 mL of magnesium acetate buffers at pH 4.8, containing different levels of sodium chloride. The separate solutions method was used for the selectivity measurements.21 Full protonation and deprotonation of the microparticles was achieved using 10 mM HCl and 10 mM NaOH, respectively. A 0.1-mL aliquot of microsphere suspension was transferred from the receiving vial to a 2.2-cm-wide microscope slide placed in a clean weighing dish (Fisher Scientific). The polymeric microspheres adhered to the glass slide after 20 min and remained on the slide throughout the rest of experiment. Approximately 25 mL of deionized water was then slowly added to each weighing dish for further curing of the microspheres prior to the exposure to the sample solution. Each glass slide with the immobilized microspheres was carefully transferred from the curing container to a sample solution and was conditioned for 20 min before the measurement. The microspheres on each glass slide were used for one sample only; for each data point, the average response was calculated from five to six randomly chosen microspheres on the same slide. Characterization. UV-visible measurements were performed on an HP photodiarray spectrophotometer (HP 8452A) with a quartz cuvette. The absorbances of the core and core-shell quantum dots were measured directly after the core QD reaction slurry was quenched by toluene or after the shell formation, respectively. A gastight glass syringe with a 15-cm-long needle was used to obtain ∼0.2 mL of sample from the nitrogen-protected container. The sample was then diluted with ∼1 mL of chloroform for absorbance measurements. A FEI/Philips CM-100 transmission electron microscope was used to estimate the size of the synthesized quantum dots. Purified QD in dry form was redispersed in chloroform and deposited on the surface of the copper grid with a sprayer. The transmission electron microscope was operated at 82 kV, with a magnification of 52000×. Fluorescence measurements were performed on a Nikon Eclipse E400 microscope (510-560-nm excitation filter) with an epifluorescence attachment (Southern Micro Instruments, Marietta, GA), and an Osram HBO 100W/2 mercury arc lamp as the excitation source. Two COHU CCD cameras were used in combination with the Pariss imaging spectrometer (LightForm, Inc., Belle Mead, NJ) for acquisition of fluorescence spectra and

Figure 1. (A) Normalized fluorescence emissions of the quantum dots synthesized and used here for microsphere sensors (λex ) 510 nm). QD1 denotes CdSe/CdS quantum dots and QD2 for CdTe/CdS quantum dots. Both QDs were capped with TOPO. (B) A TEM micrograph obtained for CdSe/CdS quantum dots (QD1).

images. A motorized stage (Prior Optiscan ES9, Fulbourn, Cambs, U.K.) was equipped with the system. The same exposure time (500 ms) was applied once in all the microsphere readouts. For the ion-sensing microspheres, the ratios of the fluorescence peak intensities from ETH 2439 to that from the internal reference dye, (either DiIC18 or CdSe/CdS quantum dots) was used in order to calculate the degree of protonation of the microsphere fluoroionophore. RESULTS AND DISCUSSION The development of QD-encapsulated polymeric sensing beads faces several challenges in terms of chemical compatibility and retaining nanocrystal optical properties. Most reported research with in situ polymerization involved a ligand-exchange step prior to polymerization in order to protect the quantum dots from intensity loss. Our goal was to circumvent this relatively complex step and to create QD-encoded polymeric microspheres in a single nonreactive step under mild conditions. Here, CdSe and CdTe quantum dots with CdS shells were synthesized. The one-pot synthesis of the core-shell quantum dots applied here is a relatively simple procedure compared to other approaches. The resulting quantum dots were capped with TOPO. Figure 1A shows the relative fluorescence emission spectra of two types of quantum dots, CdSe and CdTe with CdS shell (denoted here as QD1 and QD2), in the solvent chloroform. Emission maximums were found at 610 and 700 nm, respectively, when excited at 510 nm. The full width at half-maximum was found as ∼30 nm for QD1 and ∼45 nm for QD2, respectively, which is in good agreement with the literature.29-31 The emission spectra were found to be sharp and to have no apparent overlap. As shown in the transmission electron micrograph of QD1, the average size of the synthesized CdSe/ CdS (QD1) quantum dots was ∼5.5 nm, again corresponding well with literature reports. Purified QD1 were mixed with a variety of solvents that are commonly used in the casting of ion-sensing polymer-based microspheres. The fluorescence emission spectra recorded at the identical concentration showed that relatively nonpolar solvents such as tetrahydrofuran, chloroform, and dichloromethane yield comparable intensity ratios, while ethyl acetate causes precipitation (31) Burda, C.; Chen, X.; Narayanan, R.; El-Sayed, M. A. Chem. Rev. 2005, 105, 1025-1102.

and color change of QD1 over a time period of 30 min. Further experiments used THF and dichloromethane as solvents. A THF solution containing QD1 was also dispersed in NPOE, DOS, and DMP, common plasticizers for polymer membrane-based chemical sensors such as PVC-based ion-selective membranes and bulk optode films. While DMP and NPOE gave immediate phase separation, DOS only caused a slight change of color; the quantum dots initially mixed well but did start to precipitate after a few hours. Self-plasticized polymer matrixes, such as poly(n-butyl acrylate) (Tg ) -50 °C), copolymerized poly(methyl methacrylate-cododecyl methacrylate) (poly(MMA-DMA); Tg ) -10 °C), DOS plasticized polymers, i.e., polystyrene (containing 5-10 wt % DOS), polyurethane (5-10 wt % DOS), and PVC/DOS (1:2 by mass) were explored for preparing QD-encapsulated polymer films by solvent casting. Under the same excitation conditions, the films were compared in terms of their visible appearance under the microscope and the observed emission intensity. Poly(n-butyl acrylate) films tended to easily peel off from the glass slide due to their high surface tension and were thus difficult to characterize in this manner. Polyurethane and polystyrene films appeared to exhibit significant surface roughness and showed isolated regions of bright dots, indicating uneven QD distribution and local aggregation. Films made from PVC and poly(MMA-DMA) appeared to have a uniform distribution of quantum dots and no apparent aggregation was observed in either film, making these promising candidates for microsphere preparation. These two types of polymers were subsequently used for the preparation of monochromatic QD-polymer microspheres (doped with just one type of QDs) with an in-house-built particle caster of the type reported earlier.24 The polymer solution is delivered through a capillary on a metal support and forms a vertical stream when exiting the ceramic tip at the mixing chamber. The stream cone is surrounded and stabilized by a water sheath flow and is cut into tiny droplets by the vibration of a piezocrystal. The organic solvent partitions into a water reservoir during a curing step, forming polymeric particles from the individual droplets. Poly(MMA-DMA) microspheres were successfully doped with each QD, as demonstrated in Figure 2A for QD1 and Figure B for QD2. The two images on Figure 2, left, are 3-D renderings of a single monochromatic microsphere containing either QD1 or QD2. The particle in (A) shows a fluorescence emission maximum at 610 nm for QD1 and the one in (B) a maximum of 700 nm, both in very good agreement with the optical properties of the quantum dot solutions in Figure 1. The size of the microspheres in (A) is ∼ 9.5 µm and in (B) ∼10 µm, estimated from the 3-D images and converted to scale from the size of the pixels in the spatially resolved image of the CCD detector.32 The actual fluorescent images of the monochromatic QD microspheres are shown on the right side in Figure 2. The particles appear to be smooth and spherical, with uniform size and brightness. No leaching of QDs was observed, and the receiving solution showed only negligible background fluorescence. In contrast to the difficulty in casting poly(n-butyl acrylate) films, uniform mono-QD microspheres were successfully prepared with poly(n-butyl acrylate). QD1 was doped into PVC/DOS for microspheres as well, and the resulted loading efficiency were similar to those for poly(MMA-DMA). This is (32) Wygladacz, K.; Bakker, E. Anal. Chim. Acta 2005, 532, 61-69.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3719

Figure 2. 3-D rendering of the fluorescence emission spectra collected from a single poly(MMA-DMA) microsphere doped with one type of QD and the spatially resolved fluorescent image of several microspheres, (A) with QD1 and (B) with QD2.

evidence that the casting method we have applied here is useful for a wide range of polymer matrixes. QD-encoded microspheres may be directly fabricated from a blend of TOPO-capped QDs and a lipophilic polymer matrix. Unlike QD-latex microspheres obtained from suspension or emulsion polymerization, where phase separation occurred and the resulting microspheres had irregular shapes and sizes,18 the QD-encoded microspheres generated here showed uniform intensity and reproducible sizes. The QDs were found to be uniformly doped across the microspheres since the 3-D renderings resemble those obtained from microspheres that were doped with organic dyes in earlier reports.24 Using the casting parameters that were optimized for the preparation of single QD-encoded microspheres, poly(MMADMA) microparticles were prepared that contained both QD1 and QD2. Different loading ratios were applied in each casting, and each batch was optically characterized by fluorescence microscopy. A representative fluorescence emission spectrum obtained from a single microsphere from each batch is shown in Figure 3. In all spectra obtained here, there was no shift in the emission maximums of the two QDs. For each microsphere tested, the 3720

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

Figure 3. Relative fluorescence emissions for QD-encoded poly(MMA-DMA) microspheres containing both QD1 and QD2 at different mass ratios.

the content of the lipophilic ion exchanger. An excess amount of ionophore L is usually used to ensure the availability of the free ionophore. The degree of protonation of the chromoionophore is dependent on the extraction of both ions into the organic phase. The resulting sensor response may be described by eq 2,21 where the charge and mass balances in the organic phase are also incorporated. Equation 3 relates the normalized signal, R, to fluorescence intensities.27

aI ) RT- - (1 - R)IndT ZI R aH 1-R {LT - (RT- - (1 - R)IndT)(nI/zI)}nI

(

(zIKex)-1

)

(2) Figure 4. Plot of the apparent intensity ratio from the fluorescence emission spectra of the prepared microspheres containing two QDs vs their mass doping ratio in the casting cocktail.

with

R) observed intensity ratio at the two fluorescence emission maximums, 610 and 700 nm) were plotted versus the mixing ratio by mass (Figure 4). Each data point shown is the average of fluorescence intensities from five to six microspheres that were prepared from the same casting cocktail composition. As expected, a linear relationship between the readout ratio and the doping ratio is observed (R2 ) 0.97). Note that a significant excess of QD2 is required to yield the same fluorescence intensity as QD1, owing to differences in the quantum yield. Subsequent to achieving controllable doping of multicolor quantum dots into polymeric microspheres, the technology was applied to functional chemical-sensing beads based on bulk optode principles. Ion-selective bulk optodes possess selectivity toward target ions and are easily miniaturizable, and bulk optode microspheres were recently reported with ultratrace detection limits owing to an up to ∼1 billion-fold preconcentration factor of these miniature extractors.24 Accurate doping with multicolor quantum dots would greatly facilitate their implementation in suspension array technologies, with the doping levels giving the necessary fluorescent bar code signature for proper identification of the sensing chemistry. Ion-selective bulk optodes are known to function on the basis of a competitive or cooperative extraction of two ions into a hydrophobic polymeric phase containing all relevant sensing ingredients. For a cation-selective bulk optode based on electrically neutral carriers L that selectively bind to the primary ion Iz+, the following equilibrium is generally used to describe the ionexchange process:21

Iz+(aq) + nL(org) + zHInd+(org) + zR-(org) ) ILnz+(org) + zInd(org) + zR-(org) + zH+(aq) (1) Here two ionophores (ion carriers) are involved in the competitive extraction between two ions. The ionophore L binds to the primary ion Iz+ with a binding stoichiometry of n, and the fluoroionophore (H+-chromoionophore) Ind is selective to H+. A lipophilic cation exchanger, R-, is present to provide the necessary counterions of the extracted cations in the polymeric phase. The total concentration of ions entering the organic phase is dictated by

(

Rpro - R [Ind] )1+ IndT R - Rdep

)

-1

(3)

The symbol R is the mole fraction of unprotonated fluoroionophore and is related to the observed fluorescence: R, Rpro, and Rdep are the fluorescence intensity ratios at two wavelengths, for a giving equilibrium and at the maximum protonation and the maximum deprotonation of the fluoroionophore, respectively. The subscript T denotes total concentrations; Kex is the equilibrium constant that describes eq 1 and is a function of the acidity constant of the chromoionophore, the lipophilicity of the primary ion, the charge of the primary ion, and the formation constant for the complex between primary ion and ionophore. Choosing a chromoionophore with a suitable acidity is often useful to shift the measuring range of such bulk optode sensors.20 Established sodium-selective microsphere bulk optodes were chosen as a model example to demonstrate the feasibility of implementing TOPO-capped quantum dots in an ion-sensing system. Although polystyrene was used for QD-encoded microspheres in several literature reports, this matrix is unsuitable for bulk optodes that require a well-defined equilibration step and sufficient solubility of all sensing ingredients. PVC/DOS is a common sensing material and was compatible with quantum dots (see above) and thus was explored here as the matrix of choice for QD-encoded ion-sensing microspheres. Na ionophore X, a calix[4]arene-based ionophore was chosen because of the high selectivity for the sodium ion.22,32 Sodium-selective microsphere optodes were prepared with the Nile Blue derivative ETH 2439 because it possesses a single emission peak at 693 nm, giving less spectral overlap with the quantum dots of interest than dual emission dyes such as ETH 5294. Measurements were performed at pH 4.8 because of the lower basicity of ETH 2439 compared to ETH 5294.33 Ion-sensing microspheres were initially prepared with 19.2 mmol kg-1 ionophore, 8.8 mmol kg-1 cation exchanger, and 5.4 mmol kg-1 fluoroionophore in PVC/DOS and 1.9 wt % QD1. The resulting microspheres had reproducible sizes and spherical shapes as with the nonactive QD-bar code microspheres described above. However, the labeled microsphere ion sensors did not (33) Qin, Y.; Bakker, E. Talanta 2002, 58, 909-918.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3721

Figure 5. Fluorescence emission spectra from QD-encoded sodiumselective microsphere optodes based on plasticized PVC, in response to different sodium activities in acetate buffers at pH 4.8. The microsphere cocktail contained ETH 2439 as the chromoionophore and QD1 as the internal reference. The 10 mM HCl and NaOH were used to record the spectra of the fully protonated or nonprotonated form of the chromoionophore.

respond satisfactorily to all tested sodium concentrations, and it was difficult to fully deprotonate the fluoroionophore at sodium levels higher than 1 mM. Yet, clean fluorescent emission spectra were obtained without signal overlap. A control sensor was prepared that contained all sensing components at the same concentration, except replacing QD1 with the reference dye DiIC18 at ∼0.1 mmol kg-1, and was found to respond to sodium (10-5-1 M) satisfactorily at pH 4.8. This suggested that the quantum dots are not as chemically inert as one might have anticipated. Quantum dots may possess net charges that alter the charge balance condition in bulk optode chemistries. Indeed, Krauss et al. performed electrostatic force microscopy in dry air at room temperature and observed that positive electrostatic charge developed on a portion of CdSe QDs,34,35 the value of which seemed to be nonuniform for individual quantum dots. If quantum dots possess anion-exchange properties, an increased concentration of cation exchanger NaTFPB may be needed to eliminate this type of interference. To quantify the effect of the electrostatic charge of QDs, the response of functional QDlabeled ion-sensing microspheres is then theoretically modeled with an apparent ion-exchanger concentration, RT′, instead of the total concentration of NaTFPB used in the cocktail. Indeed, satisfactory sodium responses were observed with modified compositions. Figure 5 shows the response of the sodium-selective fluorescent microsphere sensors labeled with CdSe/CdS quantum dots capped with TOPO (QD1) that contained lipophilic cation exchanger at 14.1 mmol kg-1. The two emission peak intensities were calculated for the degree of protonation of ETH 2439 and plotted in the sodium response curve shown in Figure 6. Each data point represents the average value from five to six microspheres with the error bars indicating the standard deviations. The theoretical curve was derived from eq 2. It was estimated from the curve that logKex ) -3.3 (data summarized in Table 1). (34) Krauss, T. D.; Brus, L. E. Phys. Rev. Lett. 1999, 83, 4840-4843. (35) Krauss, T. D.; O’Brien, S.; Brus, L. E. J. Phys. Chem. B 2001, 105, 17251733.

3722 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

Figure 6. Sodium response curve with selectivity data obtained from the sodium-selective microsphere bulk optodes labeled with QD1 as described Figure 5. Each data point represents the average response from 5 to 6 microspheres equilibrated in the sample buffer solution. The separate solutions method was used for selectivity measurements.21 Table 1. Selectivity Coefficients for the Sodium Bulk Optode Microspheres Described Here at pH 4.8 and Comparison to Corresponding Sodium Bulk Optode Microspheres Containing an Internal Reference Dye DilC18 a Na+-optode microspheres with QD1b

Na+-optode microspheres without QDsc

ion J

Kex

KNa,Jopt

Kex

KNa,Jopt

Na+ K+ Ca2+ Mg2+

-3.3 -6.1 -12.7 -12.9

0 -2.8 -4.2 -4.4

-3.5 -6.4 -13.2 -13.2

-0 -2.9 -4.5 -4.5

a The selectivity coefficients were calculated for R ) 0.5 and pH 4.8. b Prepared from 9 mmol/kg NaTFPB, 19.1 mmol/kg Na X ionophore, 5.5 mmol/kg ETH 2439, and 1.9 wt % QD1 (CdSe/CdS) in PVC/DOS (1:2 mass ratio). c Prepared from 9 mmol/kg NaTFPB, 19.1 mmol/kg Na X ionophore, 5.5 mmol/kg ETH 2439, and 0.1 mmol/kg DilC18 in PVC/DOS (1:2 mass ratio).

The reversibility of the underlying sensing chemistry was confirmed with optode film measurements (data not shown). Note that the RT′ concentration used for data fitting was 8.8 instead of 14.1 mmol kg-1 NaTFPB, the residual concentration supposedly utilized to neutralize the quantum dots. In another casting, we prepared sodium-sensing microspheres that contained the same components but without QDs, at the same apparent concentration of NaTFPB (8.8 mmol kg-1) but labeled with an internal reference dye DiIC18. The two calibration curves showed good agreement, with an observed log Kex ) -3.5. The 1.9 wt % QD1 doped in the sodium-sensing microsphere was found to be equivalent to 5.3 mmol kg-1 ion exchanger and translated into an estimated positive charge of 2.8 µmol g-1 of the 5.5-nm CdSe/CdS quantum dots capped with TOPO used in this study. The error bars were found to be small for each data point in the calibration curve obtained from multiple sensing microspheres, indicating that the microspheres possess excellent signal reproducibility. The observed sodium response was accompanied with excellent discrimination to common interfering ions such as

potassium, calcium, and magnesium. The selectivity data from the microsphere ion sensors labeled with QD1 are in agreement with the QD-free microsphere ion sensors, indicating that the presence of the QDs did not influence the selectivity of the ionophore or the chromoionophore. It was not tested whether this chemical inertness of the quantum dots may be extended to transition metals or to anionic species (in anion sensors). Interestingly, the quantum dot fluorescence intensity shown in Figure 5 around 610 nm is not constant, but decreases with decreasing sodium concentration in the sample. This may be explained with energy-transfer processes between the organic chromoionophore and the quantum dot, in analogy to earlier work by the group of Kopelman on a sodium-selective bulk optode film using ETH 2439 as the fluoroionophore that acted as an inner filter on an internal reference dye DiIC18.22 Fortunately, the calculated R values from the simplified equation (eq 3) without quantitatively considering the extent of energy transfer corresponded well to the theoretical response curve. Photobleaching of the organic dye used as chromoionophore still remains a concern for continuous sensing applications since highly photostable compounds are still underdeveloped. Promising alternatives might include fluorophores that absorb in the far visible range such as cyanine dyes,36 dyes doped in porous silica nanoparticles, and photostable pH-sensitive quantum dots containing a chromogenic pH-sensitive ligand for energy or electron transfer.37,38 We expect the precision of QD-labeled microsphere bulk optodes to be further improved with chromoionophores of better photostability. CONCLUSIONS In this work, we reported on a simple method for the fabrication of QD-encoded polymeric microspheres with controllable doping ratios of multiple QDs with uniform size. We also developed the first fluorescent ion-sensing microsphere bulk optode sensor labeled with CdSe/CdS quantum dots. We started with synthesizing TOPO-capped core-shell CdSe/CdS and CdTe/ CdS QDs that emit at 610 and 700 nm, respectively, and tested (36) Rivera, L.; Puyol, M.; Miltsov, S.; Alonso, J. J. Anal. Bioanal. Chem. 2007, 387, 2111-2119. (37) Wang, L.; Yang, C.; Tan, W. Nano Lett. 2005, 5, 37-43. (38) Tomasulo, M.; Yildiz, I.; Raymo, F. M. J. Phys. Chem. B 2006, 110, 38533855.

their compatibility with select solvents, polymer matrixes, and plasticizers. Poly(MMA-DMA), poly(n-butyl acrylate), and PVC/ DOS were used to prepare multicolor QD-encoded microspheres with the particle caster, which physically breaks down the QDpolymer stream into uniform microscale droplets by the vibration of a piezocrystal. Fluorescent bar codes were obtained by doping the microspheres with different mass ratios of two QDs in the casting solution. The fluorescence readout ratio of the two QDs linearly related to the mixing ratio and indicated complete doping of the two QDs into the polymer matrix. The feasibility of using QD labels for microsphere bulk optode ion sensors in view of multiplexed analysis was demonstrated with sodium-selective fluorescent bulk optode microspheres. We found that the positive electrostatic charge of the QDs induces anionexchange characteristics that may interfere with the ion-exchange optode response function of the microspheres. A microsphere composition with a higher content of the lipophilic cation exchanger was required to overcome this potential limitation. The modified QD-labeled sodium-sensing microspheres showed satisfactory sodium response with excellent selectivity. One should note that the QD anion-exchange properties might pose challenges for the design of similar types of sensors for anionic analytes, which was not evaluated here. The effect of QD surface charge on the sensor response was estimated by comparing the response of QD-labeled microsphere Na optodes with the same sensor containing an internal reference dye. The 1.9 wt % CdSe QDs was found to be equivalent to ∼5.3 mmol/kg anion exchanger. Our future plan is to develop a microbead-based analysis system with a number of QD-labeled microsphere bulk optodes, each individually highly selective to an analyte, for multiplexed sensing. A flow cytometric study on such QD-labeled microsphere bulk optodes is currently under way. ACKNOWLEDGMENT The authors gratefully acknowledge financial support from Beckman Coulter, Inc., for this research. We also acknowledge Dr. Debbie Sherman at Purdue Life Science Microscope Facility for her assistance in TEM experiments. Received for review January 22, 2007. Accepted March 15, 2007. AC0701233

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3723