Stable Sensors with Tunable Sensitivities Based on Class II Xerogels

Anal. Chem. , 2006, 78 (6), pp 1939–1945. DOI: 10.1021/ac051657b. Publication Date (Web): February 4, 2006. Copyright © 2006 American Chemical Soci...
9 downloads 0 Views 262KB Size
Anal. Chem. 2006, 78, 1939-1945

Stable Sensors with Tunable Sensitivities Based on Class II Xerogels Zunyu Tao, Elizabeth C. Tehan, Ying Tang, and Frank V. Bright*

Department of Chemistry, Natural Sciences Complex, University at Buffalo, The State University of New York, Buffalo, New York 14260-3000

We report on the analytical figures of merit for O2responsive sensor arrays and films formed by sequestering tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) within class II organically modified silicates that are composed of tetramethoxysilane or tetraethoxysilane and monoalkylsiloxanes of the form (CnH2n+1)-Si-(OR)3 (n ) 1-12, R ) Me or Et). These sensors exhibit a reasonably linear response to gaseous and dissolved O2 (r2 > 0.99), and the sensor responses are stable for over 2 years. Sensor sensitivity can be tuned continuously by adjusting n. For gas-phase O2 detection, changes in the sensor sensitivity depend primarily on the O2 diffusion coefficient within the xerogel phase. The O2 solubility coefficient within the xerogel phase is also a factor but to a lesser degree. For dissolved O2 detection, changes in the sensor sensitivity depend on the O2 diffusion coefficient and the O2 solubility coefficient within the xerogel phase. A linear correlation also exists between the sensor sensitivity and the polarity within these xerogels. Finally, the feature size of pin-printed sensor elements was found to depend linearly on pin velocity. The results of these experiments demonstrate a new strategy for creating xerogel-based sensor arrays consisting of diversified sensor elements for the same target analyte. Developing useful chemical sensor systems is very challenging.1 For example, reliable sensors and sensor systems must be fully reversible, exhibit useful selectivity, sensitivity, and detection limits, be easy to calibrate, and be robust and stable over the long term. Rare are devices that meet all these ideals. O2 detection is important in biological,2 environmental,3 and industrial4 applications. Dissolved O2 is traditionally quantified by using a Clark electrode (CE);5 however, a CE consumes O2 and a CE can become poisoned by some anesthetics and proteins.6 Given these limitations, researchers have been developing and exploring alternative detection methods. Optical O2 sensors7 based on luminescence quenching8 by O2 on the intensity (I) or excitedstate lifetime (τ) of a luminophore (fluorophore or phosphore) are a popular alternative. * To whom all correspondence should be addressed. Phone: (716) 645-6800 ext. 2162. Fax: (716) 645-6963. E-mail: [email protected]. (1) (a) Narayanaswamy, R.; Wolfbeis, O. S. Optical Sensors: Industrial, Environmental and Diagnostic Applications; Springer: Berlin; London, 2004. (b) Diamond, D. Principles of Chemical and Biological Sensors; Wiley: New York, 1998. (c) Taylor, R. F.; Schultz, J. S. Handbook of Chemical and Biological Sensors; Institute of Physics Pub.: Philadelphia, PA, 1996. 10.1021/ac051657b CCC: $33.50 Published on Web 02/04/2006

© 2006 American Chemical Society

Previous research has shown that luminescence-based O2 sensor performance depends strongly on the matrix used to host the O2-responsive luminophore among other factors.9-23 As a result, researchers have explored a wide variety of polymeric platforms including cellulose acetate butyrate,9 fluoropolymers,10 ion-exchange polymers,11 poly(methyl methacrylate),12 poly(2) Oxygen Sensing; Sen, C. K., Semenza, G. L., Eds.; Methods in Enzymology 381; Academic Press: San Diego, 2004. (3) (a) Sojka, R. E.; Oosterhuis, D. M.; Scott, H. D. In Handbook of Photosynthesis, 2nd ed.; Pessarakli, M. Taylor & Francis: Boca Raton, FL, 2005; pp 299314. (b) Taylor, L. A.; Cooper, B.; McKay, D. S.; Colson, R. O. Miner. Metal. Proc. 1993, 10, 43-50. (4) (a) Whiffin, V. S.; Cooney, M. J.; Cord-Ruwisch, R. Biotechnol. Bioeng. 2004, 85, 422-433. (b) Kebabian, P. L.; Romano, R. R.; Freedman, A. Meas. Sci. Technol. 2003, 14, 983-988. (c) Fitzgerald, M.; Papkovsky, D. B.; Smiddy, M.; Kerry, J. P.; O’Sullivan, C. K.; Buckley, D. J.; Guilbault, G. G. J. Food Sci. 2001, 66, 105-110. (5) Clark, L. C., Jr. Trans. Am. Artif. Intern. Organs 1956, 2, 41-49. (6) Bacon, J. R.; Demas, J. N. Anal. Chem. 1987, 59, 2780-2785. (7) (a) Demas, J. N.; DeGraff, B. A.; Coleman, P. B. Anal. Chem. 1999, 71, 793A-800A. (b) Mills, A. Platinum Metals Rev. 1997, 41, 115-127. (c) Demas, J. N.; DeGraff, B. A. In Topics in Fluorescence Spectroscopy; Lakowicz, J. R., Ed.; Plenum: New York, 1994; Vol. 4, Chapter 4. (8) Lakowicz, J. R. Principle of Fluorescence Spectroscopy, 2nd ed.; Kluwer Academic/Plenum Publishers: New York, 1999; Chapters 8 and 9. (9) Douglas, P.; Eaton, K. Sens. Actuators, B: Chem. 2002, 82, 200-208. (10) Amao, Y.; Ishikawa, Y.; Okura, I. Anal. Chim. Acta 2001, 445, 177-182. (11) Vasil’ev, V. V.; Borisov, S. M.; Sens. Actuators, B: Chem. 2002, 82, 272276. (12) Mills, A.; Thomas, M. Analyst 1997, 122, 63-68. (13) Hartmann, P.; Leiner, M. J. P.; Lippitsch, M. E. Anal. Chem. 1995, 67, 88-93. (14) Hartmann, P.; Trettnak, W. Anal. Chem. 1996, 68, 2615-2620. (15) Li, X. M.; Ruan, F. C.; Wong, K. Y. Analyst 1993, 118, 289-292. (16) McDonagh, C.; MacCraith, B. D.; McEvoy, A. K. Anal. Chem. 1998, 70, 45-50. (17) (a) Carraway, E. R.; Demas, J. N.; DeGraff, B. A.; Bacon, J. R. Anal. Chem. 1991, 63, 337-342. (b) Li, X. M.; Wong, K. Y. Anal. Chim. Acta 1992, 262, 27-32. (c) Draxler, S.; Lippitsch, M. E.; Klimant, I.; Kraus, H.; Wolfbeis, O. S. J. Phys. Chem. 1995, 99, 3162-3167. (18) (a) MacCraith, B. D.; McDonagh, C. M.; O’Keeffe, G.; Keyes, E. T.; Vos, Johannes G.; O’Kelly B.; McGilp, J. F. Analyst 1993, 118, 385-388. (b) McEvoy, A. K.; McDonagh, C. M.; MacCraith, B. D. Analyst 1996, 121, 785-788. (c) Mongey, K. F.; Vos, J. G.; MacCraith, B. D.; McDonagh, C. M.; Coates, C.; McGarvey, J. J. J. Mater. Chem. 1997, 7, 1473-1479. (d) McDonagh, C. M.; Shields, A. M.; McEvoy, A. K.; MacCraith, B. D.; Gouin, J. F. J. Sol-Gel Sci. Technol. 1998, 13, 207-211. (e) Lavin, P.; McDonagh, C. M.; MacCraith, B. D. J. Sol-Gel Sci. Technol. 1998, 13, 641-645. (f) Malins, C.; Fanni, S.; Glever, H. G.; Vos, J. G.; MacCraith, B. D. Anal. Commun. 1999, 36, 3-4. (g) Mongey, K. F.; Vos, J. G.; MacCraith, B. D.; McDonagh, C. M. Coord. Chem. Rev. 1999, 185-186, 417-429. (h) McDonagh, C. M.; Kolle, C.; McEvoy, A. K.; Dowling, D. L.; Cafolla, A. A.; Cullen, S. J.; MacCraith, B. D. Sens. Actuators, B 2001, 74, 124-130. (19) (a) Klimant, I.; Ruckruh, F.; Liebsch, G.; Stangelmayer, A.; Wolfbeis, O. S. Mikrochim. Acta 1999, 131, 35-46. (b) Wolfbeis, O. S.; Oehme, I.; Papkovskaya, N.; Klimant, I. Biosens. Bioelectron. 2000, 15, 69-76.

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1939

(styrene),13 poly(vinyl chloride),14 silicone rubber,6,15 and sol-gelderived xerogels.16 Interestingly, most of the Stern-Volmer quenching plots associated with these luminophore-doped polymers are nonlinear.17-22 This behavior represents a significant shortcoming during sensor calibration. More recently, researchers23 have explored the use of class II organically modified silicates (ORMOSILs)24 as hosts for luminophore-based quenchometric sensors. These platforms yielded sensor with linear Stern-Volmer calibration plots. The relatively long-lived triplet metal-to-ligand charge-transfer state of Ru(II) diimines25 makes them attractive luminophores for quenchometric sensor development. In this paper, we report the analytical figures of merit for O2-responsive sensors formed by sequestering tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II), ([Ru(dpp)3]2+) within a series of hybrid xerogels that are composed of tetramethyorthosilane (TMOS) or tetraethylorthosilane (TEOS) and ORMOSILs of the form (CnH2n+1)-Si-(OR)3 (n ) 1-12, R ) Me or Et). The composition of these xerogels were 1:1 molar ratio (TMOS/TEOS-ORMOSIL). Specifically, we report the effects of n and storage time (in excess of 2 years) on the SternVolmer plots, the Stern-Volmer quenching constant, the [Ru(dpp)3]2+ excited-state luminescence lifetime, the O2 permeability, the O2 diffusion coefficient, and the O2 solubility coefficient within the hybrid xerogel hosts. THEORY SECTION Consider a single type of luminophore sequestered within a quencher-permeable host matrix. If the ensemble of luminophore molecules emits from largely similar microenvironments and they have accessibilities similar to the quencher molecules, one can write the generalized Stern-Volmer expression:8,21a,26,27

I0/I ) [1 + 4πgRND[Q]τ0/1000] [SQ/TQ]

(1)

where I0 is the luminescence intensity recorded in the absence (20) (a) Krihak, M.; Murtagh, M.; Shahriari, M. R. Proc. SPIE-Int. Soc. Opt. Eng. 1996, 2836, 105-115. (b) Krihak, M. K.; Shahriari, M. R. Electron. Lett. 1996, 32, 240-242. (c) Shahriari, M. R. In Optical Fiber Sensor Technology; Grattan, K. T. V., Meggitt, B. T., Eds.; Kluwer Academic: London, 1998; Vol. 4. (d) Murtagh, M. T.; Kwon, H. C.; Shahriari, M. R.; Krihak, M.; Ackley, D. E. J. Mater. Res. 1998, 13, 3326-3331. (e) Murtagh, M. T.; Shahriari, M. R.; Krihak, M. Chem. Mater. 1998, 10, 3862-3869. (21) (a) Liu, H.-Y.; Switalski, S. C.; Coltrain, B. K.; Merkel, P. B. Appl. Spectrosc. 1992, 46, 1266-1272. (b) Innocenzi, P.; Kozuka, H.; Yoko, T. J. Phys. Chem. B 1997, 101, 2285-2291. (c) Maruszewski, K.; Jasiorski, M.; Salamon, M.; Strek, W. Chem. Phys. Lett. 1999, 314, 83-90. (d) Choi, M. M. F.; Xiao, D. Analyst 1999, 124, 695-698. (e) Choi, M. M. F.; Xiao, D. Anal. Chim. Acta 2000, 403, 57-65. (f) Xu, H.; Aylott, J. W.; Kopelman, R.; Miller, T. J.; Philbert, M. A. Anal. Chem. 2001, 73, 4124-4133. (22) (a) Dunbar, R. A.; Jordan, J. D.; Bright, F. V. Anal. Chem. 1996, 68, 604610. (b) Watkins, A. N.; Wenner, B. R.; Jordan, J. D.; Xu, W. Y.; Demas, J. N.; Bright, F. V. Appl. Spectrosc. 1998, 52, 750-754. (c) Baker, G. A.; Wenner, B. R.; Watkins, A. N.; Bright, F. V. J. Sol-Gel Sci. Technol. 2000, 17, 71-82. (23) (a) Chen, X.; Zhong, Z. M.; Li, Z.; Jiang, Y. Q.; Wang, X. R.; Wong, K. Y. Sens. Actuator, B: Chem. 2002, 87, 233-238. (b) Tang, Y.; Tehan, E. C.; Tao, Z.; Bright, F. V. Anal. Chem. 2003, 75, 2407-2413. (c) Koo, Y. E. L.; Cao, Y.; Kopelman, R.; Koo, S. M.; Brasuel, M.; Philbert, M. A. Anal. Chem. 2004, 76, 2498-2505. (d) Bukowski, R. M.; Ciriminna, R.; Pagliaro, M.; Bright, F. V. Anal. Chem. 2005, 77, 2670-2672. (24) Sanchez, S.; Ribot, F. New J. Chem. 1994, 18, 1007-1047. (25) Abdel-Shafi, A.; Beer, P. D.; Mortimer, R. J.; Wilkinson, F. Helv. Chim. Acta 2001, 84, 2784-2795. (26) McDonagh, C.; Bowe, P.; Mongey, K.; MacCraith, B. D. J. Non-Cryst. Solids 2002, 306, 138-148.

1940

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

of quencher, I is the intensity at any quencher concentration ([Q] in units of molarity), g is a spin statistical factor reflecting the state type being quenched, R is the luminophore-quencher encounter distance, N is Avogadro’s number, D is the quencher diffusion coefficient (in cm2 s-1) (during the excited-state lifetime, the luminophore does not diffuse significantly within the host matrix), and τ0 is the excited-state luminophore lifetime within the host matrix in the absence of quencher. The factor of 1000 is used to keep the units correct when the quencher concentration is in molarity units (i.e., N/1000 converts molarity units to molecules/cm3). SQ and TQ reflect static quenching via complexation between the ground-state luminophore and quencher and transient quenching arising from time-dependent luminophore-quencher encounters, respectively. For O2 as the quencher at room-temperature one can write21a

[Q] ) [O2] ) 0.041(S)(pO2)

(2)

where the term 0.041 arises from the O2 Henry’s law constant,21a S is the O2 solubility coefficient in the host matrix (in mol L-1 atm-1), pO2 is the O2 partial pressure (in atm), and D × S is termed the host matrix permeability to O2 (P). Although there is some debate regarding emissive state in Ru(II) diimines,7c,25 we have assumed g to be 1/9 (emission is from a triplet state) as have other researchers,26 The 10-7 cm is a reasonable estimate for R,21a and SQ and TQ are essentially unity (vide infra) so eq 1 can be recast as

I0/I ) [1 + APτ0pO2] ) [1 + KSVO2]

(3)

where A ) 3.4 × 1012 cm. In a sensor configuration, the SternVolmer quenching constant, KSV, represents the sensor sensitivity, which arises from (i) P which depends on D and S and (ii) τ0. EXPERIMENTAL SECTION Reagents and Materials. The following reagents were used: TMOS and pyrene (Aldrich); TEOS and methyltrimethoxysilane (C1-TMOS) (United Chemical Technologies); ethyltrimethoxysilane (C2-TMOS), n-propyltrimethoxysilane (C3-TMOS), n-butyltrimethoxysilane (C4-TMOS), n-hexyltrimethoxysilane (C6-TMOS), n-octyltrimethoxysilane (C8-TMOS), n-octyltriethoxysilane (C8TEOS), n-decyltriethoxysilane (C10-TEOS), and n-dodecyltriethoxysilane (C12-TEOS) (Gelest, Inc.); 6-propionyl-2-(dimethylamino)naphthalene (Prodan) (Molecular Probes/Invitrogen); HCl (Fisher Scientific); and EtOH (Quantum Chemical). Tris (4,7′-diphenyl1,10′- phenanthroline)ruthenium(II) chloride pentahydrate ([Ru(dpp)3]Cl2‚5H2O) was purchased from GFS Chemicals and purified as described in the literature.28 Glass microscope slides were purchased from Fisher Scientific Co. Fabrication of Sensor Arrays, Films, and Monoliths. A series of sol solutions were prepared by mixing 3.4 mmol of TMOS or TEOS with 3.4 mmol of alkyltrimethoxysilane or alkyltriethoxysilane (CnH2n+1-TriMOS or CnH2n+1-TriEOS), 21.4 mmol of EtOH, and 400 µL of 0.1 M (0.04 mmol) HCl, in the order listed. The sol (27) (a) Ware, W. R.; Novros, J. S. J. Phys. Chem. 1966, 70, 3246-3253. (b) Nemzek, T. L.; Ware, W. R. J. Chem. Phys. 1975, 62, 477-489. (28) Lin, C. T.; Bottcher, W.; Chou, M.; Creutz, C.; Sutin, N. J. Am. Chem. Soc. 1976, 98, 6536-6544.

solution was mixed for 1 h at room temperature. A 100-µL aliquot of each sol stock solution was then doped with 20 µL of the [Ru(dpp)3]2+ stock solution (12.5 mM in EtOH). The 3 × 3 xerogel arrays were fabricated from each sol solution by using a ProSys 5510 system (Cartesian Technologies) pin printer.29 Arrays were formed by using SMP-3 (TeleChem) quill pins or solid tungsten pins (Cartesian). The diameters of the distal pin ends were 76 (quill) and 200 µm (solid). Substrates were clean glass microscope slides. Slide cleaning consisted of rinsing the slides with EtOH, soaking for 1-2 h in 1 M NaOH, soaking for 1-2 h in 1 M HCl, washing with copious amounts of deionized water, and drying at 80 °C. The pin-printing protocol typically used a pin velocity of 5-50 mm/s and pin-to-surface contact time of 5-50 ms. Spin-coated films were prepared by using a micropipet to deliver 50 µL of a particular [Ru(dpp)3]2+-doped sol solution onto the face of a clean glass microscope slide. Xerogel films were formed by using a spin coater (2000 rpm, 30 s). Xerogel films and monoliths were also prepared to estimate the dipolarity within each hybrid xerogel formulation. These sol solutions contained 10 µM pyrene or 167 µM Prodan. Films were prepared from 50-µL aliquots of a given sol solution (2000 rpm, 30 s). Xerogel monoliths were formed by placing 1 mL of a given sol solution into a polystyrene cuvette. The cuvettes were capped immediately. The caps were removed after 3 days and the xerogels allowed to form and age. Spectroscopy was carried out on monoliths that had aged for at least 6 months. All xerogel samples were stored and aged under ambient conditions in the dark. All experiments were conducted at room temperature. Dissolved O2 experiments were conducted in water. Instrumentation. The sensor arrays were characterized by using an epifluorescence microscope interfaced to a chargecoupled device detector. The excitation source was a He-Cd laser (442 nm). This apparatus has been described in detail in ref 29a. The time-resolved intensity decay traces were recorded by using a nitrogen-pumped dye laser, a photomultiplier tube detector, and a digital oscilloscope. This apparatus has been described in detail elsewhere.30 All O2-dependent quenching response data, O2/N2 response data, and steady-state emission spectra were recorded by using a SLM-Aminco model 48000 MHF spectrofluorometer with a Xe arc lamp as the excitation source. Film samples were oriented at ∼60° relative to the excitation beam trajectory. Film thicknesses were determined by using a field emission scanning electron microscope (Hitachi model S-4000). RESULTS AND DISCUSSION Results are reported for hybrid xerogels that consist of 1:1 molar ratios of TMOS (or TEOS) and the CnH2n+1-Si-(OR)3. In some literature,24 these are referred to class II ORMOSILs. To streamline the remaining discussion, we refer to our class II ORMOSILs as Cn recognizing that they are binary xerogels composed of CnH2n+1-Si-(OR)3 and Si-(OR)4. (29) (a) Cho, E. J.; Bright, F. V. Anal. Chem. 2001, 73, 3289-3293. (b) Cho, E. J.; Bright, F. V. Anal. Chem. 2002, 74, 1462-1466. (c) Cho, E. J.; Bright, F. V. Anal. Chim. Acta 2002, 470, 101-110. (d) Cho, E. J.; Tao, Z.; Tang, Y.; Tehan, E. C.; Bright, F. V.; Hicks, W. L.; Gardella, J. A.; Hard, R. Appl. Spectrosc. 2002, 56, 1385-1389. (30) Baker, G. A.; Wenner, B. R.; Watkins, A. N.; Bright, F. V. J. Sol-Gel Sci. Technol. 2000, 17, 71-82.

Figure 1. Typical intensity-based Stern-Volmer plots for [Ru(dpp)3]2+-doped pin-printed class II ORMOSIL-based sensor elements. (A) Gas-phase O2 responses. (B) Dissolved O2 responses.

Effects of Alkyl Chain Length on the Sensor Response. A linear Stern-Volmer plot is indicative of luminophores homogeneously distributed within a host matrix such that all luminescing molecules report from a single type of microenvironment and each microenvironment is equally accessible to quencher.7c,8,23d Figure 1 presents typical Stern-Volmer plots for our pin-printed xerogelbased sensor elements responding to gaseous (Figure 1A) and dissolved (Figure 1B) O2. These particular response profiles were from sensors that were aged/stored for 8 weeks. Each datum represents the average response from nine sensing elements. The error bars represent ( one standard deviation. The lines that pass through the data represent the best fits to eq 3. Several aspects of these results merit further discussion. First, the Stern-Volmer plots appear to be linear over the full O2 concentration range (r2 > 0.99). Thus, the SQ and TQ terms in eq 1 are unity. The linear response also suggests that the population of luminescing [Ru(dpp)3]2+ molecules within a given xerogel are largely emitting from similar types of microenvironments. (We will come back to this point when we discuss the time-resolved intensity decay profiles.) A linear response profile also simplifies calibration significantly (i.e., two-point calibrations can be used). Second, as n increases, the Stern-Volmer quenching constant, KSV, generally increases. For gas-phase O2, KSV increases by 5-6-fold as we move from the C1 to C8 xerogels. For dissolved O2, KSV increases by ∼10-fold as we move from the C1 to C8 xerogels. Third, the increase in KSV ends at n ) 8. The linear Stern-Volmer plots for these xerogels are consistent with previous results from our laboratory.23b Fourth, sensors derived from TMOS or TEOS (i.e., C8-TMOS and C8-TEOS) behave similarly. Finally, linear SternVolmer plots were also obtained for the spin-coated xerogel-based films (results not shown). Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1941

Figure 2. Summary of sensor sensitivity (A) and stability (B) as a function of Class II ORMOSIL composition. The results in (A) are for 8-week-old sensors.

We next questioned whether the sensor response depended on how the sensor elements were formed. The motivation for this study was previous results by Murtagh et al. showing that spincast thin films exhibited response profiles that were different from xerogel-based monoliths20e and the work of Gillian and Brennan showing that film-based sensors responses could be tuned depending on their composition and their processing conditions.31 Figure 2A presents the recovered KSV values for pin-printed (gaseous and dissolved O2) and spin-cast (gaseous O2) sensors. (Note: The thicknesses of these sensor elements and films are between 1 and 2 µm.) Looking first at the gas-phase measurements, we see that the Cn-dependent KSV data are equivalent within our measurement precision for the pin-printed and spin-cast sensor elements. KSV increases from C1 to C8 and decreases dramatically from C8 to C12. Additional experiments were performed to determine whether the sensor response depended on the pinprinting conditions (i.e., pin velocity and pin-to-surface contact time). These parameters were adjusted by 1 order of magnitude, and the observed KSV for a particular formulation was unaffected (results not shown). As discussed latter, there were some interesting effects of the pin velocity (vide infra). The results with dissolved O2 exhibit a similar KSV versus Cn profile when compared to the gas-phase O2 experiments. The main difference between the gaseous and dissolved O2 results is an overall decrease in KSV at a given Cn and the lack of a clear drop in KSV above n ) 8. There seems to be a plateau in KSV above n ) 8 when these sensors are used to detect dissolved O2. Taken together, these results demonstrate that one can systematically tune the sensor sensitivity by ORMOSIL choice. (31) Gillian, L. G.; Brennan, J. D. J. Mater. Chem. 2002, 12, 3400-3406.

1942

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

Sensor Element Stability. The stability of a sensor’s response is a key analytical figure of merit. Having established that our ORMOSIL-based sensor elements (pin printed or films) appear to exhibit a linear response profile (Figure 1), we set out to investigate the long-term sensor stability. Toward this end, we tracked the sensor response over more than 2 years. The results of these experiments for gas-phase O2 detection are collected in Figure 2B. Similar results were seen for the dissolved O2 responses (results not shown). The results show substantial changes in response during the first few weeks following initial sensor fabrication. This behavior arises primarily because the xerogels are not completely formed (i.e., condensation and polycondensation are not completed) until two or more weeks have passed when the samples are stored/aged under ambient conditions. This “problem” is more prevalent in the xerogels formed from the shorter chain ORMOSILs. The sensor elements are remarkably stable for the C8- and C10-based materials at all aging times. Other than the C1-based materials, the responses stabilize significantly after 4 weeks under ambient aging/storage conditions. Interestingly, if the xerogel-based sensor elements are initially aged in vacuo (∼33 µbar) at 343 K for 10-12 h after fabrication, the response (under ambient storage) then remains constant between 1 and 125 weeks (results not shown). Thus, a simple vacuum treatment step leads to highly stable sensor elements based on class II ORMOSILs. The long-term stability of these ORMOSIL-based xerogels (up to 125 weeks) is fully consistent with our previous results on the C8-TEOS/TEOS-based xerogel films (44 weeks)23b and work by McEvoy et al. on C1TEOS/TEOS- and C2-TEOS/TEOS-based xerogels (24 weeks).16 The current results argue that (i) stability is a universal trait of class II ORMOSIL-based xerogel O2 sensor platforms and (ii) stability can extend over multiple years in class II ORMOSIL-based xerogel sensor elements. The long-term stability is most likely the result of the high level of T1 silicone sites within these hybrid xerogels.32 These are inherently hydrophobic, and the covalently bonded carbon (Si-C) inductive effect makes the silicon center much less reactive in comparison to an alkoxide (Si-O-R).32 Origin of the Changes in KSV. KSV depends on (i) P which depends on D and S and (ii) τ0. To determine the origin of the changes in KSV with Cn (Figure 1) for these sensors, we recorded the time-resolved intensity decay traces for each [Ru(dpp)3]2+doped, ORMOSIL-based xerogel. We used [Ru(dpp)3]2+ dissolved in EtOH (freeze-pump-thaw degassed) as a reference lifetime standard to ensure proper system operation (τ0 ) 5.7 µs).7 Figure 3A presents a typical time-resolved intensity decay trace (points) for a dry [Ru(dpp)3]2+-doped C3 film along with the best fit (line) to a single-exponential decay model. The fit is excellent (r2 ) 0.996) over at least five lifetimes (inset). The single-exponential time-resolved intensity decay profile demonstrates that the luminescing [Ru(dpp)3]2+ molecules are largely emitting from a single microenvironment within this xerogel. This result is consistent with the steady-state Stern-Volmer plots (Figure 1). The timeresolved intensity decay traces for [Ru(dpp)3]2+ in each xerogel (32) (a) Yang, J. J.; El_Nahhal, I. M.; Maciel, G. E. J. Non-Cryst. Sol. 1996, 204, 106-117. (b) Lindner, E.; Jager, A.; Schneller, T.; Mayer, H. A. Chem. Mater. 1997, 9, 81-90. (c) da Fonseca, M. G.; Silva, C. r.; Airoldi, C. (d) Rodriguez, S. A.; Colon, L. A. Appl. Spectrosc. 2001, 55, 472-480.

Figure 3. Time-resolved intensity decay results for a dry [Ru(dpp)3]2+-doped class II ORMOSIL (C3). (A) Typical time-resolved intensity decay trace (points) and fit to a single-exponential decay model (line). (Inset) ln I vs t plot demonstrating single-exponential behavior over at least five lifetimes. (B) Effects of n on τ0 for dry and wet films.

(with or without water) were also rigorously single exponential (r2 > 0.99) over at least five lifetimes (results not shown). Figure 3B summarizes the effects of Cn on the recovered [Ru(dpp)3]2+ τo in dry and wet films. Inspection of these results shows that τ0 is on the order of 5 µs (comparable to previously reported values)7,15b,23d and essential independent of the alkyl chain length. However, τ0 is statistically lower in the wet films, but also independent of the alkyl chain length. Thus, the reason for the observed changes in sensor response (i.e., KSV) arises from changes in P, which depends on D and S. Determination of D, P, and S versus Cn. The key to determining the origin of the observed sensitivity is to first determine D, the O2 diffusion coefficient within the xerogel. Toward this end, we used a method reported by MacCraith and co-workers26 to determine the luminescence versus time of the sensor films following a step change in O2 in concert with the Schappacher-Hartmann33 protocol for recovering D from the luminescence versus time data. Figure 4A illustrates the effect of Cn on D for the dry and wet films. Several aspects of these results merit further discussion. First, D is significantly smaller within the xerogels in comparison to O2 diffusion in the gas phase or in water alone as originally reported by MacCraith and co-workers.26 Second, D is only 2-4fold greater for the gas-phase experiments in comparison to the (33) Schappacher, G.; Hartmann, P. Anal. Chem. 2003, 75, 4319-4324.

Figure 4. Origin of the tunable sensor sensitivity seen in Figure 1 as a function of Cn for dry and wet films. (A) O2 diffusion coefficient within the xerogel phase. (B) O2 permeability within the xerogel phase. (C) O2 solubility coefficient within the xerogel phase.

dissolved O2 experiments. This result suggests there is very little water actually within these xerogel pores, or the [Ru(dpp)3]2+ molecules within these xerogel are distributed in such a way that they encounter little water but can still be accessed by and quenched by O2. Third, D rises as n increases from n ) 1 to 8 and then begins to decrease for n > 8. Figure 4B illustrates the effect of Cn on the O2 permeability (P) within the dry and wet films. P is not dependent on the film state (dry or wet). D rises as n increases from n ) 1 to 8 and then begins to decrease/level off for n > 8. Figure 4C illustrates the effect of Cn on the O2 solubility coefficient (S) within the dry and wet films. For the gas-phase experiments, the O2 solubility coefficient remains constant for the C1 and C2 xerogels, S then drops by ∼50% as we progress to the C3 and C4 xerogels, and S remains n independent above n ) 4. For the wet films, S generally increases with increasing n. As n Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1943

Figure 5. Correlation between the O2 permeability within the xerogel phase and the Prodan emission maximum (A) and pyrene I1/I3 (B). Dry samples.

increases from 1 to 10, the O2 solubility coefficient within the films ranges from 0.4 to 0.2 mol L-1 atm-1 in the gas phase and from 0.4 to 0.8 mol L-1 atm-1 in the dissolved phase. Based on eq 2, when pO2 is at 1 atm, the calculated [O2] in the dry films range from 11 to 19 mM and from 18 to 33 mM in the wet films. These values are in line with, but higher than the reported [O2] in liquid hydrocarbons at 1 atm (∼10 mM).34 Correlating the Observed Sensor Response to the Physicochemical Properties within the Xerogels. Luminescent probe molecules such as Prodan35 and pyrene36 exhibit emission spectra that depend on the physicochemical properties of their local microenvironment. In the case of Prodan, the emission maximum is solvent dependent (λem,max is 401 nm in cyclohexane and 530 nm in water). In pyrene, the ratio of the I1 to I3 emission peaks (I1/I3) provides an estimate of the physicochemical properties that surround the pyrene molecule (I1/I3 is 0.6 in cyclohexane and 1.8 in water). In Figure 5, we present correlation plots for the observed P for the sensor with the Prodan emission maximum (Figure 5A) or the pyrene I1/I3. (Figure 5B). The correlations are good (r2 ∼ 0.9), demonstrating that a decrease in xerogel dipolarity is a reasonable predictor of an increase in sensor sensitivity. (34) Handbook of Chemistry and Physics, 71st ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 1990; pp 9-86-98. (35) Weber, G.; Farris, F. J. Biochem. 1979, 18, 3075-3078. (36) (a) Dong, D. C.; Winnik, M. Photochem. Photobiol. 1982, 35, 17. (b) Dong, D. C.; Winnik, M. A. Can. J. Chem. 1984, 62, 2560. (c) Waris, R.; Acree, W. E.; Street, K. W. Analyst 1988, 113, 1465. (d) Wong, A. L.; Hunnicutt, M. L.; Harris, J. M. Anal. Chem. 1991, 63, 1076. (e) Karpovich, D. S.; Blanchard, G. J. J. Phys. Chem. 1995, 99, 3951.

1944 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

Figure 6. Effects of pin velocity on the xerogel-based sensor element feature size (C8). (A) False color fluorescence images. (B) Feature size vs pin velocity.

Sensor Element Feature Size. Feature size (F) is an important factor in the fabrication of sensor arrays37 because it plays a key role in the final density of sensor elements within an array. For example, a hypothetical sensor array fabricated with a sensor element center-to-center (CTC) spacing of 200 µm (F ) 143-170 µm) has a density of 2500 elements/cm2 whereas an array with a CTC of 100 µm (F ) 71-83 µm) has a density of 10 000 elements/cm2. An array with a CTC of 15 µm would exhibit a density in excess of 444 000/cm2. However, despite its relative importance, information on F for pin-printed features and the reproducibility of such are scarce38 and all existing data are for aqueous buffer systems or aqueous buffer systems with additives (betaine or DMSO). In addition, of all the major print variables (e.g., humidity, temperature, printing solution viscosity, surface tension and density, pin geometry, wettability and surface chemistry, substrate wettabillity and surface chemistry, and pin velocity and pin-to-surface contact time), only humidity, pin size, print solution viscosity, surface wettability, and additives have been explored in any sort of systematic manner. Finally, there are no systematic feature size data on the effects of the aforementioned parameters on xerogels formed by pin printing. During the course of our initial experiments using a SMP-3 quill pin, we discovered that the final sensor element diameter (i.e., F) depended in a linear manner on the pin velocity (V) (37) Microarray Analysis; Schena, M.; John Wiley & Sons: Hoboken, NJ, 2003; Chapter 7. (38) (a) Smith, J. T.; Viglianti, B. J.; Reichert, W. M. Langmuir 2002, 18, 62896293. (b) Reese, M. O.; van Dam, R. M.; Scherer, A.; Quake, S. R. Genome Res. 2003, 12, 2348-2352. (c) McQuain, M. K.; Seale, K.; Peek, J.; Levy, S.; Haselton, F. R. Anal. Biochem. 2003, 320, 281-291.

(Figure 6):

F ) do + OV

(4)

In this expression do is the pin tip diameter, V is the pin velocity (in mm/s), and O is a term that depends on the other remaining system properties (i.e., humidity, temperature, printing solution viscosity, surface tension and density, pin geometry, wettability and surface chemistry, substrate wettability and surface chemistry, and pin-to-surface contact time). CONCLUSIONS Class II ORMOSILs with varying alkyl chain lengths (C1-C12) coupled with [Ru(dpp)3]2+ provide luminescene-based sensors for gaseous and dissolved O2 detection that exhibit linear calibration curves that are stable for more than 2 years. By changing the Cn one can conveniently tune the sensor sensitivity from a high (in

C8) to low (in C1) value. The origin of the tunability depends on the sample type. For gas-phase O2 detection, the sensor sensitivity depends primarily on the O2 diffusion coefficient within the xerogel phase. The O2 solubility coefficient within the xerogel phase is also a factor but to a lesser degree. For dissolved O2 detection, the sensitivity depends on the O2 diffusion coefficient and the O2 solubility coefficient within the xerogel phase. A linear correlation exists between the sensor sensitivity and the polarity within the xerogel. Finally, the feature size of pin-printed sensor elements was shown to depend linearly on pin velocity. ACKNOWLEDGMENT This work was generously supported by the National Science Foundation. Received for review September 16, 2005. Accepted January 11, 2006. AC051657B

Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

1945