Anal. Chem. 1997, 69, 3375-3379
Scintillator Light Source for Chemical Sensing in the Near-Ultraviolet Steven E. Hobbs,† Radislav A. Potyrailo, and Gary M. Hieftje*
Department of Chemistry, Indiana University, Bloomington, Indiana 47405
A novel chemical sensor based on a light source composed of a radionuclide and a scintillator is experimentally evaluated. Proper selection of a radionuclide/scintillator combination permits fabrication of a practical light source emitting in the ultraviolet (UV). Such a UV light source is critical for chemical sensors which utilize UV-excitable chromophores or fluorophores. Unlike conventional gasfilled discharge lamps, the developed UV source is compact, inexpensive, simple in design, stable, and highly reliable, and it does not require an external power source. The utility of the new source was demonstrated through construction of sensors for oxygen. This application was selected for experimental evaluation of the new light source since oxygen sensors have been characterized well with conventional light sources. Although the scintillator light source is less intense than conventional sources, its excellent short- and long-term stability provides a reproducibility of fluorescence measurements of about 0.35% RSD. The stability of the scintillator light source suggests its utility in simple single-beam detection configurations. Optical sensors are used in such diverse areas as medical,1-3 environmental,4,5 and industrial6-8 monitoring. Such sensors often offer analytical features that are not available from other methods.9 However, practical implementation of optical sensors is ordinarily limited to the visible and infrared regions of the spectrum; sensors utilizing the ultraviolet (UV) region of the spectrum are usually impractical because of the lack of an inexpensive, stable, compact, low-power, and continuum UV light source. This shortcoming is unfortunate, since a host of important analytes and chemical reagents requires excitation in the UV region of the spectrum. For this reason, development of a practical UV source would be extremely valuable. Technological advances over the past several years have greatly reduced the size and cost of light sources and photode†Present address: Nellcor Puritan Bennett, 4280 Hacienda Dr., Pleasanton, CA 94588. (1) Collison, M. E.; Meyerhoff, M. E. Anal. Chem. 1990, 62, 425A-437A. (2) Wolfbeis, O. S. Int. J. Optoelectron. 1991, 6, 425-441. (3) Lubbers, D. W. Sens. Actuators B 1993, 11, 253-262. (4) Matson, B. S.; Griffin, J. W. Proc. SPIE-Int. Soc. Opt. Eng. 1989, 1172, 1326. (5) Klainer, S. M.; Thomas, J. R.; Francis, J. C. Sens. Actuators B 1993, 11, 81-86. (6) Norris, J. O. W. In Techniques and Mechanisms in Gas Sensing; Moseley, P. T., Norris, J. O. W., Williams, D. E., Eds.; Hilger: Bristol, UK, 1991; pp 260-280. (7) Greenwell, R. A.; Addleman, R. S.; Crawford, B. A.; Mech, S. J.; Troyer, G. L. Fiber Integr. Opt. 1992, 11, 141-150. (8) Dao, N. Q.; Jouan, M. Sens. Actuators B 1993, 11, 147-160. (9) Cammann, K.; Lemke, U.; Rohen, A.; Sander, J.; Wilken, H.; Winter, B. Angew. Chem. 1991, 30, 516-539.
S0003-2700(96)01296-6 CCC: $14.00
© 1997 American Chemical Society
tectors for the visible, near-infrared, and infrared (IR) spectral regions. Compact frequency-doubled laser diodes and lightemitting diodes (LEDs) are available with emission down to the violet and to the blue, respectively. Miniature SiC photodiodes currently can operate down to the near-UV. Extremely sensitive micromachined photomultiplier tubes (PMTs), with roughly 1-cm3 volumes, have been available for some time. High-voltage power supplies necessary for PMTs and avalanche photodiodes are now available as battery-powered systems with dimensions of several cubic centimeters. Thus, an entire optical sensor for spectral regions ranging from the violet to the near-IR can be constructed as an inexpensive and compact instrument. Unfortunately, a small and inexpensive continuum light source is not available for the UV region of the spectrum. Many species of medical, biological, and environmental interest require such a light source for either direct spectroscopic or reagent-based chemical sensing. Such UV-based sensors can enable real-time monitoring of important medical analytes in a critical care setting and thus provide timely therapeutic information. Such sensors can also be used to indicate the presence of excessive concentrations of toxic or explosive species in a workplace or in the field, can track a source of contamination in an industrial process, and can follow the formation and movement of environmental pollutants. For these and many other applications, a small, inexpensive, and low-power UV light source is critical for the replacement of present devices such as gas-filled discharge lamps. An alternative type of light source that has been explored in the past utilizes a radionuclide and a scintillator. Application of scintillator light sources for UV absorption measurements was first suggested by Helf and White in 1957.10 Ross in 196611 used such a source for the high-precision, absorption-based quantitative determination of colored metal ions in solution. In 1971, Stevens12 suggested that a fluorophore could be directly excited with radionuclear particles without the use of a scintillator. Later, scintillator light sources were utilized for UV absorption detection in liquid chromatography and process control.13,14 Recently, Hobbs and Hieftje demonstrated the first application of a scintillator light source for time-resolved fluorometry.15,16 That source consisted of an encapsulated, β-emitting radionuclide, 90Sr, and a liquid or plastic scintillator. (10) Helf, S.; White, C. Anal. Chem. 1957, 29, 13-16. (11) Ross, H. H. Anal. Chem. 1966, 38, 414-420. (12) Stevens, B. U.S. Patent 3,612,866, Oct 12, 1971. (13) Jones, K.; Malcolme-Lawes, D. J. J. Chromatogr. 1985, 329, 25-32. (14) Rinke, G.; Hartig, C. Anal. Chem. 1995, 67, 2308-2313. (15) Hobbs, S. E.; Hieftje, G. M. Abstracts of the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Chicago, IL, Feb 27-March 4, 1994; Paper 119. (16) Hobbs, S. E.; Hieftje, G. M. Appl. Spectrosc. 1995, 49, 15-19.
Analytical Chemistry, Vol. 69, No. 16, August 15, 1997 3375
Scintillator light sources clearly fulfill the cost, size, and power requirements for field applications that other UV light sources lack. The cost of such a source is nominal, as many radionuclides are byproducts of the nuclear power industry; also, scintillator materials can consist of simple mixtures of fluorophores in an organic polymer or liquid matrix. The size of the light source used in our laboratory is about 2 cm3 and can probably be reduced to microscopic dimensions. Also, the light source requires no external power, as the energy source is the radionuclide itself. Additional characteristics of the light source that are valuable for a host of applications are its extremely stable average output,11 very narrow nanosecond pulse widths,16 high pulse rates, and extremely stable pulse-to-pulse temporal profile. Also, the light source can be configured to produce output radiation from 145 to 600 nm merely by the appropriate choice of a scintillator.16-18 Importantly, there is no risk of radioactive contamination from the radionuclide since it is bound in a ceramic matrix and encapsulated. The safety of encapsulated radionuclides is evidenced by their widespread use. Such radionuclides have been employed in household smoke detectors, for on-line thickness and density gauges in industry, as sources of X-rays in fluorescence measurements, for thermoelectric generators, and in miniature batteries inserted under the skin for heart pacemakers.19 It is the aim of the present study to evaluate radioluminescent UV sources based on plastic and liquid scintillators for use in reagent-based chemical sensors. As a representative example, we have developed two sensors for the detection of oxygen. In these sensors, the scintillator light source excites an oxygensensitive reagent, 9,10-diphenylanthracene. The fluorescence intensity of the reagent is quenched as a function of oxygen concentration. Oxygen was selected as a target analyte for two reasons. First, the new sensors can then be compared with the large number of other oxygen sensors reported in the literature. Second, fluorescent reagents are the most widely used type of indicators in optical chemical sensors.20 Thus, the performance of the oxygen sensors can be used to estimate the capabilities of a large number of other fluorophore-based sensors. In the newly developed oxygen sensors, the fluorophore was immobilized in a silicone film or silica gel, as has been widely reported in the literature.20 EXPERIMENTAL SECTION Sensor with a Plastic Scintillator. A schematic diagram of the sensor employing a plastic scintillator in shown in Figure 1. A sensing silicone film was cast from a solution of 1.52 g of silicone prepolymer (RTV sealant, Permatex 66B), 1.97 mg of 9,10diphenylanthracene (DPA, Aldrich), and 2.44 g of tetrahydrofuran (THF, Mallinckrodt, analytical reagent grade). The solution was poured onto a 10-cm2, 0.1-mm-thick polyethylene sheet and was cured for 24 h in air at room temperature. The resulting film thickness was estimated to be about 0.4 mm. No visible crystals of DPA formed in the film at this reagent concentration. The polyethylene sheet with the cured film was cut into 0.9- × 2-cm rectangles. Twelve of the rectangles were placed in a standard (17) Birks, J. B. The Theory and Practice of Liquid Scintillation; Pergamon Press: Oxford, UK, 1964. (18) Hurlbut, C. R.; Moser, W.; Flournoy, J.; Winn, D. R.; LaPierre, C.; Whitmore, M. IEEE Trans. Nucl. Sci. 1990, 37, 1776-1181. (19) Faires, R. A.; Boswell, G. G. J. Radioisotope Laboratory Techniques; Butterworths: Boston, MA, 1981. (20) Fiber Optic Chemical Sensors and Biosensors; Wolfbeis, O. S., Ed.; CRC Press: Boca Raton, FL, 1991.
3376 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
Figure 1. Schematic diagram of oxygen sensor with a plasticscintillator-based UV source.
10- × 10-mm fluorometric cuvette. Glass capillaries were used as spacers to prevent the silicone films from touching and adhering to each other. The plastic scintillator used for excitation was the commercially available Pilot U (NE Technology Ltd.). The scintillator was machined into a cylinder 0.8 cm in diameter and 3.5 cm long. The scintillator surface was polished, and silver was chemically deposited onto all surfaces but the front face. 90Sr was used as a β source (1 mCi, Isotope Products Inc.). It was bound in a ceramic matrix which was, in turn, encapsulated in aluminum. The β source and scintillator were placed in a lead block to absorb any X-rays (bremsstrahlung) that are produced when β particles decelerate near the capsule. Although 90Sr and its daughter radionuclide 90Y do not emit X-rays directly, the β particles from 90Y are energetic enough to generate photons in the X-ray region of the spectrum.16 These X-rays are infrequent and do not pose a safety hazard; however, they cause large current spikes when detected by a PMT. Thus, the X-rays can produce an appreciable background when an unshielded PMT is operated in the analog mode. The fluorescence from the mounted films was detected at 90° by a Hamamatsu R928 PMT using a long-pass filter with a cutoff at 435 nm. Sensor with a Liquid Scintillator. A schematic diagram of the sensor employing a liquid scintillator in shown in Figure 2. Thin-layer chromatography (TLC) plates were used for immobilization of DPA that served as the oxygen-sensing medium. A 4- × 4-cm piece was cut from a commercial flexible, polyesterbacked TLC plate (Whatman, PE SIL G). The thickness of the silica gel layer on the TLC plate was 250 µm. Approximately 1 mL of 8 mM DPA in toluene (OmniSolv, EM Science, spectrophotometric grade) was added dropwise onto the plate. Time was allowed for evaporation of the toluene between drops. After the plate appeared dry, a stream of dry nitrogen was passed over the
Figure 2. Schematic diagram of oxygen sensor with a liquidscintillator-based UV source.
plate surface for 10 min. The TLC plate was then bent into a cylinder with the silica surface on the inside. The cylinder was placed in a plastic scintillation vial so the polyester side of the plate was forced against the inner surface of the vial. A smaller, 2.5-mL glass scintillation vial was then placed inside the larger plastic vial. The glass vial was fitted with a 1-mm-thick plastic cap and contained a liquid scintillator, 8 g/L p-terphenyl (pTP, Sigma, scintillation grade) in toluene. The β-emitting capsule was placed above the scintillation vial with the active side down. After being capped, the entire plastic vial was placed inside an integrating sphere to collect as much light as possible from the highly scattering vial. Fluorescence was collected through a long-pass filter with a cutoff at 435 nm and detected by a Hamamatsu R928 PMT. The photometric background from X-rays was eliminated by shielding the PMT from the β source with an 8-mm-thick lead glass window (H. L. Lyons Co.). This thickness is equivalent to a 2-mm-thick lead sheet. As the worst-case example, the background signal from both sensors (with plastic and liquid scintillators) was evaluated by replacing the oxygen-sensing medium with an undoped silicone film or a TLC sheet. In these measurements, the recorded PMT photocurrents were more than twice as large as those measured with the sensing media in a pure nitrogen atmosphere. Apparently, the immobilized reagent attenuates much of the light from the scintillator, which is near the 435-nm cutoff of the long-pass filter. The data shown here have not been background subtracted, since this approach represents a real-world sensing situation. House supplies of dry nitrogen and air were used to produce zero and 21% oxygen concentrations. Oxygen (Air Products, standard grade) was used for the 100% oxygen concentration. The flow cell gas input was switched manually with a two-way valve for nitrogen and air and an on/off valve for oxygen. Gas flow rates were typically 5-10 L/min.
RESULTS AND DISCUSSION Scintillator Types. The two most important criteria used in choosing scintillators were (1) high scintillator intensity in the excitation band of the DPA fluorophore and (2) low scintillator intensity in the emission band of DPA. The former is necessary to produce strong fluorescence signals, while the latter serves to minimize the background level. The plastic scintillator (Pilot U) chosen for the present study satisfies the first criterion but fails the second; it produces a significant output at long wavelengths (maximum emission, 391 nm; bandwidth at half-maximum, from 380 to 420 nm).21 The chosen liquid scintillator (pTP/toluene) satisfies both criteria since it yields a significant output only at shorter wavelengths (peak emission, about 340 nm).22 Consequently, the plastic scintillator was employed with the right-angle excitation/detection geometry (Figure 1), since this arrangement reduced the level of scintillator light that reached the detector. The liquid scintillator was used for the on-axis sensor configuration (Figure 2), since the scattered excitation light would be blocked by the long-pass filter. It is possible to estimate the scintillator light intensity for both systems.16 For the plastic scintillator, we assume a flux of 38 million β particles/s in the direction of the scintillator, an average energy of 0.93 MeV/β particle, 10 photons/keV of particle energy, and an average spectral output of 390 nm. Based upon these assumptions, approximately 0.180 µW of light is produced from the plastic scintillator. The amount of light produced from the liquid scintillator is roughly half that from the plastic due to the lower efficiency of the scintillator and the fact that the β particles lose energy as they pass through the 1-mm-thick plastic cap of the scintillation vial. Choice of Reagent. Dynamic fluorescence quenching of polycyclic aromatic hydrocarbons (PAHs) by molecular oxygen is known to be very efficient.20 For the evaluation of the scintillator UV sources, DPA was selected because its fluorescence can be excited in the spectral range of 350-395 nm and collected at 420450 nm. Also, this fluorophore was used earlier for oxygen determinations with conventional light sources such as Xe and Hg arc lamps.23,24 In addition, an oxygen transducer based on DPA was found to be useful in the construction of a glucose biosensor.25 Importantly, no interferences have been found in the past from carbon monoxide, carbon dioxide, methane, or higher alkanes for PAH-based oxygen sensors.20 Sensor with a Plastic Scintillator. Figure 3 compares the Stern-Volmer calibration curve for the newly developed DPA/ silicone sensor (Figure 1) with that for a sensor reported by Blyler and co-workers.24,26 In the oxygen sensor of Blyler and coworkers,24,26 a transducer was fabricated by dissolving DPA in a two-part transparent poly(dimethylsiloxane) rubber. The curing process was performed by rapidly heating the prepolymer to 500 °C. The final concentration of DPA in silicone was 0.1% by weight, about the same as that in the silicone films used in our scintillatorbased sensor. Blyler and co-workers separated the 367-nm (21) NE Technology Ltd. Scintillation materials, Pilot U Data Sheet. (22) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic Molecules; Academic Press: New York, NY, 1971. (23) Cox, M. E.; Dunn, B. Appl. Opt. 1985, 24, 2114-2120. (24) Lieberman, R. A.; Blyler, L. L.; Cohen, L. G. J. Lightwave Technol. 1990, 8, 212-220. (25) Shah, R.; Margerum, S. C.; Gold, M. Proc. SPIE-Int. Soc. Opt. Eng. 1988, 906, 65-73. (26) Blyler, L. L., Jr.; Cohen, L. G.; Lieberman, R. A.; MacChesney, J. B. Eur. Pat. Appl. 0292207, 1988.
Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
3377
Figure 3. Stern-Volmer calibration curves for the newly developed oxygen sensor based on a β-emitter and a plastic scintillator (A) and a sensor based on a conventional UV light source (B). For sensor A, the 99% confidence intervals are given for four replicate measurements. The experimental points for sensor B are taken from Figure 7 of ref 24.
Figure 4. Dynamic response of oxygen sensor with a UV source based on a plastic scintillator. The oxygen transducer is a silicone film with immobilized DPA.
excitation and 430-nm emission radiation by using excitation and emission monochromators. This arrangement assured a low level of scattered excitation light to be collected by the PMT. Similar slopes of the calibration curves (Figure 3) suggest that the excitation light in the scintillator-based sensor was adequately filtered. That some excitation light reaches the PMT is evident from the slightly lower slope of curve A at high oxygen concentrations. The excellent stability of the scintillator light source resulted in a relative standard deviation for four replicate measurements of only 0.35%. The detection limit for the sensor (at S/N ) 3) calculated from the slope of the calibration curve over the lowest concentration range was 0.86% oxygen for an integration time of 0.3 s. The detection limit reported for a previous oxygen sensor24 was similar (1%); however, the integration time in that study24 was not specified. Figure 4 shows the time-dependent response for the DPA/ silicone film exposed alternately to 100% N2, air, and 100% O2. The fluorescence signal reaches 90% of its final value by about 30-35 s. The rate of the sensor response is limited by two factors: the diffusion of oxygen through the 0.4-mm-thick silicone film and the time required to purge the gas flow cell. If necessary, the response time of the sensor could be shortened by utilizing thinner silicone films. However, to achieve similar signal levels, it would then be necessary to increase the number of films or the DPA 3378 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
Figure 5. Dynamic response of oxygen sensor with a UV source based on a liquid scintillator. The oxygen transducer is a TLC plate incorporating silica gel with immobilized DPA.
concentration in the films. For example, a conservative estimate indicates that a silicone film of 50-µm thickness will yield a response time of less than 0.5 s. This calculation uses the literature27 diffusion coefficient of oxygen in silicone of 3.6 × 10-5 cm2/s. Also, simple improvements in the gas introduction arrangement and cell design should shorten the response time further. It should, therefore, be feasible to use a scintillator-based sensor for breath monitoring in the surgical theater. Sensor with a Liquid Scintillator. The fastest sensor response seems to arise when silica gel is chosen as the reagent immobilization medium.20 A real-time trace of the liquid-scintillator-based oxygen sensor with the TLC transducer is shown in Figure 5. Data were acquired at a rate of 10 points/s. The response and recovery times (90%) of the sensor were each about 200 ms. The response time was found to be a function of the gas flow rate, which suggested that it was limited in part by the time required to purge the gas lines and flow cell. Therefore, although the 200-ms response time is fast enough for real-time breath monitoring, the dynamic response might be further improved by modifications of the gas introduction arrangement and cell design. The faster response time of the TLC plate sensor (Figure 5), compared to the one based on silicone rubber (Figure 4), is a result of the high surface area of the silica gel layer. The same factor produces a higher sensitivity; the DPA fluorescence was quenched roughly twice as much in the TLC plate as in the silicone rubber. Unfortunately, TLC sensors often suffer interference from variations in humidity.28 Although this interference was not verified here, it can be overcome by embedding the modified silica gel in a hydrophobic polymer.29 CONCLUSIONS This study demonstrated for the first time that radionuclide/ scintillator combinations are well suited as UV sources for reagentbased chemical sensors. The UV sources that we developed have characteristics which are very close to those that are ideal for a sensor. The sources are small, low in cost, simple in design, and highly reliable. A host of chemical sensors require UV sources (27) Cox, M. E.; Dunn, B. J. Polym. Sci., Part A: Polym. Chem. Ed. 1986, 24, 621-636. (28) Barnikol, W. K. R.; Gaertner, T.; Weiler, N. Rev. Sci. Instrum. 1988, 59, 1204-1208. (29) Bacon, J. R.; Demas, J. N. Anal. Chem. 1987, 59, 2780-2785.
with these characteristics; only scintillators meet these requirements at present. Also, scintillator light sources can be built with emission bands from 145 to 600 nm by appropriate choice of the scintillator.16-18 They therefore constitute an attractive alternative to LEDs and diode lasers. Further, the high short- and long-term stability of the scintillator-based light sources offers a reproducibility of fluorescence measurements of about 0.35% RSD. The size of the sensor could be reduced considerably by employing more efficient shielding. The scintillator material could be used to pipe the light around a 3-mm lead shield. Further improvement of the scintillator light sources might be achieved by use of R-emitters and miniaturization to microscopic dimensions. R-emitters embedded in scintillating plastics or in encapsulated form would provide virtually hazard-free light (30) Weinstein, R. In Nuclear Engineering Fundamentals; Weinstein, R., Ed.; McGraw-Hill: New York, NY, 1964.
sources because of the very small penetration depth of this type of radiation. Even direct, external exposure to high-activity R sources is not harmful. The most energetic R particles from radionuclides penetrate only 40 µm into the 100-µm-thick layer of dead cells in human skin.30 Also, R particles do not generate bremsstrahlung, thereby eliminating the requirement of photodetector shielding. ACKNOWLEDGMENT This work was supported in part by the National Institutes of Health through Grant R01-GM 53560. Received for review December 30, 1996. Accepted April 24, 1997.X AC961296N X
Abstract published in Advance ACS Abstracts, July 1, 1997.
Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
3379
