Source of information: http://www.osl.dcu.ie/pdf-files/phase_fluorometric_dissolved_oxygen_sensor.pdf
1. Introduction
The determination of oxygen concentration is important in many areas of industry, medicine and the environment. The amount of oxygen dissolved in water is an indication of the quality of the water and careful control of oxygen levels is important in waste-water management and in fermentation processes. Optical dissolved oxygen (DO) sensors [1] are more attractive than conventional amperometric devices, because, in general, they have a fast response time, do not consume oxygen and are not easily poisoned. Sensor operation is usually based on the quenching of fluorescence in the presence of oxygen. In this work, the fluorescent ruthenium complex, [Run-Tris(4,7-diphenvl-U0-phenanthroline)]2+, (Ru-(Ph2phen)32+) was chosen because of its highly emissive metal-to-ligand charge-transfer state, long lifetime, and strong absorption in the blue-green region of the spectrum, which is compatible with high-brightness blue light-emitting diodes. The dye is entrapped in a porous, hydrophobic sol-gel film. The oxygen quenching process is described by the Stern-Volmer equations:
where / and ò are, respectively, the fluorescence intensity and excited state lifetime of the fluorophore, the subscript 0 denotes the absence of oxygen, Ksv the Stern-Volmer constant, ê the diffusion-dependent bimolecular quenching constant, and P02 the oxygen partial pressure.
Much has been published [2-4] on optical oxygen sensors which are based on intensity quenching of the fluorescence Eq. (1). It has now been established that these sensors have many drawbacks. These include susceptibility to light source and detector drift, to changes in optical path, and drift due to degradation or leaching of the dye. These effects can be minimised by operating the sensor in the time domain instead of the intensity domain [5,6]. The lifetime, x, is an intrinsic property of the fluorophore which, unlike intensity, is virtually independent of external perturbations. The lifetime is quenched (reduced) in the presence of oxygen and this quenching is described in Eq. (2). In the DO sensor reported here, the lifetime is monitored as a function of oxygen concentration using the phase fluorometry technique [7,8], where an oxygen-sensitive phase difference is measured between the modulated fluorescence signal and a modulated reference signal.
The objective of this work was to build a probe-based optical DO sensor based on the phase fluorometric principle. In this work, the main application for the sensor is in wastewater treatment. Specifications include a measurement range of 0-15 ppm, stability of 0.1 ppm per week, a limit of detection (LOD) of <10 ppb and single point calibration. In this paper, details of the optoelectronic sensor design as well as the design of the phase-measuring electronics are presented. Sensor performance is reported with respect to the above specifications, as well as the calibration procedure. Finally, advantages of the system over intensity-based sensing are highlighted.2. Film fabrication and test instrumentation
2.1. Film fabrication
A silica sol was prepared with water:precursor molar ratio (R) of 4 and pH = 1 as described previously [4]. The organically modified precursor used was methyltriethoxy-silane (MTEOS). A circular area of diameter -1 cm was coated onto the centre of a disc of PMMA (see Section 5). The film was cured at 70°C for 24 h. A black silicone rubber film (Wacker, Elastosil N189) was then coated onto the entire disc area to optically isolate the sensor film from the measuring environment.
2.2. Test instrumentation
The sensor response was measured by immersing the sensor in a water reservoir, into which mixtures of oxygen and nitrogen gases were flowed, controlled by mass flow controllers (Unit Instruments, UFC1100A). For temperature calibration, the reservoir was itself immersed in a constant temperature bath (Lauda, RE104).
3. Phase fluorometry
3.1. Principles of phase fluorometry
The fluorescence lifetime of an indicator is an intrinsic property and is virtually independent of fluctuations in light intensity, detector sensitivity and light path of the optical system [7]. In this laboratory, previous optical oxygen sensor designs [3,4] have been based on intensity quenching as described by Eq. (1). While these sensors exhibit very good signal-to-noise ratio (SNR) and repeatable response, they suffer from baseline drift due to LED output fluctuations and they are susceptible to drift due to sensor film positioning variations.
Fig. 1. Principle of phase fluorometric technique.
There is also the possibility of response variations due to dye leaching and photobleaching. Most of these problems can be overcome by using a phase fluorometric approach which involves operating in the time domain according to Eq. (2), instead of in the intensity domain. If the excitation signal is sinusoidally modulated, the dye fluorescence is also modulated but is time delayed or phase shifted relative to the excitation signal. The relationship between the lifetime, x, and the corresponding phase shift, ô, for a single exponential decay is [7]
where / is the modulation frequency. This phase shift is illustrated in Fig. 1. A disadvantage of phase fluorometry is that the SNR decreases with increasing modulation frequency, and since the phase sensitivity increases with modulation frequency [9], an optimal frequency has to be selected. This will be discussed, together with the phase measurement electronics in Section 5.
4. Sensor design considerations
A schematic of the probe design is shown in Fig. 2. A compact probe configuration was used to be compatible with the requirements of the particular application (waste-water monitoring). The probe measures approximately 15 cm in length with a diameter of 4 cm. Apart from the optoelectronics and sensor film, a compact preamplifier is also housed in the probe.
The excitation source is a blue LED (Nichia, NSPE590) and is chosen [10] for its relatively stable temperature characteristics which match those of the reference LED (see below). The detector is a silicon photodiode (Hama-matsu, S1223), which also exhibits good temperature stability. Modulated light from the blue LED is filtered using a blue glass bandpass filter (OF1: Schott, BG12) of thickness 2 mm in order to eliminate the high wavelength tail of the LED emission.
Fig. 2. Schematic of sensor probe.
The phase-shifted fluorescence from the sensor film is incident on the photodiode after passing through an optical long-pass filter (OF3: LEE-gel filter 135), to separate the excitation light from the emission.
From Fig. 2, it can be seen that the hydrophobic sol-gel sensor film, previously optimised for DO sensing [4], is coated on a PMMA disc which has an angled edge. The angle is chosen to optimise total internal reflection of the LED light into the sensor film. The PMMA was chosen as a substrate for the sensor film because of the ease of machining and with a view to ultimately designing an injection-moulded sensor cap to facilitate mass production. The sol-gel sensor layer is coated with an opaque black silicone rubber film, the thickness of which is a compromise between achieving complete opacity and minimal effect on the sensor response time. This constitutes an effective optical isolation layer which reduces background radiation due to ambient light in the measuring environment as well as avoiding LED excitation of any external fluorescent species. The second LED (Hewlett Packard, HLMA-KLOO) is part of an internal dual referencing scheme. This reference LED emits at 590 nm and is filtered by a bandpass filter (OF3: Schott, BG39). This LED is in the same spectral range as the fluorescence (610 nm), and has been carefully selected [10] to match the blue excitation LED in terms of switching time and temperature characteristics. Spurious phase shifts as a function of temperature and other fluctuations are eliminated by this dual referencing, the details of which are discussed in Section 5. A thermistor is inserted in the metal block adjacent to the sensor disc in order to monitor the measurement temperature.
5. Phase measurement electronics and data acquisition
A block diagram of the phase measurement system is shown in Fig. 3. Each LED is modulated at a frequency of 20 kHz. As mentioned in Section 3, this frequency has been carefully optimised as a compromise between reduced SNR at high modulation frequency and reduced phase sensitivity at lower frequency [10]. An additional factor in this design compromise is the LED intensity, which is reduced to a level at which photobleaching of the dye is negligible.
The modulated fluorescence signal from the photodiode is converted via a transimpedance amplifier (TR IMP) to a voltage signal. This is then amplified and bandpass filtered (AMP) to eliminate the DC component and the higher harmonics of the signal. A specific delimiter circuit (LIM) is used to prevent saturation of the operational amplifiers and overdrive of the comparator (COMP). A phase-shifted TTL signal is produced from the comparator, and is fed together with the reference excitation signal, in to the exclusive-or gate (EXOR). The output signal is then filtered by a lowpass filter (LP) giving a voltage signal proportional to the measured phase shift.
As discussed in Section 4, the sensor electronics are designed to eliminate errors in phase due to the electronics and associated temperature behaviour. This is achieved using a dual LED referencing system. As seen is Fig. 3, the excitation LED1 and the reference LED2 are alternately switched in order to determine the phase difference, ôò{, due to the electronics alone. This phase shift is subtracted in real time from the oxygen-dependent phase shift, ô^å, to obtain the specific sensor output phase shift.
Fig. 3. Schematic of phase-measuring electronics.
Most of the circuit shown in Fig. 3 is mounted separately òîò the sensor head. This unit includes a pressure sensor to nonitor ambient pressure during the measurement. The nitial amplifier stage is mounted in the sensor head Tig. 2) in order to minimise noise in the system.
Data acquisition and analysis are achieved via a PC and A/ D interface card. The phase shift, (0sig - 0ref) is recorded ogether with temperature and ambient pressure and processed in the software to give temperature and pressure-corrected calibration curves as described in the next section.6. Sensor performance and calibration protocol
6.1. Oxygen response
As discussed in Section 3, a modulation frequency of >0kHz was chosen as a compromise between oxygen ensitivity and SNR in the phase fluorometric system. A ypical sensor response curve, ô as a function of partial pressure of oxygen, is shown in Fig. 4. This response was measured over a complete oxygen concentration range of 0-100% at a fixed temperature of 20°C. It is clear that these data display good SNR and repeatability. The repeatability of the response for a particular sensor film is illustrated in Fig. 5. Here, the response of the film over a period of 3 months is expressed as a Stern-Volmer plot Eq. (2). The overlapping data indicates the stability of the response. As in Fig. 4, these data were recorded at 20°C.
6.2. Temperature correction
There are a number of contributions to the temperature dependence of the oxygen sensor response. As discussed in Section 5, any temperature dependence of the phase measurement electronics has been referenced out to a large extent by the dual LED system, resulting in the extremely low baseline temperature coefficient of 0.00087 degrees of phase/°C. The remaining contributions are the temperature dependence of the ruthenium fluorescence and the oxygen diffusion coefficient in the film [11].
Fig. 4. Typical oxygen sensor phase response.
Fig. 5. Overlay of Stern-Volmer curves for a film measured over a 3-month period.
These temperature effects are characterised by measuring the sensor response as a function of temperature. The response was measured in the range 5-30°C in steps of 2° and the data obtained are displayed in the 3-D plot shown in Fig. 6. These data are processed, as discussed in the next section, in order to generate a temperature-corrected calibration function.
6.3. Calibration protocol
Response curves as in Fig. 4 are analysed to produce the average phase value for each oxygen concentration. A cubic spline fitting procedure is used to fit the phase angle versus oxygen partial pressure data for each temperature. This is used to generate numerically a quasi-continuous set of ôdata for a continuous range of oxygen concentrations and temperatures which is stored in the software. An unknown oxygen concentration measured in the field is obtained by inputting the measured phase value which results in the generation of a corresponding temperature versus concentration data set. This is fitted as before and the correct oxygen concentration corresponding to the measurement temperature is obtained. The software also allows for adjustment of the calibration curve to correct for the ambient pressure on the day of measurement (if this differs from the pressure at the time of calibration).
6.4. Limit of detection and longterm stability
A SNR of -750 was measured experimentally. This gives rise to a very low LOD. The LOD and resolution of the sensor were measured as three times the standard deviation of the noise, while averaging for 30 s. Because of the non-linearity of the phase-oxygen response, seen in the nonlinear Stern-Volmer plots of Fig. 5, resolution varies with oxygen concentration. The LOD was measured to be 6.6 ppb or 0.15 hPa, while the resolution at 9 ppm oxygen concentration was measured to be 15 ppb or 0.34 hPa. This LOD comes well within the specification (<10 ppb) for the particular waste-water application.
Long term stability studies have been carried out on a number of sensor films, one of which has been in continuous use for over 12 months. The response of the films appears to be stable within the error of the laboratory test system. At the very least, the sensor meets the specified stability requirement of 0.1 ppm per week. The film itself is ragged and will withstand the extreme conditions of waste-water treatment plants.
Fig. 6. Phase response in the range 0-100% oxygen measured over the temperature range 5-30°C.
7. Conclusion
A high performance optical DO sensor, based on phase fluorometric detection, has been reported. The advantages of phase detection over intensity detection have been highlighted and the unique sensor head and phase measuring circuitry have been described. A dual LED referencing system and careful selection of optoelectronic and electronic components has resulted in a very stable baseline with low temperature coefficient. Sensor response is characterised by good SNR and repeatability. Sensor performance exceeds the initial specifications for the waste-water monitoring application. The LOD is <10ppb, while the long-term stability will enable reliable operation for periods of months at the very least. The temperature dependence of the sensor response has been characterised and incorporated in a calibration function. While the sensor will operate satisfactorily over the complete range of oxygen concentrations up to 100% DO, the response is particularly sensitive at low oxygen levels due to the dynamics of the optical quenching process in the heterogeneous environment of the micropor-ous film. The sol-gel sensor film is ragged and can be incorporated in a disposable cap configuration. The probe-based sensor head has been designed specifically for waste-water monitoring but the sensor can be repackaged for other potential applications. In particular, the use of miniature optoelectronic devices and the potential for micro-patterning the sol-gel film, could lead to applications in other fields such as medical and food packaging areas.
References
[1] O.S. Wolfbeis, in: S.G. Schulmann (Ed.), Fibre optical fluorosensors in analytical and clinical chemistry from molecular luminescence spectroscopy: methods and applications, Part II, Wiley, New York, 1988.
[2] E.R. Carraway, J.N. Demas, B.A. DeGraaf, J.R. Bacon, Photophysics and photochemistry of oxygen sensors based on luminescent transition-metal complexes, Anal. Chem. 63 (1991) 337-342.
[3] A.K. McEvoy, Ñ McDonagh, B.D. MacCraith, Dissolved Oxygen sensor based on fluorescence quenching of oxygen-sensitive ruthenium complexes immobilised in sol-gel-derived porous silica coatings, Analyst 121 (1996) 785-788.
[4] Ñ McDonagh, B.D. MacCraith, A.K. McEvoy, Tailoring of sol-gel films for optical sensing of oxygen in gas and aqueous phase, Anal. Chem. 70 (1) (1998) 45-50.
[5] O.S. Wolfbeis, I. Klimant, T. Werner, Ñ Huber, U. Kosch, Ñ Krause, G. Neurauter, A. Diirkop, Set of luminescence decay time based chemical sensors for clinical applications, Sens. Actuators  51 (1998) 17-24.
[6] P. Hartmann, M.J.P Leiner, M.E. Lippitsch, Response characteristics of luminescent oxygen sensors, Sens. Actuators  29 (1995) 251-257.
[7] M.E. Lippitsch, S. Draxler, Luminescence decay-time-based optical sensors: principles and problems, Sens. Actuators  11 (1993) 97-101.
[8] G.A. Hoist, T. Koster, E. Voges, D.W. Lubbers, FLOX — an oxygen-flux-measuring system using a phase-modulation method to evaluate the oxygen-dependent fluorescence lifetime, Sens. Actuators  29 (1995) 231-239.
[9] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, Plenum Press, New York, 1983.
[10] Ñ Kolle, Development and evaluation of a phase-fluorometricinstrument for luminescence based optical oxygen sensor. Ph.D. Thesis, University of Leoben, July 1999.
[11] N. Opitz, H.J. Graf, D.W. Lubbers, Oxygen sensor for the temperature range 300-500 Ê based on fluorescence quenching of indicator-treated silicone rubber membranes, Sens. Actuators  13 (1988) 159.