dual lifetime referenced trace ammonia sensors

Sensors and Actuators B 139 (2009) 132–138 Contents lists available at ScienceDirect Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb Dual lifetime referenced trace ammonia sensors K. Waich ∗ , S. Borisov, T. Mayr, I. Klimant Institute of Analytical Chemistry, Chemo- and Biosensor Group, Graz University of Technology, Technikerstr. 4/EG, 8010 Graz, Austria article info Article history: Available online 22 October 2008 Keywords: Trace ammonia Optical sensor Dual lifetime referencing (DLR) abstract Dual lifetime referencing is applied to trace ammonia sensors. Two schemes are presented to achieve luminescence lifetime based measurements in the frequency domain: the first is based on a resonance energy transfer (RET) system with eosin ethylester as indicator applying a coumarin derivative as donor. The second consists of a resonance energy transfer cascade with a coumarin derivative, sulforhodamine B and the absorption dye bromophenol blue as ammonia-sensitive dye. The large Stokes shift due to the energy transfer and the resulting spectral properties of the sensing layers enable dual lifetime referencing using an Ir(III)-coumarin complex for both ammonia detection systems. The sensing systems areapplicable for the monitoring of ammonia in the range of either 1–5000 ␮g/l or 50–50,000 ␮g/l, respectively. © 2008 Elsevier B.V. All rights reserved. 1. Introduction Many attempts have been made to meet the increasing demand of robust ammonia monitoring systems. Besides a simple employ- ment a crucial feature of these systems is a fast detection of ammonia at trace level concentrations in sea and in fresh water, because concentrations of less than 100 ␮g/l are toxic for many aquatic organisms [1]. The most suited detection method for low concentrations is the Berthelot reaction, allowing a detec- tion limit (LOD) of 10 ␮g/l of ammonia. With the integration into a microfluidic system Berthelot reaction became applicable for water monitoring, but there are some drawbacks, such as the consumption of reagents, the irreversibility of the sensor reac- tion itself and the error prone nature of the microfluidic system [2]. Established ammonia sensors for online measurements are based on a modified pH-electrode and allow the detection of ammonia with a LOD of 30 ␮g/l, but salinity affects the measure- ment [3,4]. Ion-selective electrodes for ammonium measurements with ammonium ionophores such as nonactine incorporated into a polymer layer provide the possibility to detect ammonia more selectively [5,6]. The sensing scheme for the detection of ammonia presented here is based on the deprotonation of a pH indicator and the resulting change of its optical properties. Several approaches of optical ammonia sensors employing the principle of pH mea- surement have been reported: a pH indicator is entrapped by a gas-permeable membrane and positioned at the end of an optical fibre. The ammonia penetrating causes a change of pH that results ∗ Corresponding author. Tel.: +43 316 8734324. E-mail addresses: kerstin.waich@TUGraz.at (K. Waich), torsten.mayr@TUGraz.at (T. Mayr). in a colour or fluorescence change of the indicator dye [7]. Alterna- tively lipophilized absorption indicators such as bromophenol blue and bromocresol green were blended into silicone. According to its pK a -value bromophenol blue showed lower LODs than bromocresol green [8–10]. Recently, we reported sensors for trace level concen- tration employing eosin ethylester as indicator [11]. These sensors enable a continuous monitoring of concentrations below 100 ␮g/l. The signal measured is fluorescence intensity, which is prone to be compromised by drifts in the optoelectronical setup, loss of light in the optical path and not well suited to turbid samples. Routes to eliminate these drawbacks are the application of refer- encing methods. However, potential referencing systems cannot be applied directly on the eosin ethylester based sensors: 1. Simple ratiometric dual wavelength referencing methods due to the lack of the right spectral properties of the dye. 2. Dual lifetime referenc- ing (DLR), b ecause common long-lived reference luminophores do not match the spectral properties of eosin ethylester [12–16]. In this contribution we describe the application of the dual lifetime referencing method to convert the fluorescence inten- sity information into a phase shift. This was achieved by the employment of resonance energy transfer (RET) to adjust the spec- tral properties of the ammonia-sensitive materials matching the spectral properties of the reference dye. An iridium complex was chosen as reference luminophore. The investigated sensors showed response in a range of 1–1000 ␮g/l or 50–10,000 ␮g/l of ammonia, respectively. 2. Experimental 2.1. Reagents Unless otherwise noted, materials were obtained from com- mercial sources and were used without further purification. 0925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2008.10.010 K. Waich et al. / Sensors and Actuators B 139 (20 09) 132–138 133 Cellulose acetate (MW 100,000; CAc) and cellulose acetate butyrate (CAcB) with 16.5–19% butyrate content were purchased from Sigma–Aldrich. They were washed with deionised water and dried in a drying chamber at 80 ◦ C for 48 h to evapo- rate unreacted acid. Eosin ethylester (E), sulforhodamine B (S) and bromophenol blue (B) was purchased from Sigma–Aldrich (www.sigmaaldrich.com), acetone from Merck (www.merck.de). Silicone (PP2-RG01, 2 part reprographic Silicone, Si1) was pur- chased from ABCR. The components of Si2 vinyl-terminated poly- dimethylsiloxane (PDMS) (DMS-V21, viscosity 100 cSt.), methylhy- drosiloxane/dimethylsiloxane copolymer (HMS-301), tetravinylte- tramethyl cyclotetra-siloxane (SIT 7900.0) and platinum divinyl- tetramethyl siloxane complex (PC075) were also obtained from ABCR (www.abcr.de). Poly(vinylidene chloride-co-acrylonitrile) (PViCl–PAN, MW 150,000 containing 20% wt. polyacrylonitrile) was purchased from Polysciences (www.polysciences.com). 10- (2-Benzothiazolyl)-2,3,6,7-tetrahydro-1,1,7,7-tetramethyl-1H, 5H, 11H-(1)benzopyropzrano(6,7-8i,j)quinolizin-11-one (C S-Jul )was purchased from Sigma–Aldrich and 3-(5-chlorobenzooxazol-2- yl)-7-(diethylamino)-coumarin (C O-Cl ) from Simon & Werner GmbH, Frankfurt, Germany. The synthesis of the iridium(III) complex with 3-(benzothiazol-2-yl)-7-(diethylamino)-coumarin (Ir(C S ) 2 (acac)) is reported elsewhere [17]. 2.2. Preparation of the reference microbeads Fifteen milligrams of Ir(C S ) 2 (acac) [17] and 1 g of PViCl–PAN was dissolved in 25 g of tetrahydrofurane. The obtained “cocktail” was sprayed in three portions (∼8 g each) into a glass container (V =10 L, Schott) using an airbrush HP 120 (www.conrad.at). To promote fast drying, prior to use the glass container was filled with warm air and its walls were warmed up to ∼60 ◦ C. The obtained beads were removed mechanically from the walls and washed three times with ethanol to remove the dye from the beads’ surface. The beads were then washed twice with water and freeze-dried. 2.3. Preparation of sensing membranes Sensing membranes were prepared by spreading cocktails con- taining cellulose esters (2.5% (w/w) polymer content) and dyes dissolved in acetone onto polyester films (thickness: 175 ␮m; Goodfellow, Huntingdon, UK) with a homemade knife coating device to obtain membranes of a calculated thickness of approx. 1 ␮m. Small amounts of water (0.5 and 3%) had to be added to the solutions of B and S, because these compounds were not sol- uble in pure acetone. When adding the solutions to the polymer cocktails they did not precipitate. All cocktails contained addition- ally 0.6 ␮mol of methane sulfonic acid. After the evaporation of the solvent the membranes were conditioned with concentrated gaseous ammonia and left in a drying chamber at 70 ◦ C for 12 h [11]. The ammonia-sensitive layers M1 and M3 were covered with a 5% (w/w) silicone solution in n-hexane to obtain an approx. 1 ␮m thick Si1 layer. The “cocktail” of silicone primers for Si2 was obtained by mixing tetravinyltetramethyl cyclotetrasiloxane (1.5 ␮L), 200 mg of the vinyl-terminated PDMS and 30 mg of methylhydrosiloxane–dimethylsiloxane copolymer. Then, 10 mg of the reference beads were mixed in. Finally, 3 ␮L of the platinum complex catalyst were added. The “cocktail” was stirred in a glass vial at 150 0 rpm for 1 min. It was then coated onto the ammonia- sensitive layers to obtain M2 and M4. The 10 -␮m-thick silicone layer (Si2Ir) was cured at 50 ◦ C for 5 h. For the composition and the thickness of the membranes see Table 1. Table 1 Composition and thickness of the sensing layers. Coumarin – [%] S [%] Indicator dye [mmol/kg] Polymer Protective layer, thickness M1 C O-Cl 1 E 2CAcBSi1 –1␮m M2 C O-Cl 1 E 2CAcBSi2Ir –10␮m M3 C S-Jul 2 0.1 B 2CAcSi1 –1␮m M4 C S-Jul 2 0.1 B 2CAcSi2Ir –10␮m 2.4. Apparatus 2.4.1. Spectral characterization of the sensing membranes Fluorescence measurements (spectra and intensity vs. time) were performed using a Hitachi F-7000 fluorescence spec- trometer. Sensor membranes were mounted in a fluorescence spectrophotometer equipped with a home-made flow-through cell thermostated. All measurements were performed at 25 ◦ C exposing the membranes to solutions with dissolved ammonia. 2.4.2. Dual lifetime referenced measurements Phase angle measurements were performed using following sys- tem: a dual-phase lock-in amplifier (DSP 830, Stanford Research Inc.) was used for sine-wave modulation of the LED frequency of 30 kHz and for detection. The optical system consisted of a blue LED (450 nm, Roithner Lasertechnik, Austria) equipped with a blue band-pass filter (BG12, Schott, Mainz, Germany), a bifurcated glass fibre bundle (Ø2 mm) and a red-sensitive PMT module (H5701-02; Hamamatsu, Germany) equipped with a long-pass filter (OG530 or OG560, Schott, Mainz, Germany). The fibre-bundle was placed on the flow-through cell. 2.5. Solutions and buffers Aqueous solutions were prepared from deionised water. Stock solutions were prepared dissolving ammonium chloride in a buffered solution showing pH 7. The pH was adjusted using 100 mM sodium phosphate buffer. The amount of free NH 3 in water is defined by the pH-value of the solution. The ammo- nia concentration for a certain pH can be calculated by the Henderson–Hasselbalch equation. Temperature corrected concen- trations were calculated according to data listed in Ref. [18]. 3. Results and discussion 3.1. DLR-reference The DLR method relies onsimultaneous excitation of thefluores- cent indicator and the luminescent reference dye and on measuring the overall phase shift. The overall phase shift results from the ratio of the intensities of the reference and the indicator [12–16]. The application of DLR in fluorescence sensing requires a reference luminophore and a fluorescent indicator meeting the following cri- teria: (a) the reference luminophore and the indicator fluorophore have large different decay times, (b) spectral properties includ- ing decay time, quantum yield and spectral shape of the reference luminophore are not affected by the sample, (c) the indicator flu- orophore changes its fluorophore intensity as a function of the analyte concentration, (d) the indicator and the reference can be excited at a single band of wavelength due a strong overlapping of the excitation spectra, (e) the emission of both the indicator and reference can be detected at a common wavelength or band of wavelength using a single photodetector. An Ir(III)-coumarin complex (Ir(C S ) 2 (acac)) was chosen as ref- erence dye due to its spectral properties showing an absorption 134 K. Waich et al. / Sensors and Actuators B 139 (2009) 132–138 maxima at 472 nm and 444 nm and a emission maximum at 593 nm. In addition its decay time is approx. 11 ␮s and its quan- tum yield is <0.5. However, Ir(C S ) 2 (acac) is reported as indicator in optical oxygen sensors [17]. For this reason, the Ir-coumarin com- plex is encapsulated into a polymer matrix with minimized gas permeability. Generally, several methods are used for incorporation of dyes and indicators into polymer beads: a, incorporation during the polymerization process; b, staining by swelling the beads; c, staining by precipitation from a solution; and d, covalent attach- ment of a dye to the surface of the beads bearing reactive groups. All these procedures are adequate for certain combinations of poly- mers, dyes and solvents; but they are not necessarily suitable for the preparation of the oxygen-blocking beads; and some of them are rather laborious. We have found a simple easy and simple way to prepare beads by spraying a cocktail of dissolved dye in warm air. Particularly, for the preparation of reference beads, thegas-blocking copolymer of vinylidene chloride and acrylonitrile (PViCl–PAN) is a suitable polymer matrix. The solvent, tetrahydrofurane, evaporates fast in the warm air so that microbeads are easily obtained. The size of the beads was determined to be 5–20 ␮m, which is adequate for the referencing purposes. The luminescence decay time of the dye encapsulated in microparticles is virtually oxygen-independent ( = 9.40 ␮s at air saturation and 9.48 ␮s in the absence of oxygen). 3.2. Ammonia-sensitive materials 3.2.1. Coumarin derivatives as donors in RET The spectral properties of pH-indicators suitable for the detec- tion of ammonia in trace level concentration do not sufficiently overlap with reference dyes commonly used in DLRsensors [12–14]. To circumvent this problem we employ an energy transfer system to adjust these spectral properties. Coumarin derivatives (C S-Jul and C O-Cl ) were used as donor in this energy transfer. In brief, RET is a non-radiative process. An excited state donor transfers energy to a proximal ground state acceptor through long-range dipole–dipole interactions. The rate of energy transfer is highly dependent on the extent of spectral overlap, the relative orientation of the transi- tion dipoles and the distance between the donor and the acceptor molecules [19]. Coumarin derivatives (C S-Jul and C O-Cl ) were used as donor showing emission maxima at 503 nm and 526 nm, respec- tively. Their emission results in a sufficiently large spectral overlap with the acceptor dyes. Coumarin derivatives with functional groups are known to show a pH-dependence in their spectral characteristics. Influences on the spectral properties below pH 1 were observed for C O-Cl and C S-Jul , yet do not influence the sensing scheme. 3.2.2. RET-system employing C O-Cl and eosin ethylester The absorption maximum of eosin ethylester is at 537 nm. This makes it a suitable acceptor dye for C O-Cl , which has an emission maximum at 503nm (see Fig. 1). The grey regions in Fig. 1b indicate the spectral overlap with C O-Cl . However, the energy transfer is not complete which is illustrated by the rather high C O-Cl – peak in the fully deprotonated E spectrum (Fig. 1c). The extension of the Stokes shift from 15 nm for E to 91 nm in the energy transfer system enables an excitation and detec- tion at the same wavelengths as the used reference luminophore (Ir(C S ) 2 (acac)). Although a system containing the components C S-Jul and E, shows a better spectral overlap (not shown here) than E and C O-Cl , the remaining signal change when the ammonia-sensitive material is exposed to ammonia is not high enough for the employment of the DLR method. The red-shifted emission spectrum and the higher Fig. 1. (a) Illustration of the RET system C O-Cl (donor)–E (acceptor) and absorption and emission maxima of the compounds. (b) Spectral characteristics of a RET sys- tem containing C O-Cl and E; the grey area indicates the spectral overlap of the C O-Cl emission spectrum and the E absorption spectrum. (c) M1 containing 1% of C O-Cl and 2 mmol/kg E exposed to different concentrations of ammonia dissolved in sodium phosphate buffer (100 mM, pH 7); (Abs. is absorption, Em. is emission). fluorescence intensities of C S-Jul at 553 nm diminish the intensity change of E when being deprotonated. 3.2.3. RET-cascade-system employing C S-Jul , sulforhodamine and bromophenol blue As an alternative approach an energy transfer cascade was employed (M3 and M4, see Fig. 2). The first RET occurs between C S-Jul and sulforhodamine B (S). The spectral overlap of the two components is implied by the light grey and grey region between 500 nm and 600 nm in Fig. 2b. S is used as fluorescent emitter of the energy transferred from the coumarin derivative. The intensity dependence on the ammonia concentration is the result of the sec- ond RET occurring between S and the pH indicator bromophenol blue (B). B is an absorption dye and its protonated form has a band with a maximum at 430 nm. With the extent of deprotonation the absorption band at 430 nm shrinks and the band of the deproto- K. Waich et al. / Sensors and Actuators B 139 (20 09) 132–138 135 Fig. 2. (a) Illustration of the RET system C S-Jul (donor)–S (acceptor–donor)–B (acceptor) and absorption and emission maxima of the compounds. (b) Spectral characteristics of a RET system containing C S-Jul , S and B; the light grey and grey area indicates the spectral overlap of the C S-Jul emission spectrum and the S absorption spectrum and the grey and dark grey indicates the spectral overlap of the S emission spectrum and the B absorption spectrum. (c) M3 containing 2% of C S-Jul ,0.1%ofS and 2mmol/kg B exposed to different concentrations of ammonia dissolved in sodium phosphate buffer (10 0 mM, pH 7); (Abs. is absorption, Em. is emission). nated form with its maximum at 604 nm appears. The higher the absorption band of B at 604 nm becomes, the more the fluorescence intensity of S is quenched and the less fluorescence is emitted by S. The absorption of the deprotonated form of B has a molar extinction coefficient of 75,200 M −1 cm −1 . Together with the spectral overlap (Fig. 2b, grey and dark grey between 550 nm and 650 nm) this leads to a very ef ficient energy transfer and the decrease of the fluores- cent signal of S can be used as ammonia detection signal. Combining S and B the low pK a -value of B is utilized to obtain low detection limits for ammonia and the advantages of fluorescence measure- ments can be used. In addition, smaller amounts of indicator (2 mmol/kg polymer) are needed to get a detectable fluorescence signal compared to absorption based sensors. Consequently thin- ner sensing layers can be produced which results in a faster response. Linking this ammonia indication system composed of S and B with the RET from C S-Jul the Stokes shift is enlarged to 96 nm (Fig. 2b). The excitation wavelength at 450 nm and the emission wavelength at 560 nm enable the combination with Ir(C S ) 2 (acac), as DLR reference. 3.3. Choice of polymers and sensor assembly Ammonia sensors employing cellulose esters as host polymers for pH-sensitive dyes covered with a silicone membrane to only allow ammonia to affect the output were studied, recently [11].In general, the sensitivity rises with the polarity of the cellulose ester. In this study we observed that E incorporated in CAc membranes is partly deprotonated in buffer without analyte. This leads to a cer- tain level of fluorescence emission of the indicator in absence of the analyte resulting in an insufficient total signal change. When E is incorporated in CAcB the effect of deprotonated dye in buffer is diminished, because the CAcB sensor membrane is less susceptible to deprotonation. This leads to a higher total signal change and a reasonable resolution of the DLR sensor. For this reason CAcB was used as bulk polymer for the membranes M1 and M2. B was incor- porated into CAc (M3 and M4), because the partly deprotonation in buffer was not observed. To avoid cross sensitivities through pH and salinity and to hinder leaching mainly of S and B, which are very well water soluble, the sensing membranes were coated with a protective silicone layer. 136 K. Waich et al. / Sensors and Actuators B 139 (2009) 132–138 Fig. 3. Cross-section of membrane (not to scale). (a) M1 and M3 and (b) M2 and M4. The polyester support serves as an inert and optically transparent mechanical support. For membranes M1 and M3, Si1 was solubilised in n-hexane to be applied onto the ammonia-sensitive layers (Fig. 3a). For the dual lifetime referenced sensing membranes M2 and M3 the sil- icone layer is also employed as bulk polymer to incorporate the DLR-referencing beads. However, the suspension of the reference particles in the Si1–n-hexane cocktail yielded in a swelling of the particles. This makes the Ir(III)-coumarin complex accessible for oxygen and leads to an undesired oxygen sensitivity of the sensor membrane. Therefore a silicone (Si2) with a viscosity low enough to be used without being solubilised was chosen to incorporate the referencing beads. Using this alternative, homogeneous mem- branes of a thickness of 10 ␮m were prepared (Fig. 3b). 3.4. DLR-sensors’ characteristics The spectral properties of the dyes applied in DLR sensors are shown in Figs. 4 and 5. The reference Ir(C S ) 2 (acac) is excited with a 450 nm LED together with the coumarin derivative, which trans- fers its energy to E or to the indication system containing S and B. Although the emission spectra differ, the emission of both the Fig. 4. Spectral properties of the luminescent dyes and optical components used for DLR (a) absorption and emission spectra of the Ir(C S ) 2 (acac), a schematic represen- tation of the excitation source (LED 450) and the filter used (OG530). (b) Spectral characteristics of M1 on exposure to different concentrations of ammonia. Fig. 5. Spectral properties of the luminescent dyes and optical components used for DLR. (a) Absorption and emission spectra of the Iridium (III) complex, a schematic representation of the excitation source (LED 450) and the filter used (OG560). (b) Spectral characteristics of M3 when being exposed to different concentrations of ammonia. reference and E or S can be detected with one detector using an appropriate long-pass filter. The response to different ammonia concentrations of the E- containing membrane M2 is shown in Fig. 6a. The emission of the indicator E in M2 rises in contact with the analyte. As the phase angle of the reference beads is not affected by ammonia, the decrease in the phase angle results only from the changes in the ratio of the intensities of the luminophore and the fluorophore. The fluorescence of E is enhanced by deprotonation, which results in a decrease in the phase angle. An increase in the concentration of ammonia results in smaller phase angles, because more short-lived E fluorescence can be detected along with the constant lumines- cence of the reference luminophores. An overall signal change of 4.5 ◦ was observed. The limit of detection was calculated to approx- imately 1 ␮g/l. A response time t 95 of ∼30 min on going from 25 ␮g/l to 1 00 ␮g/l and a recovery time of 50 min from 100 ␮g/l to 25 ␮g/l of are found. This response time is acceptable regarding the small concentrations. The reason for the long response time is assumed to the 10-␮m-thick silicon layer covering the sensitive membrane. Sensors not referenced by DLR such as M1 and M3 covered with a1-␮m silicone layer, showed response times of less than 10 min (data not shown). Fig. 6b shows the response curve of the sensing membrane M4 containing the RET-cascade in combination with the reference particles when exposed to different concentrations of ammonia in phosphate buffer. In contact with the analyte the fluorescence intensity of S is quenched by B and therefore the emission overlap integral of the reference beads is diminished. As a result the emis- sion is dominated by the luminescence of the Ir(C S ) 2 (acac)-beads with an increasing ammonia concentration yielding in an increas- ing phase shift. A response time t 95 of ∼20 min on going from 100 K. Waich et al. / Sensors and Actuators B 139 (20 09) 132–138 137 Fig. 6. Response curves of (a) M2 and (b) M4 on exposure to various concentration of ammonia in phosphate buffer (100 mM, pH 7.0). to 500 ␮g/l and a recovery of ∼40 min were found. A signal change of ∼10 ◦ was observed, which significantly higher compared to the sensing membranes describ ed above. However, the calculated LOD is 50 ␮g/l. In previous experiments we showed that there is no cross sen- sitivity through pH and ionic strength, what makes the sensor materials suitable for measurements in sea and in fresh water. Nevertheless, there is a cross sensitivity through amines. We inves- tigated the response of the sensing membranes to trimethylamine, which is the most abundant amine in the target measurement envi- ronment. The response was in the same order of magnitude as the response to ammonia. However, the fact that the sensor per- formance is not altered after exposure to TMA makes the sensor suitable for the target applications in spite of this interference. It quickly indicates a critical ammonia andamine concentration, what makes it possible to act fast enough to avoid the aquatic life to be effected [11]. 4. Conclusion A powerful sensing scheme for the determination of ammonia applying dual lifetime referencing is presented. Coumarin deriva- tives were employed as donors in a resonance energy transfer to pH-indicators enabling the DLR method. Employing this scheme is beneficial because signals are not affected by drifts in the optoelec- tronical setup, loss of light in the optical path and difficult in the case of turbid samples. In addition, the sensors exhibit dynamic ranges of 1–500 0 ␮g/l or 50–50,000 ␮g/l, respectively. For this rea- sons, the presented sensors are suitable for continuous detection of ammonia in sea and in fresh water, as warning devices at con- centrations toxic for aquatic habitat. Acknowledgements Financial support by the Austrian Nano Initiative of the Aus- trian Science Fund FWF (Research Project Cluster 0700 – Integrated Organic Sensor and Optoelectronics Technologies – Research Project 0701 and 0702) is gratefully acknowledged. References [1] B. Timmer, W. Olthuis, A. van den Berg, Ammonia sensors and their applications, a review, Sensor and Actuators B 107 (2005) 666–677. [2] A. Daridon, M. Sequeira, G. Pennarun-Thomas, H. Dirac, J.P. Krog, P. Gravesen, J. Lichtenberg, D. Dermond, E. Verpoorte, N.F. de Rooij, Chemical sensing using an integrated microfluidic system based on colorimetrics: a comparative kinetic study of the Bertholet reaction for ammonia determination in microfluidic and spectrophotometric systems, Sensors and Actuators B 76 (2001) 235– 243. [3] R.F. Thomas, R.L. Booth, Selective electrode measurement of ammonia in water and wastes, Environmental Science and Technology 7 (1973) 523–526. [4] M.D. Love, H.L. Pardue, G. Pagan, Evaluation of transient responses of ammonia- selective potentiometric electrodes for quantitative applications, Analytical Chemistry 64 (1992) 1269–1276. [5] M.E. Meyerhoff, Y.M. Fraticelli, J.A. Greenberg, J. Rosen, S.J. Parks, W.N. Opdy- cke, Polymer-membrane electrode potentiometric sensing of ammonia and carbon dioxide in physiological fluids, Clinical Chemistry 28 (1982) 1973– 1978. [6] J.S. Benco, H.A. Nienaber, W.G. McGimpsey, Synthesis of an ammonium ionophore and its application in a planar ion-selective electrode, Analytical Chemistry 75 (2003) 152–156. [7] M.A. Arnold, T.J. Ostler, Fiber optic ammonia gas sensing probe, Analytical Chemistry 58 (1986) 1137–1140. [8] C. Preininger, G.J. Mohr, I. Klimant, O.S. Wolfbeis, Ammonia fluorosensors based on reversible lactonisation of polymer-entrapped rhodamine dyes and the effects of plasticizers, Analytica Chimica Acta 334 (1996) 113–123. [9] M. Trinkel, W. Trettnak, F. Reininger, R. Benes, P. O’Leary, O.S. Wolfbeis, Analytica Chimica Acta 320 (1996) 235–243. [10] M. Trinkel, W. Trettnak, F. Reininger, R. Benes, P. O’Leary, O.S. Wolfbeis, Study of the performance of an optochemical sensor for ammonia, Analytica Chimica Acta 320 (1996) 235–243. [11] K. Waich, T. Mayr, I. Klimant, Fluorescence sensors for trace monitoring of dis- solved ammonia, Talanta 77 (2008) 66–72. [12] I. Klimant, C. Huber, G. Liebsch, G. Neurauter, A. Stangelmeyer, O.S. Wolfbeis, in: B. Valeur, J.C. Brochon (Eds.), New Trends Fluorescence Spectroscopy: Appli- cation to Chemical and Life Science, Springer, Berlin, 2001, p. 257 (Chapter 13). [13] C. Huber, I. Klimant, C. Krause,O.S.Wolfbeis, Dual lifetime referencing as applied to a chloride optical sensor, Analytical Chemistry 73 (2001) 2097–2103. [14] T. Mayr, I. Klimant, O.S. Wolfbeis, T. Werner, Dual lifetime referenced optical sensor membrane fort he determination of copper(II) ions, Analytica Chimica Acta 462 (2002) 1–10. [15] G. Liebsch, I. Klimant, C. Krause, O.S. Wolfbeis, Fluorescent imaging of pH with optical sensors using time domain dual lifetime referencing, Analytical Chem- istry 73 (2001) 4354–4363. [16] C. von Bueltzingsloewen, A.K. McEvoy, C. McDonagh, B.D. MacCraith, I. Kli- mant, C. Krause, O.S. Wolfbeis, Lifetime-based optical sensor for high-level pCO 2 detection employing fluorescence resonance energy transfer, Analyst 127 (2002) 1478. [17] S.M. Borisov, I. Klimant, Ultrabright oxygen optodes based on cylometa- lated Iridium(III) coumarin Ir(CS)2(acac)es, Analytical Chemistry 79 (2007) 7501–7509. [18] http://edis.ifas.ufl.edu. [19] K.E. Sapsford, L. Berti, I.L. Me dintz, Materials for fluorescence energy trans- fer analysis: beyond traditional donor–acceptor combinations, Angewandte Chemie 118 (2006) 4676–4704. Biographies Kerstin Waich worked as a PhD student at the Institute of Analytical Chemistry and Radiochemistry at the Graz University of Technology (Austria) from 2005 to 2008. Her research interests are trace ammonia sensors based on pH indicators and luminescent nanomaterials. Sergey M. Borisov received his PhD degree in chemistry from the Herzen State Pedagogical University (St. Petersburg, Russia) in 2003. In 2004–2006 he was a post-doctoral fellow in the University of Regensburg (Germany). Since 2006 he is a post-doctoral fellow at the Institute of Analytical Chemistry and Radiochem- istry at the Graz University of Technology (Austria). His research interests are in the chemistry of porphyrins, in the application of luminescent probes in optical sensing, and in the use of luminescent micro- and nanomaterials in bioanalytical methods. 138 K. Waich et al. / Sensors and Actuators B 139 (2009) 132–138 Torsten Mayr received his PhD in chemistry from the University of Regensburg (Germany) in 2004. In 2002–2004 he was a post-doctoral fellow at the Karolinska Institute in Stockholm (Sweden). Since 2004 he is assistant professor at the Institute of Analytical Chemistry and Radiochemistry at the Graz University of Technology. His research interests include optical chemical sensors, biosensors, functionalized micro- and nanoparticles and microfluidic systems. Ingo Klimant received his PhD in chemistry in 1993 from the Karl-Franzens Univer- sity in Graz. Since 2001 he is full professor at the Institute of Analytical Chemistry and Radiochemistry of the Graz University of Technology (Austria). His areas of interest include optical chemical sensors, analytical methods in biotechnology and molecule spectroscopy. . 2008 Keywords: Trace ammonia Optical sensor Dual lifetime referencing (DLR) abstract Dual lifetime referencing is applied to trace ammonia sensors. Two schemes. B: Chemical journal homepage: www.elsevier.com/locate/snb Dual lifetime referenced trace ammonia sensors K. Waich ∗ , S. Borisov, T. Mayr, I. Klimant Institute