OXYGEN OPTICAL SENSORS BASED ON LUMINESCENCE QUENCHING OF ORGANOMETALLIC COMPLEXES EMBEDDED IN POLIMERIC MATRIXES

The subject of the present research deals with optical sensors for detecting oxygen. They are based on the quenching by oxygen of the luminescence of organometallic complexes embedded in polymeric matrixes. Excitation light is provided by a LED source and a photodiode is employed as detector. Optical sensor may substitute electrochemical ones, because they allow in situ, real time, non destructive measurements. They are more robust than electrochemical ones reducing the need of frequent calibration and membrane replacement. Attention has been focused on luminescence-intensity-based sensors, instead of lifetime-based ones, because they are more promising to build low cost industrial sensors. Final aim is the realization of a sensor working in a wide concentration range and up to 90°C. No commercial sensor with such characteristics is available. Stern Volmer model describes dynamic quenching, and oxygen concentration may be obtained from luminescence quenching according to: I0/I=1-K'sv*%O2 where I and τ are the luminescence intensity and excited-state lifetime of the luminophore, respectively. I0 and τ0 denote the same parameters in the absence of oxygen. The Stern Volmer constant K’SV is proportional to the luminophore lifetime in the absence of oxygen, τ0, oxygen diffusion coefficient in the polymeric membrane, and oxygen solubility into the membrane, . Various luminophores having various lifetimes in the absence of oxygen has been tested in order to optimize sensor analytical performances ruthenium tris-(4,7-diphenyl-1,10-phenanthroline) bis(octylsulphate) (Ru(dpp)OS, τ0=6μs), 5,10,15,20-Tetrakisphenyl-21H,23H-porphine platinum(II) (PtTPP, τ0=50μs), platinum 5,10,15,20-Tetrakis(pentafluorophenyl)-21H,23H-porphine platinum(II) (PtTFPP τ0=70μs), 5,10,15,20-Tetrakis(pentafluorophenyl)-21H,23H-porphine palladium(II) (PdTFPP, τ0=850μs) and 2,3,7,8,12,13,17,18-Octaethyl-21H,23H-porphine palladium(II) (PdOEP, τ0=990μs). They have been embedded in polysulfone (PSF) or polyvinylchloride (PVC). Dip-coating and spin-coating deposition procedures have been optimized. Stern Volmer model has been modified to take into account contributions to light intensity not considered in Stern Volmer model. In particular, a new procedure may determine light emission contribution from luminophores enclosed in sites where oxygen cannot enter. This correction allowed to demonstrate that curvatures of the SV calibration, often cited in the literature, come just from the cited contribution. In particular, SV calibration of three different PSF membranes embedding Ru(dpp)OS, PtTPP and PdTFPP, are linear. Luminophore degradation influence over luminescence drift has been analyzed, and a correcting algorithm has been developed. At 30°C, Ru(dpp)OS has a luminescence drift of -1,01•10-4 s-1 whilst, for porphyrins, it has been proved negligible at room temperature. Drift becomes more influent at higher temperature, because of luminophores thermal degradation. A PtTFPP-PSF sensor was studied in details to determine activation energies of the involved processes: Stern Volmer constant, K’SV, light emitted intensity in absence of oxygen, I0, and sensor response time, t1, have been determined at various temperature. Employing a suitable physical model like Arrhenius equation, free activation energies, ΔG‡, of diffusion and non radiative decays processes have been obtained. They are 2.8(0.3) kJ/mol and 16.5(0.5) kJ/mol, respectively. ΔH relative to the solubility in the membrane has been calculated too, obtaining 13(3) kJ/mol. Membrane sensitivity, K’SV, and maximum detectable oxygen percentage , has been calculated for various membranes. Most sensitive membranes are characterized by lower maximum detectable oxygen percentage. In order to extend the sensor working interval to higher oxygen percentages maintaining high sensitivity, two routes have been followed: 1) dynamic calibration; 2) construction of a “bi-label” sensor. 1) The dynamic calibration model is based on the transient intensity light profiles vs. time instead of equilibrium intensities. As theoretically demonstrated and experimentally confirmed, transient intensity during oxygen exit from the membrane has a sigmoidal shape. The parameters of this sigmoid do not vary with oxygen starting concentration, and the only variable is the inflection time, which may be employed as analytical quantity instead of equilibrium light intensity. The great advantage is that inflection point time may be determined for each %O2 value even with very sensitive membranes. Experimental verification has been performed on Ru(dpp)OS, PtTPP and PdTFPP membranes embedded in PSF. The precision of “classic” measurement, based on the light intensity measurement at equilibrium, is almost constant with increasing %O2, and equal to 3.5, 0.7 e 0.4 % for Ru(dpp)OS, PtTPP and PdTFPP, respectively. In dynamic calibration model, precision decreases with increasing %O2. The dynamic model is preferable to classical one for low oxygen concentration (less than 97%, 9.2%, e 7.2%, for membranes containing Ru(dpp)OS, PtTPP or PdTFPP, respectively). Classical measurements have been proved more sensitive than dynamic measurements for large oxygen percentages and membranes with high K’SV, while the opposite is valid for low oxygen percentages and membranes with low K’SV. The oxygen percentage where the two methods have the same sensitivity is 60%, 6% e 2% for Ru(dpp)OS, PtTPP and PdTFPP respectively. Dynamic calibration model is better than classical for low oxygen concentration determination and for application fields requiring an extended working range. Emission profiles measurement, however, is more complicated than equilibrium intensity measurement, and requires a reference gas (i.e. nitrogen) limiting its applicability in the portable sensor field. 2) A sensor embedding two luminophores in the same polymeric matrix (“bi-label sensor) has been prepared. Sensor behaviour has been theoretically rationalized and experimentally verified in two cases: Ru(dpp)OS and PtTPP embedded in PSF and PtTPP and PdTFPP embedded in PVC. The two luminophores have been demonstrated to behave independently into the matrix. A working graph has been obtained in order to predict optimal membrane composition to extend the working range up to required oxygen concentration optimizing sensor sensitivity. In the considered cases, for a working range from 0 to 100% , luminophores with K’SV near to 0.02 (K’SV (Ru(dpp)OS in PSF)= 0.014, K’SV(PtTPP in PVC)=0.019) and to 0.2 (K’SV (PtTPP in PSF)= 0.14, K’SV(PdTFPP in PVC)=0.27) have been chosen. Luminophores optimal molar fraction realized the condition that emission intensity fraction due to PtTPP is 0.45 e 0.31 of overall emitted intensity, for PSF and PVC, respectively. A commercial sensor prototype has been built. In order to obtain a robust sensor, whose response is not influenced by most of instrumental factors, pulsed light source have been employed to reduce photodegradation and optical fibers allowed to isolate light sources and detectors from temperature change in the analyzed mixture. A software has been developed in order to control simultaneously all the instruments (flow meters, oven, pulse generator, etc.) and to automate measurements and data elaboration. Sensors embedding PtTFPP have been tested continuously 24 hours a day for one month. If test is carried at room temperature, the luminescence decrease is close to 7.1% and measurement repeatability is very good. If the same test is carried at 90 °C, luminescence decrease is equal to 28.7% but measurement repeatability, using drift corrected calibration algorithm results very good. Finally, a portable sensor has been built for a particularly complex application: oxygen continuous monitoring in composting urban wastes, with temperature up to 80°C. Sensor precision, estimated from standard deviation, results <0.3% O2 for every oxygen concentration. Sensor accuracy, expressed as relative error of mixtures with known oxygen concentration, is always <4%.