OptochemicalSensorsBased onLuminescence...294 OptochemicalSensorsBasedonLuminescence index of the...

www.aspbs.com/eos

Encyclopedia of

SENSORS

Optochemical Sensors Basedon Luminescence

Merima Čajlaković, Alessandro Bizzarri, Christian Konrad, Hannes VorabergerInstitute of Chemical Process Development and Control, JOANNEUM RESEARCH,

Forschungsgesellschaft m.b.H., Steyrergasse 17, 8010 Graz, Austria

CONTENTS

1. Introduction2. Detection Principles for Luminescence

Based Sensors3. Instrumentation for the Phase

Measurement Technique4. Sensor Schemes for Lifetime-Based Optochemical

Sensors Using Phase-Modulation MethodGlossaryReferences

1. INTRODUCTIONIn the past two decades researchers have shown a greatinterest in the area of optical chemical sensors resultingin rapid development of such sensors with applications infields such as industrial, environmental and medical. In gen-eral, the transducer in optochemical sensors is convertedinto a optical signal for real-time and on-line informationabout the presence of specific compounds or ions in com-plex samples. Depending on the origin of the optical sig-nal, the devices are classified as absorbance, reflectance,luminescence, Raman, light scattering, or chemolumines-cence sensors [1, 2]. Whereas at the beginning of the eraof optochemical sensors the detection of absorption or flu-orescence with conventional spectrometer was the center ofinterest, the developments in telecommunication have pavedthe way for cheaper and miniaturized sensor readout equip-ment. These developments have also been over the past twodecades the driving force for the success of luminescencebased sensors, to which this chapter is dedicated to.

Optical-sensing techniques have been widely used forquantitative measurements of various analytes such as H+

[3–5], carbon dioxide [6, 7], oxygen [8–11], metal ions[12, 13], ammonia [14, 15], glucose [16, 17] and humidity[18, 19] in environmental, industrial, clinical, medical andbiological applications. Optochemical sensors in generaloffer many advantages over other sensing techniques [20, 21]

in the way that optical chemical sensing does not consumeanalytes, no reference is required and the signal is insensi-tive to sample flow rate, stirring speed and exterior inter-ferences. Additionally, optical sensors have the potential forminiaturization, remote sensing and easy installation whenoptical fibers are used [12, 22].

Among the various optical methods, fluorescence detec-tion seem to offer the advantages of high sensitivity andion-selective fluorescence probes. During the past severalyears there has been increased interest in lifetime-basedsensing [23], which is preferred over intensity-based meth-ods because the lifetime is mostly independent of the probeconcentration and can be unaffected by photobleaching orwashing out of the probe. As an analytical tool for life-timebased sensors there is an increasing interest in the use oftime-resolved fluorescence [23]. The basic idea is to identifyfluorophores or sensing schemes in which the decay time ofthe sample changes in response to the analyte concentration.Such lifetime-based sensing is most often performed usingthe phase-modulation method or frequency domain. The useof phase angles or decay times are mostly independent ofthe signal level and can be measured in turbid media.

The aim of this contribution to the Encyclopedia of sen-sors is to provide the reader the basic principles of lifetime-based optochemical sensors focusing on phase-modulationfluorometry as the measurement principle. Section 2 dealswith detection principles for luminescence based sensors,where the differences between intensity and luminescencelifetime (time-domain and frequency-domain) measure-ments are described. Section 3 discusses instrumentation forthe phase measurement technique focused on the princi-ple set-up of phase measurements, different phase detec-tion principle and limiting factors for performance of phasemeasurements as well as optical components in instrumen-tation set-up. Last Chapter 4 describes sensor schemes forlifetime-based optochemical sensors using the phasemodula-tion method. This chapter reviews chemical detection prin-ciples used in lifetime-based sensors for monitoring pH,oxygen and carbon dioxide.

ISBN: 1-58883-063-2Copyright © 2006 by American Scientific PublishersAll rights of reproduction in any form reserved.

Encyclopedia of SensorsEdited by C. A. Grimes, E. C. Dickey, and M. V. Pishko

Volume 7: Pages (291–313)

292 Optochemical Sensors Based on Luminescence

2. DETECTION PRINCIPLES FORLUMINESCENCE BASED SENSORS

2.1. Luminescence Lifetime versus IntensityMeasurements

Luminescence is observed when the energy of an electroni-cally excited state species (luminophore or dye) is releasedin form of the light. Luminescence is divided into twosubcategories, fluorescence and phosphorescence. They dif-fer from the photophysical processes which occurs in themolecule during excitation and emission [1]. Depending onthe whether the excited state is singlet or triplet, the emis-sion is called fluorescence (in this case the molecule returnsdirectly to the ground state) or phosphorescence [24, 25].This results in completely different lifetimes of these twocategories. Fluorescence lifetime or decay time typicallyoccurs over tens of nanoseconds (10−9 to 10−7 s), phospho-rescence occurs over much longer time periods (millisec-onds; 10−5 to 10 s). One has to differentiate between thecases in which the luminophore is the analyte itself (emit-ting the luminescence signal for detect) from the cases inwhich the analyte quenches the luminescence signal of theluminophore (in this case the luminophore acts as an indi-cator for the analyte).

In opto-chemical luminescence based sensors a reductionof luminescence caused by the analyte is detected. Two char-acteristics of the luminophores can be applied for measuringthis reduction of luminescence. On the one hand there is thepossibility to measure luminescence intensity, which can begenerally applied in sensors with all available luminophores.In principle the measurement of the luminescence intensityoffers an easy and cost effective sensor. However, the mea-surement of luminescence intensity, suffers from optic inter-ferences caused by changes of turbidity, refractive index orcolor of the sample, fiber bending and microbending of thefiber tip. Intensity signals generally are strongly disturbed byfluctuations in the opto-electronic system (drifts of the lightsource or photodetector).

Furthermore, degradation of the indicator/dye in the sen-sitive membrane caused by photobleaching and leaching isvery critical. Consequently for practical applications someimportant properties of such instrumentations have to beconsidered to get a stable system. Gruber et al. [26] intro-duced an instrumentation based on luminescence inten-sity measurement which tries to compensate opto-electronicinterferences with a second photodiode. Consequently devi-ations of the excitation light, caused by temperature effectsor ageing of the light source, can be compensated. Opti-cal interferences and interferences resulting from the sen-sitive membrane, however, cannot be compensated by thistechnique.

On the other hand there is the possibility to overcomemost of these effects by measuring the luminescence life-time1, as a parameter which is almost independent of theabsolute signal height. Intensity changes in the excitationlight as well as deviations in the optical path do not causechanges in the measurement signal and furthermore the

1 Lifetime is defined as the average time a molecule remains in theexcited state.

influence of the photobleaching behavior is reduced dramat-ically. Unfortunately, luminescence lifetime based sensorsare restricted only to a few dyes with certain characteristics.However, sensing based on luminescence lifetime providescertain benefits [27]. This approach is able to overcome theproblems which affect intensity-based sensors such as indica-tor concentration, photodegradation or leaching of the dye,ageing of the light source or any changes in the optical pathwhich also influences the intensity of the detected lumines-cence. In particular, photobleaching of the dye is one ofthe main factors which limits the stability of optochemicalsensors in practical applications. With the assumption thatthe resulting photoproducts do not gain luminescence andif the changes in microenvironment of the dye are not rele-vant, the photobleaching causes only a decrease of lumines-cence intensity but does not show an effect on the lifetimebehavior of the sensor. Developments in the last years haveenabled some of these restrictions to be overcome and con-sequently luminescence lifetime based sensing schemes areavailable for a significant amount of analytes today.

In Table 1 the major distinctions between intensity andlifetime measurements from an instrumentation point ofview are summarized. It can be concluded that lifetimemeasurement is the more advantageous detection principlebecause of its insensitivity to fluctuations of instrumental

Table 1. Major distinctions between intensity and lifetime measure-ments with particular consideration to properties which are relevant forpractical applications.

Intensity Lifetime (phase-)Property measurement measurement

Excitationintensity

Proportional toexcitation intensity

Theoretically independent

Dependent ondeviations of theexcitation lightsource (temperature,ageing)

(� � � � intrinsic dye property)

Opticalarrangement

Deviations in theoptical path (spatialdistribution, soiling)results directly intoan intensity change

Theoretically independent

Sensorproduction

Luminescence isproportional to theamount of dye andthe thickness of thesensing layer

oxygen characteristic showslittle dependence on dyeconcentration in thepolymer

Calibration 2 point calibrationnecessary

In-production and 1 pointre-calibration

Backgroundsensitivity

Direct relationship Vector like addition

Photobleaching Direct proportional Theoretically independent,practically lower than inintensity measurements(problems with backgroundsignal)

Signal/noiseratio

Worse (higher measurementfrequencies)

Resolution Worse at the sameirradiation intensity

Optochemical Sensors Based on Luminescence 293

artifacts and the lower photobleaching effects together withthe benefits concerning sensor production and calibration.In contrast to absorption methods, no reference measure-ment is necessary, and, in contrast to fluorescence-intensitymeasurements, no compensation for variation of the men-tioned instrumental parameters is necessary.

2.2. Luminescence LifetimeMeasurement Methods

The luminescence lifetime of a sample is the mean durationof time the luminophore remains in the excited state afterexcitation of the luminophore with a short light pulse (seeFig. 1). Following pulsed excitation, the intensity decays ofmany luminophores are single exponential [28]:

I�t� = I0e−�t/�� (1)where I0 is the intensity at t = 0 and � is the lifetime.

For the detection of the luminescence lifetime there arein principle two available techniques. Firstly, the direct(time domain) method and secondly, the indirect (frequencydomain also referred as phase modulation) method. The timedomain method monitors the emission signal in time domainand evaluates the decay profile of the signal. On the otherhand the frequency domain method uses sinusoidally mod-ulated excitation signals at certain frequency and detectsthe time delay (phase shift) of the emission signal of thesensor dye to calculate the lifetime. Both approaches pro-vide the same information for an assumed single-exponentialdecay characteristic, because of their Fourier-transformrelationship.

2.2.1. Time Domain Measurements(Direct Method)

Measurement of the time dependent emission which fol-lows excitation with a brief pulse of light as shown in Fig. 1is the so called time domain or direct method. For theseshort excitation pulses flash-lamps, lasers or synchron radi-ation sources are commonly used as light sources. Further-more, detectors (photomultiplier tubes PMTS) with a wide

00

1

I/I m

ax

Time

1/ e

τ

Pulse excitation

Luminescence response

Figure 1. Characteristic luminescence decay curve after pulseexcitation.

bandwidth (high speed) are necessary in combination withfast signal sampling circuits (oscilloscope). Another variationof the time domain measurement is the so called step-measurement. In this case instead of the short-pulse excita-tion the sensor is illuminated with a rectangular signal andthe responding signal from the sensor follows the step withan exponential rise. Instead of the fact that in this case canbe used slower components for the excitation, the detectionelectronics has to remain the same and consequently thehigh cost instrumentation is inevitable. In addition to thislarge and costly instrumentation it is often necessary to usedeconvolution treatment to extract the decay curve if theexcitation pulse duration and the signal transition times ofthe detectors are not negligible [29]. All these facts pointto an instrumentation which uses frequency domain mea-surements which can be realized with less cost and printedcircuit board (PCB) size, which is quite useful for applica-tion of optochemical sensors in industrial and biochemicalenvironments.

2.2.2. Frequency Domain Measurementsor Phase-Modulation Fluorometry(Indirect Method)

In contrast to time domain measurements in phase-modulation fluorometry an intensity modulated (sinu-soidally) light source at certain frequency can be usedinstead of pulse sources. Because of the time lag betweenabsorption and the emission of the dye, the emission isdelayed in time relatively to the excitation signal. This delayis described by the phase shift between these two signals(see Fig. 2).

With the following equations the relationship betweenexcitation signal and emission signal of the system isdescribed.

The excitation function E�t� can be expressed as the sumof a constant intensity term and a sinusoidally varying inten-sity term [30].

E�t� = E0�1+ME sin��t�� (2)where the steady state DC-fraction with an angular fre-quency of c0 is represented by E0 and ME is the modulation

tt

bφ

a A

B

Intensity Intensity Modulation, m = B/Ab/a

Figure 2. Scheme of the phase-modulation fluorimetry for lifetime-based sensing: the fluorophore is continuously excited by an intensity-modulated light beam. The emission is consequently modulated at thesame frequency. Due to the finite fluorescence lifetime of the fluo-rophore, the emission is phase shifted (�) and demodulated (m =bA/aB) with respect to the excitation. The fluorescence lifetime iscalculated from either the phase shift or the demodulation values.


index of the excitation signal. It is defined as ratio of theAC-amplitude to the DC-intensity and can be set ME ≤ 1.

The emitted time dependent luminescence response com-ing from the sensor remains unchanged in frequency and issinusoidal as well. However, the luminescence signal is timedelayed (phase shifted) relative to the excitation signal. Thiscan be expressed as follows:

L�t� = L0�1+ML sin��t + �� (3)Modulation index of the luminescence signal Mi, will be lessthan that of the excitation signal. On the subject of the inten-sity of the signal L0 it must be said that it is smaller butproportional to E0 and is determined by the absorption coef-ficient, concentration and quantum efficiency of the lumi-nescent dye, the used matrix and the thickness of the layerand as well by an instrumental factor representing the opti-cal arrangement.

2.2.3. Phase-Lifetime RelationshipThe relationship between the luminescence lifetime and themeasurement of the phase shift is described by the followingequations.

If a single-exponential decay is assumed for the generalcase, the impulse response of the luminescence signal a�t�is given as follows:

a�t� = � · e−�t/�� (4)where � is the intensity at time t = 0. To describe thistime-domain behavior in the frequency domain, the Fourier-transformation can be used and following equation isobtained:

A�j�� = � · �1− j� · � =

� · �1+ �2 · �2 − j ·

� · � · �21+ �2 · �2 (5)

where � = 2�f is the circular frequency. If this expression’srelationship with signal processing theory is considered, thisfunction describes a first order low-pass filter.

Consequently the phase shift can be derived as follows:

� = arctan Im�j��Re�j��

= arctan�� · �� (6)

This equation represents the basic relationship of phase shift� and lifetime � of the system.

The magnitude of the signal can be written as:

�A�j�� = � · �√1+ �2 · �2 (7)

The magnitude of the signal also depend on the frequency.This behavior can be expressed by the degree of modula-tion M :

M = 1√1+ �2 · �2 (8)

After rearrangement of Eq. (6) and incorporation of this inEq. (8), the following relationship between the modulationand phase shift � is obtained:

M = 1√1+ �2 · �2 =

1√1+ tan2 �

= cos� (9)

These are the relationships of lifetime, phase and modula-tion data for ideal single-exponential decay behavior. In thiscase the determination of lifetime with phase or modulationmeasurement must be equal and independent of the mod-ulation frequency [31]. Unfortunately, for practical sensormeasurements an ideal single-exponential model can rarelybe assumed. If the emission is characterized by multi expo-nential decays, the above calculations have to be modified.Therefore, the lifetimes cannot be determined using a singlemodulation frequency. In order to analyze the lifetime frac-tions, multi-frequency phase and modulation measurementsare required. For a general case the multi-exponential sys-tem can be represented by a sum of N exponential decays:

��t� =N∑i=1

�i · e−�t/�i� (10)

where �i is the contribution to the amplitude of the i-thcomponent and �i is the lifetime of the i-th component.

This is a shortly description of the theoretical backgroundfor implementation of a phase measurement system. For realworld applications, the instrumentation for the phase mea-surement technique in detail is described and presented witha schematic block diagram in the next section. Afterwards,the major components and the key factors for a successfulrealization of such an instrument are presented.

3. INSTRUMENTATION FOR THE PHASEMEASUREMENT TECHNIQUE

The main objective of the phase-sensitive detection forphase fluorometric applications is the determination of thephase shift of the luminescence signal relative to the mod-ulation signal. Practically that means, that the measurementis performed at known frequencies, which are generated bycircuits in the same instrument. In consequence all systemsconsist of a signal generation unit, a signal detection unitand one unit to compare these two signals to obtain thephase shift. A principle set-up of such a phase fluoromet-ric instrument is shown and described in Section 3.1, withremarks concerning some practical aspects like temperaturedependence and component tolerances which can be elimi-nated by a referencing system.

Since the heart of a phase fluorometric instrument isthe phase detector, several principle setups with particularblocks are described in Section 3.1.

3.1. Principle Setup of PhaseMeasurement Systems

A short diagram overview of the setup is given in Fig. 3 toidentify all the relevant components of a phase fluorometricinstrument. Based on this scheme the characteristics andrelevant specifications of the used functional blocks areexplained.

A frequency generator provides a signal of a certain fre-quency to modulate the intensity of a light source using anoptimized driver circuit. The light illuminates the sensitivedye via an optical excitation filter. A photodetector in com-bination with another optical filter collects the luminescence


ΦSDrivercircuit

Lightsource

Emissionfilter

Pulsedetector

Frequencygenerator

Amplifiercircuit

Lightdetector

Sensitivedye

Excitationfilter

Figure 3. Principle set-up of a phase fluorometric instrument.

coming from the sensitive dye. These obtained signals arethen amplified with appropriate amplifier circuits and arefed afterwards to the phase detection unit, which determinesthe phase shift of the incoming signal relative to the modu-lation signal produced by the frequency generator.

Considering the signal path of Fig. 3, the measured phaseshift of the system, which occurs in the set-up between sig-nal generation and multiplication with the reference signal,consists of the combination of:

• the phase shift due to the modulation of the lightsource. It depends on the time response characteris-tic of the used light source (e.g., LED) in combinationwith a suitable driver circuit;

• actual phase shift due to change at the sensitive dye;• phase shift of the detection circuitry including the pho-

todetector (e.g., photodiode), photodetector amplifier,additional amplifier and signal conditioning circuits;

• phase shift of the multiplier stage in the phase detec-tion unit (in particular the phase shifts resulting fromgeneration of the in-phase and quadrature referencesignals which can be neglected, when using a digitalimplementation).

Furthermore the phase shifts caused by electronic andoptoelectronic components are changing with temperature,ageing and also the component tolerances cause errorswhich cannot be neglected. So there is the need for a refer-encing system to eliminate all these undesired phase shiftsof the system. In principle therefore several implementationstrategies can be used to overcome this problem:

• The use of a second photodetector and amplifier cir-cuit, which exclusively detects the excitation light. Thephase shift of the sensitive dye can be calculated by sub-tracting this second phase shift from the phase shift ofthe luminescence detector. Unfortunately this methodonly works under the assumption that the detection cir-cuits behave exactly equally. In reality this is difficult toimplement, because of component tolerances and tem-perature behavior of the sensitive components.

• The measurement of the excitation light by removingthe optical emission filter in front of the photodetectorand preventing that the light can excite luminescence ofthe sensor. This is a laboratory method, and can hardlybe applied in practical setups.

• Usage of a luminescence standard (sensor foil with aspecified fluorescence time independent of the analyteconcentration) instead of the indicator to calculate theinstrumentation phase shift. For this method a simi-lar restriction for practical setup occurs as mentionedabove.

• The use of a second light source (e.g., a referenceLED) with an emission spectrum in the range of theluminescence emission spectrum. The light of this sec-ond light source does not excite any luminescence, butcan easily pass the emission filter in front of the pho-todetector. This method allows the exclusive determi-nation of the detector circuitry alone. It keeps thephase shift of the light source-driver circuit as well aswith the light source itself, which characteristic is easierto handle and therefore this referencing method, whichis illustrated in Fig. 4 as a good choice for a real setup.

With an alternatively phase measurement of the reference��S� as well as the signal ��R�, the phase shifts caused bythe system can be eliminated by subtraction of the signaland reference signals:

� = �S −�R (11)In case of using this referencing method the developer has tokeep in mind that the signal and the reference light sourceshave to be very carefully selected, because different char-acteristics of the switching times (also over temperature)of these two components may drastically affect the phasemeasurement.

3.2. Phase Detection Principles

Since the beginning of phase measurement in the late 19thcentury many systems based on different technologies andprinciples have been developed. A detailed historical per-spective is given by C. Kolle in Ref. [32]. In present timesanalogue instruments have been almost replaced by digitalbased measuring devices. However for every application oneshould be mindful to maintain the complexity and there-fore the costs as low as possible for the requirements ofthe system. Recent developments in logic-cells, �-Controllerand digital signal processor (DSP) technologies caused ashift of the detection techniques to digital techniques witha significant improvement on resolution and detection lim-its. Nevertheless on this point a short overview of the most


ΦS

ΦR

Sig - LED

Emissionfilter

Excitationfilter

Frequencygenerator

Phasedetector

Amplifiercircuit

Lightdetector

Sensitivedye

Ref - LED

Drivercircuit

Figure 4. Set-up of a phase fluorometric instrument with a second LED for a system reference measurement.

common phase detection principles will be given in followingsections.

3.2.1. Zero-Crossing DetectionThe principle of the zero-crossing detection is based onthe measurement of time shift of the zero-crossings ofDC-less modulation and luminescence signals. Therefore,comparators can be employed as zero-crossing detectorsto produce rectangular digital signals and out of them thephase shift measurement can be reduced to a precision timemeasurement.

In the majority of cases for time measurements count-ing devices with a precision time reference (quartz oscilla-tors) are used. Another approach is the simple techniqueof conversion of the time shift into a duty-cycle using one“exclusive-or” gate. If the phase shift is zero, the output sig-nal is zero as well, otherwise if the phase shift is 180� thenthe output signal is constant one. All phase shifts in betweenthese extremes are representing a pulse-width modulatedsignal, which is proportional to the phase shift of the lumi-nescence signal in comparison to the modulation (reference)signal. The principle is also shown in Fig. 5.

The signal coming from the optical detector is amplifiedand band-pass filtered (A) to eliminate the DC fraction andthe higher harmonics of the signal. With the comparator(C) a phase shifted transistor-transistor-logic signal (TTL) isproduced, which is fed into an “exclusive-or” gate (EX-OR)as well as the excitation square wave. This acts as an phasediscriminator whose output gives a duty-cycle signal. This

C

LP

SIG

REF

Phi

Phi

UDC

A B

=1EX-OR

Figure 5. Basic principle of phase shift measurement using an“exclusive-or” gate.

out-coming pulse train can be averaged by a low-pass filter(LP) whose output represents a voltage signal correspond-ing to the measured phase-shift. In general this principleis called as “time-to-amplitude converter” and represents asimple and powerful method for a phase sensitive detectionsystem in several applications [33, 34].

3.2.2. Synchronous DemodulationAnother promising technique for the fulfilling the demandsof phase sensitive detection is the synchronous demodula-tion. The principle of this method is based on a high correla-tion of the signal of interest and a reference signal, which islocked to the signal of interest and therefore the term “lock-in system” is used for such systems. This detection is oftenapplied to recover signals from noise, because the methodcaptures only a very narrow band signal energy containedin the fundamental harmonic of the signal. A comprehen-sive and practical insight into this technique was given byMeade [35] and Mandelis as well [36]. All lock-in systemscan be decomposed into the basic structure which is shownin Fig. 6.

In the figure shown above the signal of interest s�t� is fedinto two multiplier stages which are the key elements of thissystems. The goal is to determine the phase shift of signals�t� to an reference signal r�t� which has to be precisely syn-chronized with the signal of interest. The signal s�t� is on theone hand multiplied by an in-phase reference signal ri�t� and

90°

r(t) = Aref . sin(ωt)

s(t) = A . sin(ωt + Φ)

ri (t)

rq (t)

vi (t)

vq (t)

×

×

LPF

LPF

Figure 6. Structure of the synchronous demodulation principle.


on the other hand by an quadrature reference signal rq�t�,which is shifted by 90� with respect to ri�t�. The outputs ofthe two multipliers are then introduced into low-pass filtersto firstly eliminate the second harmonic signal of the multi-plier and secondly to operate as integrator which allows verynarrow bandwidth detection. The two output signals vi andvq constitute the in-phase and the quadrature componentsof a vector-representation which can be described by a mag-nitude and a phase. Mathematically this circuit is describedwith following equations:

vi�t� = s�t� · r�t� = A · sin��t +�� ·Aref · sin��t� (12)

With assumption of Aref = 1, one obtains

vi�t� =A

2· �cos��− cos�2�t +�� (13)

The signal averaging, which is performed by the low-passfilter, eliminates in the case of ideal lock-pass filter the sec-ond harmonic signal at the frequency 2�. Hence the outputsignal is expressed as:

vi�t� = 2 · cos� (14)

The similar treatment of the quadrature reference signalrq�t� yields to

vq�t� = s�t� · r�t� = A · sin��t + �� ·Aref · cos��t� (15)

and eventually

vq�t� =A

2· sin�� (16)

From these results the magnitude R and the phase � canbe calculated using the following equations:

R =√v2i + v2q (17)

� = arctanvqvi

(18)

In opposite to other methods the synchronous detection hasthe essential advantage in providing the determination ofa phase shift as well as a value for the signal magnitude,which is able to deliver important information on the sys-tem. During many stages of the development of the opto-electronic, optic and chemical parts of the system and alsoduring operation of the optochemical measurement system,the magnitude can provide information on quality and sta-bility of the optic/optoelectronic parts, ageing of the opticalcomponents (LEDs), quantum yield and photobleaching ofthe sensitive dye.

3.2.3. Cross Correlation TechniqueA common approach, which is employed for phasemeasurements with high frequencies, is so called the

“cross-correlation” technique. This method is characterizedby multiplication of the high frequency signal with a sig-nal, whose frequency is only slightly different. The differentfrequency, which is called the cross-correlation frequencymaintains the phase shift information. The actual phase shiftdetection can then be performed at a low frequency withimproved sensitivity and precision.

Such a system, which operates at two frequencies, was firstimplemented by Spencer and Weber [37]. Another continu-ous multi-frequency cross-correlation phase and modulationfluorometer was reported by Gratten and Limkemann [38],where a detailed mathematical treatment of this cross cor-relation technique was given. This cross-correlation methodis now state of the art in all frequency-domain fluorome-ters and modulation frequencies of several Gigahertz can beachieved using this technique [39].

However this heterodyning technique is dominant at highfrequency applications with high-speed optical detectors likePMTS. In practical sensor applications the frequencies arelower and low-cost solid state photodetectors can be used.In 1995 Gruber et al. [40] described a heterodyning demod-ulator in which the polarity of the photodiode is switched,which corresponds to a multiplication with +1 and −1.

3.2.4. Phase Locked DetectionThe phase locked detection of fluorescence lifetimes dif-fers in several aspects from common synchronous detectionschemes. It is based on phased locked loops and the lifetimeis converted to a repetitive output signal, which serves asmodulation signal as well. In this approach the frequency ofthe signal is directly proportional to the luminescence life-time. The frequency or the period of the resulting signal isa parameter which can be easily determined with high pre-cision. However, this technique is only suitable for lifetimesgreater than 1 �s because of the restriction by the electron-ics available.

3.3. Performance Limiting Factors for PhaseMeasurement Systems

As already mentioned at the beginning of this documentthe using of a phase measurement instrumentation insteadof an intensity measurement has several advantages. Nev-ertheless for such instrumentations some special propertiescaused by the sensitive dye and by the measurement at acertain frequency have to be considered to get an optimizedinstrument. Therefore in this section some essential proper-ties will be pointed out.

3.3.1. Optimum Modulation FrequencyThe determination of the optimum modulation frequency isoften carried out in an easy experimental way with regardto the maximum phase shift at the output. However thisoptimization procedure does not automatically imply thatthe resolution of the system is maximized, because of dif-ferent signal to noise ratios (SNR) at different measure-ment frequencies. The SNR depends to a large value on theexcitation light intensity and unfortunately a high illumina-tion intensity causes fast photobleaching of the dye. So the


major goal is to optimize the modulation frequency to max-imize the SNR and keep the excitation light intensity as lowas possible.

For the theoretical approach an ideal single-exponentialdecay of the sensor luminescence is assumed. After rear-ranging the Stern-Volmer equation

Cx =�0/� − 1kq/k

(19)

the luminescence lifetime as a function of the quencherconcentration can be expressed as:

� = �01+KSV · Cx

(20)

where �0 and � denotes the lifetimes in the absence andthe presence of quencher (e.g., oxygen), respectively, kq thetransition due to radiationless deactivation by a quencher,k decay rate, Cx the concentration of the quencher (e.g.,oxygen) and KSV is termed as Stern-Volmer quenchingconstant.

The lifetime sensitivity (i.e., phase change due to lifetimechange) can be derived as:

s�O2 =d�

dO2= − �0 ·KSV

�1+KSV ·O2�2= −�

2 ·KSV�0

(21)

For the special case at 0% of quencher the sensitivitysimplifies to:

s�O2∗ = −�0 ·KSV (22)

These are the basic relationships calculating the lifetime sen-sitivity when the Stern-Volmer constant and the quencherfree lifetime are known. In a phase-fluorometric measure-ment the phase sensitive information is required. Using theEq. (6) the phase sensitivity s�� (i.e., phase change due tolifetime change) can be expressed as follows:

s�O2 =d�

d�= �

1+ �2 · �2 (23)

From this equation it can be seen that the sensitivity notonly depends on the lifetime but also on the modulationfrequency. In order to find the optimum frequency for max-imum sensitivity at a specific lifetime one can differenti-ate Eq. (23) with respect to co and set the result equal tozero, i.e.,

d

d�

(d�

d�

)= 1− �

2 · �2�1+ �2 · �2�2 = 0 (24)

As a result one obtains �opt� = 1. From this follows a phaseshift of 45� �= arctan1�. Hence the optimum frequency fopt1is given by

fopt1 =1

2 · � · � (25)

This relationship is optimized to obtain the maximum phaseshift for a given luminescence lifetime. Practically � varieswith analyte concentration and therefore an optimum fre-quency across a lifetime range must be chosen. In order to

select a frequency for a given measuring range limited by thelifetimes �1 and �2, the optimum frequency for a maximumphase shift difference &� can be derived as follows:

&� = arctan�� · �1�− arctan�� · �2� (26)d�d��

d�= �1 − �2 + �

2 · �22 · �1 − �2 · �21 · �2�1+ �2 · �21 � · �1+ �2 · �22 �

= 0 (27)

Therefore, the optimum modulation frequency fopt1 to max-imize the phase difference across the measuring range isgiven by:

fopt1 =1

2 · � ·√

1�1 · �2

(28)

3.3.2. Phase Noise Using the SynchronousDemodulation

Actually the resolution of the system is limited by thenoise of the detected luminescence signal and the sensitivity,whereby in case of phase-fluorometry can be defined as:

s�Cx =d�

dCx(29)

That means the phase shift change per quencher change andthis property is given by the used quencher dye. In practicethe noise is described by its root mean square value or thestandard deviation, respectively. However the noise distri-bution is problematic and therefore the maximum peak-to-peak value, which determines the resolution, cannot exactlybe given. Therefore it can be described by a probability thatthe actual signal will exceed nominal peak-to-peak values.A very common specification is a confidence interval of±3' whereby 0.27% of all data points will exceed the limit,whereby 99.73% remain within the bounds. While in ampli-tude measurements the root mean square value of phasenoise is defined by the system alone, in phase measurementsit is a function of the actual amplitude of the signal.

In order to find the relationships of amplitude and phasenoise a phasor representation is displayed in Fig. 7 for themodulated luminescence signal A.

As it can be seen from Fig. 7 the noise phasor n�t� canbe represented by an in-phase ni�t� and an quadrature nq�t�

A

r(t)

φ(t)n(t)

nq(t)

ni(t)

ω

Figure 7. Phasor diagram of the sum of signal and narrowband noise.


part. The signal phasor A and the noise phasor n�t� togetheryield in the measured signal r�t�, which can be described as:

r�t� = )A+ ni* · sin�t − nq · cos�t = R�t� · sin��t + ��t��(30)

This is allowed because the synchronous demodulation cap-tures only the very narrowband signal energy contained inthe fundamental harmonic of the signal.

The result of the vector like addition of the signal and thenoise phasor is the introduction of both, amplitude noise [inR�t�] and phase noise [in ��t�]. Thus the Eq. (30) can alsobe written in form:

r�t�=√)A+ni�t�*2+n2q�t�·sin

(�t+arctan

(nq�t�

A+ni�t�))

(31)

Assuming that the noise is small ni�t�, nq�t��A, the ampli-tude can be approximated by

r�t�≈A+ni�t� (32)and the value of the phase noise angle is

��t�≈arctannq�t�A

≈ nq�t�A

(33)

The phasor diagram in Fig. 7 is a “snapshot.” Consideringthat the noise parts ni�t� and nq�t� are gaussian randomprocesses, their mean values are zero and that the noisephasor spends an equal amount of time in any place withinthe envelope of the circle, it is clear that the mean valuesof them are equal. Furthermore with assumption of uncor-related noise the following relationship can be established:

n2�t�=n2i �t�=n2q�t� (34)or

'2n ='2ni ='2nq (35)what means that the variances (mean square values orpower) of n�t�, ni�t� and nq�t� are equal.

The SNR of the amplitude, which is represented by theratio of the root mean square value of the signal to the rootmean square value of the noise, can be defined as:

SNRV =A

'n(36)

Considering the variance (mean square value) of the phasenoise (Eq. (33)) one can express it as:

'2�='2nA2

(37)

Thus the root mean square value (standard deviation) ofphase noise yields to:

'�='nA

= 1SNRV

(38)

From this relationship it can be easily seen that the phasenoise is directly linked to the amplitude of the signal.

3.3.3. Influence of Background LightAlthough an optical filter is introduced directly in front ofthe photodetector to prevent the detection of the excita-tion light (set as primary light), still some influences of theprimary light cannot be avoided. In particular, the resid-ual primary light, which is not blocked by the optical fil-ter, and the fluorescence light of the optical filter, whichis also excitated by the light source, give rise to the so-called “falselight.” Because the fluorescence light of theoptical filter typically shows short lifetimes, both contribu-tions give rise to a background signal which is in-phasewith the excitation and contribute to the phase shift. Thefalsification of the measured values is demonstrated inFig. 8.

The background signal (BG) is drawn slightly exagger-ated to highlight the effects. It can be easily seen that dueto the background signal the amplitude is increased fromA to ABG and the phase shift (�) changes correspondinglyto a lower value ��BG�. Another effect, which is related tothe background signal, is that the error on the phase shiftincreases when the signal amplitude decreases. This effectis also shown in Fig. 8: the decrease of the original ampli-tude (A) to a new amplitude �N� causes that the measuredphase shift decreases from � to �BG+N , so that the error��−�BG+N � increases. This is observed for example in thecase of photobleaching of the dye.

In this case the task of the developer is to optimize thesystem in a way to reduce the influence of this effect to aminimum.

3.4. Practical Aspects for Implementation(Aspects for Designing PhaseMeasurement Periphery)

As mentioned in the previous sections the various aspectshave to be considered for an optimized instrumentationof an optochemical sensor. Considering Fig. 4, it can beseen that an accurate phase measurement block is just oneimportant part of such a complex instrumentation. To opti-mize the resolution of the system the developer also has topay attention to the analogue parts of the system, becauseevery noise source which is introduced in this sensitiveparts limits the SNR and therewith the resolution of thesystem. The goal of this section is to show some impor-tant aspects in selection of the optoelectronic components

A

BG

ABG

ABG + N

ΦBG + NΦBGΦ

N

ω

Figure 8. Influence of background signal on the phase shift perfor-mance.


and designing the analogue parts of the phase measurementsystem.

Basically the requirements of the optical set-up do notconcern only the resolution and accuracy of the mea-surement, but also the temperature behavior of the opto-electronic components and the properties of long timestability of the sensitive element containing the luminophore[32]. The time stability of the sensitive element dependson its photo-stability and this is affected by the dose ofthe exciting light which is received by the luminescent dye(photo-bleaching). This means that the photo-stability ofthe sensitive element increases with decreasing excitationintensity [24, 32, 41]. However it is not possible arbitrarily todecrease the intensity of the excitation light source withoutdecreasing the signal-to-noise ratio and worsening the res-olution of the measurement. As a consequence, an opticalset-up is demanded, which is able to efficiently collect theluminescence signal with an excitation intensity as minimumas possible.

The optical set-up of an opto-chemical sensor typicallyconsists of the following parts:

• an excitation module, consisting of a suitable lightsource for the excitation of the luminescent signal (exci-tation source) and one or more optical filters set infront of the excitation source to separate the excita-tion radiation emitted by the excitation source from theluminescent signal (excitation filter);

• a reference module, which may consists of a secondlight source (reference source) or of a second pho-todetector, to account for measurements errors due tochanges with temperature of the operational param-eters of the electronic components and eventually oneoptical filter set in front of the reference source toselect the wavelength emission in the same range as theluminescence signal (reference filter);

• a detection module, consisting of one photodetector todetect the luminescent signal emitted by the sensitiveelement and of one optical filter set in front of the pho-todetector to separate the luminescent signal emittedby the luminophore from the excitation light and otheroptical background contributions (emission filter);

• a collecting optics, which may consist of refractive opti-cal components (prisms or lenses) and/or light-guides(glass rods or optic fibers) to carry the excitation lightto the luminophore and the luminescent signal to thephotodetector.

Additionally, together with the selection and optical char-acterization of the light sources and of the photodetector,the following tasks must be also fulfilled for the optimizationof the measurement system:

• selection and characterization of the driver circuits forthe chosen light sources and

• selection and characterization of amplifier circuits forthe chosen photo-detector in order to maximize theSNR of the setup.

All these components have to be fitted together in a appro-priate housing with respect to interferences and cross-talkto get an user friendly instrument.

3.4.1. Selection of Light Sourcesa. Excitation Sources Many kinds of different opticalsources are nowadays available for applications in the sen-sor field, like for instance wideband continuous spectrumsources (e.g., incandescent lamps), light emitting diodes(LED), laser diodes or lasers. The choice of the propersource depends on several factors: the spectral character-istics of the luminescent dye (luminophore) which is usedfor sensing purposes, the luminescence parameter whichis going to be measured (e.g., intensity, lifetime, polariza-tion), the measuring technique that is going to be adopted,like for instance intensity measurement, direct recording ofluminescent decay times by pulsed excitation technique orindirect determination of luminescence lifetime by phasetechnique. Other properties which have to be taken intoconsideration for choosing the light sources are the lumi-nescence lifetime, the quantum yield and photo-stability ofthe luminophore. The latter parameters are also impor-tant in order to determine the amount of exciting radiationwhich the luminophore can tolerate without degradation[24, 25, 32].

Lasers due to their extremely high optical output powerand beam collimation are used for excitation of molecularsystems which show narrow absorption bands or peaks withlow quantum yield and high photo-stability [25, 42]. They aresuitable light sources in all those applications in which fastswitching time or high modulation frequency of the excitinglight are necessary. Typically such features are highly val-ued in direct intensity measurement of luminescence or indirect decay time measurement by pulsed excitation of lumi-nescence from luminophores with short lifetimes (order ofmagnitude of a few nanoseconds). However lasers can bequite voluminous and expensive, therefore, for some appli-cations laser diodes offer an interesting alternative and arepreferred, showing a good monochromaticity and beam col-limation, fast switching time and possibility of high modu-lation frequency like lasers, but being less voluminous andexpensive. Besides they are typically tunable light sources.

Contrarily to lasers and laser diodes, LEDs exhibit anon-monochromatic incoherent emission (characterized bya much lower optical output power and poor beam collima-tion) and they do not allow fast switching time and modula-tion frequency as lasers and diode lasers. Nevertheless LEDsare suitable light sources for excitation of luminophores withbroader absorption bands and with typical luminescent life-time in the order of milliseconds up to microseconds. Incomparison to lamps (Tungsten, Tungsten-halogen, Xenonand Quartz-halogen lamps) LEDs are advantageous due totheir lower power consumption, high stability and reliability,small size, robustness and lower costs. Moreover incandes-cent lamps are typically modulated by mechanical chopper,while semiconductor sources can be easily modulated up toseveral hundred kHz by simple modulation of the appliedvoltage or current. LEDs have already proved to be particu-larly suitable for instrumentation based on the indirect mea-surement of the luminescence lifetime of a large variety ofopto-chemical sensors by phase technique [32]. An exampleof such opto-chemical sensor is given by a Palladium (Pd)-porphyrin complex immobilized in a polystyrene matrix,whose absorption and luminescence spectra are shown inFig. 9.


0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0350 400 450 500 550 600 650

Wavelangth (nm)

Abs

orba

nce

(a. u

.)

6000

500000

1000000

1500000

2000000

2500000

3000000

3500000

4000000

4500000

5000000

5500000

620 640 660 680 700 720 740

Wavelength (nm)

Inte

nsity

(co

unts

/s)

Figure 9. Absorption (top) and luminescence (bottom) spectra of a Pd-porphyrin complex immobilized in a solid matrix of polystyrene. Theluminescence spectrum has been recorded under excitation at 522 nmby a Shimadsu spectrometer.

The absorption spectrum of this dye (top picture) showsthree main absorption bands: a principal one with maxi-mum around 400 nm and two minor bands centered at about520 nm and 560 nm, respectively. The emission is shiftedwith respect to the absorption to longer wavelengths (Stokesshift). The luminescence emission of this luminophore foran excitation at 517 nm shows a broad band with maxi-mum at about 670 nm (see bottom picture). The so-calledStokes shift of this dye is therefore of the order of 150 nm.For excitation of an optochemical sensor based on a Pd-porphyrin complex dye like in this example, LEDs with emis-sion bands matching the absorption bands of the dye shouldbe used. The excitation at 400 nm is particularly efficientdue to the high absorbance shown by the dye. As a con-sequence the dose of excitation light to produce a measur-able luminescence signal can be kept lower for this bandthan for the other two. However it must be also considered

that the excitation efficiency of the luminescence is a wave-length dependent function which increases with decreas-ing wavelength. Therefore, at excitation wavelengths round400 nm, also other luminescence emissions from materialsclose to the dye can be efficiently excited. Materials whichform together with the luminophore the sensitive element orwhich are just close to it, as well as materials of the opticalset-up (optical filters for instance) can emit a photolumi-nescence which increases the optical background. For thisreason a blue-green LED, which matches the second absorp-tion band of the dye at about 520 nm, is often preferred asexcitation source for this luminophore. A typical emissionspectrum of one blue green LED from Nichia is shown inthe Fig. 10. The emission band is centered at about 517 nmand has a width of about 30 nm at half peak maximum.

The choice of the excitation source requires a detailedinvestigation not only of its spectroscopic properties, butalso of its electronic characteristics. This is generally impor-tant for all kinds of light sources but in particular forLEDs, since these devices are developed for applicationsin displays, lighting, indicator lights, traffic lights, etc. andthey are not optimized with particular regard to emissionand wavelength stability and temperature stability. Unfortu-nately this kind of information is not given by the producerin the LED data sheets. That is why dedicated investiga-tions are normally carried out to characterize the spectraland electric behavior of the source (e.g., peak wavelength,current–voltage characteristics, junction capacitance) at var-ious operating conditions (temperature and current).

b. Reference Source To compensate the measured lumi-nescence for the effect produced on the excitation sourceby changes of the operating conditions, it is necessary touse a reference signal to monitor the behavior of the lightsource [32].

In many opto-chemical sensor instrumentation, a refer-ence concept is adopted, which is based on the use of asecond light source. It is important to select the reference

400 450 500 550 600 650 700 750 800−500

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Am

plitu

de (

a.u.

)

Wavelength (nm)

Figure 10. Emission spectrum of a LED used as excitation source forthe sensitive element based on Pd-porphyrin luminophore. The mea-surement was performed at a forward current of 20 mA and at constantambient temperature of 25�C.


source with an emission spectrum in the same spectral rangeas the luminescence signal, because in such instrumenta-tion the photo-detector is typically equipped with one setof long-pass optical filters to prevent the detection of theexciting light and consequently to reduce the optical back-ground. Moreover, this second light source must not exciteany luminescence from the sensitive element. The selectionof a proper reference light source must keep into consid-eration also its electric properties. In particular, the switch-ing time characteristics and the temperature behavior of theswitching time of this light source must match those of theexcitation source as best as possible. By the assumption thatthe two driver circuits of the light sources behave equally,the contribution to the luminescence phase shift due to thedetection circuit can be determined.

In case of a sensor based on Ruthenium-diphenyl phenan-throline (Ru-dpp) dye, for example, the most suitable exci-tation source is a blue LED emitting at 460 nm. Possiblereference sources are LEDs with an emission peak at about620 nm (the luminescence peak of this dye). Consequently,LEDs emitting at 590, 594, 645 and 650 nm were selectedand investigated. Unfortunately, a very good matching ofthe switching times of signal and reference LED is hardlyobtainable. However, the changes of the switching time overthe temperature of both LEDs is more important than theabsolute values. Table 2 shows the rate of change of theswitching time of the excitation LED with temperature andrate of change of four candidates as reference LED. It canbe easily seen that the temperature behavior of the LEDemitting at 594 nm better matches that of the excitationLED. The worst candidates are the red LEDs emitting at645 and 650 nm, since their switching time decrease withincreased temperature, whereas the excitation LED has anincreasing switching time with increasing of temperature.

If one LED has more than one emission band, the tem-perature behavior of all emission bands, and in particularof their switching time, must be investigated, because it mayoccur that they behave differently from each other. This isfor example the case of the LED emitting at 594 nm, whichaccording to Table 2 represents the best candidate as refer-ence LED for Ru(dpp) based sensor instrumentation. ThisLED shows a second emission band at about 850 nm with amuch slower switching time. Therefore, if this LED has tobe used as reference, it is necessary to prevent that the lightof this second emission band reaches the photo-detector, forexample by filtering the light by means of a proper opticalfilter.

c. LED Modulation Circuits The major property of theLED driver circuits is to provide the LEDs with an adequate

Table 2. Rate of change of LED phase with changing thetemperature in the range 10–70 �C.

Reference Rate of change of theLED Emission peak phase with temperature

Excitation 460 nm +6�7×10−4�/�CLED 1 590 nm +1�1×10−4�/�CLED 2 594 nm +5�0×10−4�/�CLED 3 645 nm −1�0×10−2�/�CLED 4 650 nm −3�5×10−3�/�C

prepared square wave or sinusoidal signal from the fre-quency generation unit of the phase detection unit. For aoptimum performance (accuracy and stability) the absoluteswitching times are not that important. The key of success inthis case is a reproducible switching of both circuits (Signaland Reference).

In an ideal case the LED emits light directly after assert-ing the signal. In practical there is a delay which is depen-dent on:

• temporal activity of LED itself is limited by the carrierlifetime in pn junction,

• junction capacitance of the LED has to be loadedbefore emitting light. This behavior can be affectedwith the LED modulation circuit,

• reproducibility of switching time.

3.4.2. Optical FiltersOne important task that any optical set-up in sensor instru-mentation has to perform, is the separation between theluminescence signal, which has to be collected and transmit-ted to the photo-detector and all other optical contributions,such as for instance the exciting radiation, other lumines-cent emissions due to materials close to the luminophore orcomposing the optical set-up, ambient light.

Usually when a luminophore is characterized by a broadluminescence band and by a Stoke shift large enough, opti-cal set-ups based on optical filters are used. It is possiblealso to use dispersive instrumentation [34, 43], often in com-bination with an array of photodiodes as a photo-detector.This instrumentation comprises prism-based or holographicgrating-based spectrometer. Sometimes it is also possibleto use spectrometers with more than one dispersive ele-ment, such as a holographic grating sandwiched betweentwo prisms, in order to increase the spectral resolving power(the minimum separation between two spectral lines thatcan be resolved). However the use of optical filters is oftennecessary also in case of a sensor instrumentation based ondispersive optical components.

For all those applications (and among these opto-chemical sensor instrumentation) where the stokes shift islarge, the optical filter based solution is preferred to the useof dispersive optical components due to its higher robust-ness, stability, smaller size and lower costs.

The separation of luminescence signal from the excitationcan appear at the first sight a rather easy task, since theluminescence emission is shifted toward longer wavelengthswith respect to the exciting radiation. However, light sourcestypically show a broad emission. Even in the case of LEDs,the emission at longer wavelengths beside the nominal peakare significantly high. For this reason one or more short-passoptical filters (excitation filter) must be used in front of theexcitation source to cut the emission at wavelengths closeto the luminescence band of the luminophore. In additiona long-pass optical filter or a set of long-pass optical filters(emission filter) should be set in front of the photodetectorto avoid that the radiation of shorter wavelengths than theluminescence band is also detected.

For example, in case of a Ru-dpp based sensor withabsorption spectrum at around 460 nm and emission around600 nm, a blue LED emitting at 460 nm is used as excitation


source. So an optical high-pass filter with an edge wave-length of 550 nm is used to separate the excitation from theemission light. Consequently, the filter at the exciting LEDhas to suppress the light of the LED at wavelengths higherthan 550 nm, the so called “red-tail” of the LED.

Even more complicated is the reduction of the contribu-tion of the optical background coming from luminescenceemission of other material than the luminophore. Sinceluminescence emissions typically occur in the red spectralregion, it is not really possible to distinguish these contri-butions from the luminescence signal. So the only possiblestrategy to cope with the luminescence contribution fromother materials is to avoid their excitation.

The choice of the excitation and emission filters is of fun-damental importance for the good functioning of the opticalset-up and it must be performed taking into account notonly the spectral characteristics of the excitation source andluminophore, but also other properties of the filters, such asthe self-fluorescence (in particular for the emission filter),the thermal and mechanical stability and the chemical inert-ness against the organic solvents typically used for cleaningof optic components (e.g., acetone, ethanol, isopropanol).

A wide range of different commercially available opticalfilters exist. They can be classified according to their func-tion in:

1. band-pass filters, which allow required wavelengthranges to pass through selectively;

2. long-pass filters, which block light of short wave-lengths and transmit light of long wavelengths;

3. short-pass filters, which block the light of longerwavelengths and transmit light of short wavelengths;

4. neutral density filters, which attenuate the visibleradiation in the same way for all the wavelengths ofthe spectrum.

Filters can be classified according their composition inabsorption based (glass and gelatine filter) or interferencebased filter.

a. Glass Filters Colored glass filters are characterized byselective absorption in a defined range of the visible spec-trum. They are obtained by adding impurities to glass mate-rial, which are otherwise transparent. This can be achievedin different ways, by adding heavy metal ions or ions of rareearths to obtain ionically colored glass, or by temperaturetreatment of particular color carriers (impurities) present inthe colorless glass to obtain colloidal colored glasses. In bothcases the color of the filter depends on the type and concen-tration of the added species and of the color carriers. In thecase of colloidally colored glasses the basic glass used andthe course of the temperature during the activation processplay also a significant role in determining the final propertiesof the filter.

Glass filter can be produced in different size and thick-ness, typically of one or more millimeters. They show avery good properties in terms of mechanical and tempera-ture stability and a high grade of chemical resistance againstmany solvent, but due to the presence of colorants they canexhibit a fluorescence emission under UV or visible (blue)excitation.

The measurement of the filter transmission and its even-tual self luminescence can be performed by measuring first

the transmission with the filter far away from the detec-tor (in this case the contribution of the eventual lumines-cence of the filter is negligible) and second by repeating themeasurement with the filter placed in front of the detec-tor. The apparent transmission, which is measured in thissecond case, is the sum of the real transmission and of theluminescence contribution emitted by the filter. Figure 11shows one example of the apparent transmission of threelong-pass glass filters used in opto-chemical sensor instru-mentation under direct green excitation from a typical LEDused as excitation source.

b. Gelatine Filters Gelatine filters are manufactured bydissolving suitable organic dyes in liquid gelatine and bycoating the proper amount of solution onto a glass surface.After the drying of the coating, the gelatine film is strippedfrom the carrier material and coated with a lacquer.

This kind of filters have a typical thickness of about0.1 mm and a maximum working temperature of 180 �C.Because of their uniform thickness, gelatine filters have anexcellent optical quality and are suitable for precise worksin which an increment in length of the optical path cannotbe tolerated. However, like glass filters they exhibit fluores-cence emission under exposition to visible radiation of shortwavelengths.

Moreover the dyes used in gelatine filters may alter theirspectral properties with time and may be degraded by heat,humidity and light (for example UV radiation). Gelatinefilters are attacked by most solvents and are dissolved byacetone.

c. Interference and Dichroic Filters Interference fil-ters are obtained by depositing a multi-layer structure ofthin films of dielectric material, called spacer, between tworeflecting surfaces. The light entering the filter undergoesmultiple reflections and the reflected beams interferes witheach other, experiencing constructive or destructive inter-ference depending on their wavelengths. The condition forconstructive interference is determined by the thickness ofthe spacer. Therefore, by selecting the proper thickness ofthe spacer it is possible to decide which wavelengths can passthrough the filter and which on the contrary are blocked bydestructive interference.

If the back layer is totally reflective, then the arrange-ment is called dichroic filter. These devices let the light ofa selected spectral region pass through and reflect the lightbelonging to the complementary range. They combine there-fore the behavior of the filter with that of a wavelengthselective beam-splitter.

In comparison with colored glass and gelatine filters,interference filters exhibit both narrow transmission bandsand they show very sharply defined transmission bands likelong-pass and short-pass filters. The blocking capability ofinterference filters (the measure of the transmission outsidethe band-pass region) is therefore much better than in caseof glass or gelatine filters. Since no light absorption takesplace in the material of filter, they are also characterizedby no luminescence emission. Beside they exhibit very sta-ble spectral characteristics with respect to temperature andhumidity changes, a shift of the nominal transmission bandoccurs when the filter is tilted with respect to the directionof incidence of the light. In particular it is found that as


300 400 500 600 700 800 900 100060

80

100

120

140

160

180

Wavelength (nm)

RG 665 (thickness 2 mm)A

mpl

itude

(a.

u.)

300 400 500 600 700 800 900 100060

80

100

120

140

160

180

Wavelength (nm)

RG 645 (thickness 2 mm)

Am

plitu

de (

a.u.

)

300 400 500 600 700 800 900 100060

80

100

120

140

160

180

Wavelength (nm)

RG 630 (thickness 2 mm)

Am

plitu

de (

a.u.

)

Figure 11. Apparent transmission of three long-pass glass filters withedge wavelengths at 665, 645 and 630 nm, respectively. The appar-ent transmission is measured by setting the filter directly in front ofthe photo-detector and by detecting the transmitted light of a propersource. In the present case a LED emitting at 522 nm was used as a lightsource. The recorded spectra show the emission peak of the source,which is transmitted by the filter, and a second broad emission, whichis given by the photoluminescence of the filter itself.

the tilt angle increases, the center of the transmitted bandmoves toward shorter wavelength.

The selection of the most suitable filters for the sen-sor instrumentation is principally subjected to optical andeconomical criteria: on the one hand the blocking capability

(the ability of the filter to block the radiation of wavelengthsoutside the nominal transmission band) and the reduction ofself-luminescence (especially if filters are set directly in frontof the photo-detector) are highly valued, and on the otherhand, the cost of the sensor instrumentation must be keptas low as possible. The best optical properties are shownby interference filters, which are the most expensive com-ponents, while the most affordable price is that of gelatinefilters, which on the contrary are the most critical compo-nents in term of mechanical, chemical and temperature sta-bility of their optical properties. For many applications glassfilters represent therefore a good compromise between opti-cal and economical aspects.

3.4.3. PhotodetectorsThe selection of the proper detector must take into accountthe following properties [25, 42]:

• the spectral relative response, which determines thespectral range where the detector can be used;

• the absolute spectral sensitivity, defined as the ratio ofthe output signal (a voltage or a current according toconfiguration in which the detector is used) to the inci-dent radiation power at the different wavelengths;

• the signal to noise ratio which is also a function of thewavelength;

• the maximum intensity range where the detector has alinear response (where the output signal of the detectoris linearly dependent on the incident radiation);

• the time response of the detector, which is a propertyparticularly important for applications where the inci-dent radiation is modulated (the output signal of thedetector typically fall with increasing modulation fre-quency of the incident light).

Existing different optical detectors can be classified intotwo classes: (a) the thermal detectors, such as bolometers,pyroelectric or thermoelectric detectors, and (b) the photo-detectors, like for instance photo-multiplier tubes, photo-conductors, junction photodiodes or avalanche photodiode[25]. In thermal detectors the energy absorbed from the inci-dent radiation raises the temperature and causes changes inthe temperature dependent properties of the detector. Ther-mal detectors are characterized by a wavelength indepen-dent sensitivity and therefore are very useful for calibrationpurposes, like for example for absolute measurements of theradiation power of lasers.

Photo-detectors are based either on the emissionof photo-electrons from photo-cathodes (photo-multipliertubes) or on changes of the conductivity of semiconduc-tor due to the incident radiation, or either on photo-voltaic devices where a voltage is generated by internalphoto-effect. Generally they show a spectral response whichdepends on the work function of the emitting surface oron the band gap in semiconductors. Photo-multiplier tubesare often used in laser spectroscopy and are suitable for allthe applications in which very low radiation powers must bedetected. A possible alternative for photo-multiplier tubesis represented by avalanche photodiodes. They are reversedbiased semiconductor diodes in which the free carriersacquire sufficient energy in the accelerating field produced


by the applied bias voltage to produce additional carriers oncollisions with the semiconductor lattice. The advantage ofthe avalanche photodiodes is their fast response time whichdecreases with increasing bias voltage.

However, photomultiplier tubes and avalanche photodi-ode are often excluded for application in sensor instrumen-tation due to economic considerations, especially when theapplications do not required extreme performance of thephoto-detector in terms of sensitivity and response time.Silicium (Si)-photodiodes have proved to be particularlysuitable for instrumentation based on the indirect measure-ment of the luminescence lifetime of opto-chemical sen-sors by phase-shift technique [32]. They actually show agood sensitivity at the operating wavelength (the lumines-cence maximum of the luminophore emission is typically inthe range 600–800 nm depending on the luminophore) anda response time fast enough for such applications. Morestringent for Si-photodiode in opto-chemical sensor instru-mentation are other properties, such as detector capaci-tance (especially for measurement by phase shift technique),noise, homogeneity of the active area, temperature behav-ior and stability of the photodiode [32]. In particular, theactive surface of the photodiode has to be as large as pos-sible to get as much light as possible, but the capacitanceof the photodiode must be as low as possible to reduce thenoise. Unfortunately, these two requirements are in contrast,since the capacitance is directly proportional with the areaof the active surface and a compromise has to be found.For example, for a homogeneous distribution of the lumines-cence light on the photodiode surface, a large active surfacesgive a better signal-to-noise ratio than smaller ones withlower capacitance. That is why in this case it is importantto use a photodiode with a high spatial homogeneity of theactive area. It is known that, specially at the margins of thephotodiode surface, deviations in the spectral sensitivity andresponse time from their values in the center are possible.Spatial homogeneity is a very important parameter, since itis impossible to have a change of spatial distribution of theluminescence on the photo-detector with time and tempera-ture, as a consequences of changes in the light source emis-sions or in the optical path: lens, filters, light guide, opticfibers, and in general any optical component of the measur-ing set-up are affected by temperature variations and maychange with time. If the spectral sensitivity and the responsetime of the photo-detector change as a consequence of thechange of the luminescence distribution, the measure of theluminescence intensity or lifetime (phase shift) also changes.

a. Photodiode Amplifier Circuits One key componentof the instrumentation for optical chemical sensors is theconversion of the optical signal coming from the fluores-cent dye back into electrical domain. Therefore several pho-ton devices exist to perform this conversion. Photodiodesare the most common devices for this purpose and have agood price/performance characteristics for the application inoptochemical sensors.

The goal of the photodiode amplifier circuits is to con-vert the signal from the photodiode into an adequate voltagesignal which can be further processed at the analog-digitalconverter (ADC) stage. In fact, this is the key to success foran optochemical sensor based on phase measurement tech-nique, because every noise which is introduced in this part

of instrumentation is amplified through all stages and theSNR can not be risen later without losing bandwidth. On theother hand this part should be as sensitive as possible to beable to reduce the intensity of the illumination light as muchas possible and consequently to reduce the photobleachingof the dye.

Concerning these assignments the photodiode amplifiercircuit has to feature the following requirements:

• high signal to noise ratio of the photodiode amplifiercircuit,

• phase shift independent of the signal magnitude,• no or low gain peaking to avoid undesired oscillations,• no or low influence of power supply noise,• no or low influence of electrostatic, magnetic and radio

Frequency Interference (RFI) coupling.

To achieve all these goals a consideration of all the possiblenoise sources would help to find the selection criteria forthe components and circuits.

In case of using the photodiode in photoconductive modethe noise sources of amplifier can be classified as:

• Thermal noise, intrinsic behavior of feedback resistorswhich can only be reduced by cooling the device.

• Voltage noise of the operational amplifier inputs.• Current noise of the operational amplifier inputs and

the photodiode itself.• Power supply noise which can be suppressed by an

high power-supply-rejection-ratio (PSRR) of the oper-ational amplifiers.

• Electrostatic, magnetic and RFI coupling noise, whichcan be reduced by a careful layout and adequate shield-ing of the circuits.

• In the case of using the synchronous demodulation,cross-talk of the demodulation signal is normally pro-duced on the same PCB and adds to the signal from thedye in a vector like addition and falsifies the output in asimilar way as the background signal (see Section 3.3.3Influence of Background Light).

Furthermore, in order to optimize the performance of thesystem several methods can be used to stabilize the photodi-ode parameters and to avoid undesired changes of the phasemeasurement. In this context the photodiode capacitancechange as a function of temperature has to be especiallyobserved.

4. SENSOR SCHEMES FORLIFETIME-BASED OPTOCHEMICALSENSORS USINGPHASE-MODULATION METHOD

Optical sensors based on the measurement of the lumines-cence intensity suffer from interferences by changes of tur-bidity, refractive index or color of the sample. Changes inthe optoelectronic system such as drifts of the light sourceand the photodetector, bending of the optical fibers and dis-placement or delamination of the sensing layer may alsocause signal changes to occur. Furthermore, degradation ofthe indicator caused by photobleaching and leaching are crit-ical. Decay-time sensing is advantageous over the intensity-based sensing because such measurements are independent


of probe concentration, photobleaching and drifts in lampintensity, of inner filter effects, all of which are major limita-tions of current fluorescence-intensity-based optrodes. Themeasurement of the luminescence decay time as a param-eter which is almost independent of the absolute signalheight can solve such problems and therefore has substan-tial advantages in practice. The advantages of lifetime-basedsensing (time-domain and frequency-domain sensing) areillustrated in Fig. 12 [44]. In contrast to intensity measure-ments, lifetime measurements depend on the signal duringa short period of time, 1–20 ns, depending on the probe’slifetime. The decay time is obtained from the slope of theintensity decay following pulsed excitation (Fig. 12, middle).The intensity-based sensing depends on reliable measure-ments of probe’s intensity. However, lifetime measurementsbased on pulse method are presently too costly and com-plex. Fluorescence lifetimes can be conveniently measuredby the phase-modulation technique, where the sample is

λ

∆IC

∆IH

Inte

nsity

Inte

nsity

Fluo

resc

ence

Tim

e do

mai

nFr

eque

ncy

dom

ain

Tim

e re

solv

ed τ = (slope)−1

τ = t (Φ)

Φ

Φ

Cuvette

Cuvette

Cuvette

Blood

Blood

Blood

Syringe

Syringe

Syringe

t0 t (ns)

Figure 12. Intensity, time-domain and frequency-domain sensing, asapplied in the laboratory, a cuvette and blood sample in a clinicalsetting. Reprinted with permission from [44], J. R. Lakowicz, “ProbeDesign and Chemical Sensing in Topics” in “Fluorescence Sensing,”Vol. 4, p. 6. Plenum Press, New York, 1994. © 1994, Plenum.

excited by light which is intensity-modulated at frequencies�f � ranging from 1 to 200 MHz. The possible mechanismsof lifetime-based sensing have been reviewed [23] and havebeen described for a large number of analytes, including pH[45], oxygen [46, 47, 48], carbon dioxide [49 ], NH3 [50] andglucose [51].

4.1. Optical Sensing of pO2 Using aPhase-Modulation Fluorimetry

The simplicity and robustness of phase fluorimetry is bestexemplified by the optical oxygen sensor. The availabil-ity of long-lifetime synthetic probes has been particularlyinstrumental in reducing costs in this case, permitting low-frequency modulation using simple electronics. As of now,lifetime determination (both time or frequency-domain) ofdyes with long-lived excited state is the method of choice forhigh-accuracy oxygen measurements [47, 48].

Oxygen optosensing is in general based on the efficientquenching of a photoexcited luminophore immobilized ina polymer matrix used as a solid support. If quenchingoccurs only by a dynamic (collisional) mechanism, then theratio �0/� is equal to I0/I and is described by the clas-sic Stern-Volmer equation (see Eq. (20)) [31], which pre-dicts a first-order type response to oxygen tension resultingin higher oxygen sensitivities at low oxygen concentrations.This equation is directly related to the intensity and can beexpressed as:

I0I= �0

�=1+KSV ·pO2 (39)

where, I0 and I are the luminescence intensities in theabsence and the presence of quencher (oxygen).

The overall dynamic quenching constant KSV (Stern-Volmer constant) can be described by:

KSV=kq ·�0 (40)where kq is the bimolecular quenching rate constant which,for a diffusion-limited reaction can be described by theSmoluchowski equation:

kq=4�Np�DA+DB� (41)where N is the number molecules per millimole, p is a factorrelated to the probability of each collision causing quenchingand to the radius of interaction between the donor and thequencher. DA and DB are the diffusion coefficients for thedonor and acceptor, respectively.

Using Eq. (33) it is possible to relate intensity or life-time of phosphorescence (or fluorescence) to the oxygenpartial pressure. The oxygen sensing properties of phos-phorescence intensity based sensors are characterized bythe ratio I0/I100, where I0 and I100 represent the detectedphosphorescence intensities from the film exposed to argonand oxygen saturated conditions, respectively (Fig. 13) [52].The phosphorescence intensity changes under various oxy-gen pressures and Stern-Volmer plots are shown in Fig. 14[52]. The KSV value is obtained from the slope of I0/I versuspO2. The highly sensitive oxygen sensor has a larger I0/I100and KSV, respectively.


I100

I0

760

pO2 = 0 Torr. . .

Phos

phor

esce

nce

inte

nsity

(a.

u.)

Wavelength (nm)

Figure 13. Typical phosphorescence spectrum changes of probemolecule immobilized in polymer film under various oxygen partialpressure conditions. Reprinted with permission from [52], Y. Amao,Microchim. Acta 143, 1 (2003). © 2003, Springer.

The first phase fluorimetric sensing of oxygen based onabove described principle was reported by Woflbeis andco-workers [53]. This phase fluorimetric oxygen sensor mea-sures the quenching by oxygen of the transition metal com-plex, tris[4,7-diphenyl-1,10-phenanthroline] ruthenium(II)2+

(Ru-dpp)2+. The fluorophore was immobilized in a siliconerubber membrane, which is an ideal matrix for oxygen deter-mination because of its high oxygen solubility. The siliconematrix selects for gaseous analytes because of its hydropho-bic nature, adding to sensor specificity. In addition, thehydrophobic matrix and the water insolubility of the ruthe-nium complex add to long-term sensor stability by prevent-ing the leaching of the probe.

The sterilizable oxygen sensor with new polymer (poly-sulfone (PSU) or polyetherimide (PEI)/dye (Ru(II)-tris(4,7-diphenyl-1,10-phenanthroline) combinations, was developedat Joanneun Research [54] for use in bioreactors. The mea-surement system detected the luminescence lifetime of thedye in microsecond range by means of the phase-modulationtechnique. The light source was continuously modulated ata frequency of 20 kHz. Schematic optical set-up and compo-nents are illustrated in Fig. 15.

Beside Ru(II) complexes, a variety of luminescent speciescan be used as indicators, particularly those exhibitinglong-lived singlet or triplet states for maximum sensitiv-ity to the analyte [27]. Among the most frequently usedluminophores for O2 monitoring are the polycyclic aro-matic hydrocarbons (PAHs) [55], because of their longexcited-state lifetimes (40–300 ns) and high solubility in theextremely oxygen permeable silicone matrices. However,their UV excitation wavelengths and easy photodecomposi-tion have prevented so far a wide usage. The strong room-temperature phosphorescence ��m>0�1� and long emissionlifetimes (>10 �s) of Pd(II) and Pt(II) porphyrin complexes[56, 57] make them promising indicator dyes, but they oftenundergo facile oxidation when illuminated in the presenceof O2. Highly luminescent Ru(II) complexes [58–60] withpolyaza heterocyclic chelating ligands are currently the mostwidely used oxygen-sensitive indicators. Their advantages

00

00

2

4

6

8

10

12

14

16

760570380190

760570380190

0.2

0.4

0.6

0.8

1.0(a)

(b)

Rel

ativ

e in

tens

ity (

a.u.

)

Oxygen partial pressure (pO2) [Torr]

Oxygen partial pressure (pO2) [Torr]

(I0/

I)−1

Figure 14. (a) Phosphorescence intensity changes and (b) Stern-Volmerplot of probe molecule immobilized in polymer film under variousoxygen pressures. Reprinted with permission from [52], Y. Amao,Microchim. Acta 143, 1 (2003). © 2003, Springer.

include a strong absorption in the blue, intense lumines-cence quantum field �em up to 0.4 in the 550–800 nmregion with a large Stokes shift (>150 nm) due to the long-lived (0.1–7 �s) metal-to-ligand charge transfer (MLCT)excited state, close to diffusion-limited O2 quenching andthe possibility of tuning their photophysical and immobiliza-tion properties by a judicious choice of the chelating lig-ands(typically 2,2′-bipyridine (bpy) or 1,10-phenanthroline(phen)).

In fluorescence sensing there is an entirely differentclass of methods for measurements, based on ratio deter-minations. Some of them do not require modulatingthe excitation light. The use of such methods requiresoxygen-sensitive dye with specific characteristics. Two-dyemolecules have been employed for similar purpose [61];however a true dual-emitting dye is required to over-come the variations in bleaching [62]. The synthesis ofa new oxygen sensitive dye [(dppe)Pt{S2C2(CH2CH2-N-2-pyridinium)}] [BPh4] [63], that exhibits both fluores-cence and phosphorescence emissions, made ratio oxygen

308 Optochemical Sensors Based on LuminescencePr

eam

plif

ier

mod

ule

Photodiode

Emission filter

Oxygen sensitive film

Luminescence signal

Excitation light

Reference signal

Optical filters

Reference LEDExcitation LED

Figure 15. Optical set-up and components.

measurements more feasible. Recently, the same dual-emission dye immobilized in cellulose acetate, has beenemployed for development of ratio-based oxygen sensor[64]. The immobilized dye exhibits two emissions withgreatly different lifetimes. The triplet intensity in film is sig-nificantly quenched by oxygen; this makes the dye suitablefor for oxygen ratiometric measurements. It allows the peakratio to be measured through any of the existing fluores-cence methods (intensity, lifetime or polarization).

Room temperature phosphorescence (RTP) quenching-based sensors have become the focus of recent attentionas they posses several advantages over those based on flu-orescence. The longer excited-state lifetimes of phospho-rescent indicators give rise to high quenching efficiencyby oxygen. The long excitation and emission wavelengthsare more compatible with available optical monitoringtechnology. Most recently, RTP quenching-based sensorshave utilized metal chelates such as tetrakis (pyrophos-phito)diplatinate(II) [65] and 8-hydroxy-7-iodo-5-quinolinesulfuric acid (ferron) chelates [66] and a range of modi-fied Pd(II) and Pt(II) porphyrins [67]. When the porphyrindye is used for oxygen measurement in vivo by intra-venous infusion, a palladium series, especially Pd-meso-tetra-(4carboxyphenyl)-porphyrin (Pd-TCPP), is employed[68]. Its phosphorescence is sensitive to oxygen, it is chem-ically stable, its lifetime is rela

Date post:	27-Jan-2021
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

OptochemicalSensorsBased onLuminescence...294 OptochemicalSensorsBasedonLuminescence index of the...

Documents