Developing a laser-based biosensor with plasmonic accuracy

MScThesis in Europhotonics Technical University of Catalonia UPC-BarcelonaTech Centre for Sensors, Instruments and Systems Development CD6 Developing a laser-based biosensor with plasmonic accuracy Author: Athanasios Vidakis MSc Project supervisor: Dr. Santiago Royo Presented in Terrassa, Spain, July 8th 2015 LaserBased BioSensor Abstract This work evaluates a newly developed laser-based sensor which detects small changes of the refractive index. Currently, the state of the art in high-accuracy refractive index sensing are biosensors based in surface plasmon resonance (SPR). However, the high cost, the complexity and the sensitivity of plasmonic set-ups has not allowed yet this sensors to be part of the market . On the other hand, the technique proposed in this project , based on Differential Self Mixing Interferometry (DSMI), counts some low-cost, stability and compactness advantages when compred to SPR. Theoretical models and simulations, which have been developed along this research, are presented showing that the technique is capable to measure refractive index with an accuracy of 10−7. Two experimental setups for validation have been built along this thesis, and a series of experimental tests have been performed. Results show good agreement between theory and experiments, with a reasonable reduction in performance due to instability issues and environmental constraints. The first experimental set-up has shown an accuracy of 10−5in refractive index-sensing and the final set-up showed an resolution of 0.25g/L in detection of glucose concentration in water, while it is capable to improve its accuracy with some minor modification and a better stabilized environment. I LaserBased BioSensor II LaserBased BioSensor Acknowledgement First and foremost, I would like to show my deepest gratitude to my supervisor, Dr Santiago Royo and to his PhD student Francisco J. Azcona who have provided me with valuable guidance at every stage of my research. Secondly, I shall extend my thanks the School of Electrical and Computer Engineering of Aristotle University of Thessaloniki and especially the professors: Emmanouil Kriezis , Georgios Sergiadis, Leontios Hadjileontiadis and Antonios Papagiannakis who taught me what the word "Engineer" means and provided me with all the necessary knowledge to work through this goal. Last but not least, I would like to thank my family and my friends who encourage and support me all these years. III LaserBased BioSensor IV LaserBased BioSensor Contents Abstract VI Acknowledgement VI Table of contents VI List of figures VIII List of tables IX 1 Introduction 1 1.1 Motivationofthework.............................. 1 1.2 Lasers and Interferometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.3 Goal-Outcome................................... 2 1.4 ThesisOrganization................................ 3 2 Theoretical Background 5 2.1 Self-Mixing-Interferometry . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1.1 Operating principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1.2 External Cavity Model . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.3 Self-Mixing Signal-Processing . . . . . . . . . . . . . . . . . . . . . . 8 2.1.4 SMI Applications & State of the Art in refractive index-sensing . . 9 2.2 Differential-Self-Mixing-Interferometry . . . . . . . . . . . . . . . . . . . . 11 2.2.1 PrinciplesofDSMI............................ 11 2.2.2 Steady State Capturing . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2.3 Self-Referenced DSMI . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2.4 DSMIApplications............................ 14 2.3 Three external Cavity Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3.1 Theoretical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3.2 Simulation................................. 17 3 Proof of Concept 23 3.1 1st ExperimentalSet-up.............................. 23 3.2 Diagnostics..................................... 26 3.3 Preparation .................................... 28 3.4 Processing ..................................... 29 3.5 Results ....................................... 31 4 Towards Bio-Sensing 33 4.1 Idea......................................... 33 4.2 Theoreticalresearch................................ 34 4.2.1 Brewster’s angle technique . . . . . . . . . . . . . . . . . . . . . . . 34 4.2.2 Matrix Theory for Multilayer Optics . . . . . . . . . . . . . . . . . . 35 4.2.3 Simulation................................. 37 4.2.4 GlucoseLiterature ............................ 39 4.3 Glucose solutions preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 40 CONTENTS V LaserBased BioSensor 4.4 ExperimentalWork................................ 41 4.5 Results ....................................... 43 4.6 DSMIvsSPR.................................... 45 5 Conclusion and Future work 47 5.1 ResearchContributions.............................. 47 5.2 FutureResearchTopics.............................. 48 Bibliography 49 VI CONTENTS LaserBased BioSensor List of Figures 2.1 Conventional self-mixing configuration using a LD. Source: Adapted from [1].......................................... 5 2.2 How the the feedback parameter C influence the SM signal . Adapted from [2].......................................... 6 2.3 Theoretical representation of the LD with external feedback and its equivalent model of the two mirror cavity: Adapted from [3] . . . . . . . . . . . 7 2.4 (a) Electronic signal processing block scheme for fringe-counting displacement interferometer. (b) Upper trace: experimental self-mixing signal obtained for a sinusoidal target displacement of 3.3 µmp p amplitude and 1 kHz frequency; lower trace: analogue derivative of self-mixing signal, showing up/down pulses. Timescale: 100 µs/div Adapted from [1] . . . . 9 2.5 StandardDSMISetup[4]............................. 11 2.6 Fringe position change caused by a target displacement smaller than λ/2 fortwolaser[5]................................... 11 2.7 SteadyStateCapturing ............................. 13 2.8 Three external Cavity Fabry-Perot Model. . . . . . . . . . . . . . . . . . . . 15 2.9 EquivalentCavity.................................. 15 2.10Simulatedsetup................................... 18 2.11 Results of simulation for a temperature change in water for a range of ±0.5°C and C∼0.5. A) For a 0.5mm cavity. B) For a 1.0 mm cavity. C) For a 3.0 mm cavity. D) For a 4.0 mm cavity. . . . . . . . . . . . . . . . . . . . . 20 2.12 Results of simulation for a temperature change in water for a range of ±0.5°C and C∼1. A) For a 0.5mm cavity. B) For a 1.0 mm cavity. C) For a 3.0 mm cavity. D) For a 4.0 mm cavity. . . . . . . . . . . . . . . . . . . . . . 20 3.1 DSMI experimental prototype setup. . . . . . . . . . . . . . . . . . . . . . . 23 3.2 The program which controls the used instruments (oscilloscope, signal generator) and also captures the SM signals via the oscilloscope . . . . . . 24 3.3 Aluminium target design for 4 different cavities for liquids (1, 2, 3 or 4[mm]). 25 3.4 SM signal in DSMI technique C∼1(yellow). The blue line represents the displacement of the piezoactuator. . . . . . . . . . . . . . . . . . . . . . . . 25 3.5 Plot of Lasing Wavelength vs. Case Temperature from the specification sheet for the Hitachi HL7851G laser diode . . . . . . . . . . . . . . . . . . . 27 3.6 Phase fluctuations of the fringe maximum during a 600-s-long measurement[6]....................................... 27 3.7 The phase shifting of the fringes is translated to displacement in [nm] (∆d). Each measurement shows the continuous displacement/phase-shifting for 1 second (time of a ramp). (a) In case that the lights are on, the live measuring show an average absolute error of 20 [nm] (b)In case that the lights are offthe live measuring show an average absolute error of 5 [nm] 27 LIST OF FIGURES VII LaserBased BioSensor 4 1.4. THESIS ORGANIZATION LaserBased BioSensor 2. Theoretical Background This charter gives the theoretical Background of this thesis. The first part includes the basic theory for SMI. In the second part the SMI theory is expanded for Differential SMI (DSMI) development. In the third part is presented the "Three external cavity model" which was necessary in order to simulate the potential DSMI set-up for refractive indexsensing. The third part contains also the simulation of the refractive index-sensor and its results. 2.1 Self-Mixing-Interferometry Self-mixing interferometry (SMI), also known as optical feedback interferometry (OFI), is a well characterized method capable of measuring displacement related phenomena. The SMI effect was first observed in the decade of 1960, however it was disregarded and considered a nuisance for the development of laser based communications. Later, in the early 1980’s, the interest of characterizing the behaviour of the laser diodes (LD) gave rise to some of the early studies that can be related to the SMI method. Of those studies, probably the most representative is the study performed by Lang and Kobayashi [17]. The study describes and tests a mathematical model for a single-mode LD subjected to feedback and the appearance of a modulation in the LD optical output power (OOP) related to the amount of feedback. Later work on the subject proposed other models such as double Fabry-Perot cavity [18] with a single reflection in the second cavity to account for the modulation produced over the OOP. 2.1.1 Operating principle Figure 2.1: Conventional self-mixing configuration using a LD. Source: Adapted from [1] In short, SMI can be defined as the modulation of a LD-OOP caused when back-reflected light from the target interferes with the light already present in the cavity (fig.2.1) and depending on the time delay (phase) of the back-reflected beam, the LD (Laser Diode) threshold condition is diversified. Concerning that the pump current is held constant, the emitted power changes with small displacements of the target ( the displacements are translated to variation on the time delay /phase). A slight variation is also implied in the emitting wavelength because of the change in the LD threshold [19]. The changes on the OOP are then monitored by an internal photo-diode (PD), typically located in the back-facet of the electronic package. 5 LaserBased BioSensor The complete analysis of optical feedback in LDs can be performed by using the Lang and Kobayashi equations [17]. From Lang and Kobayashi equations derived the following simple and general expression for the power emitted by the LD: P(φ)=P0[1 +mF(φ)] (2.1) where P0is the power emitted by the unperturbed LD, m is the modulation index and F(φ) is a periodic function of the interferometric phase φ=2kD, of period 2πwhere k=2π/λ. The modulation index m and the shape of the function F(φ) depend on the so-called feedback parameter C (after [19]): C=κ·D·√1+α2 l·n(2.2) where αis the LD linewidth enhancement factor, l is the laser cavity length, D the distance between the LD and the target, n is the cavity refractive index and κis given by: κ=√Rext √A 1−R2 √R2 (2.3) where R2is the LD output facet power reflectivity (fig. 2.1), Rext the power reflectivity of the remote target and A>1 is the optical power attenuation in the external cavity. Practically this attenuation is necessary especially when the remote target is close to the laser (less than half a meter) and can be regulated with density filters or with the adjustment of the laser’s lens as we will see in the experimental chapters. Figure 2.2: How the the feedback parameter C influence the SM signal . Adapted from [2] The value of the C parameter depends on both the amount of feedback and, interestingly, on target distance D. The C parameter is of great importance, because it discriminates between different feedback regimes (fig.2.2): •For C 1, we have the very weak feedback regime. The function F(φ) is a cosine and the modulation index m is inversely proportional to √A. •For 0.1 <C<1, we have the weak feedback regime. The function F(φ) gets distorted, showing a non-symmetrical shape while the modulation index m is again inversely proportional √A. If this nonsymmetrical shape is to the right as in the figure or to the left we can export if target is moving away or approaching the target respectively. •For 1 <C<4.6, we have the moderate feedback regime. The function F(φ) becomes three-valued for certain values of the phase φ,i.e. the system is bistable, with two stable states and one unstable . The modulation index m increases for decreasing √A, but it is no longer inversely proportional to it. The interferometric signal becomes sawtooth like and exhibits hysteresis. 6 2.1. SELF-MIXING-INTERFEROMETRY LaserBased BioSensor •For C >4.6, we have the strong feedback regime. The function F(φ) may become fivevalued , and it is experimentally tested that not all the specimens of LD’s remain in the self-mixing regime; rather, in some cases the LD enters the mode-hopping regime and interferometric measurements are no longer possible. [1, 14] 2.1.2 External Cavity Model In order to proceed to the three external cavity model approach, which is presented in the last section of this chapter, we have to go a-bit deeper to Self-Mixing Theory and get the idea of the basic external cavity model. Figure 2.3: Theoretical representation of the LD with external feedback and its equivalent model of the two mirror cavity: Adapted from [3] So, a simple approach to understand how SMI works, is the one of an equivalent model of two mirror cavity as it is shown in the figure (2.3). The first step for this model is the estimation of the equivalent reflectivity r2eq of the second mirror. Considering that the feedback level is not so strong, we take into account only the "first" reflection from the target. That way, the equivalent reflectivity r2eq can be estimated as: r2eq(ν)=r2+(1 −r2 2)r3e−j2πντD =r2[1 +κe−j2πντD](2.4) where: •r2,r3is the amplitude reflection coefficient of the LD’s facet and the target’s respectively, •νis the optical frequency emitted by the LD under feedback conditions, •τD=2Dn/cis the external round trip delay,with c the speed of light and nthe refractive index of the media between the LD and the target (usually air), •D the distance between the LD and the target, which must be smaller than the half-coherence length of the LD (around 4 m), •κ=(1 −r2 2)r3/r2is the feedback coupling coefficient. As the LD’s pump current is held constant and above its threshold, it has to satisfy the phase and the amplitude criterion for emission. Expressing the equivalent reflectivity r2eq as: r2eq =|r2eq|∠φeq the amplitude criterion is: r1|r2eq|e(gc−αs)L=1 (2.5) 2.1. SELF-MIXING-INTERFEROMETRY 7 LaserBased BioSensor and the phase creterion: 2βl+φeq =2πm(2.6) Applying the two last equations in the case that there is back-reflected beam, and the case that there is not (r2eq =r2), from phase equation we can end up to : ∆φ=2πτ0(ν−νth)+√1+α2κ1sin(2πντD+tan−1α)=0or ν−νth +C 2πτD κ1sin(2πντD+tan−1α)=0(2.7) and from the amplitude we can result the optical output fluctuations: ∆P∼gc−gth ∼ −1 lκcos(2πντD) (2.8) where •νis modulated optical frequency due to the feedback, •νth is in case of no feedback, •αis the linewidth enhancement factor of the LD, •Cis the so-called feedback parameter which is referred in the previous section. The optical output fluctuations are function of τD, which is changing according to the target’s displacement, and the modulated optical frequency ν, which depends on τdas it is the solution of the equation (2.7) for a given τd. If we look carefully at the equation (2.7) we can see that if the feedback parameter is C>1 the equation has more than one solutions while for values C≤1 there is only one solution for the modulated frequency ν . Finally if we connect the interferometric phase of the previous section φ=2kD with the time delay C=2Dn/cwe have: τd=nφλ/πc, and we can result that the periodic function of the equation (2.1) is : F(φ)=co[(2πν(φ)n 2π λ cφ] (2.9) 2.1.3 Self-Mixing Signal-Processing This part presents, in a few words, the signal-processing we implement in the captured photodiode’s signal in order get the necessary informations about the target’s position or about the optical path’s length between the LD and the target. First of all we have to define how the optical output fluctuations are connected with the displacement of the target. From the phase of the equation (2.8) we can get the "displacement period" as: ∆(2πντD)=2π⇒∆D=λ/2 (2.10) That means that a full-period of SM signal on time corresponds to a λ/2 displacement . In case we are working we the feedback parameter C close to 1 the SM signal looks as the one in the figure (2.4b). Working with the C∼1 it is easier to get the direction of the target. So the steps of SM signal processing are: •Amplifying The incorporated photodiode in the LD structure gives a really weak signal which has to be amplified . •Derivative The signal is differentiating in order to detect the rapid changes of the SM signal on time. For the purpose of clarity, on the next sections the word "fringes" will be used to describe those rapid changes. 8 2.1. SELF-MIXING-INTERFEROMETRY LaserBased BioSensor (a) (b) Figure 2.4: (a) Electronic signal processing block scheme for fringe-counting displacement interferometer. (b) Upper trace: experimental self-mixing signal obtained for a sinusoidal target displacement of 3.3 µm p p amplitude and 1 kHz frequency; lower trace: analogue derivative of self-mixing signal, showing up/down pulses. Timescale: 100 µs/div Adapted from [1] •Polarity Discrimination Then we check which fringes where "up" and which "down". Up fringes show a displacement towards the laser and the down the opposite (moving away). •Up-Down Counter and Display Finally , counting the up and down fringes we can get easily the target’s displacement with an accuracy of λ/2. (The distance of two same fringes on time corresponds to a λ/2 displacement .) 2.1.4 SMI Applications & State of the Art in refractive index-sensing SMI has already provided well established metrological applications with low-cost, stability and compactness advantages. Those applications have been reported in [3, 1] and include: •Displacement measurement with accuracy from 1µmto a few nm. •Velocity measurement with error less than 5% [13] •Distance measurements with the accuracy of 0.5 mm at 60cm distance •Vibration measurement with amplitude larger than λ/2 for modal analysis, vibration and noise testing, characterization of loudspeakers and piezoceramic transducers. •Operation on diffusive targets for displacements of a few mm in order to avoid speckle effects. Bio-Sensing Apps Because of the strong interest in SMI technique bio-sensing application implemented in a short time. Most of them are based on the previous application.[2] •Vibro-cardiographic signal (VCG) which is similar to ECG and it is optically picked-up without electrical connection nor any physical contact with the patient. •Respiratory sounds were also revealed by using SMI without a physical contact. •Blood Velocity detection by In vivo application of SMI with an optical fiber in a needle. 2.1. SELF-MIXING-INTERFEROMETRY 9 LaserBased BioSensor State of the Art in refractive index-sensing with SMI: •The first efforts for thickness and refractive index measurement were reported in [20, 21] with the an accuracy of 0.02 and 1% respectively. The methods are based on the optical phase shift measured by a self-mixing interferometer (SMI) as a function of the angle of incidence on the sample. •The last two years (2013-14) techniques were reported where SMI is enhanced by the contribution of birefringent material or acousto-optic modulators achieving high accuracy sensing with the cost of complexity, 0.00002 on refractive index and 0.6 1µmon thickness [22, 23]. 10 2.1. SELF-MIXING-INTERFEROMETRY LaserBased BioSensor 2.2 Differential-Self-Mixing-Interferometry Differential Self-Mixing Interferometry technique came up to surpass the limited accuracy of SMI. Theoretical results showed that this technique can measure sub-nanometer scale displacements with resolution within the angstrom scale. The experimental setups that have been demonstrated till now reached the accuracy of 3-5 nm due to mechanical limitations and instability issues [4]. 2.2.1 Principles of DSMI DSMI technique is based on the time delay of two SM signals, which are produced on purpose with the contribution of a displacement stage. A typical setup of DSMI is shown in figure (2.5) where the measurement laser Lmis mechanically attached to a linear displacement piezoelectric actuator, and the reference laser Lris placed facing the actuator. The DSMI setup is adjusted so that the feedback parameter Ccan range around one (C∼1). That way the SM signal has a sawtooth shape which makes easier the fringe’s tracking. Figure 2.5: Standard DSMI Setup [4] Figure 2.6: Fringe position change caused by a target displacement smaller than λ/2for two laser [5]. In detail, the linear displacement of the piezoelectric actuator will provoke the creation of SM signal in both lasers. Supposing that: •the target and the environmental conditions are stable, •that the two laser have the same wavelength, •and that the displacement of actuator is linear (constant velocity), the time difference between two fringes will be stable and equal for both Lasers: ∆tm= ∆tr. Both ∆tmand ∆trcorrespond to λ/2 displacement of the piezoelectric actuator. If the target starts moving the fringes of measurement laser will be appeared faster or delayed from the reference. Thus, the target’s displacement can be revealed as a function of the time delay: ∆t= ∆tm−∆tr. Due to the differential nature between displacement and velocity, it is possible to calculate the discrete displacement with two lasers that have a small difference of wavelength as: ∆d=λr 2·∆t ∆tr(1 −∆λ λr)−∆t(2.11) where ∆λ=λm−λris the wavelength difference of the lasers Lmand Lrrespectively. In case that the measurement and reference laser have almost the same wavelength : 2.2. DIFFERENTIAL-SELF-MIXING-INTERFEROMETRY 11 LaserBased BioSensor ∆λλr,m, the displacement can be expressed as: ∆d=λ 2·∆t ∆tr−∆t(2.12) and for really small displacements: ∆d=λ 2·∆t ∆tr (2.13) DSMI works as a sampler on time, with the sampling period being the time that the piezoactuator needs to be displaced λ/2.To calculate the sampling frequency then, we need to know the velocity,of the piezoactuator. If vis the velocity of the actuator and Ts the time for a λ/2 displacement then Tsis also the sampling-period and we have: v=λ 2Ts⇒fs=2v λ(2.14) Although, usually we do not know the velocity of the actuator, but the amplitude (A) and the frequency (fr) of the triangle signal that controls it. In that case the velocity vfor the up or the down ramp can be expressed as: v=2Afr(2.15) Combining the last two equations the sampling frequency is: fs=4Afr λ(2.16) So in the case we know the maximum frequency (ft), of the target’s displacement, from the Nyquist theorem it is easy to say that we need the double sampling ratio: fs=2ft(2.17) We have to consider also that the piezoactuator’s frequency is limited and in case of an non-deformed triangular displacement the maximum fris around 1kHz. This technique, of continuous displacement tracking, lacks in-between the time that the actuator’s ramp is chancing from down-ramp to up and upside-down. The sign of the target’s displacement (∆d) is different if we are tracking the target’s motion in the up or the down ramp of the piezoelectric actuator: •If the up-ramp is the one in which the laser Lmis approaching the target and the fringes are delayed on time (∆tm>∆tr), means that the target is moving away from the laser while when the fringes precede on time (∆tm<∆tr) means that the target is moving towards the laser. •In the opposite direction if the down-ramp is the one in which the laser Lmis moving away from the target and the fringes precede on time (∆tm<∆tr), means that the target is moving away from the laser while when the fringes are delayed on time (∆tm>∆tr) means that the target is moving towards the laser. 2.2.2 Steady State Capturing Except the continuous tracking which was presented above as the basic principle of DSMI technique, there is also one bit-different way to capture the target’s displacement. The idea is to capture the SM signal for the up-ramp (or the down-ramp) and capture it again after a time interval (e.g. 1 sec ). For that, we have to know that the target’s displacement for the capturing time (Tr/2) is insignificant and can be considered as stable. In case of a 12 2.2. DIFFERENTIAL-SELF-MIXING-INTERFEROMETRY LaserBased BioSensor small displacement in the interval time between two captures, the SM signal will "shift" on time giving the desired ∆tas it shown in the figure (2.7). In this kind of tracking is used mostly the measurement laser (Lm) while the reference laser (Lr) can be used to detect changes in optical path, because of temperature instabilities for example, giving the undesirable time difference ∆ter, with the same exactly way and decrease the error. In Figure 2.7: Steady State Capturing that case, the diplacement can be revealed from the equation: ∆d=λ 2·∆t−∆ter ∆tr (2.18) The basic drawback of this method is that can capture absolute displacement smaller than λ/2 or smaller than λ/4 when sign sensitivity is required. Although we can take more "captures" or "measurements" and track larger displacements as long as the displacement, in-between the time intervals of the captures, does not excess the fore-mentioned limitations. 2.2.3 Self-Referenced DSMI The self-referenced method uses a simplified model of the double laser method . In this case, two consecutive measurements are needed. First, a calibration is performed in order to capture the time difference between fringes when the target is fixed. During this step, a C value calibration is also performed. The reference signal is acquired and processed according to the typical fringe detection in the SMI analysis. Once the reference signal is acquired, one or several samples of the target displacement can be taken. As in the double laser case, with the reference and measurement wavelengths to be equal, the DSMI formula was described by the equation (2.12), it can be also applied for the Self-Referenced method: ∆d=λ 2·∆t ∆tr−∆t(2.19) for the continuous tracking and: ∆d=λ 2·∆t ∆tr (2.20) for the capturing method. A drawback of Self-Referenced DSMI is that once the setup is calibrated then it does not take into account changes as air’s refractive index variation . In the case of the continuous tracking, environmental changes that affect the refractive index of the air, are not important as they are changing slowly, but in the case of the capturing method they might introduce large errors. 2.2. DIFFERENTIAL-SELF-MIXING-INTERFEROMETRY 13 LaserBased BioSensor 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (a) @ T=20 ∆n=3.9534×10−5@ T=19.5 ∆n=−4.083×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (b) @ T=20 ∆n=4.1477×10−5@ T=19.5 ∆n=−4.2774×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (c) @ T=20 ∆n=4.1153×10−5@ T=19.5 ∆n=−4.2935×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (d) @ T=20 ∆n=4.0829×10−5@ T=19.5 ∆n=−4.2773×10−5@ T=20.5 Figure 2.11: Results of simulation for a temperature change in water for a range of ±0.5°C and C ∼0.5. A) For a 0.5mm cavity. B) For a 1.0 mm cavity. C) For a 3.0 mm cavity. D) For a 4.0 mm cavity. C∼1Model The excess phase equation (2.42) was solved using Schröder’s method for C∼1 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (a) @ T=20 ∆n=4.1148×10−5@ T=19.5 ∆n=−4.1791×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (b) @ T=20 ∆n=4.147×10−5@ T=19.5 ∆n=−4.2112×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (c) @ T=20 ∆n=4.1309×10−5@ T=19.5 ∆n=−4.2112×10−5@ T=20.5 0 0.01 0.02 0.03 0.04 −1 −0.5 0 0.5 1 time [sec] Normalized O.O.P (d) @ T=20 ∆n=4.1041×10−5@ T=19.5 ∆n=−4.2112×10−5@ T=20.5 Figure 2.12: Results of simulation for a temperature change in water for a range of ±0.5°C and C ∼1. A) For a 0.5mm cavity. B) For a 1.0 mm cavity. C) For a 3.0 mm cavity. D) For a 4.0 mm cavity. Cavity Length (L3) [mm] 0.5 1 2 3 ∆T=0.5°C Expected ∆n[10−5] : −4.2049 −4.2049 −4.2049 −4.2049 Measured ∆n[10−5] : −4,1791 −4,2112 −4,2112 −4,2112 Error [10−5] : 0.0258 −0.0063 −0.0063 −0.0063 ∆T=−0.5°C Expected ∆n[10−5] : 4,1148 4,1469 4,1309 4,1041 Measured ∆n[10−5] : 3.9534 4.1477 4.1153 4.0829 Error [10−5] : −0.0101 0.0219 −0.0058 −0.0209 Table 2.4: Results for a ±0.5°C change on a water cavity of different length with C ∼1. 20 2.3. THREE EXTERNAL CAVITY MODEL LaserBased BioSensor The reason we present the same simulation for C values 0.5 and 1 is that the first one which is based only on the equation (2.41) runs in a few seconds while the simulation for C∼1 runs in a few minutes. So in case we want to get a first image, for some different values of the table (2.1) it’s more efficient to run the simulation for C∼0.5. Both simulations showed that the DSMI set-up has the potential to measure refractive index changes with an accuracy larger than 10−6and especially in the case of C∼1 the accuracy could reach up to 10−7. 2.3. THREE EXTERNAL CAVITY MODEL 21 LaserBased BioSensor 22 2.3. THREE EXTERNAL CAVITY MODEL LaserBased BioSensor 3. Proof of Concept This chapter describes in detail how the DSMI experimental set-up, for refractive index sensing, was build, its instability issues, the processing we follow and the experimental results. 3.1 1st Experimental Set-up In order to test the attainable resolution of the DSMI method in laboratory conditions, a prototype was built (Fig. 3.1). The prototype setup is composed by a Fabry-Perot HL7851G laser diode of Hitachi. The LD is mounted on a Thorlabs LT230220P-B collimating tube equipped with a C230220P-B aspheric lens pair with numerical aperture NA = 0.55/0.25. The lens is adjusted to produce SM signal in the boundary between weak and moderate feedback regimes in order to avoid any possible fringe loss. The tube is then attached using a mechanical aluminium support to a PI LISA (P-753.2CD) PZT stage with a maximum travel distance of 25µm. Electronically, the LD is connected to a custom made electronic circuit which includes a controlled current source and a conditioning stage for the SMI signals. The SMI signals are digitized using a Tektronix DPO2024B oscilloscope with 125kS window, which directly sends the information to a remote PC for further processing. The displacement of the target was chosen as a triangular motion with fixed peak to peak amplitude of 10µmand frequency of 0.5 and 1Hz. For the control of the piezoactuator was used a Signal Generator connected with Digital Piezo Controller. (a) (b) Figure 3.1: DSMI experimental prototype setup. 23 LaserBased BioSensor Control In order to capture the SM signal one or more times, the oscilloscope and the signal generator were connected to a pc which runs the program shown in the figure (3.2). •The first part of the program controls the oscilloscope’s window for 4 different channels. For this experiment only the first two channels were used, one for the SM signal and one for the signal of the piezoelectric actuator. •The second part is the one that controls the signal generator and consequently the piezo displacement. We can choose the waveform type, the Amplitude, the Offset, the frequency and the Phase of the piezo. The 10 V correspond to the maximum displacement of the PI-LISA which is 25µm. For 10µmdisplacement the required voltage is 4V. •The third part sets the necessary parameters for the measurements, such as the interval time between two captures, and the total duration of the test, which finally determine how many captures we will get. Figure 3.2: The program which controls the used instruments (oscilloscope, signal generator) and also captures the SM signals via the oscilloscope Cavities The laser is emitting towards to a brushed aluminium target which is designed for 4 different length cavities of water (1, 2, 3 or 4[mm]). The cavity length (L3) is defined from the aluminium target and cut microscope glasses which are placed on the especially designed part as is shown in figure (3.3). In the same part there is a place for the temperature sensor which contacts with the cavity by a small hole. The cavity target can be placed 7.5, 10 or 12.5 [cm] away from the laser. Finally the 10 [cm] was chosen. After the installation of the target in a specific distance, the cavity length is chosen and a suitable glass is placed into the target to define the desirable cavity. Afterwards the cavity is filled with distilled water. Adjustment-Calibrating The basic calibration that SMI technique requires, is the adjustment of the lens. By the adjustment of the lens we can increase the diameter of the 24 3.1. 1ST EXPERIMENTAL SET-UP LaserBased BioSensor Figure 3.3: Aluminium target design for 4 different cavities for liquids (1, 2, 3 or 4[mm]). laser’s beam and that way change the feedback parameter Cof the setup. Finally we try to calibrate the setup with a Cvalue around 1 in order to have a reliable and easy fringes tracking. In that case, the SM signal looks like the captured one in the figure (3.4). Figure 3.4: SM signal in DSMI technique C ∼1(yellow). The blue line represents the displacement of the piezoactuator. 3.1. 1ST EXPERIMENTAL SET-UP 25 LaserBased BioSensor 3.2 Diagnostics Before the measurement part it is important to verify that a steady state we are looking for in order to capture at least two of them, exists. A steady state means that the fringes have to be in a specific position in time as the laser is moving according to the triangular motion of the piezo. The factors that can introduce errors are presented: Sound The first thing we observed was that the sound waves, which were produced by loud voices in the laboratory, closing doors, moving chairs in the floor above the lab and other environmental sound, introduced instability issues to the setup. For that reason we tried to shield the set-up as much as possible. Finally in order to avoid the noise the setup was moved in a laboratory where there are no other people present. Enviromental Light The vertical orientation of the set-up resulted that the target is facing not only the laser’s light but also the Fluorescent lamp lights of the laboratory. A part of the Fluorescent light is back-reflecting into the laser’s cavity. The fluorescent light is not continuous, it works with "beat effect", so it interferes randomly with laser’s cavity field causing random shifting to the fringes. Measurements with lights off, showed a great improvement as the maximum fringe shifting between two measurements reduced from 300 [nm] to 40 [nm] (3.8). "Shifting" is translated into displacement for better understanding: as the laser’s wavelength is λ=787.9[nm] and λ/2 corresponds to phase shift ∆φ=2π, 394[nm] means 2πshift and the other way around. Air conditions According to [27], the factors that may affect the air refractive index and their typical index variations for air are presented in the Table (3.1). To minimize the error because of air refractive index changes we had the air-conditioning of the room on and at the same time we were capturing the temperature with a sensor. The sensor showed a stable temperature at 22.3±0.5°C. ∆ ∆nDaily change ∆T1°C −9.3×10−70.1°C ∆P1hPα2.68 ×10−720hPα ∆%H2O1hPα−3.7×10−81hPα ∆ppmCO2100ppm 1.45 ×10−8100ppm Table 3.1: Relevance order for refractive index and daily variation Laser instability Varations in the LD’s temperature and the supplied current to LD can introduce erros. First of all changes larger than half or one Celsius degree are able to change the emitted wavelength, fig.(3.5), while in case of stable wavelength, temperatures and current variations larger 0.01°C and ±10µArespectively, are suspicious for the phase shifting shown in figures (3.6) and (3.8b) [6]. Due to limitations, this setup tested without a temperature stabilizer for the laser. Measurements in laboratory conditions showed that the wavelength of the HL7851G Laser was locked at 787.6 [nm] with a bandwidth of 1 [nm]. Lens calibration In the current set-up in order to calibrate the C value around to one we are increasing the diameter of the laser’s beam adjusting the lens. This probably affects the coherence of the light beam as a part of the "sees" a larger optical path length. 26 3.2. DIAGNOSTICS LaserBased BioSensor Figure 3.5: Plot of Lasing Wavelength vs. Case Temperature from the specification sheet for the Hitachi HL7851G laser diode Figure 3.6: Phase fluctuations of the fringe maximum during a 600-s-long measurement [6] 0 0.2 0.4 0.6 0.8 1 −40 −20 0 20 40 60 time [s] ∆d [nm] (a) Lights on 0 0.2 0.4 0.6 0.8 1 −10 −5 0 5 10 15 time [s] ∆d [nm] (b) Lights off Figure 3.7: The phase shifting of the fringes is translated to displacement in [nm] (∆d). Each measurement shows the continuous displacement/phase-shifting for 1 second (time of a ramp). (a) In case that the lights are on, the live measuring show an average absolute error of 20 [nm] (b)In case that the lights are offthe live measuring show an average absolute error of 5 [nm] 0 0.2 0.4 0.6 0.8 1 −300 −200 −100 0 100 200 300 ∆fr [nm] time [s] (a) Lights on 0 0.2 0.4 0.6 0.8 1 −60 −40 −20 0 20 ∆fr [nm] time [s] (b) Lights off Figure 3.8: This figure shows the difference in fringe positions of two consecutive measurements translated in displacement [nm]: (a) In case that the lights are on, the steady state capturing show an average absolute error of 300 [nm] (b)In case that the lights are offthe live measuring show an average absolute error of 30 [nm] 3.2. DIAGNOSTICS 27 LaserBased BioSensor 3.3 Preparation In order to have the required difference in water’s temperature and in refractive index as well, a circuit composed of 4 resistances in parallel, 10Ωeach and Rtotal =2.5Ω, was used to heat the water. The circuit was placed below the target of figure (3.3) as it is shown in the figure (3.9). The radiator circuit was connected to a DC power supply and a usb-thermometer was used to capture the water’s temperature. In room conditions the water’s temperature was stabilized at 22.31 ±0.05°C. As the radiator circuit was turned on, we tried manually to balance the flow of the heat energy and the losses of the metal components to the environment in order to stabilize the water’s temperature 0.5−1°C above the initial temperature, by regulating the voltage and the current of the power source. Finally to stabilize the temperature about 1 degree above the room’s temperature the voltage was set at V=3.2V and the current at I=0.8A providing about 2.5W for a few minutes. Figure 3.9: A 4 resistance set in parallel (Rtotal =2.5Ω) placed bellow the target to heat the water which is in the other side of the target Measurements After the water’s temperature is stabilized at a desirable temperature we start capturing the SM signal for the up ramp each 8 seconds. After a few captures, around 15, we turn offthe heating circuit in order to let the water cold and return to its initial state. Meanwhile we continue the capturing until the temperature sensor show the stabilized temperature of the room and for some minutes after. 28 3.3. PREPARATION LaserBased BioSensor 3.4 Processing As it is already mentioned each capture includes the self-mixing signal which is produced when the laser is moving on the up ramp of the piezo. As the piezoactuator is moving with an amplitude of A=10µmthe SM signal of each capture includes 10−6/(λ/2) ∼25 fringes. •At first we pre-process each SM signal in order to track its fringes on the time scale of 0 to 1 second, time of the up ramp. •Then we check that each signal has the necessary fringes and if someone is missing (because of noise) we replace it, following the principle that in steady state two fringes are in distance that correspond to λ/2. •If there was no instability issues we could just get a measurement that represents the "hot" water and another one that represents the "cold" water and get the time difference of the fringes as in the simulation part. However, the instability issues led us to a more static technique. •We represent each measurement with one "median fringe". Actually we get the mean value of the fringes of each measurement. If we have N fringes (N=odd) and "frN−1 2+1" is the position of the median fringe on time and ∆trthe time that corresponds to λ/2 displacement we can write: fr1,··· ,frN−1 2,frN−1 2+1,frN−1 2+2,··· ,frN∼ frN−1 2+1−N−1 2∆tr,··· ,frN−1 2+1−∆tr,frN−1 2+1,frN−1 2+1+ ∆tr,··· ,frN−1 2+1+N−1 2∆tr⇒ frmeas =       mean(fri)∼frN−1 2+1,N odd mean(fri)∼frN 2+Deltatr 2,N even,i=1,2,3···N(3.1) •The next step is to synchronize the values of temperature that were captured from the usb sensor with the measurements that represent the "hot" and the "cold" water. As we track them we choose the most representative value for each state and then we calculate the median value of the fringes that are less than 30[nm] far from the representative value. 30 [nm] is the typical deviation of the fringes as it is shown in the figure (3.8b). •In the figure (3.10), the blue line corresponds to the frmeas (Eq.3.1)of each measurement translated to change in optical path starting measuring from the first measurement. The green line represent the water’s temperature on time. The red circles show the frmeas values that are less than 30[nm] far from the representative value of the "hot" water and the magenta circles respectively for the "cold" water. •Let’s explain the so-called "tranlation of the fringes position to optical path change". The represantative fringe of measurement frmeas is detected originally on time and if tmeas is the position of frmeas on time we can write: frmeas,1=λ 2 tmeas,1 ∆tr frmeas,2=λ 2 tmeas,2 ∆tr ∆l=frmeas,2−frmeas,1=λ 2 tmeas,2−tmeas,1 ∆tr =λ 2 ∆t ∆tr (3.2) where ∆lis the change in optical path between two steady state capture. The last equation is the same with the equation 2.20 of DSMI theory. 3.4. PROCESSING 29 LaserBased BioSensor Figure 4.5: (a) Reflections of a single incident oblique wave from the boundaries of a multilayered medium. (b) In each layer, the forward waves are lumped into a collected forward wave, and the backward waves are lumpdes into a collected wave. Adapted from [7] Figure 4.6 "U(+) 1 U(−) 1#="A B C D#"U(+) 1 U(−) 1#(4.2) The matrix M, whose elements are A,B,C, and D, is called wave-transfer matrix or transmission matrix and it depends on the optical properties of the layered medium between the two planes. Defining the properties of a multilayer structure by two layers, the total transmission matrix can be expressed as: Figure 4.7 M=MN×···×M2×M1(4.3) where the elements 1,2,...,N are numbered from left to right as shown in the figure. A wave transmitted through a planar boundary between media of refractive index n1 and n2at angles θ1and theta2, satisfying Snell’s law (n1sinθ1=n2sinθ2), can be descriped by an Mas: 36 4.2. THEORETICAL RESEARCH LaserBased BioSensor M="A B C D#=1 2α21 ˜ n2"˜ n1+˜ n2˜ n2−˜ n1 ˜ n2−˜ n1˜ n1+˜ n2#(4.4) This expression is applicable for both TE(s) and TM polarized waves with the following definitions: TE :˜ n1=n1cosθ1,˜ n2=n2cosθ2, α12 =α21 =1, TM :˜ n1=n1secθ1,˜ n2=n2secθ2, α12 =cosθ1/cosθ2=1/α21 =1. Thestructureofthe set-up shownin the figure(4.4) has five planes: Air1,Glass1,Water,Glass2,Water2, that means four transitions, and four transmission matrices need to be calculated for each step of the simulation: MA−>G,MG−>W,MW−>G,MG−>A. After the calculation of the total transmission matrix of the structure: Mtotal =MG−>A×MW−>G×MG−>W×MA−>Gwe need to relate it with the reflection and transmission coefficients. For that we can use the transformation of the Transmission Matrix to Scattering which is directly related with the reflection and transmission coefficients: S="t12 r21 r12 t21#=1 D"AD −BC B −C1#(4.5) where the quantities t12 and r12 are the forward (1->2) amplitude transmittance and reflectance respectively, while t21 and r21 are the backward (2->1) amplitude transmittance and reflectance respectively and A,B,C,Dthe elements of the M-matrix. Finally the reflection and transmission coefficients given from: R=|r12|2=|r21|2(4.6a) T=t12 ×t21 (4.6b) 4.2.3 Simulation In order to check the stability of the Brewster’s angle set-up for the TE(s) and the TM(p) polarization the Transmission coefficient (T) is calculated for: •Different values of the distance between the two glasses of double-glass cavity (Fig.4.8a) •Different incidence angle θtaround the value 55o(Fig.4.8e,4.8c,4.8c) •Different values of the medium layer refractive index , around the water value n=1.3287 (Fig.4.8b) The results of the simulation are presented in the figure (4.8) and in the table (4.1). The simulation results show that in the case of the "s" polarization incidence, the optical power of the final refracted beam (Air->Glass->Water->Glass->Air) fluctuates between the 100 and 51% of the incidence beam while in the case of "p" polarization the fluctuations are between 100 and 98.5 %. If we also take into account that the beam returns back to the multilayer structure the fluctuations for "s" polarization are from 100 to 25%, while for "p" from 100 to 97%. These simulations made clear that the "p" polarization case is much more stable end effective as: •stable amount of energy is back-reflacted to the laser cavity, 4.2. THEORETICAL RESEARCH 37 LaserBased BioSensor 0.995 1 1.005 60 65 70 75 80 85 90 95 100 Distance between the two microscope glasses [mm] Transmission [%] For θt = 55o P−Polarization S−Polarization (a) 1.3284 1.3285 1.3286 1.3287 1.3288 1.3289 1.329 50 60 70 80 90 100 Refractive Index of the Medium Layer (Water) Transmission [%] For θt = 55o P−Polarization S−Polarization (b) 54 55 56 57 58 59 60 70 80 90 100 θt varation [o] Transmission [%] P−Polarization (c) 54 55 56 57 58 59 60 70 80 90 100 θt varation [o] Transmission [%] S−Polarization (d) 54.99 54.995 55 55.005 55.01 60 70 80 90 100 θt varation [o] Transmission [%] P−Polarization S−Polarization (e) Figure 4.8: Simulation results for the Transmission coefficient for the variables: (a)Distance between the two microscope glasses of the double-glass cavity, (b)Refractive index of median layer, (c,d,e) incidence angle θt. Variable From-To Range Tp,min[%] Ts,min[%] Glasses Distance [mm] 0.995-1.005 0.001 99 62 nmedianlayer 1.3284-1.3290 0.0006 99.5 55 θt[o] 54.99-55.01 0.02 99 56 θt[o] 54-59 5 98.5 51 Table 4.1: Simulation results for the Transmission coefficient for different variable with the transmission coeficient to fluctuate from 100% to the values is shown in this table. 38 4.2. THEORETICAL RESEARCH LaserBased BioSensor •there are no reflection in-between the layers to interfere with the refracted beam and affect its phase. So in case that the laser beam incident to the double-glass cavity, filled with the water, with incidence angle θt=54 −58o, the double-glass cavity can be considered as transparent for the laser beam. 4.2.4 Glucose Literature Glucose is an optically active substance that is changing the refractive index of a media, and it has also the ability to rotate the plane of a linear polarized light when it passes through it, as the left-circularly polarized and right-circularly polarized components experience a bit different refractive index [29]. However the optical activity of glucose will not bother us for two reasons: •The light beam ,in the proposed set-up, is passing through the glucose-solution twice and with different directions each time, so a potential rotation would be cancelled •According to the literature [30] the specific rotation [α] of glucose is 73o/(dmg/dl) at 514 nm and 41.9o/(dmg/dl) at 656 nm and for the 787.6 nm is expected smaller. Taking into account that the cavity length is 1 mm and that we worked with solutions with a maximum concentration of 1 g/dl , the rotation phenomenon looks insignificant in this case. However, the optical activity of glucose shouldn’t be skipped without enough investigation. In order to measure refractive index difference in different concentration glucose solution we have to know more or less the relation between the two sizes, nand C respectively, as we have maximum limitations in the case of Steady State Capturing (∆nshould corresponds to a change in optical path smaller than λ/2 or λ/4 for sign significance). For these reasons the following expressions were extracted from figures of the documents [30] and [31] which include linear graphs between refractive index and glucose concentration in water: Wavelength [nm] unknown 930 1120 ∆n/∆C0.000119 0.000136 0.000143 Table 4.2: Potential relation between refractive index and concentration of glucose in water solutions change. Moreover to define the limits, the length of the optical path that laser follows inside the cavity is needed. According to the figure (4.4) the laser beam is refracted in the medium layer with an angle around 38,5. Measuring the distance between the two glasses of the cavity with a vernier we defined it as 1.1 mm. So the desired length is given as: L=1.1/cos(38.5o)=1.4 [mm] and according the table (4.2) the limitations in ∆Care in the table (4.3). ∆n/∆Cλ/2λ/4 0.000119 2.36 1.18 0.000136 2.06 1.03 0.000143 1.67 0.93 Table 4.3: Maximum ∆C can be detected with capturing method of DSMI technique in [g/L] 4.2. THEORETICAL RESEARCH 39 LaserBased BioSensor 4.3 Glucose solutions preparation Before the experimental work we should prepare some glucose solutions taking into account the limitations that were set before. First we made the solution with the biggest concentration which is 10g/Lby diluting 2 grams of pure glucose in powder in 200 mL distilled water. Then the rest solution of lower concentration were made from the dilution of the initial solution of 10g/L. Finally we made six different solutions 0.5, 1, 1.5, 9, 9.5, 10 g/L of 50 mL each as it is shown in the figure (4.9). Figure 4.9: Glucose solutions 40 4.3. GLUCOSE SOLUTIONS PREPARATION LaserBased BioSensor 4.4 Experimental Work For the "DSMI -Brewster’s angle" set-up we used exactly the same components and the same instruments as it is described in the section (3.1).The steps to set the new set-up were: Figure 4.10: Brewster’s angle set-up •First of all the double-glass cavity was made with two microscope glasses placed in parallel with 1 mm distance (turned to be 1.1 mm in practical). in the one of the small sides of the double-glass cavity two needles were placed in order to circulate later the different solutions. This short side and the large ones were sealed with silicon while the other short side remained open to the air. •After that, the double-glass cavity was placed in order to satisfy the Brewster’s angle criterion as it is shown in the figure (4.10) with a deviation of ±1oat 55.5o. •Next step was to place the laser in order to have TM (p) incidence. This step finally proved very easy. If we turn the laser around its axes ,we could see the reflection beam "shining" and fading to zero. So when the reflection is zero we are in the position for the TM polarization. •A reflective target was also placed at 10 cm facing the laser. Before that, it was checked that there is no SM signal when the laser is displaced and there is no target in front of it except the cavity structure. •Then the cavity is filled with water and the set-up is calibrating adjusting the lens distance from the LD. This adjustment as it is also referred in the previous chapter, increase or decrease the diameter of the laser’s beam and decrease or increase respectively the back-scattering light, and finally adjusting the feedback parameter C around to 1 (the DSMI signal should look sawtooth). 4.4. EXPERIMENTAL WORK 41 LaserBased BioSensor •Finally we empty the cavity and connect the two needles with one syringe each via some tubes. The two syringes that contain glucose-solutions with slightly different concentrations (in-between the limits of the table (4.3)) can fill and empty the cavity with each solution and meanwhile we can capture the DSMI fringes. The syringes were placed on the metal piece as it is shown in the figure (4.11) in order to keep the solutions temperature stable and unaffected from the heat of the hands. Figure 4.11: Bio-sensing set-up Piezoelectric actuator Settings For this experimental set-up the piezo displacement was set from 0 to 2.5 µmcontrolled by a triangular signal with amplitude of 1V volt, dc-offset 0.5V and frequency of 50Hz . This displacement produces 2.5µm/(λ/2) ∼6 fringes in the up or down ramp. Norm of Capturing •The interval time between the captures was set in 5 sec (12 captures/minute). •The capturing was starting while one of the solutions was inside and we were changing the solution with the syringes each 10-12 captures. •The total measurement includes 30-50 captures and 2 or 3 changes of solution as after the 3rd change the solutions start to mix. •A basic practical issue of this measurement is that when we empty manually the cavity with the syringe, we have to stop exactly before the solution is out of the cavity, otherwise air will be in the needle causing bubbles and acoustic noise in the next change. •The signal processing for the fringes detection and their displacement in time are exactly the same as the in the 1st experiment. 42 4.4. EXPERIMENTAL WORK LaserBased BioSensor 4.5 Results The new set-up based on Brewster’s angle finally proved to have bio-application. Although the results seem to be unstable after processing and observation we could result a relation between the change in refractive index and the change in glucose concentration. Before the analysis of the results we have to review the factors that are responsible for the instability which is observed in the figures (4.12) and (4.13). •Enviromental light. A small amount of light in the lab was necessary as the changes of the solution were manual. Although even a small amount of "beat" light (fluorescent lamps) could shift the SM signal. •Laser instability. Temperature changes in the laser diode must be the most "guilty" factor for the "fringe shifting" observed in the figures (4.12) and (4.13). Comparing these results with the ones of the previous set-up we can say that the shifting is worse in the current set-up. The basic difference between the two experiments, as far as the temperature issues, is that the first experiment tested on the last weeks of March (15°C outside) while this experiment the last week of May (30°C ). In both cases the lab’s temperature was stable because of the air-conditioning but the way our body mainly but also the walls are radiating thermal power is totally different. The presence of person was necessary for the experiment to change the solutions. •Bubbles and noise In some cases while we were changing the solutions, as it is said before, some bubbles can appear and cause acoustic noise (mechanical oscillation) changing that way a few nm the optical path. 0 20 40 60 80 100 120 400 500 600 700 800 900 1000 1100 1200 1300 1400 X: 66 Y: 968.8 Measurements [12/min] Fringe position in [nm] X: 60 Y: 746.9 C=1.5g/L C=0.5g/L ∆l = 220 nm Change between 60 − 65 Figure 4.12: 12 minutes capturing (120 Captures) the fringe position. Around the measurements 60-65 we change the glucose solutions, from 1.5g/L to 0.5g/L , and we observe an 220nm change in optical path which could correspond to the ∆C=1g/L value. The rest of the captures show the "fringe shifting" Analysis Some of the results of this experimental set-up are shown in the figures (4.14). Because of the intense "fringe shifting", the change in the optical path length detected between one or two captures before change the glucose-solution and one or two after. Although is not so official, it seemed to work. The intense "fringe shifting" did not allow a statistic approach with a lot of captures as in the first experiment and that was the main reason we chose to take only 10 captures for each solution. After the detection of the changes in optical path length we translated in change in refractive index with the equation: ∆l= ∆n×L⇒∆n=∆l L(4.7) 4.5. RESULTS 43 LaserBased BioSensor 0 20 40 60 80 100 120 450 500 550 600 650 700 750 X: 50 Y: 528.5 Measurements [12/min] Fringe position in [nm] C=0g/L C=1g/L Figure 4.13: In that measurement we had bubble noise while we were the changing the solutions making impossible to observe an result. Although we can observe the shifting. 0 20 40 60 80 100 120 −200 −100 0 100 200 300 400 500 X: 43 Y: 328.3 X: 57 Y: 71.71 Measurements [12/min] Fringe position in [nm] Change at 52−56 ∆l = 260 nm C=0g/L C=1.5g/L (a) 0 5 10 15 20 25 30 35 550 600 650 700 750 800 850 900 950 X: 23 Y: 696.9 Measure 4 Fringe position in [nm] Measurements [12/min] X: 20 Y: 911.5 X: 13 Y: 907.1 X: 9 Y: 596.6 C=1.5g/L C=0g/L C=0g/L ∆l =290 nm ∆l =210nm (b) 0 5 10 15 20 25 30 35 40 45 150 200 250 300 350 400 Measurements [12/min] Fringe position in [nm] C=0g/L C=0.5g/L C=0.5g/L C=0g/L ∆l =110 nm ∆l =120 nm ∆l =110 nm (c) 0 10 20 30 40 50 60 600 650 700 750 800 850 900 Measurements [12/min] Fringe position in [nm] C=0.5g/L C=0.5g/L ∆l =130 nm C=1g/L C=1g/L ∆l =110 nm ∆l =90 nm (d) 0 10 20 30 40 50 60 850 900 950 1000 1050 1100 1150 1200 1250 Fringe position in [nm] Measurements [12/min] C=1.5g/L C=1g/L C=1.5g/L ∆l =160 nm ∆l =80 nm ∆l =70 nm C=1g/L (e) Figure 4.14: Experimental results, change in optical path length for different concentration glucosesolutions. 44 4.5. RESULTS LaserBased BioSensor 0.5 1 1.5 2 2.5 3 0.4 0.6 0.8 1 1.2 1.4 1.6 ∆C [g/L] ∆n [10−4] Experimental results Figure 4.15: The relation between the detected ∆n and the known ∆C Where ∆lis the change in the optical path length and L the cavity the length that lights "sees" which has already calculated as L∼1.4mm. In the figure (4.15) we can see the relation between the detected ∆nand the known ∆C. The results do not look really accurate or precise, however they show that we are moving to the right direction. Although we can say that we have an resolution close to 0.25g/L in glucose concentration in water solutions. Taking into consideration all the results of this experiment we can extract that the relation between the change in refractive index (∆n) and the change in glucose concentration is: ∆n ∆C=0.0001425 which is really close to values of the table (4.2). Probably with more measurements we could have an even more accurate approach. 4.6 DSMI vs SPR Literature related to Surface-Plasmon-Resonance sensor [31, 32, 33] evidence that the resolution of SPR sensor reaches up to 10−6RIU (Refractive Index Units) which corresponds to 0.01 g/L resolution in glucose concentration. According to the experimental results the proposed DSMI set-up can measure with a 0.25 g/L resolution. Although this resolution can easily be improved. In order to improve the resolution of the current set-up we can use a larger double-glass cavity. For example for a 2 mm cavity the resolution will be 0.125 g/L. However that way we decrease the measurement range of the table (4.3) The limit in measurement range can be surpassed if use a circulation system where the glucose concentration is changing gradually. In any case the final proposed set-up showed one more time that DSMI is capable for high accuracy bio-sensing. 4.6. DSMI VS SPR 45 LaserBased BioSensor [15] M. Wang and G. Lai, “Displacement measurement based on fourier transform method with external laser cavity modulation,” Review of Scientific Instruments, vol. 72, no. 8, pp. 3440–3445, 2001. [16] D. Guo, M. Wang, and S. Tan, “Self-mixing interferometer based on sinusoidal phase modulating technique,” Optics express, vol. 13, no. 5, pp. 1537–1543, 2005. [17] R. Lang and K. Kobayashi, “External optical feedback effects on semiconductor injection laser properties,” Quantum Electronics, IEEE Journal of, vol. 16, no. 3, pp. 347–355, 1980. [18] W. Wang, W. Boyle, K. Grattan, and A. Palmer, “Self-mixing interference in a diode laser: experimental observations and theoretical analysis,” Applied optics, vol. 32, no. 9, pp. 1551–1558, 1993. [19] G. Acket, D. Lenstra, A. J. Den Boef, B. H. Verbeek et al., “The influence of feedback intensity on longitudinal mode properties and optical noise in index-guided semiconductor lasers,” Quantum Electronics, IEEE Journal of, vol. 20, no. 10, pp. 1163–1169, 1984. [20] M. T. Fathi and S. Donati, “Thickness measurement of transparent plates by a selfmixing interferometer,” Optics letters, vol. 35, no. 11, pp. 1844–1846, 2010. [21] M. Fathi and S. Donati, “Simultaneous measurement of thickness and refractive index by a single-channel self-mixing interferometer,” IET optoelectronics, vol. 6, no. 1, pp. 7–12, 2012. [22] L. Xu, S. Zhang, Y. Tan, and L. Sun, “Simultaneous measurement of refractiveindex and thickness for optical materials by laser feedback interferometry,” Review of Scientific Instruments, vol. 85, no. 8, p. 083111, 2014. [23] W. Chen, S. Zhang, and X. Long, “Thickness and refractive-index measurement of birefringent material by laser feedback technique,” Optics letters, vol. 38, no. 6, pp. 998–1000, 2013. [24] M. Koelink, F. De Mul, A. Weijers, J. Greve, R. Graaff, A. Dassel, and J. Aarnoudse, “Fiber-coupled self-mixing diode-laser doppler velocimeter: technical aspects and flow velocity profile disturbances in water and blood flows,” Applied optics, vol. 33, no. 24, pp. 5628–5641, 1994. [25] C. H. Henry, “Theory of the linewidth of semiconductor lasers,” Quantum Electronics, IEEE Journal of, vol. 18, no. 2, pp. 259–264, 1982. [26] A. N. Bashkatov and E. A. Genina, “Water refractive index in dependence on temperature and wavelength: a simple approximation,” in Saratov Fall Meeting 2002: Optical Technologies in Biophysics and Medicine IV. International Society for Optics and Photonics, 2003, pp. 393–395. [27] S. J. A. G. Cosijns, Displacement laser interferometry with sub-nanometer uncertainty. [28] R. Kadlec, E. Kroutilová, and P. Fiala, “Multiple reflection from layered heterogeneous medium,” PIERS Online, vol. 7, no. 7, pp. 633–636, 2011. [29] Y.-L. Yeh, “Real-time measurement of glucose concentration and average refractive index using a laser interferometer,” Optics and Lasers in Engineering, vol. 46, no. 9, pp. 666–670, 2008. 52 BIBLIOGRAPHY LaserBased BioSensor [30] B. D. Cameron and Y. Li, “Polarization-based diffuse reflectance imaging for noninvasive measurement of glucose,” Journal of diabetes science and technology, vol. 1, no. 6, pp. 873–878, 2007. [31] W. Lam, L. Chu, C. Wong, and Y. Zhang, “A surface plasmon resonance system for the measurement of glucose in aqueous solution,” Sensors and Actuators B: Chemical, vol. 105, no. 2, pp. 138–143, 2005. [32] J. Homola, “Surface plasmon resonance sensors for detection of chemical and biological species,” Chemical reviews, vol. 108, no. 2, pp. 462–493, 2008. [33] X. Guo, “Surface plasmon resonance based biosensor technique: a review,” Journal of biophotonics, vol. 5, no. 7, pp. 483–501, 2012. BIBLIOGRAPHY 53