Religious & Philosophical

Optical fiber sensing devices based on organic vapor indicators towards sensor array implementation

Description
Optical fiber sensing devices based on organic vapor indicators towards sensor array implementation
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Sensors and Actuators B 137 (2009) 139–146 Contents lists available at ScienceDirect Sensors and Actuators B: Chemical  journal homepage: www.elsevier.com/locate/snb Optical fiber sensing devices based on organic vapor indicators towardssensor array implementation Cesar Elosua a , ∗ , Candido Bariain a , Ignacio R. Matias a , Francisco J. Arregui a ,Elena Vergara b , Mariano Laguna b a Departamento de Ingeniería Eléctrica y Electrónica, Universidad Pública de Navarra, Campus de Arrosadia s/n, 31006 Pamplona, Spain b Instituto de Ciencia de Materiales de Aragón-CSIC, Universidad de Zaragoza, 50009 Zaragoza, Spain a r t i c l e i n f o  Article history: Received 15 August 2008Received in revised form18 November 2008Accepted 11 December 2008Available online 25 December 2008 Keywords: Optical fiber sensorVolatile organic compounds sensorVapochromic materialElectrostatic Self-Assembly (ESA) method a b s t r a c t A family of vapochromic complexes has been used to develop optical fiber sensor transducers; thesematerialsshowreversiblechangesintheiropticalpropertiesinpresenceofsomesolvents,sotheycanbeusedtodetectvolatileorganiccompounds(VOCs).Thematerials’commonchemicalstructureisdescribedby the formula [Au 2 Ag 2 (C 6 F 5 ) 4 L  2 ] n , where L is a ligand molecule that varies for each material, and deter-mines its sensitivity to specific organic vapors. The optical architecture used as transducer was a dopednanocavity built onto a cleaved ended fiber by Electrostatic Self-Assembly (ESA) method. These sensortransducers were used in a reflexive set-up, and their response was characterized in terms of reflectedoptical power at 850nm and absorbance spectra. Up to 4 different sensors were built following this pro-cedure. Absorbance spectra were measured to confirm the response of all the sensors in a qualitativeway; additionally, it was studied quantitatively exposing them to different solvents, showing a linearrelationshipbetweenreflectedpowerandthevaporconcentrationforeachVOC.Finally,thefoursensorswere exposed to three different beverages, registering in each case different absorbance spectras. Thishighlights the potential use of these sensors in an array system to identify aromatic alcoholic beverages.© 2008 Elsevier B.V. All rights reserved. 1. Introduction Artificial systems for odors recognition have important appli-cations in several fields, such as safety at work, environmentalapplications,andfoodindustry.Theyareveryhelpfulintaskswhereacontinuousand/oriterativeexposuretocertainodorswouldseri-ously damage the human olfactory system. These devices needsensors with overlapping VOCs selectivities, which means, eachsensor should respond in a distinct way to a variety of individualsolvents [1]. Most of the systems proposed so far have been based on electronic sensors [2–4].In recent years, many studies were focused on the develop-ment of an opto-electronic array sensors system [5]; these types of devicesofferinterestingadvantagescomparedtoelectronicones,asforexample,lowweight,remotemeasuringcapability,electromag-netic immunity, and several signal multiplexing; this last featurewould help to implement a compact multi-sensor system with thepossibility of self referring techniques [6]. Furthermore, they do not need to be biased for operating, so no electric signal is neces-sary,whicheliminatesanyexplosionriskinthedetectionofspecificVOCs. ∗ Corresponding author. Tel.: +34 948169328; fax: +34 948169720. E-mail address:  cesar.elosua@unavarra.es (C. Elosua). Some optical fiber VOCs sensors are based on gas spectroscopy:alightbeamatawavelengthmatchinganabsorptionlineofthegasofinterestisguidedthroughthefiberreachingthetarget.Thisinter-action can take place in intrinsic [7], or extrinsic way [8]. Another approach consists of the deposition onto the fiber of a sensitivelayer which reacts with the gas or vapor in a reversible or non-reversible way. In all cases, the transduction produces changes inthe optical properties of this layer that modulate the light guidedthrough the fiber, depending on the concentration of the gas orgases to be detected. This way, different sensors have been devel-oped based on luminescence measurement, optical power changeor absorbance spectra, among other types. A good review of thissensing technology is included in [9].In the work described in this article, a family of vapochromiccomplexeshasbeenusedtoimplementopticalfibersensors.Thesematerials show interesting sensing properties because they sufferreversiblechangesofrefractiveindex,aswellascolor,whentheyareexposedtosomeVOCs.Allthesecomplexesshareacommonchem-icalstructure,buteachonehasacertainligandthatdeterminestheselectivity of its response to different vapors. As they display dif-ferent responses to a wide variety of VOCs, they could be used toimplement an array system with sensor heads with different andoverlapping selectivities [1]; the information obtained by this sen- sor system could be used to identify vapors mixtures by followingdifferentmathematicalalgorithms(PCA,ANN)[10,11].Inthiswork, 0925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.snb.2008.12.037  140  C. Elosua et al. / Sensors and Actuators B 137 (2009) 139–146 thesesensitivematerialshavebeenfixedontocleavedendedfibers,andtoachievethis,theElectrostaticSelf-Assembly(ESA)depositiontechnique has been employed. This procedure allows the build upnanocavitiesontoseveralsurfaces,yieldingthicknessesthatareonthe order of the working wavelength.Four optical sensor heads have been fabricated, each using adifferent sensing material to dope the nanocavity. The changesundergone by these materials have been studied in terms of reflected optical power at a wavelength of 850nm, as well aschanges in the general absorbance spectra. Linear relationshipshave been found relating the concentration of different individualvapors and variations in the reflected power. For all the sensors,changes in absorbance spectra have been registered for 6 differentVOCs, and also in case of exposing them to three different types of wines (wine, vintage wine and vinegar). 2. Materials and ESA technique  2.1. Vapochromic complexes The behavior of variation of changes in absorption and emis-sion maxima characteristic in presence of specific organic solventsis known as solvatochromism [12]: specifically, the case that it occurs with organic vapors it is called vapochromism [13]. Theoptical properties such as refractive index, color or fluorescence,changeinpresenceoforganicsolvents;thesechangesarereversiblein most of cases. In this work, vapochromic complexes with thegeneral chemical formula [Au 2 Ag 2 (C 6 F 5 ) 4 L  2 ] n  have been used toimplement the sensors heads (Fig. 1). This type of material hasbeen successfully used in previous works to develop optical fibersensors [14,15]. Each of the complexes that share this general for- mula has a concrete ligand molecule (L), which provides stabilityto the chemical structure, and also determines the sensivity todifferent VOCs. In solid state, these materials have a polymericstructure, being the monomers bounded by gold–gold unions. Inpresence of organic solvents, this union is broken, thus elimi-nating the polynuclear structure; after this, the vapor moleculescoordinate with the silver atoms, which is induced by the rup-ture of the gold–silver bonds. This new structure has a differentcolor and refractive index (Fig. 1). When the organic vapors areremoved from the surrounding media, the srcinal chemical struc-ture is recovered. More specific information about this process canbe found in [16]. In this work, four vapochromic materials hav- ing ligands of ammonia, 1/2 diphenyl-acetylene, 1,3-Dimethylureaand 1,10-Phenantroline, respectively, have been used to build thesensors.  2.2. ESA technique TheESAmethodisaniterativeprocessthatallowsthedepositionof several materials in different substrates [17], producing opti-callyhomogeneoussensitivenano-films.Inthecaseofusingopticalfiberassubstrate,thistechniquepermitsfabricationofFabry-Perot Fig.1.  Molecularstructureofthevapochromicmaterialanditsreactioninpresenceof methanol vapors. nanocavities, with lengths of less than a micrometer [18], ontocleaved ended pigtails. This method has been shown to be simpleand highly reproducible [17].Asacyclicprocess,theESAmethodinvolvesthefollowingsteps:the first one consists of cleaning the substrate (optical fiber) andinducing an electric charge on its surface; after this, the substrateis alternatively dipped into solutions of cationic and anionic poly-mers, building a multilayer nano-film. The polymers used in thiswork were PAH (poly allylamine hydrochloride) as polycation, andPAA(polyacrylicacid)asthepolyanion.Thechargedensityofthesepolymers, once solved, can be controlled varying several parame-ters as the pH [19]. In this case, pH has been set to 8, following the results obtained in a previous work [20]. In both cases, concentra- tion of polymeric solutions was 10mM. 3. Experimental set-up The sensor heads were addressed using a standard optical mul-timodefiberpigtail(62.5and125  mcoreandcladdingdiameters,respectively)ontothecleavedendonwhichthenano-structurewasconstructed.Aprecisioncleaver,SiemensS46999-M9-A8,wasusedto cut at one end the pigtails. The configuration is shown in Fig. 2:as reflection set-up, the main device is a Y coupler (50:50), whichhasthesamecore/claddingdimensionsofthefiberinordertomin-imizeinsertionlosses.Theopticalsourcewasconnectedtoport1of thecoupler,andthelightguidedtoport2,werethefiberwithsen-sor head was connected; this way, the optical signal interacts withthe sensing deposition, guiding back the modulated reflected lightthrough the coupler to port 3, where the optical receptor is con-nected.Theopticalsourceandreceptordependsonthemagnitudeto be measured: when recording the variations in reflected opticalpower, a LED at 850nm and an optical detector (Rifocs 675 RE) areused;ontheotherhand,toanalyzethecolorchangeofthedeposi-tion,theabsorbancespectraismeasuredusingawhitelightsourceandtwospectrometers.Inthiscase,thereflectedopticalsignalwassplit to two spectrometers, one operating in the range from 400to 900nm and the other from 900 to 1720nm (models S2000 andNIR-512 from Ocean Optics respectively). This way, the responsefor each sensor was studied in the Near Infra Red region where thesecond and third telecommunications windows are placed.Using this configuration is quite simple to measure the wave-length response of the sensors, but is affected by potentialfluctuations from the optical source when the reflected opticalpower is measured to characterize the sensor response. The sim-plestwaytosolvethisproblemwouldbeemployinganotheropticalpathtoactasreferencesignal,althoughotherself-referencingtech-niques could be used as well [6]. These approaches will be used in further works. Fig.2.  Set-upimplementedtoanalyzetheresponseoftheopticalfibersensorhead.Dotedlinesrepresenttheconnectionsusedwhenmeasuringtheabsorbancespectra.  C. Elosua et al. / Sensors and Actuators B 137 (2009) 139–146  141 A cylindrical receptacle was used to expose the sensor headsto the organic vapors. The VOCs were injected in liquid state intothe chamber where they evaporate. To avoid leaks, the chamberwas hermetically closed once the sensor head was placed in. Thevolumeinsidewasaround500cm 3 ,whichallowssmallamountsof thedifferentsolventstobeinjectedthatgetcompletelyevaporated.This chamber has a thermal jacket that maintains the temperatureconstant at 30 ◦ C, by flushing water at this temperature constantlythrough it. Besides, as the temperature is the same in all cases, thetime that each solvent needs to become vapor is the same for eachVOC,minimizingtheeffectthatthisfactorhasonthetimeresponseof the sensor [21]. 4. Sensing material deposition process The vapochromic complexes used in this work, beingorganometallic, are highly apolar, and because of this, they havebeen fixed onto optical fibers by sol–gel [22], dip coating [23] or Langmuir–Blodgett[24].Comparedwiththesetechniques,theESAmethod offer reproducibility, automation, and the possibility of deposition with the influence of the geometrical properties of thesubstrate.Besides,theESAmethodisimplementedatroomtemper-ature [25], which avoids thermal degradation of the vapochromiccomplexes. As the sensing materials are non-water soluble, fixingthem with the ESA method is a challenge which has been studiedinpreviousworks[20,26].Themostpromisingapproachindealingwith this problem is preparing a solution of the vapochromic com-plexinethanol,andthen,addthistoultrapurewater[27].Thispresolution(althoughthesensingmaterialisreallyinsuspension)wasfirstaddedtothepolycationsolutionpriortoitsdepositionontothefiber [20]; the next step was to dope either the poly cationic or thepolyanionicsolution[26].Inbothcases,thepHwassetto8,inorderto avoid degradation of the vapochromic complex due to reactionswiththepolyelectrolytes.Thisway,sensorheadstodetectdifferentVOCs were implemented, by alternatively dipping the optical fiberinto the non-doped and doped polymeric solutions. In all cases thesame vapochromic material was used.In this work, we have employed another alternative thathas been successfully used to deposit non-polar material, suchas TiO 2 , with the ESA method [25]. Te complex to be fixedis pre solved but not added to either of the polymeric solu-tions, dipping the substrate in it as another step of the ESAprocedure (Fig. 3). The final multilayer structure constructedis [PAH + /PAA − ] 5 [PAH + /PAA − /(Ethanol+Vap)/PAA − ] 15 . The initial 5bilayers ensure that the nanocavity begins to grow, and after this, Fig. 3.  Main steps of the ESA method followed to deposit the sensing material. Thesymbolsareidealizedandnotintendedtorepresentexactlytheconformationofthepolyelectrolyte chains. the vapochromic solution was added to the process. The fiber wasimmersedineachpolymericsolutionfor2minandfor3minincaseof the vapochromic one. After each monolayer is formed, the fiberis rinsed in pure water for 1min to remove the excess polymericmolecules that are not bound, avoiding this step after immersingthe fiber in the vapochromic solution. Finally, the sensor head wascured for 24h at ambient temperature (25 ◦ C), removing ethanolresiduals from the sensor surface. As the pH is the same that theoneusedin[28],thethicknessofthefinallayerissimilartotheoneobtained then, around 250nm.Following this alternative method, 4 different vapochromiccomplexeshavebeendepositedontothefiber.Theratesforthepresolution in each case were 1.5mg/ml for ammonia ligand complex(Sensor 1), 1.6mg of 1/2 diphenyl-acetylene ligand complex per 1ethanolml(Sensor2),2.5mg/mlincaseof1,3-Dimethylurealigandcomplex (Sensor 3) and 1mg/ml when using 1,10-Phenantrolineligand complex (Sensor 4). Finally, 2.5ml of each of these solu-tionswereaddedto10mlofpurewater.Differencesbetweentheserates are due to the different ligand molecules, as they also deter-mine the vapochromic complex solubility in ethanol [16]. Thesesolutions show the same color when the complexes are in solidstate, so it is possible to measure the changes in color and refrac-tive index once they are fixed onto the fiber. It is important tokeep in mind that the vapochromic complexes are in suspen-sion, so it has to be shaken about every hour to counteract thesensing material precipitation; besides, the materials also showa degradation over time, reported in a previous study [20], dueto oxygen and sun light; to minimize this effect, solutions haveto be sealed when not in use and kept away from direct sunlight. 5. Results and discussion The devices implemented were studied with the set-updescribedbeforeinFig.1.Firstly,theywereexposedtosomeorganicvaporswithdifferentpolaritiesinordertocheckthecolorchangeof the sensing material deposited onto the fiber. After this, one of thesensors was studied in detail registering its response at a certainwavelength. Besides, the influence of relative humidity was alsostudied at this wavelength. To conclude, the absorbance spectrafrom all the sensors were recorded in presence of vapors of differ-entalcoholicbeveragesinordertoobtainsomedistinctivebehaviortrends. 5.1. Qualitative sensor response The color change induced by the sensing materials was mea-suredbyusingawhitelightsourceandaspectrometerwhosespanis in the visible spectral region. To calculate it, the spectra emittedfromthesensorheadinpresenceofnovaporswasusedasreferenceand it was compared with the one when the devices were exposedto the organic vapors. A dark reference was also used to minimizethe background error from the spectrometer. More details aboutthis procedure can be found in [23]. A color change is denoted byanon-planarabsorbancespectrum,asthismeansthatsomewave-lengthsaremoreattenuatedthanothers.Anotherspectrometerwasalso used in parallel to register the variations in the Near Infra Redregion; this was done to check the possibility of working with thesensors at telecommunication wavelengths, towards developmentof a lower cost sensor systems.The devices were exposed separately to saturated atmospheresof six different VOCs: methanol, ethanol, isopropanol, acetone,acetic acid and dichloromethane. These compounds were chosenbecause they are emitted by foods and beverages, and they arealso used in chemical industry. This was done in order to assessif the different selectivities of the sensing materials as charac-  142  C. Elosua et al. / Sensors and Actuators B 137 (2009) 139–146 terized by the absorbance spectra in terms of their shape andamplitude.In Fig. 4a the spectra from Sensor 1 is shown: the non-planarshape is evident for every VOCs in the visible range, which meansthatacolorchangeisregisteredandconfirmsthatthesensingmate-rial has also been properly deposited as well. Besides, the spectraare different for each VOC: it is a distinct feature that in all casesthe absorbance is negative except in case of acetic acid: this factcould be use to identify it. Finally, in all cases the signal amplitudeat 850nm, where the first telecommunication window is placed,is high enough to measure changes in reflected optical power, asit has been done in previous works. In the NIR spectral region, theshapes are smoother and offer also the possibility of working withwavelengths within this range.In case of Sensor 2, the spectra obtained can be seen in Fig. 4b:the shape is quite similar for all the VOCs, showing an absorptionvalley at 575nm between two peaks located at 475 and 775nmrespectively. The amplitude of the curves decreases steadily fromthis peak until around 1100nm, where it shows a slight positiveslope.Forthissensor,themethanolwouldbetheeasiestcompoundto distinguish as its spectrum has the highest amplitude and itssecond peak is located at 825nm. Anyway, in all cases, an opticallight source at 850nm could be used to interrogate the sensors. Itsresponse was registered in terms of changes in the reflected out-put optical power. In the NIR region the amplitude is lower, butcould be enough to work at 1310 or 1550nm, second and thirdtelecommunication windows respectively.The results obtained for Sensor 3 are plotted in Fig. 4c: in thiscase,thespectraareeasiertoidentifyintermsofshapeandampli-tude. Looking into the VIS region, a curve denotes color change inall cases, showing a similar peak-valley-peak for dichloromethaneand acetone (which are the solvents with the lowest polar behav-ior)thantheonedescribedabove.Thesesolventsproducenegativeabsorption spectra, and the rest of VOCs have a positive one insome spectral ranges. Beyond the VIS region, the volatile com-poundscouldbedetectedintheNIRregionbutnotincaseofethanolisopropanol, for which its spectra amplitude is around zero.Finally, the spectra registered for Sensor 4 can be observed inFig. 4d: in this case, only two VOCs are detectable: acetic acidand dichloromethane (with the highest and lowest polar behav-ior respectively). For the rest of the organic solvents, the spectralcurves are quite similar both in shape and amplitude. However,aceticacidanddichloromethaneareeasytoidentifyasthefirstoneshowsapositiveabsorptionvaluesandthesecondanegativeones.The curves in the VIS region also demonstrate color change in thiscase; in case of the NIR region, only two of these solvents could bedetectedwhenmeasuringvariationsinreflectedpowerwithalightsourceatacertainwavelength(secondorthirdtelecommunicationwindows).ItisclearthatthesensorshavedifferentVOCselectivities,defin-ing this as having device the ability to distinguish between thedifferent solvents. The selectivity depends on the ligand moleculeof each sensing material employed; and is also related with thepolar behavior of the organic vapors. A summary of these results ispresented in Table 1. 5.2. Quantitative sensor response analysis Once all the sensors were characterized by their absorbancespectra, Sensor 1 was studied in detail using a LED at 850nm anda photodetector. This way, the variation in refractive index of thesensor can be studied as the change in reflected optical power canbe estimated by Fresnel’s Law [29], which lowers cost and allowseasier implementation in a real application. Firstly, the sensor wasexposedtosaturatedatmospheresof5differentVOCsindividually:theresponsesregisteredareshowninFig.5.Foreachorganicvapor, Fig. 4.  Absorbance spectra obtained for Sensor 1 (a), Sensor 2 (b), Sensor 3 (c) andSensor 4 (d); in each case, the sensor is exposed to six different VOCs.  C. Elosua et al. / Sensors and Actuators B 137 (2009) 139–146  143  Table 1 Summarize of the selectivities shown by each sensor to the different VOCs.Acetic acid MetOH EtOH IsoPropOH Acetone CH 2 Cl 2 Sensor 1 High Low Medium Medium Medium LowSensor 2 Low Medium Low Low Low LowSensor 3 Medium Low Low Low Medium MediumSensor 4 High Low Low Low Low High  Table 2 MainparametersofthetimeresponseforSensor1.Alldynamicrangevaluesareabsolute;byFresnelLaw,therefractionindexofthedepositionisaround1.68,anditsvariationis expressed in absolute values as well in each case.Dynamic range (dB)   n  Time response (min) Recovery time (min) Max concentration (ppm)Acetic acid 3.14  − 0.05 1.8 0.04 38MetOH 3.01 0.11 3.09 0.18 205EtOH 2.7 0.10 5.01 0.15 137IsoPropOH 4.95 0.20 5.08 0.14 137Acetone 4.75 0.19 8.46 0.03 715 two exposition cycles were done in order to check the precision of the response. No changes due to the degradation of the sensingmaterial were found, although in a future work, this will be deeplystudy. The change in reflected optical power is reversible, whichmatches with the property of the sensing material. It is clear thatin all cases the reflected optical power is increased but in case of aceticacid,agreeingwiththeresultsobtainedfromtheabsorbancespectra of this sensor. The dynamic range is different for each VOC.Timeresponseismainlytheresultofthetimethatthevaporneedsto get vaporized (it is injected in liquid state), so it not a featureof the sensing element, but from the whole measuring system. Intheworstcase,timeresponseisaround12minforacetone,andthefastest one is 1.8min for acetic acid. When the sensor is extractedfromthechamber,thevaporsleavethesensorheadquicklyandtheresponse time is very fast in all cases (always below 1min), so thisisafeatureofthesensingelementtruly.ResultsaresummarizedinTable 2.After these measurements, Sensor 1 was exposed to differentconcentrations of organic vapors individually. The maximum con-centration for each vapor (expressed in ppm) can be estimated bythe ideal gases law, and depends on the vapor pressure of the sol-vent and the temperature. From this value, it can be calculated themaximum amount of liquid volume, for each VOC, that gets com-pletelyevaporatedinsidethechamber(seeTable2,column5).Thisway, different liquid volumes below this value were injected to getdifferentvaporconcentration.TimeresponseofSensor1forsixdif-ferentacetoneconcentrationsisshowninFig.6a.Theconcentrationis expressed as a percentage of the maximum concentration thatcan be reached with this VOC. It is clear that the higher concentra- Fig.5.  TemporalresponseofSensor1,intermsofreflectedopticalpowerat850nm,in presence of five VOCs. tion, the higher change in reflected optical power. This procedurewasrepeatedforethanolandaceticacid,showingasimilarbehavior(Fig.6b).Finally,thevariationinopticalpowerwasrepresentedver-susthevaporconcentration,findingthattherewasalinearrelationship between them. The sensitivity (dB/  % concentration) of eachsensoristhelinearfactorofeachapproximation.Theseregressionswere calculated by Least Squares Method, and can be evaluatedby the  R 2 parameter: the closer this value is to 1, the better is theapproximation (in the worst case, its value is 0.99). These resultscan be seen in Fig. 7 for these three VOCs; inside the figure, linearapproximations and  R 2 values are indicated, and also in Table 3.Thelowerdetectionlimitwascalculatedfollowing[30],andcanbeobserved in this table as well.Inalltheexperimentstemperatureisundercontrol,butnottherelativehumidity.Inordertostudytheeffectoftherelativehumid- Fig. 6.  Time response of Sensor 1 when exposed to different concentrations of ace-tone (a), ethanol (b) and acetic acid (b) vapors; these concentrations are relative tothe maximum one possible.

rhgehgdeh

Feb 17, 2019

25-28

Feb 17, 2019
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks