文献翻译——回顾电子鼻技术

第页1外文文献资料ReviewofElectronic-noseTechnologiesAbstract:Novelmobileelectronic-nose(e-nose)devicesandalgorithmscapableofreal-timedetectionofindustrialandmunicipalpollutants,releasedfrompoint-sources,recentlyhavebeendevelopedbyscientistsworldwidethatareusefulformonitoringspecificenvironmental-pollutantlevelsforenforcementandimplementationofeffectivepollution-abatementprograms.E-nosedevicesareidealinstrumentsformeasuringandmonitoringcarbonandgreenhouse-gasemissionsduetotheirsensitivitytoawidediversityofvolatileorganiccompounds(VOCs).Alargenumberofe-noseinstrumenttypes,basedonawidediversityoftechnologiesandoperationalmechanisms,areavailabletomonitorgaseousandparticulatepollutantsreleasedintotheatmosphere,orliquidanddissolvedorganicpollutantsreleasedintomunicipalandindustrialwaste-watersystems.Somecommonlyusede-nosetechnologiesincludeconductingpolymers(CP),metal-oxidesemiconductor(MOS),quartzcrystalmicrobalance(QCM),andsurfaceacousticwave(SAW)sensors.Potentialpollution-detectionapplicationsofe-nosesrangefromatmosphericpollutant(gas-leak)detectionofcarbonemissionsfrombiofuelproductionplantsandfossil-fuelproductionsourcesintheoilandgasindustrytoVOC-releasesfromnumerousotherindustries.E-nosetechnologiesarepotentiallycapableofmonitoringallphasesofindustrialmanufacturingprocessestominimizeproductionofpollutantsandmaintainefficient,cleanproductionlines.E-nosedevicesarealsousefulindesigningmoreenvironmentally-friendly,cleantechnologiesforenergyproduction,variousindustrialprocessesandproduct-manufacturingsystems.Thispaperisareviewofrecentnovelelectronic-nosesystemsandalgorithms,developedoverthepastdecade,thathavepotentialapplicationsfordetectingspecifictypesofharmfulVOCpollutantsintheenvironmenttomeetcarbon-captureandemission-reductiontargetsofworldwideenvironmentalprotectionagencies.第页2Keywords:Electronicaromadetection;E-nosedevices;Environmentalprotection;Pollution;Multi-sensorarrays;Volatileorganiccompounds1.IntroductionThedetectionandcontroloftoxicchemicalpollutionintheenvironmenthasbecomeamajorchallengetomoderndevelopedanddevelopingcountriesoftheworld.Billionsoftonsoforganicandinorganicchemicalpollutantsarereleasedintotheair,water,andsoilannuallyresultinginwidespreadpotentialhealthhazardstoplants,animals,andhumansworldwide.Monitoringofenvironmentalpollutionisnecessaryandrequiredtoprotectthepublicinordertomitigatethemanypotentialnegativeeffectsonenvironmentalqualityandhumanhealth.PotentialadverseeffectsonecologicalprocessesandfunctionswithintheUnitedStateshavenecessitatedtheestablishmentofstrictregulationsbytheU.S.EnvironmentalProtectionAgency(U.S.EPA)onallowabletolerancelevelsforspecificchemicalcontaminantsbasedonconcentrationofemissionswithinair,water,andsoilresources.Similarregulationshavebeendevelopedbycomparableenvironmentalprotectionagencieswithinmanydevelopedcountriesoftheglobalcommunity.Currentconventionalmonitoringmethodsformostchemicalcontaminantsarecostly,timeintensive,andinvolvelimitedsamplingandanalyticaltechniques.Thus,thereisanincreasingneedforcheap,improvedandreliablemethodsfortherapid,accuratedetectionandquantificationofenvironmentalchemicalpollutantsinordertofacilitatetheeffectivemanagementandmitigationofharmfuleffectsonsocietyandtheenvironment.Electronic-nose(e-nose)technologieshavebeenwidelyexploredwithinthepastdecadetoevaluatetheefficacyofutilizinge-nosedevices,basedondiverseoperationalprinciples,andtomonitorchemicalpollutioninvariousenvironmentalsettingsorapplications.Theimpetustowardevaluationofelectronic-nosedeviceshasresultedfromtherapiddevelopmentofgasmulti-sensortechnologiesthathaveprovidedrelativelycheap,portableelectronic-detectiondeviceswiththeflexiblecapabilitiesofdetectingawiderangeoforganicandinorganicgaseoussubstances,includingchemicalpollutants.Thispaperprovidesareviewofnumerouspotentialapplicationsofe-nosedevices,developedbyscientistsworldwideoverthepastdecade,forthedetectionandquantificationofenvironmentalchemical第页3pollutants.2.Electronic-nosegassensorsforenvironmentalpollutionThepotentialapplicationsofelectronic-nosedevicesintheareaofenvironmental-pollutionmonitoringaremanyandvaried.Someofthemoreimportantpotentialutilitiesofe-nosesindetectingpollutionrangefrommonitoringairquality1,2,theearlyorreal-timeareamonitoringofdiurnalurbanpollution-emissioneventsviasensormonitoringnetworks(outdoorpollution)3,localizationofstationary(point-source)pollutionsources4,5,andmappingofchemicalplumes6,7,todetectionoffiresatchemical-storagefacilities,maintainingchemicalsecurityatharborentrancesorimportationports5,detectionofleaksoftoxicorhazardousmaterialsfrompipelinesorindustrialplants,andearlywarningoftheaccumulationoftoxicfumessuchassolventsorexplosivefumes,carbonmonoxideorcarbondioxidewithinenclosedareasofbuildingsormines(indoorpollution).Thefollowingdiscussionpointsoutsomeofthedifferencesbetweenelectronic-nosedevicesandcurrentconventionalspectrometricmonitoringdevices,someoftheadvantagesanddisadvantagesofe-nosesrelativetoconventionalpollutionmonitors,andasummaryoftheefficaciesofe-noseinstrumentsforvariouspollution-monitoringapplicationsasdeterminedfromrecentscientificresearch.2.1.Electronic-nosesvs.conventionalpollution-detectiondevicesMonitoringofurbanpollutioncurrentlyiscarriedoutprimarilybymeansofspatially-distributednetworksoflimitedandfixedmonitoringtations8,9.Avarietyofchemical-detectiontechnologiesareutilizedinthesemonitoringstationssuchasionmobilityspectrometry(IMS)10,gaschromatography(GC)11,andmassspectrometry(MS)12.Industrialspectrometersprecludethedeploymentofadequatelydensepollution-detectionnetworksduetothelargesizeandhighcostsofthesedeviceswhichrequiresecurefixedstations.Complexairpollutiondiffusioninurbanareasisheavilyaffectedbyatmospheredynamicsthatoftenleadtoinaccurateormisevaluatedmeasurementsofpollutantlevelsasaresultoflimitednumbersofmeasurementnodes8.Electronicnosesarelikelytoplayanincreasingfutureroleinraisingthedensityofpollution-monitoringnetworksduetotheirlowcosts,light第页4weight,mobility,andcapabilityofmeasuringawidediversityofvolatileorganiccompounds(VOCs)andtoxicinorganicchemicals.Alargediversityofelectronic-nosetechnologieshavebeendevelopedincludingacousticsensors,suchasQuartzcrystalmicrobalance(QMB);surfaceandbulkacousticwave(SAW,BAW),Carbonblackcompositedetectors(CBCD),Catalyticbead(CB),Catalyticfield-effect(MOSFET),Calorimetric,Complementarymetaloxidesemiconductor(CMOS),Conductingpolymer(CP),Electrochemical(EC),Electricalporoussiliconsensor(EPSS),Fluorescence,(FL),Infrared(IR),Metaloxidessemiconducting(MOS),andOpticalsensors13,14.Amongallofthesee-nosetypes,MOSsensorsarethemostwidelyusedclassofgassensorsforenvironmentalpollutiondetectionbecausetheyarecapableofdetectingbothorganicandinorganictoxins.MOSe-nosesalsohavegoodreproducibility(precision)andlimitedmanufacturingcosts.2.2.Efficacyofelectronic-nosesforpollutiondetectionElectronic-nosegassensorshaveanumberofadvantagesovercurrentconventionalpollution-monitoringsystemsinthattheyarerelativelycheap,portable,highlyversatile,andeasilytrainedinstrumentsthatcanbeemployedforawidevarietyofpollution-detectionapplications.Low-coste-nosesreducelong-termoperatingcostsallowingincreaseddensityofmonitoringnetworksandreducedproblemswithsecurityandcapitallossesthatareassociatedwiththepotentialdamageofexpensivespectrometerinstruments.Overheadcostsforsecurebuildingsandinstrument-maintenance,relatedtomore-expensiveanalyticalinstruments,alsoarelargelyavoided.BecauseE-nosesarerelativelycheapinstrumentscomparedtospectrometers,theycanbechangedoutrapidlyandreplacedwithfreshunitsmorefrequentlytomaintainthepollution-monitoringnetworkinfulloperation.Extrabackupunitsmaybeheldinequipmentinventorytoreplacefieldunitsthatstopfunctioning,needrepairs,maintenanceorrecalibration.Electronicnosedevicesalsohavesomedisadvantagesthathavehithertolargelylimitedtheiruseforreal-timepollution-detectionmonitoringfunctions.ThetoxicityofvariousorganicandVOCpollutantsisconcentrationdependentandthusenvironmentalmonitoringforthesepollutantsrequiresprecisedetectionintermsof第页5concentrationsmeasuredintheenvironment15.Somee-nosedevicesareverygoodatdetectingandidentifyingthetoxiccompoundspresent(qualitativemeasurements),butarenotsoeffectiveatmeasuringconcentration(quantitativeinformation)oftheindividualcompoundsdetected,particularlyincomplexgasmixtures.Thislimitationmustbeovercomeife-nosesaretobeusedsuccessfullyforquantifyingtoxiccompoundsintheenvironment.Complexgasmixturesalsomaycontainhighconcentrationsofinterferinggasescomparedtothelowconcentrationoftargetcompounds.ThetypicallackofintrinsicselectivityofMOSe-nosesforspecifictypesoftoxicchemicalsisanotherlimitingfactor15.Bycontrast,CPe-nosescanbecustomizedtobemuchmoreselectiveindetectingcertainclassesofVOCssimplybythechoiceofspecifictypesofsensorsutilizedinthesensorarray.ThesensorarrayofCPe-nosescancontainawidearrayofsensortypesthataresensitivetoalargediversityofVOCchemicalclasses.TheflexibilityofCPe-nosesmaybeextendedevenfurtherbythecapabilityofselectingdifferentspecificcombinationsofsensorsthatareinoperationatanyonetime.Thus,thecombinationofsensorsthatareturnedonformonitoringcertaintypesofpollutantsatoneinstance(ormonitoringepisode)canbefreelychangedsothatanothercombinationofsensorsisusedinthesameinstrumenttodetectadifferentsetofpollutantsatanotherinstance.Duetocalibrationproblemsassociatedwithchangingsensorcombinationsbeingused,amorelikelyscenarioistohavedifferentCPe-nosesinthemonitoringnetworktobededicatedandtunedtoadifferentclassofpollutantsinordertocovertheentirespectrumofpollutantclasseslikelytobepresent.Unfortunately,CPe-nosesareonlysensitivetoVOCsandareunabletodetectmostinorganictoxiccompoundsthatmightbepresent.Thus,acombinationofdifferente-nosetypeswithinthepollution-monitoringnetworkmaybenecessarytocoverallcategoriesoforganicandinorganicchemicalpollutantsthatmaybepresent.Otherdifficultiesassociatedwithusingelectronic-nosedevicesforenvironmentalpollutionapplicationsincludetheinabilitytoeffectivelyidentifyindividualchemicalgascomponentspresentinverycomplexgasmixtures.Thesesituationsoftenariseincaseswhereemissionsfrommicrobialfermentationarecommonsuchasinmonitoringmalodorousemissionsfromsanitarytreatmentplants,第页6waste-treatmentfacilities,animal-productionyards(feedlots),aswellasindustrialcompostingandfermentationsystems,andotherindustriesreleasingeffluentswithorganicbyproducts16.Inthesecases,anaddedproblemisthattheconcentrationsatwhichodoroussubstancesbecomeoffensivetoresidentsinurbanareasarefarbelowthelegalthresholdlimitvaluesthatareacceptablebyregulationsincetheseconcentrationsarenotdirectlytoxicorharmfultohumanhealth.Theidentityofcomponentgasespresentincomplexgasmixturesbecomeincreasinglydifficultfore-nosestodetectastheconcentrationsdecrease.Nevertheless,regulatorybodiesrequirereliablemethodstoevaluateodorimpactoriginatingfromindividualindustrialinstallationseventhoughverylowconcentrationsofanalytescomplicatetherecognitionofemissionsoriginatingfromthemonitoredsource17.MOSe-nosesgenerallyoperateathightemperatures(rangingfromabout300Cto550C)duetotherequirementforhigh-temperatureoxidationsinthedetectionprocess13.Attemperaturesbelow100C,thelowvaporpressureofwatermoleculesinhibitsoxidativechemicalreactions18.Associatedwiththishightemperaturerequirementishighpowerconsumptionthatincreaseslong-termoperatingcosts.CPe-nosesoperateatroomtemperatureandthusmayhavelowerlong-termenergycosts,butMOSsensorsgenerallyhavemuchlongeroperationallifecomparedwithCPsensorsthattendtohaveshorteroperationallifebutaregenerallymoresensitivetosensoryoverloadbycertainchemicalsubstancesthatresultininactivationandlossoffunction.MOSe-noseshaveveryhighsensitivity,butlimitedsensingrange,rapidresponseandrecovery,butaresusceptibletosulfur&weakacidpoisoning,sensitivetohighhumidity,andhavelowrecoveryforhighmolecularweightanalytecompounds.CPsensorshaveshortresponsetime,diversesensorcoatings,areinexpensive,resistanttosensorpoisoning,butarealsosensitivetohumidityandtemperaturefluctuations.Othere-nosetypeshavevariablelistsoffeaturesandcharacteristicsofferingadvantagesanddisadvantagesformonitoringenvironmentalpollution-typeapplications.Quartzcrystalmicrobalance(QCM)e-noseshavegoodprecision,diverserangeofsensorcoatings,andhighsensitivity,buttheyhavecomplexcircuitry,poorsignal-to-noiseratio,andaresensitivetohumidityandtemperature第页713.Surfaceacousticwave(SAW)e-noseshavehighsensitivitytovirtuallyallgases,goodresponsetime,anddiversesensorcoatings,aresmallandinexpensive,butaretemperaturesensitiveandcertainanalytecompoundsareaffectedbythepolymeric-filmsensorcoatings.Opticalsensorshaveveryhighsensitivity,arecapableofidentifyingindividualcompoundsinmixtures,andhavemulti-parameterdetectioncapabilities,butaremoreexpensivetooperateandhavelowportabilityduetodelicateopticsandelectricalcomponents.Electrochemicalsensors(EC)operateatambienttemperatureswithlowpowerconsumptionandareverysensitivetodiverseVOCs,butarebulkyinsizeandhavelimitedsensitivitytosimpleorlowmolecularweightgases.2.3.E-nosealgorithmsforcalibrationanddata-analysisE-nosealgorithmsanddata-analysismethodsareplayingseveralimportantkeyrolesinmakingelectronic-nosedevicessuitableandcompetentinstrumentsforreal-lifepollution-monitoringapplications.Theintrinsiclong-termstabilityandselectivityofsolid-statee-nosesensorscanseverelylimittheirreliability9.Whene-noseunitsarecalibratedforsingle-analytequantificationinthelaboratory,theyexhibitpoorperformancewhenoperatinginreal-lifefieldconditionswithcomplexmixtures19,20.In-fieldcalibrationstrategiesareneededtomakesurefielde-noseunitsareproperlycalibratedtothetypesofanalytepollutantmixturesthesensorswillmostlikelycomeincontactwithatanyonepointintime.Sinceairpollutionmixtureschangefrequentlyduetounpredictablerandomemissionsfromvariouspollutionpointsources,theneedforrelativelyfrequentin-fieldcalibrationsbecomeobvious.Suchin-fieldcalibrationsarepossiblebyperiodicallysubjectingfielde-noseunitstosensor-proximalreleasesofpreviouslypreparedandpreciselyknownpollutiongas-mixturestandardsduringcalibration.Thedetectiondatacollectedfromthesecalibrationstandardsmaybeusedincombinationwithspecializedinstrumentperformancealgorithmstodeterminehowwellthee-noseperformedduringcalibrationandwhatadjustmentsareneededininstrumentsettingstooptimizeperformanceofthesensorarray.Inthisway,e-nosealgorithmscanpreventlong-termdegradationininstrumentperformancecausedbytemporalmodificationsinanalytespeciescausingdivergencesbetweenabsoluteandrelativeconcentrations.Algorithm第页8basedrecalibrationproceduresarealsousefulformanagingproblemsduetosensor-stabilityissues,oftencorrectablebyreplacingindividualsensorsinthesensorarray.In-depthanalysisofin-fieldcalibrationresultsusinge-nosealgorithmscansignificantlyhelpinstrument-feasibilityassessments,guidethedevelopmentofmonitoringstrategies(devicepositioninginthefieldrelativetopollution-emissionsources,calibrationmethodology,andlong-termperformanceassessments),aswellasinstrumentdesignandthevalidationprocess9,21.Field-recordeddataisessentialforachievingeffectivefieldcalibrationmethodologieswithsyntheticpollutionmixturestandardsusingsensor-fusionalgorithmsthatareproperlytunedviasupervisedtraining8,19.Pattern-recognitionalgorithmsareessentialforclassifyingandquantifyingchemicalconstituentsofpollutionmixturesbasedonsignal-signatureoutputsgeneratedbythesensorarray.Classificationalgorithmscombinedwithdata-analysistechniquessuchasPrincipleComponentAnalysis(PCA)enabletherecognitionofdifferentsampletypesviaaggregationofsimilaremissionsintoclustersrepresentingcompoundsfromrelatedchemicalclasses.Thisclusteringofdatapointsinthesensoroutputfollowingdataanalysisbyalgorithmsismadepossiblebythevariablesensitivityofindividualsensorsinthearraytodifferentclassesofchemicalcompounds.Differentgaseousmixturesofpollutantselicitvariableresponsesfromthecross-reactivesensorarray.Thus,thecapabilityofdifferentiationofdifferentmixturesofpollutantsinagaseoussampleisdeterminedbyallofthepossiblesensor-intensitycombinations(permutations)thatarepossibleforthesensoroutputswhichisproportionaltothenumberofsensorspresentinthearray,theoverlapincross-reactivitytodifferentclassesofcompounds,andthediversityofchemicalclassesthatthecombinedsensorarrayiscapableofdetecting.Whensensor-fusionalgorithmsarecombinedwithmultivariatecalibrationandotherstatisticalmethods,problemsofconcentrationestimationforcomplexgasmixturescanbemoreeffectivelyresolved,particularlywhenArtificialNeuralNetworks(ANNs)areutilizedwithe-nosetrainingforanalyterecognitionofcomplexmixtures8,13,22-24.3.Pollution-detectionapplicationsofe-nosesTheprimaryapplicationsofelectronic-nosedevicesformonitoringenvironmental第页9pollutionmaybecategorizedintofourmainareasbasedontypesofchemicalpollutantsbeingdetected,including1)inorganicchemicals,suchasheavymetals,andoxidesofcarbon,nitrogen,andsulfuraswellasreducedformsofnitrogenandsulfur;2)volatileorganiccompounds(VOCs);3)point-sourcechemicalpollutants;and4)gaseswithmultipletoxiccompoundscontainingbothorganicandinorganictoxicchemicals.Sugimotoetal.25furthersubdividedVOCanalytesintotwogroups:polarVOCsandnonpolarVOCs,basedonthecharge-polarityoftheseorganiccompounds.Thefollowingsectionssummarizee-noseapplicationsfordetectingandmonitoringenvironmentalpollutantsthatfallintothesefourmaincategories.4.ConclusionsThemonitoringofenvironmentalpollutionisaverypromisingnewfieldforthedevelopmentofnewapplicationswiththeelectronicnose.E-noseshavethepotentialofprovidingmanynewmonitoringcapabilitiestofillnichesthatarenotcurrentlypossiblewithconventionalexpensivespectrometers.Amultitudeofmonitoringtasksmaybegreatlyimprovedbytheuseoftheserelativelycheapcross-reactivegas-sensordevicesthatofferthecapabilityofincreasingthedensityofsensingstationswithinpollution-monitoringnetworks.Thisfunctionalonewillprovidesignificantlyimprovedmonitoringdataallowingmoreeffectivereal-timemanagementofmanycomplexurban-pollutionproblems.Utilizinge-noseassistancetoon-linecontinuousmeasurementsofpollutionemissionsshouldprovidesuperiordata,corroboratedbypollutionmeasurementsfrommultiplee-nosesinthenetwork,whi