Photoacoustic characterization of different food samples

Photothermal ExperimentsonCondensedPhase SamplesofAgricultural Interest: OpticalandThermal Characterization Promotoren: dr.J.Reuss emeritushoogleraarindemolecuul-en laserfysica KatholiekeUniversiteit Nijmegen dr.G.P.A.Bot hoogleraarindetechnischenatuurkunde Landbouwuniversiteit Wageningen Co-promotor: dr.D.Bicanic universitair hoofddocent, departementAgro-,Milieu enSysteemtechnologie Landbouwuniversiteit Wageningen N NoHl&\ Z'^'i"^ Photothermal Experiments onCondensed Phase SamplesofAgricultural Interest: OpticalandThermal Characterization JanPaulFavier proefschrift terverkrijging vandegraadvandoctor opgezagvanderector magnificus vandeLandbouwuniversiteit teWageningen, dr.C.M.Karsen, inhetopenbaarteverdedigen opvrijdag 31oktober1997 desnamiddagstevieruurindeAula. CIP-DataKoninklijke Bibliotheek, DEN HAAG Favier J.P. Photothermal Experiments on Condensed Phase Samples of Agricultural Interest: Optical andThermal Characterization / J.P. Favier. [S.I. : s.n] Thesis Wageningen Agricultural University. -With ref. -With summary in Dutch ISBN 90-5485-768-4 Cover: Nazca lines inPeru, the spider (46 meters), the monkey (90 meters) and the hummingbird (110 meters) WßüOTHr.FK LANDBO''VvlJNi\TRs:T"rT V.AGFN'iNGKIV /vjAiO^lO' , 1 3 3 ? Stellingen 1 De standaard fotopyroelektrische techniek maakt het mogelijk om snelen nauwkeurig dewarmtevereffeningscoëfficient (mV1) tebepalenvan zowel vaste stoffen, vloeistoffen alspasta's. Daderlatera/., Chem. Phys. Lipids 82, 115-123 (1996) dit proefschrift 2 Het concept van het optothermisch venster levert een elegante en eenvoudige methode omvloeistoffen spectroscopisch te analyseren, dit proefschrift 3 Fotothermische technieken zullen incombinatie met moderne scheidingstechnieken de detectielimieten aanscherpen. Tran CD. et al..Anal. Chem.Acta 299, 361-369 (1995) Franko M. et ai, ingediend Chromatography (1997) 4 Indien entropie intuitiefwordt gedefinieerd alsmaat voor de hoeveelheid wanorde, isdit inintegenspraak met het hardebollen systeem, aangezien de entropie van dit model, bijvoldoende hoge dichtheid, inde geordende kristallijne fase groter isdan indewanordelijke vloeistoffase. B.J. Aider and T.E. Wainwright,J. Chem.Phys. 27, 1207 (1957) M. Dijkstra, proefschrift Universiteit Utrecht (1995) R. van Roij, proefschrift Universiteit Utrecht (1996) 5 Door de lage structuurstabiliteit van het Fefragment enderelatief hoge structuurstabiliteit van hetF(ab')2 fragment zal debindingsactiviteit van geadsorbeerd IgGvoor hCGbehouden blijven onder verschillende adsorptie omstandigheden. J. Buijs. proefschrift Landbouwuniversiteit Wageningen (1995) 6 Differential Scanning Calorimetry iseen"cantankerous device"dat onderzoekers meer problemen geeft dan oplost. V.E. Sweat, Chapter 2,Engineering properties offoods, MA. Rao and S.S.H. Rizvi (1994) 7 VolgensFelicia Huppert hebben vrouwen eenbeter geheugen dan mannen. Dat betekent dat bijverdergaande emancipatie het fenomeen "verstrooide professor" zal verdwijnen. Intermediair 20-2-1997 8 In detoekomst zalbodemverontreiniging vooral aangepakt worden door gefundeerd niets doen. 9 Alshet aantal dodelijke slachtoffers van malaria indeWesterse wereld een voelbare fractie zou zijn van dat indetropen, dan zou er harder naar een oplossing worden gezocht. 10 Voetbal zou een stuk aantrekkelijker worden alsmenmet dezelfde frequentie regelwijzigingen zou doorvoeren alsbij het hockey. 11 Erkennen dat menzichheeft vergist, isslechts constateren dat men vandaag meer inzicht heeft dan gister. 12 Kindertelevisie isvolwassen geworden envolwassenentelevisie kinds, dat betekent dat televisie inieder geval met zijntijd meegaat. Stellingen behorend bij het proefschrift "Photothermal Experiments on Condensed Phase Samples ofAgricultural Interest: Optical and Thermal Characterization" van JanPaul Favier, Landbouwuniversiteit Wageningen, 31 oktober 1997 TheArticulatingPhotophone.TheSeleniumReceiver Contents 1 Introduction toPhotothermalScience 2 PhotoacousticSpectroscopyforOpticalCharcterizationofDifferent Samples 2.1 PhotoacousticCharacterization ofDifferent FoodSamples 2.2 NewandVersatilePhotoacousticCellforStudiesofPowderedSpecimens AcrossBroadSpectralRange 2.3 OrganicCompoundsMeasuredwithInfrared (3.39pm)Photoacoustics 3 OptothermalWindow Spectroscopy forOpticalCharcterization ofDifferent Samples 3.1 Detection ofTotalTransFattyAcidsContentinMargarine: anIntercomparison StudyofGLC,GLC+TLC,FTIR, andOptothermalWindow(OpenPhotoacousticCell) 3.2 C0 2 LaserInfrared OptothermalSpectroscopyforQuantitative Adulteration StudiesofExtraVirginOliveOilinBinary Mixtures 3.3 OptothermalDetectionofInfrared Radiation-Induced Absorptionin AqueousSolutionsofCarbohydrates:LactoseandCornStarch 3.4 Compact,OpenandGeneralPurposeCellofVariableEffectivePathlength: DirectAbsorptionMeasurementofS042"inWater 4 ThermalCharacterization ofDifferent Samples 4.1 ThermalDiffusivity ofHardBoiledCandyObtainedby PhotothermalBeamDeflection andStandardPhotopyroelectricMethod 4.2 Photopyroelectric(PPE)MeasurementofThermalDiffusivity in LowDensity Polyethylene(LDPE)andPolyvinylChloride(PVC)Foils Summary 1 9 11 21 29 35 37 49 57 65 69 71 83 97 Samenvatting 101 Dankwoord 105 CurriculumVitae 107 ListofPublications 109 1 IntroductiontoPhotothermal Science History Photothermal (PT) science is a cumulative name for a class of phenomena that involve the generation of heat caused by the absorption of modulated orpulsed radiation. The first report on photoacoustic (PA) effect, the oldest among PT phenomena, dates back to 1880.1 The discoverer Alexander Bell and his coworker Tainter were ahead of their time when using the photophone to transmit the 'speech' by modulated light over a distance of 213 meter. Rayleigh, Röntgen, Mercadier, and Tyndall were among the scientists who were also involved in studying the new phenomena. Due to practical limitations of hearing tubes used as detectors in the early experiments, the PA effect remained a scientific curiosity, but some interesting conclusions could be made. For example, Mercadier who performed PA spectroscopic studies onvarious materials, came to the conclusion that "the maximum effect was found to be produced by the red rays and by the invisible ultra-red rays". Likewise, Röntgen stated that "the sounds inquestion areduetothebending of theplatesunder unequal heating". It was only at the beginning of the 1970s that the PA effect was rediscovered mainly due to the event of lasers (well defined and strong radiation sources) and the development in electronics (microphones, diodes etc.). The introduction of the Rosencwaig-Gersho theory2'3 that has described the PA effect and the concept of the 'thermal wave' led to numerous, new PT detection schemes and various applications. In general one distinguishes two classes of PT methods, i.e. thosefor optical characterization and thermal characterization of a sample. Thephotothermalexperiment Inthe November 4 1880 issue of Nature onereads the following: "A beam of lightfrom the sun orfrom apowerful artificial source, such asan electrical lamp, falls upon a mirror and is reflected through a large lens, which concentrates the rays to a focus. Just at the focus is interposed a disk pierced with holes, forty or so in number, arranged in a circle. This disk can be rotated so that the light is interruptedfrom one to five orsix hundred times asecond. The intermittent beam thusproduced is received by a lens, or apair of lenses upon a common support, whosefunction is to retender the beam once more parallel, or to concentrate it upon the disk of ebonite placed immediately behind, but not quite touching them. From the disk a tube conveys the sound to the ear. Wemay remind our readers here that this apparent direct conversion of light intosound takesplace, asProf. Bell found, in disks of all kinds of substances hard rubber, zinc, antimony, selenium, ivory, 2 Chapter1 parchment, wood,and that he has latelyfound that disks of carbon and of thin glass, which heformerly thought exceptions to thisproperty, do alsobehave thesame way." Figure 1 Photophone used by Bell and Tainter where light was reflected by mirror M on a large lens L which focused the light on a rotating disk with holes R. The lens system T concentrated the beam on the sample (not in figure)1 Modern experimental set-up for PA studies resemblesBell's photophone; the radiation source is eitherapowerful lamp (with a monochromator) or a laser, thepressure wave is detected by a sensitive transducer and the modulated signal processed by a phase sensitive lock-in amplifier. RecentPTtrends inagriculturalandenvironmentalsciences Worldwide developments of new PT detection schemes have also had a substantial impact in the field of agricultural and environmental sciences. " In the Netherlands most significant progress was made in highly sensitive and on-line PA concentration measurements of atmospheric andbiological gases.5'9"11 The concept of intermodulated Stark PA spectroscopy9 proved the unique and elegant approach to suppress the unwanted interferences without the loss in sensitivity. Further development of PT techniques and their applications for optical as well as thermal characterization of condensed phase samples seemed the next logical step. The generally recognized lack of data about thermal properties and in particular their temperature dependence is another important topic worth investigating. To what it concerns the choice of condensed phase samples, foods in general deserve considerable attention. At the moment numerous (analysis) techniques are under development and existing techniques optimized. Analysis and quantitative testing of product composition Introduction toPhotothermal Science represent an important research area in the analysis of (food) compounds. Another active researchareaisthatassociatedwiththequalityandsafety offoods. Target sampleslikepowders,opaqueliquidsandgelsofspecial interest asthey aregenerally difficult to study with available techniques. There are thehigh absorption andthe scattering resulting in low amount of radiation throughput, that constitute major problems in transmission spectroscopybecausetheinformation aboutabsorbanceofthesampleisderived from the measurement of transmitted energy. The essential feature of PTtechniques is that the signal isproportional tothe energy absorbed by the sample and therefore PTtechniques don'thavethisproblem. Photothermal techniques can be regarded as instrumental methods for investigationof materialphysicalproperties.Inthisresearch newphotothermal methodshavebeen proposed and tested as candidate techniques for obtaining thermal and optical properties of agricultural products. Exploitation of qualitative and quantitative photothermal studies was concentrated on (food) samples not easily amenable (e.g. high scattering and strong absorption) to traditional methods. Examples include trace analysis, studies of liquids and solids, non-destructive sampling, adulteration etc.. The new techniques and instrumentation are characterized by their reliability, sensitivity, ease of operation, speed, compactness, $mall quantity of sample needed for analysis anda relatively low cost.Recent progress inthedevelopment ofnewradiation sources,transducers and data acquisitionjustified the studies described in this thesis. It is anticipated that in the near future photothermal methods will be generallyacceptedasavaluableanalyticaltool. ThisPhDthesisdescribestheapplications of several newPTtechniques to condensed phase samples. Spectroscopic studies were made on flours, oils, fatty acids (margarine), aqueous solutions of carbohydrates and water pollutants. Thermal investigations concerned mainly concentrated sugarsystems(candies)andpolymermaterialsusedforfoodpackaging. Two different PA techniques were used here for optical characterization are described in following twochapters.Inchapter2PAspectroscopy wasusedfor investigation ofpowdered and liquid samples. Initially, spectra ofvarious flours (differing in colorand grain size)and spiceswere studied inthevisiblepart ofthe electromagnetic spectrum. Then, anewPA cell for infrared measurements was constructed and a spectrum of Bovine Serum Albumin was recorded inthe 10Pwavelength rangeof the C0 2 laser.The samePAcellwasusedwithan infrared He-Nelaser (wavelength 3.39 urn)tostudyvarious compounds (alcohols, liquidand solidcarboxylicacids)asafunction ofthechainlength. 4 Chapter 1 The concept of optothermal window, an elegant and new approach to obtain the absorption coefficients of "difficult" to study specimens (such as optically opaque and highly viscous samples), is discussed in chapter 3. The feasibility of the new technique was demonstrated andtheresultscompared tothese obtained by otherestablished techniques suchasFTIRand GLC. In this thesis, the use of optothermal window was extended to 5-6 \xm and 9-11 |im spectral regionsboth rich in characteristic absorption bands of different molecules. Initially, thecontent oftransfatty acidsinmargarinewasdetermined. Then, the extent of adulteration (with known adulterants) in virgin olive oil was measured. Since most biological systems contain water, optothermal measurements were also performed on aqueous solutions (i.e. sulfate and carbohydrates). Traditional infrared analysis of such systems isprecluded dueto anintrinsically strongabsorptionofwaterinthesameregion. The standard photopyroelectric method (PPE), a newly proposed technique for thermal characterization of condensed phase samples is described in chapter 4. Its usability was demonstrated by obtaining thermal diffusivity over a wide temperature range (including the region of glasstransitions) for ahardboiled candy and different polymer foils used for food packaging. Finally, the PTBD zero crossing method was used as an alternative and nondestructive technique to determine thermal diffusivity of a hard boiled candy at room temperature. Basic facts about ofPT methods As stated already the PT phenomena rely on the conversion of absorbed excitation energy intoheat. Although initial absorption ofthe energy canbe selective (spectroscopy), thenonradiative relaxation rises the temperature of the sample. The magnitude and nature of the generated PT signal depend upon many parameters, among which sample's absorption coefficient, the excitation power, the efficiency of radiation to heat conversion andthermophysicalpropertiesofthesample.5'12"14 Common to all PT phenomena is that the strength of PT signal is directly related to the energy absorbed by the sample. Changes in temperature, pressure and density may occur simultaneously due to optical absorption are the basis for the PT methods. The choice of a mostsuitablePTtechniqueanddetection schemedependsonthenatureofthesampleandthe purposeofthemeasurement. InPA,theoldestamongthePTtechniques,periodicheatingandcoolingofthesamplecauses aconsequentheatingandcoolingofthesurroundinggaslayer.Thesubsequent expansionand contractionofthegasgivesrisetoanacousticwave,whichisdetectedwithamicrophone. Before considering the PT signal in more detail, it is useful to introduce the concept of thermaldiffusion length \xdefined as(2aaf')"1/2 wherea isthethermal diffusivity (mV) and Introduction to Photothermal Science co (rads"') is the modulation frequency. Due to the dissipative character of diffusion, the amplitude of PT signal decreases with increasing distance from the heat source. At a distance H away from the source, the amplitude of the thermal wave is attenuated to e" of its initial value and therefore heat originating in a depth layer larger than \i will not contribute substantially to the signal. In general, mathematical relations describing the amplitude and phase of the PT signal are knotty and their interpretation is neither straightforward nor simple. However, physical insight may be gained by examining some special cases resulting in simpler equations. These are classified according to optical opaqueness of the samples as determined by the relation of the optical absorption length lß = ß" (ß is the optical absorption coefficient (m 1 )) to the physical thickness of the sample ls.Three cases are distinguished for each category of optical opaqueness depending on the relative magnitude of \ias compared to lsand lp. For example, if fx < ls the sample is said to be thermally thick, while for \i > ls the sample becomes thermally thin. Aanalogy appliesfor optical opacity and transparency. The amplitude R and the phase 9 of a PT signal predicted by the general RosencwaigGerscho theory in a case of thermally thick sample canbe approximated by the expressions R= , A I ° ^ 2 J(ßu) +(ßM-+ 2) (1) 2 and ( <p= tan 1 (2) In Eq. (1) I0 is the intensity of incident radiation and A is the instrumental constant that depends on both sample and the geometry. From Eq. (1) it is obvious that PT response, nonlinear in ßm canbe influenced by changing u.through variation offrequency f (chapter 3). The potential of PT method, for sensitive trace analysis (weak absorption) in gases and liquids follows from Eq. (1)that in such a case (ß(o.< 1)reduces to r. AI0ß|J. R =— ^ V2 indicating a linear relationship between R and ß (and hence also the concentration of the sample). At another extreme i.e. ßn > 1,the PT signal saturates and R becomes independent of the sample's absorption, since R = AI 0 (3) (4) Chapter1 diode laser glass window thermometer sensor CU support Figure2 PPEset-up used inchapter4 An example of a PT method that directly measures the temperature rise in the sample is the photopyroelectric calorimetry (PPE) the use of which is described in chapter 4 (Fig. 2). The periodic temperature rise induced in a sample is detected by a pyroelectric sensor being in intimate thermal contact with the sample. Depending on the configuration used, the sensor canbe irradiated either from a front or a rear side. When pyroelectric materials, exhibiting a permanent internal dipole moment are periodically heated, the induced temperature fluctuations cause a corresponding polarization change, a variation of the surface charge and hence the current change. The average temperature rise <AT>of the sensor is given by 1 p P o <AT>= — [T(x,t)dx L (5) Introduction toPhotothermal Science where T(x,t) is the time dependent one dimensional temperature distribution in the sensor, andLpisthethicknessofthesensor.Duetoitsability toprovidethetemperature dependence of the thermal parameters, the PPE method can also be used to detect phase and glass transitionsaswellasotheranomalousbehaviorofmaterials.' Photothermal beam deflection (PTBD)makesuseof aheatingpumplaser (Gaussian profile) andprobing ofthe changes intherefractive indexn induced inthe sample orthe contacting fluid (Fig. 3). Since the refractive index of a medium is itself temperature dependent, a refractive index gradient is generated. The latter isprobed by a second (probe) laser beam andthedeflection Dismeasuredbyaposition sensitivedetector. Thedeflection oftheprobe beamisrelatedtotheamountofabsorbedpumpbeamradiationandcanbecalculated from 1dn fvT(r,t)xdl D=— — n dt J where T(r,t) istheradialtemperaturedistributionand1 isthedistanceoverwhichpumpand probebeamoverlap.12"14Avariant ofPTBD,the zero crossing methodwasused in chapter4 forthermal characterization. Figure 3 PTBD set-up: 1Ar laser pump beam, 2 mirrors, 3 chopper, 4 diaphragm, 5 lens, 6 movable mirrors, 7 He-Ne laser probe beam, 8 sample, 9 lens, 10 quadrant diode (6) 8 Chapter 1 References 1 Bell'sPhotophone,Nature 23,15-19Nov.4(1880) 2 Rosencwaig A. and Gersho A., Theory of the Photoacoustic Effect in Solids, J.Appl. Phys.47,64-69(1976) 3 Rosencwaig A., Photoacoustics and Photoacoustic Spectroscopy, John Wiley & Sons, NewYork(1980) 4 Mandelis A.,Principles andPerspectives ofPhotothermalandPhotoacoustic Phenomena, Progress in Photothermal and Photoacoustic Science and Technology, Vol. I, Elsevier SciencePublishingCo.,NewYork(1992) 5 Mandelis A., Non-Destructive Evaluation, Progress in Photothermal and Photoacoustic ScienceandTechnology, Vol.II,PTRPrentice-Hall,NewJersey(1994) 6 Mandelis A. and Hess P., Progress in Photothermal and Photoacoustic Science and Technology, Vol. Ill: Life and Earth Sciences, SPIE Optical Engineering Press, WashingtonDC(1997) 7 Bicanic D. (Ed.), Photoacoustic and Photothermal Phenomena III, Springer Series in Optical Sciences69,SpringerVerlagHeidelberg,Berlin(1992) 8 BicanicD., ChirtocM.,AsseltK.van, GerkemaE., Jalink H., SaurenH., Groot T., Torfs T. and Haupt K., New Trends and Perspectives of Photoacoustic and Photothermal Spectroscopies in Agricultural andEnvironmental Sciences,Acta Chemica Slovenica 40, 175-202(1993) 9 Sauren J.J.A.M., Ammonia Monitor Based on Intermodulated C0 2 Laser Photoacoustic Stark Spectroscopy,PhDthesis,WageningenAgriculturalUniversity (1992) 10 Vries H.S.M, de, Local Trace Gas Measurements by Laser Photothermal Detection: PhysicsMeetsPhysiology,PhDthesis,Nijmegen University(1994) 11Bijnen F.G.C, Refined CO-Laser Photoacoustic Trace Gas Detection: Observation of AnaerobicProcessesinInsects,SoilandFruit,PhDthesis,Nijmegen University(1995) 12 Amond D.P. and Patel P.M., Photothermal Science and Techniques, Chapman & Hall, London(1996) 13Bialkowski S.E., Photothermal Spectroscopy Methods for Chemical Analysis, Chemical Analysis:ASeriesofMonographsonAnalyticalChemistryandItsApplications Vol.134, JohnWiley&Sons,NewYork(1996) 14 Sell J.A., Photothermal Investigation of Solids and Fluids, Academic Press, London (1989). 2 Photoacoustic Spectroscopy forOptical Charcterization ofDifferentSamples 2.1 Photoacoustic Characterization of Different Food Samples 2.2 New andVersatile Photoacoustic Cellfor Studies of Powdered Specimens Across Broad Spectral Range 2.3 Organic Compounds Measured with Infrared (3.39 \im) Photoacoustics 2.1 Photoacoustic Characterizationof Different FoodSamples basedon JanPaulFavier,JosBuijs, AndrasMiklós,AndresLörinczandDaneBicanic ZeitschriftfürLebensmitteln-Untersuchungund-Forschung 199,59-63(1994) Abstract Photoacoustic spectroscopy in the 350-700 nmrangeproved a useful tool for discriminating between a variety of opaque, light-scattering samples. Spectral features originating from powdered food specimens of different colour and grain size were observed. These results suggestthefeasibility ofphotoacousticsforquality controlinthefood-processing industry. Photoakustische Charakterisierung von verschiedenen Lebensmitteln Zusammenfassung Die photoakustische Spektroskopie im Wellenlängenbereich von 350-700 nm hat sich als nützliches Instrumentarium zur Unterscheidung verschiedener undurchsichtiger, Licht streuender Proben erwiesen. Spektrale Eigenschaften pulverisierter Nahrungsmittelproben unterschiedlicher Korngrößen undFarbewurden beobachtet. Die Ergebnisse zeigen, daßder Einsatz der Photoakustik in der Qualitätskontrolle für die Nahrungsmittelindustrie vielversprechend ist. 12 Chapter 2.1 Introduction In the past photoacoustic spectroscopy (PAS) was used in studies of inhomogeneous and light-scattering powdered samples. " Since de-excitation processes in these samples proceed along non-radiative channels, thermal excitation spectra of powders correlate well with their absorption spectra. Biological specimens (such as food products) are complex mixtures of either chemically diverse compounds or chemically related molecules, often possessing significant differences in physical properties. As many food products are also powders, PAS might be useful for quality control in the food-processing industry.3'7"9 Establishing more objective selection and characterization criteria for flours is an example of potential PAS application in practice. Likewise, cosmetic qualities of foods have become increasingly important during recent years. As a result colour (just as quality and nutritional factors) has achieved a more preeminent position for the consumer. New techniques and instrumentation for color measurements are therefore considered asuseful additions to the laboratory. It is well-known1' ' ' that the magnitude of the PA signal is directly proportional to the fraction of energy absorbed by a sample. The absorbed energy in its turn is a function of light intensity distribution in the sample, which, when multiple scattering plays an important role, might differ from that predicted by Beer's law. Light scattering, which is also wavelengthdependent, reduces the optical penetration depth of the radiation into the sample.13 In photoacoustics (PA), light scattering affects the PA signal in two different ways. In the first place there is a scattering on the cell walls that leads to correspondingly increased intensity of light inthe cell and gives rise to an acoustic signal.Moreover, light intensity distribution is influenced by internal scattering due to reflections and scattering on the surface of grains. When grain particles are loosely packed and relatively big, the light can reach deeper layers (under the sample surface) at distances that exceed the effective illumination depth (or effective optical penetration depth) ofpowdered samples (i.e.thickness across which the light intensity has decreased to apractically negligible value). It is the effective illumination depth that characterizes the intensity distribution in the cell. Although this parameter is important whenever comparison of PA spectra of powdered samples of different grain size is anticipated, surprisingly no reliable method for its determination has been proposed so far. The main objective in undertaking the study described in thispaper was to obtain PA spectra (350-700run range) of different flour and otherpowdered food specimens and to collect more data onparameters involved intheprocess ofPA signal generation. Zeitschriftfür Lebensmitteln-Untersuchung und-Forschung 199, 59-63 (1994) PhotoacousticCharacterization ofDifferent FoodSamples Materials andmethods 1. chopper 2. lenses 3. mirror 4. microphone 5.housing (steel) 6. teflon block 7. sample holder (quartz fitting) 8.battery 9.powder Figure 1 The set-up used for measuring photoacoustic (PA) spectra of powdered flour samples: ADC, analogue todigital converter; PC,IBMcompatible computer The experimental set-up used to record the spectra of flours (all specimens were kindly provided by Gabona Tröszt, Budapest, Hungary) and samples of commercially available instant coffee and spices is shown in Fig 1. It comprises a 400 W Xenon lamp (ILC Technology) thepolychromaticbeamwasmechanically chopped (1)before it entered JobinYvon H-20 monochromator (spectral resolution of 8 nm). The beam was focused into a quartzcell (seedetail inFig. 1) bymeansoftwoglasslenses (2)andaplanemirror (3).The sample holder (7) isa cup 10mmin diameter and 2mm deep (distance measured from the bottom).Thecellwindow andsamplewindow areseparatedbyadistance of 2mm.Thecell design allows easy loading (quantity of sample used in each measurement was kept as constantaspossible)andenablesremovalofsample(9)withoutaneedtochangetheposition of the cell. This greatly faciliated performance of the measurements since no problems associated with alignment were experienced. Pressure fluctuations were measured with a Knowles EA-1954 microphone (4) powered by a battery (8). The signal was fed into Standford Research SRS preamplifier, the output of which was connected to a Standford Zeitschriftfür Lebensmitteln-Untersuchungund-Forschung 199,59-63(1994) 13 Chapter2.1 14 Research SR510lock-inamplifier. Theoutput signal ofthe lock-in wasinterfaced to aIBMcompatible computer. The scan speed of the step motor was 1nm/s; 40 data points were taken each second. The arithmetic averages of these 40 points were used to construct the spectra displayed inFigs.4-8.Thechoppingfrequency used in all the measurements was70 Hzbecause the signal to noise ratio was found optimal at that value. The thermal diffusion lengths(|x)ofcarbonblackpowder, airandflours atthisfrequency are640, 313and 20(j.m, respectively.Forcarbonblackthethermal diffusion length ismuchlargerthanthedimension ofthegrainswhereasforflours itisofthesameorderofmagnitude. 600 > .1» 400 CO 200 400 500 600 700 wavelength(nm) Figure2 Photoacoustic spectrum of carbon black 400 500 600 700 wavelength(nm) Figure3 Power spectrum of radiation incident ontothe PAcell (used for normalization ofthe PAsignal) In photoacoustics it is customary to normalize the PA spectra in order to eliminate wavelength dependence of the source. In previous studies,2' ' 9'1112 the PA spectra of powders were normalized to the PA spectrum of carbon black taken as a reference sample. Thebasic ideabehind suchanapproach isthe assumption that carbon black absorbs equally stronglythroughouttheentirespectralrange.Howeverourownmeasurementsindicate(Figs. 2 and 3) that the PA signal of carbon black and the power spectrum of the lamp show the same trend for wavelengths shorter than 500 nm. Above this wavelength the PA signal of carbonblack decreases (Fig.2)whereasthepowerofthexenonlamp(showing severalpeaks between 400and500nm)remainsnearly constant (Fig.3).Below 350nmthe measurements are no longer accurate because of a substantially reduced signal to noise ratio due to lens absorption. ThePA spectra (Figs. 5, 7and 8)were normalized by ratioing their magnitudes (recorded at 5 nm increments) directly to the power spectrum of xenon lamp. The reproducibility of the set-up before each measurement wastested consistently by measuring the PA signal from carbon black at two preselected (maxima at 400 and 467 nm) wavelengths. ZeitschriftfürLebensmitteln-Untersuchungund-Forschung 199,59-63(1994) PhotoacousticCharacterization ofDifferent FoodSamples 15 Results Barium sulphate(awhitepowder) wasinvestigatedfirst.Forsuchasamplethe effective gas volume is very close to that used in actual measurements. The spectrum (normalized to a power spectrum of the xenon source) is wavelength independent in the spectral range 350700nm(Fig.4A),incontrasttothatnormalizedtothePAspectrum(Fig.4B)ofcarbonblack (conventional normalization). Since wavelength dependence in Figs. 4 resembles that expectedforthewhitesample,thenormalizationproceduresuggestedhereseemsjustified. ^"800 A •^800 B 600 I400 >600 I §400 rzi p£ 200 ^W^^^ 0 400 500 600 700 wavelength(nm) Figure 4A SpectraofBaS04 powder normalized tothe method proposed here g;200 0 4Ö0 ' 500 ' 6Ó0 ' 7Ö0 wavelength(nm) Figure4B SpectraofBaS04powder normalized ratioed tocarbon black Powder spectra of flours Theflour specimensinvestigated alldiffered incolourandgrain size.Therewereafew white flours such asfineand normal bread, pastry and rice flours. The soya flour, corn grits and cornflour areyellow, whereaswhole wheat flour, wheat flour and ryeflour contained some browncoloured grains.Finally,driedpeaflourhasagreenishcolour. Figure5AshowsPAspectraofwhitebreadflours;all hadabsorption bands around 370,385 and410nm.Above410nmthePAsignal decreasesrapidly anddropstoanearly zeroat700 nm.Thegrain sizeaffects themagnitudeofthesignal;asanexamplenormalbreadflour can be discriminated from fine bread flour (Fig. 5A). The signal from fine bread flour is lower thanthatobtainedwith normalbreadflourbecausethe grain diameter (d)issmallerthanthe thermal diffusion length (d<n);therefore lessheat is deposited into grains. Figure 5Bshows the PA spectra of yellow coloured specimens. Again, absorption bands common to all specimenswereobserved near 370,385and410nm. Ingeneral, theoverall spectral features ofyellowfloursappeared broaderthan forwhiteflours.The effect of different grain sizeon amplitude of thePA signal can also be seen when inspecting spectra of corn grits and corn flour. Ifthegrain size is(much)largerthanthethermaldiffusion length, i.e. d»n(suchasfor Zeitschriftfür Lebensmitteln-Untersuchungund-Forschung 199,59-63(1994) Chapter2.1 16 corn grits), the signal decreases because fewer grains are irradiated. In addition the surface/volume ratiodecreasedwith increasinggrain size.Figure5CshowsthePAspectraof three differently coloured (yellow, green and brown) flours. The dried pea flour is theonly sort that hasa maximum signal at 410 run. Soyaflour exhibits abroader spectrum whereas rye flour resembles that of the bread flour and also produced the highest signal of all the samples. 3000 bread flour finebreadflour cornflour corngrits ^2000 400 500 600 700 wavelength(nm) 400 5Ô0 6Ö1T 7Ö0 wavelength(nm) 6000 - ryeflour _ soyaflour . fried peasflour 14000 £ 2000 400 Figure 5 5Ó0 6Ó0 ' 7Ó0 wavelength(nm) Photoacoustic spectra of flours. (A) Bread flour and fine bread flour. (B) Corn flour and corn grits. (C) Rye flour, soya flour and dried pea flour: a.u., PA signal divided by the power of the xenon lamn (Fia. 3} Inordertoquantitatively comparedifferent flours, themagnitude SofthePAsignalsat385, 410and475nmwerenormalizedtothesignalsobtainedwithnormalbreadflour atthesame wavelengths, yielding thedimensionless ratios (S')shown inthefirstthree columnsof Table 1. Careful inspection of all the spectra suggests that most differences are observed near475 nm. ZeitschriftfürLebensmitteln-Untersuchungund-Forschung 199,59-63(1994) PhotoacousticCharacterization ofDifferent FoodSamples 17 Table 1 Results ofthemeasurements onpowdered flour samples.Themagnitude ofthesignals (S')andtheir relative magnitudes atthree different wavelengths (385,410,475 nm)aregiven. No. flour brand (color) S'î85 410 S'475 (-) (-) (-) 1.00 0.77 0.61 0.58 1 2 3 4 5 6 7 bread flour (w) fine bread flour (w) pastry flour (w) wheat flour (b) rye flour (b) whole wheat flour (b) corn flour (y) 1.00 0.85 0.59 0.61 2.13 1.59 1.54 g 9 10 corn grits (y) soya flour (y) dried peas flour (g) rice flour (w) 1.19 1.22 1.18 0.52 11 a 2.02 1.40 1.39 0.99 1.22 1.29 0.43 a S'410 S'385 S 475 ö 475 S'410 1.00 0.71 0.53 0.59 2.46 1.59 1.90 1.00 1.20 1.11 1.03 0.87 1.00 1.00 1.08 1.15 0.98 1.00 1.10 0.97 1.05 1.05 1.14 1.11 1.30 1.58 1.50 0.92 0.77 0.79 1.18 0.82 0.88 0.73 0.76 0.77 0.86 0.98 0.44 385 0.81 1.20 1.00 0.91 1.21 w,white;b,brown;y,yellow; g,greenish From Table 1 there is a direct relationship between the magnitude of the PA signal and the colouroftheflour: compare,for example, cornpowders (7and 8)towheat (4)andrice(11) flours. Compared to coloured flours, white flours yielded only a small signal. The spectrum shape also provides some information about the colour of samples. For example yellowcoloured flours (corn and soya) can be discriminated from other coloured specimens (columns 4 and 5). The ratio S'3gs/S'475, (column 4 in Table 1) decreases according to the colour of the sample. The reproducibility of the measurements was satisfactory (error was typicallywithin5%). Thelight distribution in powders Both internal reflections within the grains as well as the wavelength dependent reflectance (R)at the surface of flour grains affected14 the magnitude of the PA signal significantly (as showninTable2). Table2 Thewavelength-dependent reflectance (R)for atypical flour specimen14 wavelength (nm) R(-) 400 500 600 700 0.75 0.87 0.94 0.97 Next, therolethatthe effective illumination depth oftheradiation plays inthegenerationof PA signal in various flour specimens was explored in more detail. This was done by Zeitschriftfür Lebensmitteln-Untersuchungund-Forschung 199,59-63(1994) Chapter2.1 18 comparing the PA signals measured with the cell loaded with a specific flour to those obtained from a thin layer («0.5 mm) of the very sameflourspecimen placed on top of a relatively thick backing layer of carbon black. Pure carbon black produced typically a maximumPA signal of 500 (J.V at 467nm (Fig. 2)whereas the signal originating from pure flours atthesamewavelengthvariedfrom 10to20^.V (Fig.6). Inatwolayersample(flourontopofacarbonblack)anincreaseinthePAsignalisexpected for larger effective illumination depths. Even a minor fraction of transmitted light is sufficient toproduce aPA signal with a magnitude that might be comparable to that of the flour. Thecontributionofcarbonblacktothesignal maybeestimatedby comparingthenonnormalized signals from pureflourto signals obtained from thevery sameflouron top ofa thecarbonblack.Twospectraaredepicted inFig.6.Forsoyaflour(Fig.6A),the differences were not very obvious at shorter wavelengths andthe contribution of carbon black (trace a) became significant above 500nm.Forlargergrains, carbonblack strongly prevailed (tracea in Fig. 6B) over that of wheat-meal (trace b in Fig. 6B). The largest PA signal due to the contribution of carbon black (approx. 50 \xV at 467 nm) was about 10%smaller than that obtained when the cell was loaded with pure carbon black. For small grains (Fig. 6.A) the amplitude decrease was substantially lower (about 2%). Scattering also reduces the optical penetration depth and at shorter wavelengths scattering is larger than at longer wavelengths asseeninFig.6. 400 500 600 700 wavelength(nm) Figure 6A Photoacoustic spectra of carbon black covered by athin layerofsoya flour (a)and ofsoyaflour (b) 400 500 600 700 wavelength(nm) Figure 6B Photoacoustic spectra of carbon black covered byathin layer of flour wheat(a)and ofthe flour wheat(b) The above measurements suggest that due to a short effective illumination depth the PA signal is generated in the uppermost layer (0.5 mm) of powdered sample (the effective illumination depth increases for larger grains as concluded by careful inspection of tracesa andb(Fig.6)).Thisimpliesthatwiththepresent cell designradiation doesnotreachthecell Zeitschriftfiir Lebensmitteln-Untersuchung und-Forschung 199, 59-63 (1994) Photoacoustic Characterization of Different Food Samples 19 walls; which has two important consequences. First, the contribution of the cell walls is minimal and secondly, only small quantities of flour are needed to produce satisfactory spectra. Photoacoustic spectra of spices and coffee black pepper hot pepper . pizza spices 20000 CS — Coffee Paloma - -Nescafe (CAP) 330000 -^ .§20000 SS £ 10000 10000 400 500 600 700 wavelength (nm) Figure 7 Photoacoustic spectra different spices of three 400 500 600 700 wavelength (nm) Figure 8 Photoacoustic spectra of two different instant coffees In order to demonstrate the practibility of the PA method, spectra of a few more food products were measured. Figure 7 features spectra of hot pepper, black pepper and pizza spices whereas Fig. 8 shows PA spectra of two instant coffee brands: Paloma, Compack Budapest and Nescafe (CAP) Collombie (100% arabica beans). On visual inspection, the samples appear very much alike, but their PA spectra are quite different. We conclude that the PA technique appears capable of producing reproducible spectra of powdered food samples (discrimination of different flours based on origin, colour and grain size is possible), demonstrating itsusefulness for quality control purposes. It was shown that light intensity distribution near the surface increases due to the scattering and that a large fraction of incident light is reflected and leaves the sample. For this reason no energy canreach thebottom of the sample; in addition only small quantities of sample are required. Theuse of quartz optics will further improve the sensitivity of the PA method. This might eventually also allow the use of PA method for determination of basic amino acids, present inalmost all biological samples. Acknowledgements. Credit is due to dr. Otto Doka for his advices and help during this work. We also thank Lâszlo Détâri for his suggestions in making the computer programme operational. The skilled drawing work of Kees Rijpma and Mees Schimmel from the Illustration Dept. (WAU) isgladly acknowledged. Zeitschriftfür Lebensmitteln-Untersuchung und-Forschung 199, 59-63 (1994) 20 Chapter 2.1 References 1 Zafer A.Y., Jackson W.B. and Amer N.M., Photothermal Spectroscopy of Scattering Media,Appl. Opt. 21,21-31 (1982) 2 Doka O., Biró T., Lörincz A., High-exposure Dosimetry with LiF (TLD-100) by Photoacoustic Spectroscopy,Appl. Phys. D 21, 820-825 (1988) 3 Doka O., Kispeter J. and Lörincz A., Potential Value of Photoacoustic-Spectroscopy for determining Iron Content Determination of Milk Protein-Concentrates, J. of Dairy Research 58,453-460 (1991) 4 Alebic-Juretic A., Güsten H., and Zetzch C , Absorption Spectra of Hexachlorobenzene Absorbed on Si0 2 Powders,FreseniusJ.Anal. Chem. 340, 380-383 (1991) 5 Moreira-Nordemann L.M., Lucht L.A.M., Muniz R.P.A., Photoacoustic Spectroscopy and Surface Temperature Measurements of Tropical Soils,Soil Science 139, 538-546 (1985) 6 Rosencwaig A.; Photoacoustics and Photoacoustic Spectroscopy, Chemical Analysis Vol 57, Wiley, New York (1980) 7 Martel R., N'Soekpoe-Kossi C.N., Paquin P. and Leblanc R.M., Photoacoustic analysis of someMilk Products inUltraviolet and VisibleLight.,J. Dairy Sei. 70, 1822-1827 (1987) 8 Sadler A.J., Horsh J.G., Lawson E.Q.,Hamatz D.,Brandau D.T. and Middaugh CR. Near Infrared Photoacoustic Spectroscopy of Proteins.,Anal. Biochem. 138,44-51 (1984) 9 N'soukpoé-Kossi C.N., Martel R., Leblanc R.M. and Paquin P., Kinetic Study of Maillard Reactions in Milk Powder by Photoacoustic Spectroscopy, J. Agric. Food Chem. 36, 497501 (1988) 10 Clydesdale F.M., Food Analysis, Chapter 3 Colour measurement, Dekker M inc. New York (1984) 11 Helander P. and Lundstrom I. Light Scattering Effects in Photoacoustic Spectroscopy, J. Appl. Phys. 51,3841-3847 (1980) 12 Monchalin J.P., Bertrand L. and Rousset G., Photoacoustic Spectroscopy of Thick Powdered orPorous Samples atLow Frequency,J.Appl. Phys. 56, 190-210 (1984) 13 McGovern S., Royce B.S.H, and Benziger J.B., The Importance of Interstitial Gas Expansion in Infrared Photoacoustic Spectroscopy of Powders, J. Appl. Phys. 57, 17101718 (1985) 14 Dwight E.G. (ed), American Institute of Physics Handbook, 3 e ed., McGraw-Hill Book Company (1982). Zeitschriftfür Lebensmitteln-Untersuchung und-Forschung 199, 59-63 (1994) 2 NewandVersatilePhotoacoustic Cell j •^ forStudiesofPowderedSpecimens Across BroadSpectral Range basedon JanPaulFavier, AndresMiklósandDaneBicanic ActaChimicaSlovenica40, 115-122(1993) Abstract Photoacoustic spectroscopy (PAS) is insensitive to scattering since the generated photoacoustic signal (PA) is proportional only to the fraction of energy absorbed by the sample. It is therefore suitable for the analysis of powdered samples. A new PA-cell for applicationwithinabroad spectralrangewasdevelopedand itsdesignoptimized intermsof both,performance andtheuseraspects. Provzetek Fotoakusticna spektroskopija (PAS) ni obcutljiva na sipanje, kerje nastali fotokustiöni (PA) signal sorazmeren le z delom energije, kijo absorbira vzorec. Zatoje metoda primerna za analizo praSkastih vzorcev. Avtorji so razvili novo PA-celico, uporabno v âirokem spektralnem obmocju, ter jo optimirali tako s staliäca uöinkovitosti kot prakticnosti pri uporabi. 22 Chapter2.2 Introduction Photoacousticspectroscopy isoneoftechniquesfor studyingtheabsorptionfeatures ofgases, liquids and solids in the wide spectral range.1 The method proved uniquely suitable for strongly scattering and opaque samples (e.g. powdered and porous specimens) often intractable for other spectroscopies. The traditional methods for spectral analysis of inhomogeneous and light scattering samples provide satisfactory results only under certain limited conditions.2"4 The majority of spectroscopic data for powders reported so far in the scientific literaturewerecollected intheUV,visibleandthenearinfrared region. Asmany food products arepowders, thePAS could potentially be useful for quality control inthefood industry; one example istheprocessing of milk (to control directly thepresence ofprotein peak in diary products or to determine the concentration of protein and moisture content in skimmed milk5"7). Other PA work onpowdered samples includes applications in studies of high exposure dosimetry8, adsorption9 as well as the structural investigations of biological10andsoilsciencespecimens.11Intheworkdescribedhere,anewandversatilecell isbeingproposedanditsfeasibility demonstrated (inthe 10micronsregion)withprotein asa specimenunderinvestigation. The heart of each PA based instrument is the cell, actually a chamber that accommodates both, the sample and the microphone. The incident periodically chopped radiation of a suitable wavelength is coupled into the cell, where upon interaction with the absorbing sample, heat isgenerated and eventually anacoustical signal produced. In order to increase the sensitivity of the apparatus, it is important to minimize the effect of various noise sources. This implies the suppression of unwanted, non-sample contributions to the PA signal, finding of the optimal position for the microphone and finally the optimalization of thecellsize,shapeanddegreeoftheacoustictightness.Someofthesearediscussedbelowin moredetail. Thenon-samplecontributionstothePAsignal There are several sources of the non-sample PA signals; for example scattered radiation (directly andindirectly)andtheabsorption ofradiationbythecellwindowsall aresensedby themicrophone. Indirectly scatteredradiationreachesthe cellwallby meansofdiffuse light reflections in the sample and produces a PA signal. In order to eliminate this effect it is recommended to use transparent material in the cell construction. Metals having favourably large thermal diffusivities are frequently being used as cell construction materials. Nevertheless,relatively largebackground signalshavebeenmeasured (for examplein caseof stainless steel) in the near infrared; in addition cellsfabricated from metals are not easy to clean.Finally,itisalsoadvantageoustousethewindowsexhibitinghightransmission inthe wavelengthregionofinterest. ActaChimicaSlovenica 40, 115-122(1993) Newand Versatile Pholoacoustic Cellfor StudiesofPowdered Specimens Thepositionofthemicrophone To avoid scattered radiation from reaching the microphone membrane, it is necessary to separate the microphone from the sample volume. This is usually done by connecting the microphone and the cell via a small circular opening. Ensuring a proper degree of acoustic matchingbetweenthemicrophoneandsamplingvolumeoffers anopportunity to improvethe signaltonoiseratio considerably. Thesize,shapeandtheacoustictightnessofthecell Numerous experiments have established that the magnitude of the PA signal is inversely proportional to the cell volume. However, the cell dimensions may not be decreased unlimitedly, asthe cell mustalwaysbe largerthan thethermal diffusion length of the gasin thecell. WheneverdesigningaPAcell,oneshould considertheaspects suchas loadingand replacement of the sampleaswell as, the ease of cleaning and the maintenance. Sealingof the cell shouldbe done in a quick, simple, reliable and reproducible manner. The cell must be acoustically insulated from the ambient in order to reduce the effect of external noise. However, the sealing must not be too tight, sothat an equilibrium between the external and theinternalpressurecanexistavoidingtherebyapressurebuildupatthemicrophone. Experimental C0 2 laser beamsplitter % frequency analyser Ichopper lock-in PA-cell Figure 1 Set-up used for PA measurements of powder samples The experimental setup used for the measurements of powder spectra is shown in Fig. 1.It consists of ahome made cwC02-laser in conjunction with thePA cell (Fig. 2),the chopper and theelectronics neededforprocessing ofthe microphone signal (Fig. 1).TheZnSebeam splitterinsertedintotheradiationpathservedsimultaneouslytoreducethepowerofthelaser beam and to couple a fraction of laser radiation into the spectrum analyzer (Optical Engineering) for identification of laser emission. The radiation was mechanically chopped ActaChimicaSlovenica 40, 115-122(1993) 23 24 Chapter2.2 (PTI 4000 Optical Chopper) and focussed into the cell by means of an off-axis parabolic reflector (6) (Melles Griot 02 POA 015). The output of the microphone was fed into the Princeton AppliedResearchlock-inamplifier (model 128A)for synchronousdetection atthe modulationfrequency. Thecellmountedonarevolvingdrum(4)wasaffixed tothemagnetic foot (1)sothatcellpositioncouldbemaintainedunchangedduringthemeasurements. Figure 2 Exploded view of the PA-cell, see the text for explanation The exploded view of the cell is shown in Figure 2. The revolving drum (4) and detection unit (7-15) are the two major parts of the PA cell. The drum is an aluminum circulair platform that carries three identical and symmetrically distributed sample holders. Each holder consists of a cup (13) that accomodates the sample (the cup is actually a ZnSe flat (25.4mmdiameter, thick 3mm)providedwiththe central spherical cavity (13mmdiameter and2mmdeep))restingintheretainingteflon ring(14)supportedbyametalplate(15).The plateisboltedtotheplatform bymeansofthreescrews. The detection unit includes the metal adapter (7) fitting closely (by two "O"rings) into the microphone assembly (9) that on itsturn isglued to the aluminum housing (8).The adapter ActaChimicaSlovenica40, 115-122(1993) NewandVersatilePhotoacoustic Cellfor StudiesofPowdered Specimens 25 (7) isterminated by aZnSeflat (12) window (12 mm diameter and 2 mmthick). Detection unit as a whole is screwed against the fixed auxiliary plate (15). When put together, the detection unit and one single sample holder form the PA cell. The detection unit actually servesasacoverfortheabovedescribedsampleholderleavingalayerofair300 umthickin between.Theacousticwaveisnegligibly damped across 300 umlength whiletheheatingof themicrophoneisprohibited. The radiation of appropriate wavelength is collected by the Rhodium electroformed paraboloidal reflector (6)withthe clearapperture of 50mm,deflected andfocused (distance from reflector mechanicalaxistothefocal pointis66mm)intothesampleholder(13). The microphone assembly is equipped with the two miniature electret microphones for sensingofpressurefluctuations. Oneof them, the "signal" microphone (11)ismounted ina distant channel(2mmdiameter, height4mm) 13mmfrom thevertical symmetryaxisofthe cell. Another transducer (10), the "reference" microphone, was used in a differential arrangement (for suppression of acoustical and electrical noise). Such a configuration allowedtheautomaticsubtractionofthebackgroundsignal.Themetalhousing(8)shieldsthe microphonesfrom thepossiblesourcesofelectricalnoise. Such cell design allowed an easy (un)loading of the sample and the replacement of the cup (13-14) and of the entrance window (12) whenever needed the cell suitable for studies at other wavelengths aswell. The three sample holders on the drum permit to complete more measurementswithin a given time interval, sincethe cleaning ofthe cup (removed from the supportingplate) cantakeplacesimultaneously withtheinvestigationof thenewspecimen. With the position of the detection unit unchanged all the time, the drum can be manually revolve around the vertical rod. The spring load (2) helps to lock the drum in any of three distinctpositions.Thelockisachievedbythreestainlesssteelballs(3)fixed inthedrumthat fitintosemispherical cavitiesdrilledinthebody(5). Astothemeasurements themselves, the acoustic characteristics ofadifferential PAcellwas determined first using the Briiel&Kjaer dual channel signal analyzer 2032. The major objective of this study was to obtain information about the intrinsic sensitivity and the frequency response of both microphones. This is important since proper functioning of a differential arrangement isensuredonly ifthe sensitivity figures andtheresponce curves for both microphones are comparable. The results indicate the presence of response peaks near 3136 Hz and the second harmonic of this frequency. However, the sensitivities of both microphonesdiffered substantially (theratioofmicrophonesignalmagnitudetothatrecorded by the reference microphone was three). Presumably, this is due to the fact that the microphonesusedwerenotofthesameorigin. ActaChimicaSlovenica40, 115-122(1993) Chapter 2.2 26 1000 /*"*-. £> E 4 -—' o 1003 o O o ^ 10100 1000 10000 frequency (Hz) Figure3 Thefrequency dependence oftheamplitude ofthePAsignal ofBSA As the next step, few preliminary experiments served only to illustrate the versatility of this new PA cell design wereperformed. Figure 3 shows 1/fthe frequency dependence of the PAsignal amplitude obtained from BSA (bovine serum albumin) with the C02-laser tuned to a 9P(10) transition at 9.473 um. The systemperformed well and ina reproducible manner up to thefrequencies ashigh as 1500Hz. With the laserblocked thenoise never exceeded 2 uV. '10.4810.49'10.5110.53'10.55'lO.ÏT'lO.S^lO.ei'lO.óa'lO.eS'lO.e?' 10.7 C02-laser lines in the 10Pbranch (Um) Figure4 Measured PAspectrum ofBSAatdifferent C0 2 laserlines Figure 4 shows the PA spectrum of BSA ratioed to the incident laser power (in the 10P branch of the C02-laser). This latter was determinded by an independent measurement using thepower meter (Photon Control model 35502)placed infront of the PA cell. The true power reaching the cell is, however, not known because the potentially present reflection and absorption losses (paraboloid mirror and of the entrance window), were not considered here. With the laser blocked the noise in this experiment was 1uV (chopping frequency of 233 Hz and tcof 300 ms). Acta Chimica Slovenica 40, 115-122 (1993) New and Versatile Photoacoustic Cell for Studies of Powdered Specimens 27 In conclusion, the usefulness of the new PA cell for spectrocopic studies of powders in the middle infrared was demonstrated. The use of microphone preamplifiers could not only further improve the sensitivity of a system but also assist in matching the sensitivities of both microphones. In addition to the above discussed factors (that influence thepower), an attempt must also be made to determine experimentally the amount of radiation reflected directly from the surface of the sample, so that the actual level of radiation input power used in generating thePA signal could be calculated. An improved PA device could eventually become a low cost, multiwavelength, and sensitive instrument suitable for non-destructive analysis of powdered samples over a wide wavelength range. Work on this matter is currently inprogress. References 1 Rosencwaig A.; Photoacoustics and Photoacoustic Spectroscopy, Chemical Analysis Vol 57, Wiley, New York (1980) 2 Helander P. and Lundström I. Light Scattering Effects in Photoacoustic Spectroscopy, J. Appl. Phys. 51,3841-3847 (1980) 3 Monchalin J.P., Bertrand L. and Rousset G., Photoacoustic Spectroscopy of Thick Powdered orPorous Samples atLow Frequency,J. Appl. Phys. 56, 190-210 (1984) 4 McGovern S., Royce B.S.H, and Benziger J.B., The Importance of Interstitial Gas Expansion in Infrared Photoacoustic Spectroscopy of Powders, J. Appl. Phys. 57, 17101718(1985) 5 Martel R., N'Soekpoe-Kossi C.N., Paquin P. and Leblanc R.M., Photoacoustic Analysis of Some Milk Products in Ultraviolet and Visible Light., J. Dairy Sei. 70, 1822-1827 (1987) 6 N'soukpoé-Kossi C.N., Martel R., Leblanc RM. and Paquin P., Kinetic Study of Maillard Reactions in Milk Powder by Photoacoustic Spectroscopy, J. Agric. Food Chem. 36, 497501 (1988) 7 A.Miklós and D.D.Bicanic, (submitted 1993). 8 Doka O., Biró T., Lörincz A., High-exposure Dosimetry with LiF (TLD-100) by Photoacoustic Spectroscopy,Appl. Phys. D 21, 820-825 (1988) 9 Alebic-Juretic A., Glisten H., and Zetzch C , Absorption Spectra of Hexachlorobenzene Absorbed on Si0 2 Powders,FreseniusJ.Anal. Chem.340, 380-383 (1991) 10 Sadler A.J., Horsh JG., Lawson E.Q., Hamatz D.,Brandau D.T. and Middaugh CR. Near Infrared Photoacoustic Spectroscopy of Proteins.,Anal. Biochem. 138,44-51 (1984) 11 Moreira-Nordemann L.M., Lucht L.A.M., Muniz R.P.A., Photoacoustic Spectroscopy and Surface Temperature Measurements of Tropical Soils,Soil Science 139, 538-546 (1985). Acta ChimicaSlovenica 40, 115-122 (1993) 2.3 OrganicCompounds Measuredwith Infrared(3.39urn)Photoacoustics basedon JanPaulFavier,DaneBicanic,KeesvanAsseltandAndrasMiklós JournaldePhysique IV4,495-497(1994) Abstract A variety of carboxylic acids, alcohols and alkanes were studied using photoacoustic spectroscopy (PAS) with an excitation wavelength of 3.39 ^m. The analytical potential of this method was estimated with different organic compounds dissolved in chloroform. A detectionlimitof0.25%,withasignaltonoiseratioof6,wasachieved. Chapter 2.3 30 Introduction The high infrared absorption coefficients of (semi-)liquid samples such as edible oils, fats and alkanes make their quantitative analysis by means of transmission spectroscopy a difficult task. They are normally examined in very thin layers, either neat or dissolved in an infrared inactive solvent. The thickness of such layers, and therefore the pathlengths investigated, are often difficult to determine accurately. The development of more reliable methods is therefore desirable. Experimental In this work, the photoacoustic (PA) signals from various carboxylic acids, alcohols and alkanes were compared at an excitation wavelength of 3.39 mn. Additionally, the analytical potential of PA at that wavelength was estimated by varying the concentrations of these chemicals in chloroform. The measurements were performed by exciting the C-H stretch bond at 2967 cm"1 with a 2 mW HeNe laser and using a chopping frequency of 82 Hz. The cell has been employed as described in chapter 2.2 ! ; it can be aplied over a wide spectral range on either liquid or powdered samples. Its design facilitates easy and reproducible sample positioning andusesa parabolic mirrortocouplethe incident laser radiation intothePA cell. Results Figures 1and 2 show the dependence of the PA signal on carbon length. Although the signal itself was not linear with the number of C-H bonds (Fig. 1), after it was normalized to the number ofmolecules a stronger linear tendency wasobserved (Fig.2). PA s i g n a l / n u m b e r C-H bonds 140 o (a.u. 120 100 '-' fatty acide 80 O alcohols 60 40 20 0 2 5 8 11 14 17 20 number of C-atoms Figure 1 Photoacoustic signal, normalized to the number of C-H bonds, versus number of C-atoms in the sample molecule Journal dePhysique IV 4, 495-497 (1994) OrganicCompoundsMeasuredwith Infrared (3.39\xm) Photoacoustics Asshown inFig. 1,theratioofPAsignal tonumberofC-Hbonds,decreased with increasing number, n, of C-atoms in the molecules being investigated. For n > 10 the ratio reached a constant value.The signal wasthen normalized tothe number ofmoleculesby multiplying it bythe samples' molecular weight and dividing it by its density and is shown in Fig. 2 (this wasdonebecause the volume of all samples was constant). The normalized PA signal hasa lineardependenceontheincreasingnumberofC-Hbonds. PA signal x Mol w/density (a.u.) D alcohols + liquid carboxylic acids O solid carboxylic acids 5 10 15 20 number of C-atoms Figure 2 Photoacoustic signal normalized to the number of molecules versus the number of C-atoms in the samples Figure 3 indicates the analytical potential of the PA method. The relationship between PA signal and the concentration of organic/chloroform solutions was linear (R=0.997) in the 010%concentration range. PA signal from alcohols and carboxylic acids showed maxima at around 25%concentration. This was due to dimer formation in the solutions, resulting in a shift of the absorption band it unresolvable from the hydrogen bonded 0-H...0 band.2 For octane,thesignal increased upto 100%concentration, sincetherewasno influence from the OHband.Athighconcentrationsthecurveflattened asthePAsignalbecamesaturateddueto thelargeabsorptioncoefficients (>1000cm"1)asexpectedfromtheory.3 Additionally, the thermal properties ofthe solutions change over such a large concentration range.Differences inthe thermal properties ofthe organic compounds and chloroform could cause differences in the PA signals as large as 200-300% (calculated from the thermal conductivity, heatcapacityanddensityofthecompounds3'4). Similar behaviour was found for an optothermal sensor as described in ref. 5; the signal exhibited pronounced linearity for concentrations up to 25% and saturation at higher concentrations.6ThePAmethodwasfound superior (detection limitfor oleicacid0.25%with S/N=6)totheoptothermal method (detection limitfor C18:l 0.35%S/N=3). Theoptothermal JournaldePhysique IV4,495-497(1994) 31 Chapter 2.3 32 method, however, is often more convenient because its measurements are independent of the amount ofsample. PA signal x Mol w/density 0 (a.u.) • caprylic acid • octane • octanol o oleic acid 20 40 60 80 100 concentration inchloroform [% (w/w)] Figure 3 Photoacoustic signal normalized to the number of molecules versus concentration of octane, octanol, and caprylic and oleic acids Conclusions These experiments show the feasibility of a photoacoustic study of organic liquids and solids in the infrared. It forms an important addition to optical spectroscopy, as it allows measurement of absorption coefficients two or three orders of magnitude greater than is accessible by conventional spectrometers. The possible use of different radiation sources would improve the usefulness ofthe technique. References 1 Favier J.P., Miklós A., and BicanicD, New and VersatilePhotoacoustic Cell for Studies of Powdered Specimens AcrossBroad Spectral Range,Acta Chem. Slov. 40, 117-122 (1993) 2 O'Conner R.T., Field E T . and Singleton WS., The Infrared Spectra of Saturated Fatty Acids with Even Number of Carbon Atoms from Caproic, C6, (Hexanoic), to Stearic, C18 (Octadecanoic), and their Methyl and Ethyl Esters, J. Amer. Oil Chem. Society 28, 154160(1951) Journal dePhysique IV 4,495-497 (1994) Organic Compounds Measured with Infrared (3.39 ^.m)Photoacoustics 33 Rosencwaig A., Photoacoustics and Photoacoustic Spectroscopy, John Wiley & Sons, inc. (1980) Parry J.H, Chemical Engineers' Handbook 4th edition, McGraw-Hill book company New York (1963) Helander P., AReliable Optothermal Sensor, Photoacoustic and Photothermal Phenomena III, Springer Series in Optical Sciences 69 ed.D. Bicanic, 562-564 (1992) Bicanic D., Chirtoc M., Chirtoc I., Veldhuizen B., Favier J.P. and Helander P., New Technique for Measuring Absorption Coefficients of Strongly Absorbing Liquids: Optothermal Study of Sunflower Oil, Oleic Acids and Its Chloroform Solutions at 3.39 Microns, Spec.Lett. 28, 101-110 (1995). Journal dePhysique IV 4,495-497 (1994) 3 OptothermalWindowSpectroscopy for OpticalCharcterization of Different Samples 3.1 Detection of Total Trans FattyAcids Content in Margarine: an Intercomparison Study of GLC,GLC+TLC, FTIR, and Optothermal Window (Open Photoacoustic Cell) 3.2 C0 2 Laser Infrared Optothermal Spectroscopy for Quantitative Adulteration Studies of ExtraVirgin Olive Oil in Binary Mixtures 3.3 Optothermal Detection of Infrared Radiation-Induced Absorption inAqueous Solutions ofCarbohydrates: Lactose and Corn Starch 3.4 Compact,OpenandGeneral PurposeCellofVariable Effective Pathlength:DirectAbsorption Measurement ofS042"inWater 3.1 DetectionofTotalTrans FattyAcids Content inMargarine:an Intercomparison Study ofGLC,GLC+ TLC, FTIR,andOptothermalWindow (OpenPhotoacoustic Cell) basedon JanPaulFavier,DaneBicanic,PetervandeBovenkamp,MihaiChirtocandPerHelander AnalyticalChemistry68,729-733(1996) Abstract Four techniques, i.e., gas-liquid chromatography, gas-liquid chromatography + thin-layer chromatography, and two spectroscopic methods, Fourier transform infrared spectroscopy and optothermal window, a variant of the open photoacoustic cell, were intercompared to determinetheirpotential todetect thetotaltransfatty acid contentinmargarine. Atthesame time, this study represents a first application of the optothermal window technique at long wavelengths (10 \xm).The total trans fatty acid data obtained by different methods show good mutual agreement. Besides offering several attractive advantages above conventional methods, the optothermal window also proved suitable for measuring total trans fatty acid contentaslowas2%. 38 Chapter3.1 Introduction Oilsandfats areprimary sources of lipidsthatprovide a major portion of the energy supply in the human diet. The unsaturated constituents of most natural vegetable and marine oils contain only nonconjugated or isolated double bonds in cis configuration; a sizable fraction of these are converted to trans isomers during theprocess of hydrogénation needed to give fatsandmargarineabetterconsistency andstability.1 Sincetheimportanceofmonounsaturatedfatty acidsinreducingsaturated fat intake,thereby lowering the serum level of the atherogenic low-density lipoprotein cholesterol, has been emphasizedbyvarious studies,2the content oftotaltransfatty acids in edible oil isregarded as an important issue. It was shown that the effect of trans fatty acids (TFA) on the serum lipoprotein profile is at least as unfavorable as that of the cholesterol-raising saturated fatty acids. TheTFAnot only raise the low-density lipoprotein cholesterol levels, but also lower the high-density lipoprotein cholesterol levels.2 The recommended reduction of saturated fatty acidsmightleadtoincreased consumptionofTFA.SincetheTFAarethebest substitute for saturated fatty acids in aproduction process of semisolid and solid fats in the edible oil industry,2 the demand for the availability of a reliable and rapid on-line method capable of determining TFAisself-evident. Atpresent, most commonly used methodstodetect TFAin margarine utilize chromatography and infrared spectroscopy, each having its own specific prosandcons. In their infrared spectra, oils and fats feature several characteristic, absorbing bands with large absorption coefficients ß (m"); the one centered at 966 cm"1 is recommended and widelyusedfor measurementsofTFA. Thisinturnrequirestheuseofacellwith short(<10 |j.m)and difficult to reproduce path lengths, thereby often complicating quantitative studies of such samplesby means of traditional transmission spectroscopy. However, this problem canbecircumventedbyusingthenovel concept of optothermal window (OW) spectroscopy. This paper reports on the first application of the OW technique for measurement of TFA contentinsamplesofmargarineandcomparesitsperformance tothatofexistingmethods. Theory With the exception of the OW method, all other methods used here are well-established laboratory techniques;for this reason, only OW isdiscussed in moredetail. The operational principle of the OW technique, actually a variant of conventional photoacoustic spectroscopy, isasfollows: a modulated (laser) radiation passes through the OW cell before impinging on the sample. The OW cell is actually an optically transparent disk (window), having a large thermal expansion coefficient. Its rear side is provided with an annular piezoelectric transducer. Due to the absorption of radiation, the sample's temperaturerises AnalyticalChemistry68,729-733(1996) DetectionofTotalTransFattyAcidsContentinMargarine:anIntercomparisonStudy 39 and the generated heat diffuses into the disk (being in a good thermal contact with the sample), which expands.Theinduced stress isthen detected at the modulation frequency by thepiezoelectrictransducerinconjunction withthelock-in amplifier. When compared to conventional photoacoustic spectroscopy, the OW method offers some attractive features. Atfirst,accommodating the sample in the sealed cell is no longer an impetus. The OW signal remains unaffected by thermal expansion of the sample and in additionisalsolesssusceptibletothe effect ofothersample'sthermalparameters.Finally,as long as it exceeds the sample's thermal diffusion length u (m), the thickness d (m) of the sample is not relevant, making the OW technique more practical for quantitative infrared (1700cm"1)analysisofstronglyabsorbingfluids andsemifluids.8'9 At a given modulation frequency f (s1), the magnitude of the OW signal was shown5 tobe proportional totheamplitudeofthetemperatureoscillation originatingonethermal diffusion lengthbelowthesample-window interface. Ingeneral,thermaldiffusion length \iisrelatedto themodulationfrequency fvia (1) frcpc where K(Wm'K"), p(kgm")andc(Jkg"K" )arethermal conductivity, density, and specific heat,respectively. For an optically opaque (ß"1 <d) and thermally thick ()i<d) sample, making good thermal contact with athermally thick (m^do«,<dwindow)window, the amplitude of the normalized, dimensionlessoptothermal signal Sisrelated5totheproductofß|j.: +i V(ß^ ) (2) 2 + l 1+. •'window wheree(Ws"1/2K'm"2)isthethermal effusivity, defined ingeneralase=^/icpc.TheplotofS from Eq. 2 versus ßn shown in Fig. 1 was computed for distilled water used as a test specimenandthewindow madeofzincselenide. Asconcludedfrom Eq.2,thesensitivity dS/d(ß|j.)increasesfor decreasing ß\i. Consequently, when studying strongly absorbing (large ß) samples by the OW method, short n (high modulation frequencies) arerequired. Onehastobear in mind, however, that the magnitude ofthephotothermal signaldecreaseswithincreasingmodulation frequency. Analytical Chemistry68,729-733(1996) Chapter.1.1 40 Figure 1 PlotofScalculated from eq2 of ßu. for distilled water(e= 1577Ws'^K^m"2) used as asample and theZnSeoptothermalwindow (ew^j^ =5800Ws"1/2K"'m"2,respectively. Expressing ßu from Eq. (2) asafunction of Sone obtains (3) -2 2 xS"* x 1+ ^ l-i c window ' This equation constitutes the basis for obtaining the absorption spectrum of the sample under investigation, provided optical and thermal properties of the reference sample at a given wavelength and modulation frequency areknown. To do so, SinEq. 3isreplaced by S= s rrf v "rrf In Eq. 4 Sref is the calculated (Eq. 2) OW signal for a reference substance having known values for ß at selected wavelength and n at a given modulation frequency. The Vref is the measured signal at the same wavelength and modulation frequency as Srefand normalized to the incident laser power. Likewise, V is the measured signal (normalized to laser power) obtained from an arbitrary sample under identical conditions. Upon substitution of Eq. (4) in Eq. (3) one can compute ß if n is known. By repeating this procedure at other laser wavelengths, the absorption spectrum of the sampleunder investigation canbe constructed. Analytical Chemistry 68,729-733 (1996) (4) Detection ofTotalTransFatty AcidsContentinMargarine:anIntercomparison Study 41 ExperimentalSection Samples The samples of margarine used for the intercomparison study all contained relatively high TFA (40-60%) content (mainly elaidate). They were originally prepared for the purpose of another independent experiment aimed at studying theinfluence ofhighTFA concentrations on low-density lipoprotein and high-density lipoprotein cholesterol levels in healthy subjects.10 In order to reduce the effect of spectral interferences, the triglycerides in the margarine were first converted into the methyl esters of their component fatty acids (FAME).11 The calibration curves were prepared using methyl elaidate (C18:l trans) and methyl oleate (both Nu-Chek-Prep Inc.,Elysian,MN),andresults expressed asapercentage of methyl elaidate in methyl oleate. In the actual experiment, 200 mg of the oil fraction (preparedfrom margarinesamplescontainingbothtransandcisbonds)wasdissolved in2ml ofhexane.Inthefinal phaseofourstudy,twodifferent margarines(onewithalow(3%)and another with a high (40%) TFA content) were investigated using only the OW method. All spectroscopic studies were carried out according to the instructions specified in AOAC method964.34.3 Gas-Liquid Chromatography The gas-liquid chromatography (GLC) analysis of margarine was performed on a HewlettPackard 58902plus,withaflame ionization detectorandthe split injector (1|xLof7mgof FAME/mL ofpetroleum ether; splitflow 200 ml/min). The separation wasachieved usinga SIL88column(100mx0.25 mmChrompack No.007488).12Thetemperatureofthe injector was consistently maintained at 275°C;that of the detector was 250°C. The inlet pressure of the hydrogen carrier gas was 167kPa. The initial temperature (150°C) was raised to 155°C using a fixed heating rate (2°C/min) and was maintained constant for 60 min. Then, a 40°C/minheatingratewasappliedtoreachandkeepthetemperatureat200°C.Thetimeofa singlerunwas89min. Ag-Thin-Layer Chromatography Thin-layer chromatography (TLC)wascarriedoutbydepositing200[iL ofsolution (7mgof FAME/mLofpetroleumether)ona 10%silvernitrate-coated silicaplate(20x20cm,500um Alltech Silicagel GFNo. 729012) and eluting withpetroleum ether/diethyl ether 19:1(v/v). After drying, theplates were sprayed with Rhodamine 6G (25 mg/100 mL of ethanol) and examined under UVlight. Threebands associated with cisfatty acids,transfatty acids,and saturatedfatty acidswereobserved;thesewereremovedandseparated, andtheFAMEeluted withdiethyletherbefore GLCexamination.Thetemperatureprogramwasidenticaltothatfor the GLC detection mentioned above. Recovery values of 90% from the silica plate were found;thecompletedescription isgivenelsewhere.13 Analytical Chemistry68,729-733(1996) Chapter 3.1 42 Fourier transform infrared spectroscopy Fourier transform infrared spectroscopy (FTIR) is apowerful technique that has gained much in popularity during the last ten years. ' The Biorad BST-7 FTIR spectrophotometer (resolution 4 cm"1) with 12 urn thick NaCl cell was used here for spectral studies of margarine samples in a 880-1040 cm"1 range. All samples exhibit an absorption peak with a maximum at 966 cm"1. The regression coefficient of the calibration graph (prepared with methylelaidate) was 0.9998. Optothermal window cell The experimental arrangement for OW studies (Fig. 2) comprises a homemade c.w. C0 2 waveguide laser, the radiation of which was modulated (typically 223 Hz) by a chopper (EG&G Model 179). Moveable mirrors (MM) were inserted to divert the laser radiation toward the spectrum analyzer (SA) and the homemade 25 \xm thick PVDF pyroelectric detector (P,), with a typical response of 1.5 V/W at 500 Hz. A diaphragm (»1.5 mm diameter) was used to reduce the size of the laser beam, so that it could enter (from below) andpassunobstructedly through theuncoated ZnSe disc (thickness 1.5 mm, diameter 20 mm) that served as the OW cell. The annular lead-zirconate-titanate PZT piezotransducer (impedance 3MO at 100Hz) ring was glued to thebottom of the ZnSe window. A droplet of sample (thickness typically exceeding 3mm) wasdeposited onthe rear side of the ZnSe disc. -m DMMK- sample ZnSedisc lock-in PZT SA chopper MM CC^laser ^ MM Adiaphragm H^5 |DMMf Figure 2 Experimental arrangement used for in this OW study: with moveable mirrors (MM), pyroelectric detectors (P, and P2), PZT piezotransducer, spectrum analyzer (SA), and digital multimeter (DMM) The strength of the OW signal was measured via current to voltage amplifier using a twophase lock-in amplifier (Stanford Research Model SR850 DSP) and then normalized to the Analytical Chemistry 68,729-733 (1996) DetectionofTotalTransFatty AcidsContent inMargarine:anIntercomparison Study 43 incident power measured by P, placed behind the iris diaphragm (Fig. 2). A similar pyroelectric detector (P2) mounted above the OW cell was used to monitor the power transmitted through the sample (for strongly absorbing samples, no power is transmitted). The signals from P, and P2 were read off the digital multimeter (DMM). Cleaning theOW cell iseasily accomplished bywiping out thesample (using cotton swabsandapapertowel) andremovingresidualoildepositsbywashingwithpure chloroform. The optical alignment of the OW cell is relatively easy to perform. While positioning the OWassembly (mounted onaadjustable x-y-z-platform), oneattemptsto minimize thesignal from an empty OW, oralternatively, theOWloadedwithaweakly absorbing liquid such as chloroform, whilemaximizingtheradiation throughputbehindtheZnSediscasmeasuredby P2. The alignment and the signal level were regularly checked between succesive measurementsandfollowing eachcleaningofthecell. Asexplainedpreviously, normalized OWsignalsallow intheprinciplefor direct calculation (eq3and4)ofthesample's ßandthusfor recording ofitsabsorption spectrum.However, it was experimentally verified that the uncoated front surface of the ZnSe disk occasionally could lead to the occurrence of undesirable wavelength-dependent effects, such as multiple interferences for example, and hence false peaks in the spectrum. It is for this very reason thatwithourpresentOWthemeasurementswereperformed onlyatasinglewavelength (966 cm"1correspondingtothe 10R6C0 2 laserline). Results Distilled water (withwell-known thermalproperties 6)wasusedasareference sampleat293 KtocalibratetheOWcellatf=223Hz(\imter« 14\xm). Sincenoaccurate ßofwaterat966 cm' wasavailableintheliterature,thenearest17value ß=79200m"1atthe 10P10(953cm"1) C02 laserlinewasusedinstead;thisgivesSref=0.377from eq 1 and2.Thecalibration curve (seeExperimental Section) was constructed by computing the normalized lock-in signalsV (series of standard reference solutions containing 0-60% TFA) and Vref (distilled water), followed by calculation (eq 3and 4) of ßn for TFA-containing samples. For distilled water oneobtains ß=65800m"1at966cm'1:adecreaseof ßatshorterwavelength isinagreement with the expected general trend. The calibration curve is characterized by the average regression coefficient of 0.9993 (a =0.004). The limit of detection for the present system, calculated as aratio of three times the standard deviation obtained from the blank solution, i.e.,methyloleateinhexane,andtheslopeofthecalibrationcurveis 2%. Asan illustrative example, at a laserpower of 300 mW(signal from P, measured onDMM was 0.173 mV), the normalized signal strength V (for sample containing 32.7%TFA) was 2.061(a =0.028andlock-insignalwas0.356mV).Forwater,oneobtained Vref= 10.05(a = 0.050, lock-in signal was 1.658 mV) when DMM voltage was 0.165 mV. Under similar Analytical Chemistry68,729-733(1996) 44 Chapter3.1 conditions,theempty cellproducedanormalized signalof0.657(a =0.040). Thisvaluewas consistently subtracted from the readings obtained from both samples and reference when calculating ß(i. Table 1 TFAContent(%) indifferent samplesofmargarinebyvarious methods margarine sample (code) FTIR OW GLC GLC+ TLC 6337 59.3 64.9 62.2 63.6 42.0 50.5 53.6 35.9 58.3 62.7 41.8 42.4 39.8 41.0 6338 6339 6340 42.9 41.8 Table 1 displays the percentage of the total TFA content found in different samples of margarine (each datum represents the arithmetic mean of average values obtained from duplicate measurements, and each of these in turn implied a series of three consecutive measurements)byvariousmethods.Thecorresponding concentrationswerederivedfrom the calibration graphbyinterpolation. Dataobtainedbydifferent methodsaregenerally inagood mutualagreement: someminordiscrepancies are explainable.Forexample,theTFAcontent determinedby GLCisusually lowerthanthat obtainedby othermethods:presumably thisis because some of the trans peaks are masked by cis peaks and hence cannot be resolved. Consistently highervaluesfound by spectroscopic methods(OWandFTIR)areprobably due to the presence of polyunsaturated fatty acids in the sample (spectroscopic measurements providean estimateofthetotal numberoftransbondsinthe sample).Thus,ifknowledgeof thetotal TFA content isapriority, OWandFTIR arepreferred. When,however, the profile ofTFAisdesired, thecombined GLC+TLC method, capableofdetecting constituents inthe mixture (includingvarious stereo andpositional isomers) isaproper choice. Adisadvantage ofGLC+TLCisitstime-and labor-consuming character. Therelative errorinall techniques istypically2%. Table2 TFAContent(%)indifferent samplesofmagarine measured ondifferent days sample age (days) sample A 3xA+C A+C C 0 6.1 15.7 27 42 6 6.3 15.7 27.2 44.6 13 6.3 15.7 28 45.7 As a next step, the potential of the OW method to detect total TFA in margarine at concentration levelsmorelikely tooccurinpracticewasinvestigated. Two(arbitrarily called AnalyticalChemistry68,729-733(1996) DetectionofTotalTransFatty AcidsContentinMargarine:anIntercomparison Study 45 A and C) different samples of margarine and their combinations (A:C mixed in mass proportionality ratios 1:1 and 3:1) werestudied.Basedonresults ofthe GLCmeasurements, sampleAhasalow (3%)TFAcontentwhilesample Cisahigh-TFA (40%)mixture. A new calibrationgraph(0-40%TFA)wasprepared inthe sameway asdescribed above. Sinceone ofobjectives ofourstudieswasalsoto checktherepeatability andreproducibility oftheOW method (measurements were performed on different days), samples were stored at 5°C (to avoid evaporation). As seen in Table 2, expected and measured (each datum represents the arithmetic average of three consecutive measurements) values compare well (e.g., (44.6+6.3)/2=25.5% is close to a measured TFA content of 27.2%);this is also true in the lowerTFAconcentrationrange.Minordifferences mightbeduetothefact thatthemixingof samples was not homogeneous (only » 200 mg was taken from the mixed margarine). The TFAcontentinall samplesshowsatendency toincreaseslightlywithage,whichisprobably causedbyuncontrollable evaporationofthehexane. Conclusions Thefeasibility of the OW method asa candidate technique for quantitative analysis (at 966 cm"1)of total TFA content (2-60%) in realistic margarine samples was demonstrated in this intercomparison study. Compared to otherwell-established methods, the OW concept offers intrinsically someeconomicandpracticaladvantages suchaseaseandspeed ofoperation, as wellassimpleandrapidsampleloadingandcleaningofthecell. Currentdetection limit(2% TFA) for OW is restricted by the slope of the calibration curve rather then by the relative experimental error (0.028/2.061 =0.13%);the reproducibility of the measurement wasvery good. Further enhancement of detection sensitivity is anticipated by reducing the level of background absorption (proper choice as well as specific treatment (e.g., coating) of the window material). Changing the modulation frequency could also lead to an improved sensitivity. Inconclusion, unlikeintraditional IRanalysiswherecellthicknessistherestrictingfactor in dealingwith strongly absorbing samples,the magnitude ofthe OW signal depends solely on the product ßy., which, in principle, can be manipulated by the choice of the modulation frequency. Atthis stage,wavelength-dependent multiple interference effects inthe currently usedZnSewindow(itwasnotpossibletomountitataBrewsterangle)precludetheaccurate determination of absolute optical absorption coefficients ß for our samples over a broader spectralrange(seechapter3.2).Nevertheless,theOWmethod isthoughtuseful for potential applications in many other research areas including absorption-concentration studies in analytical chemistry, process control, and analytical studies of intact specimens in the life sciences. Analytical Chemistry68,729-733(1996) 46 Chapter 3.1 Acknowledgments M.C. expresses his gratitude to NWO (Dutch Organization for Scientific Research) for a fellowship received. Credit is to B. van Veldhuizen (Department Organic Chemistry WAU), for his advise and assistance provided in the FTTR experiments and to H. Boshoven for manufacturing components used in the experimental setup. Gratitude is to Mrs. J. Bos and Mrs. Kosmeijer-Schuil (bothDepartment of HumanNutrition) for preparing the samples and provided valuablehelpduring GLCand GLC+TLC measurements. References 1 OsborneB.G., Fearn T. and HindleP.H., Practical NIR Spectroscopy with Applications in Food andBeverage Analysis 2nded.,Longman Scientific & Technical Singapore (1993) 2 Mensink R.P. and Katan M.B., Effect of Dietary Trans Fatty Acids on High-Density and Low-Density Lipoprotein cholesterol levels in healthy subjects, New Engl. J. Med. 323, 439-445 (1990) 3 AOAC Method 965.34 Official Methods of Analyses of the Association of Official Analytical Chemists; AOAC Washington DC, VolXV, 969-701 (1990) 4 Helander P., Signal Processing in Optothermal Spectroscopy, J. Appl. Phys. 59, 33393343 (1986) 5 HelanderP., AMethod for the Analysis of Optothermal and Photoacoustic Signals, Meas. Sei. & Technol. 4, 178-185 (1993) 6 Tarn A.C., Applications of Photoacoustic Sensing Techniques, Rev. Mod. Phys. 58,381427 (1986) 7 Wetsel G.C. Jr. and McDonald F.A., Photoacoustic Determination of Absolute Optical Absorption Coefficcient, Appl. Phys. Letts. 30, 252-254 (1977) 8 Bicanic D., Chirtoc M., Chirtoc I., Veldhuizen B., Favier J.P. and Helander P., New Technique for Measuring Absorption Coefficients of Strongly Absorbing Liquids: Optothermal Study of Sun Flower Oil, Oleic Acid and Its Chloroform Solutions at 3.39 Microns,Spec. Letts. 28, 101-110 (1995) 9 Bicanic D., Chirtoc M., Chirtoc I., Favier J.P. and Helander P., Photothermal Determination of Absorption Coefficients in Optically Dense Fluids: Application to Oleic Acid and Water at COLaser Wavelength,Appl. Spec. 49, 1954-1959 (1995) 10 Zock P.L. and Katan M.B.,Hydrogénation Alternatives:Effects of Trans Fatty Acids and Stearic Acid Versus Linoleic Acid on Serum Lipids and Lipoproteins in Humans,J. Lipid Res. 33,399-410 (1992) 11 Metcalfe L.D., Schmitz A.A. and Pelka J.R., Rapid Preparation of Fatty Acid Esters from lipidsfor GasChromatographic Analysis,Anal. Chem. 38,514-155 (1966) Analytical Chemistry 68,729-733 (1996) Detection of Total Trans Fatty Acids Content inMargarine: an Intercomparison Study 47 12 Pfalzgraf A., Timm M. and Steinhart H., Content of Trans-Fatty Acids in Food, 1. Ernahrrungswiss. 33,24-43 (1994) 13 MorrisL.J., Separations of Lipidsby Silver Ion Chromatography, J. Lipid Res. 7, 717-731 (1966) 14 Nyquist R.A., Chrzan V., Kirchner T., Yurga L. and Putzig C.I., Studies of CarbonylContaning Compounds in Mixed Solvent Systems by Application of Infrared Spectroscopy,Appl. Spec. 44, 243-263 (1990) 15 Bertie J E . and Eysel H.H., Infrared Intensities of Liquids I: Determination of Infrared Optical and Dielectric Constants by FT-IR using the Circle ATR Cell, Appl. Spec. 39, 392-401 (1985) 16 Dwight E.G. (ed); American Institute of Physics Handbook, 3° edition, McGraw-Hill Book Company, New York (1982) 17 Hale G.M. and Querry MR., Optical Constants of Water in the 200-nm to 200-nm Wavelength Region,.%>/. Optics 12,555-63 (1973). Analytical Chemistry 68,729-733 (1996) 3.2 C0 2 Laser Infrared Optothermal Spectroscopy for Quantitative AdulterationStudies inBinary MixturesofExtraVirginOliveOil submittedtoJ.Amer.OilChem.Soc. JanPaulFavier,DaneBicanic,JanCozijnsen, BebvanVeldhuizenandPerHelander Abstract Optothermal window spectroscopy at C0 2 laser infrared wavelengths,wasusedto detect the extent of adulteration of extra virgin olive oil by sunflower and safflower oils. A good linearity between the strength of optothermal signal and the concentration of each adulterating compound wasfound. Predicted limits of detection presently attainable by this new method are 6% (w/w) and 4.5% (w/w) for extra virgin olive oil adulterated with safflower oil and sunflower oil respectively, were confirmed experimentally; the corresponding relative errors were 0.3%and 0.18%. Interference effects are comparable to thoseencounteredinotherspectroscopicmethods,atthesamewavelength. Keywords: adulteration, infrared, olive oil, optothermal spectroscopy, photoacoustic spectroscopy 50 Chapter3.2 Introduction Consumption of olive oil has increased over the past few years, a trend which may be attributedto its characteristicflavor andpotential healthbenefits. Oliveoil isobtained from thefruit oftheolivetree(Olea europaea I.)andmayneitherbemanipulatednorsubjected to anytreatmentnotapprovedbytherecognized internationalstandards.1 Substitution or adulteration of extra virgin olive oil by a cheaper ingredient is not only a major economic fraud, but can alsohavepotential health implications.2'3 The authentication of extra virgin olive oil is a major analytical challenge, but is often time-consuming and laborious.3 Traditional analytical techniques used for characterizing extra virgin olive oil, suchasdetermination ofiodineandsaponification values,measurement ofdensity,viscosity, andrefractive index,allrequiresomekindofsamplepretreatment.1'2 Modern high performance liquid chromatography (HPLC) and gas-liquid chromatography (GLC) techniques provide adequate separation and good sensitivity,4 but require substantial know-how to interpret the results. As an example, adulteration of extra virgin olive oil by canola oil could be detected down to 7.5% (w/w). Likewise, mass spectrometry in combination with an artificial neural network was shown capable of detecting 5% (v/v) adulterationofextravirginoliveoilbysoya,sunflower, cornandpeanutoil.6 Spectroscopy,ontheotherhand,offers thepossibility for directandnon-destructiveanalysis. For example, UV spectroscopy (at 210 and 315 nm) was successfully used to detect adulteration of extra virgin olive oil by refined oil at a level as low as 6%.3 Near infrared spectroscopy wasrecently utilized to estimatethe level (5%(w/w)of corn oil, sunflower oil and raw olive residue oil in virgin and extra virgin olive oils using principal component analysistechniques.7 Although mid infrared (MTR) spectroscopy (4000-400 cm"1 region) is often employed for studiesofoilsandfats,itsapplicationfordetection ofadulteration inoliveoilhasbeenrarely reported.3 In contrast to the IR spectra of other seed oils, the spectrum of olive oil israther flat in the 900 to 1200 cm"1 spectral region.1'2 Recently performed8 quantitative analysis of extravirginoliveoiladulteratedbywalnutoilandrefined oliveoilusingFT-IR spectroscopy combinedwithattenuatedtotalreflection (ATR)yieldedadetectionlimitof2%(w/w). In this paper, the optothermal window (OW) method, a spectroscopic technique related to photoacoustic spectroscopy, exploided to determine the levels of the adulterants safflower andsunflower oilsinextravirginoliveoil. submittedtoJ.Amer.OilChem.Soc. C0 2 LaserInfrared OptothermalSpectroscopyforQuantitativeAdulteration Studies 51 Experimental section Samples used in this study, were purchased fresh from a local store. A standard series of blendscontainingdifferent proportions(analyticalbalance)ofsafflower andsunflower oilsin extra virgin oil was prepared. During the initial phase of the research the percentage of adulterant inthemixturevariedfrom 0-100%(w/w)instepsof20%. Fatty acid profiles of the methyl esters9 (Table 1) were obtained by GLC with splitless injection (Carlo Erba 4160 GLC, Carlo Erba Strumentazione Milano, Italy). The separation was achieved on a DB-225 (50% Cyanopropylphenyl-methyl polysiloxane) column (15mx0.53mm,filmthickness 1.0 (xm;J&W Scientific, Folsom, CA).The linearvelocity of the helium carrier gas was 50 cms"1. The initial temperature (150°C), was programmed to 210°Cataheatingrateof2.5°C/min.Theinjection temperatureandthedetectortemperature were250°Cand275°C,respectively. Prior to the actual OW experiments, the spectra (900-1150 cm") of all adulterated and pure samples were recorded on a Biorad BST-7 FT-IR (Digit Lab. Dev. Cambridge MA 02139) spectrophotometer (resolution2cm') withadeuteratedtriglycinesulfate detector. The basic principles behind the OW spectroscopy and a detailed description of the experimental set-up aregiven elsewhere. ' The optothermal window constituted a 1.5 mm thickZnSediskandtheannularpiezoelectriccrystal(PZT)wasbondedtotherearsideofthe disk. A liquid sample («50 (xL)was deposited directly atop the disk, that on its turn was exposed to modulated C0 2 laser radiation. Following the absorption of radiation in the sample, the generated heat diffused into the disk causing its expansion that is sensed by a PZT. The OW cell used in this study was an improved version of a previous design." To avoid Fabry-Perot resonances in the ZnSe disk, the latter was provided with an antireflective coating on the side facing the incident laser beam. Additionally, to prevent laser radiation from directly striking the PZT detector which would give rise to false signals, a reflective gold layer (5^.mthick)wasdeposited ontothevery samesurface insuchawayasto leavea clear,uncoated, central circularareaabout4mmindiameter. Results and Discussion The composition of pure extra virgin olive oil and of adulterated samples were investigated byGLCmeasurementsandareshowninTable 1.TheGLCdatashowedalinearrelationship between the extent of adulteration andthe content of oleicand linoleic acid. Thepeak areas were expressed as percentages of the total areas (response factors were not used) and a typical errorforthesemeasurementsis3%(foracontentof36%theerroris1.1%). submittedtoJ.Amer.OilChem.Soc. 52 Chapter3.2 The FT-IR experiments were performed first in order to determine most appropriate excitation wavelengths of the C0 2 laser (emitting between 931-1084 cm"1). The comparison ofFT-IRabsorption spectraobtained from adulterated samples and thepure olive, safflower and sunflower oil revealed only small differences. In the 900 to 1050 cm"1 region olive oil featured arelatively flat absorption asexpected.2 Theactual OWstudieswere carried out (at 293K) atfiveselected C0 2 laserwavelengths: 931 cm', 953 cm"1, 966 cm'1, 1041 cm"1and 1079 cm"1. Table1 The fatty acid profiles (determined by GLC) FAMEs of olive and safflower oils and of their mixtures.Thecomposition ofsunflower oil isalsogiven. C16:0 C16:l C18:l C18:2 C18:3 safflower oil (% C18:0 C20:0 w/w) 1 1 0% 13 2 73 9 <0.5 1 20% 11 2 62 23 1 <0.5 <0.5 37 <0.5 40% 11 2 49 <0.5 10 <0.5 36 51 <0.5 60% 2 <0.5 8 <0.5 80% 2 23 66 <0.5 <0.5 7 77 100% 2 11 <0.5 <0.5 7 <0.5 sunflower oil 4 18 70 <0.5 <0.5 C20:l <0.5 <0.5 <0.5 <0.5 <0.5 <0.5 <0.5 Because of its well known optical and thermal properties, distilled water was used to calibrateoftheset-upfor OW measurements.11 The953 cm"1and 1041cm"1 C0 2 laserlines wereusedfor thecalibration;theabsorption coefficients ßfor wateratthesewavelengthsare 7.92x104m"1 and5.94x10 m" respectively.12Theopticalalignmentaswellasthemagnitude (typically < 1 n.V) of a background signal obtained from the empty OW window were checked between successive measurements and after completing the cleaning procedure of thecell(with chloroform). The plot showing the OW signals at 953 cm"1 calculated from experimental data obtained from mixtures containing known proportions of the safflower oil in olive oil is displayed in Fig. 1.Inorderto calculatetheabsorption coefficients, theOWsignalswerefirstnormalized tothelaserpowermeasuredbehindtheOWcell.'' Each data point in Fig. 1 represents an average of twenty successive measurements performed at221HzwiththeC02 lasertunedto953cm"1line;atthisparticularwavenumber therelative errorwas0.3%. Ingeneral,the laterwasfound todepend onapower stabilityof the laser at a specific transition; i.e. 0.7% at 931cm'1, 0.4% at 966 cm"1and 0.2%at 1079 cm".Forolive oil adulterated with safflower oil,the limit of detection (LOD),here defined astheratioofathreefold standard deviation obtainedfrom theblank (oliveoil)andtheslope ofthecalibration curve,is6%(w/w);theregression coefficient rfor datainFig. 1 is0.9995. submittedtoJ.Amer.OilChem.Soc. C02 LaserInfrared Optothermal Spectroscopy forQuantitative Adulteration Studies The 8% (w/w) and 6% (w/w) LOD's (regression coefficients r = 0.997) obtained when studyingtheabovetest samplesat931cm' and 966cm" areascribedtoapoorer stabilityof the laser at these two lines. At 1079 cm"1,as expected, no difference wasbetween theOW signals obtained from safflower and olive oils. On the other hand, differences measured at other lines were found to be due to the optical, rather than thermal changes induced in the sample. 0 20 40 60 80 100 contentofsafflower oiladulterantinoliveoil(%w/w) Figure 1 Experimentally determined absorption coefficient ß for olive oil adulterated by safflower oil. The studywasperformed at953cm"1transition oftheC0 2 laser 5700 20 40 60 80 100 contentofsunflower oiladulterantinoliveoil(%w/w) Figure2 Experimentally determined absorption coefficient ß for olive oil adulterated by sunflower oil. The study wasperformed at 1041cm"'transition oftheC0 2 laser submittedtoJ.Amer.OilChem.Soc. 53 54 Chapter3.2 Resultsfor oliveoil adulterated by sunflower oil shown inFig.2,wereobtained in thesame way asthoseshowninFig. 1,butwiththe C02 lasertunedtothe 1041cm"1lineinstead.The relative errorwas0.18%,givinga4.5%LODandaregression coefficient rof0.999. When deriving the above stated LOD's (Figs. 1and 2), the concentration range between 0 and 20%(w/w) wasoriginally extrapolated. Thisdeficiency waslater rectified by preparing extra standards (5%(w/w)) of each adulterant. Both of these samples produced measurable OW signals. As an example, for the extra virgin olive oil adulterated with 5% (w/w) safflower oil,absorption coefficient (at953cm")calculatedfrom calibration curve(Fig. 1) is 3604m"1,whichisvery closeto experimentally found 3590m"1.Similarly, onefinds 5059 m 1 (at 1041cm"1)compared to expected 5051m' for a specimen containing 5%of sunflower oilintheextravirginoliveoil. The OW method used here proved suitable to rapidly detect reasonably small amounts of known adulterants (safflower and sunflower oils) in the extra virgin olive oil; LOD's presently attainable are 5%(w/w). Generally, the obtained LOD's are of the same orderof magnitude as those attainable for similar adulterants by traditional methods (FTIR, GLC, HPLCandthemassspectrometry). Anenhancement ofLODisanticipated foramorestable, discreetly tunable C02 laser. The level of abackground noise for a OW device used in this study was significantly below that of the OW sensor used previously.11 Manufacturing the thinner window from a material having better optical and thermal properties than currently usedZnSe(e.g.diamond),isexpectedtoleadtoanimprovementofsensitivity. In conclusion, it has to be emphasized here that the OW study was performed with a discretely tunable C0 2 laser on two binary mixtures containing only adulterants, the quantities and the identities ofwhich wereprecisely known. Obviously, in order to identify and quantitate the adulterant of unknown nature in a complex mixture, one needs continuously tunable radiation. The presently emerging generation of compact and high powerinfrared diodelaserswillthereforeboostboththespecificity andsensitivity ofthe OW method, simply because the excitation wavelengths characterized by the optimal spectral contrast will become available. As for the effect of unwanted spectral interferences at a selected wavelength in the OW approach, this latter is comparable to that normally encounteredwithanyotherspectroscopictechnique. Oneshouldalsobearin mindthat OWdevicerepresents auniversal detection scheme, i.e.it is not wavelength limited. This means that one and the same device can be used within a widerangeofwavelengths,providedwindowremainstransparenttotheexcitingradiation. When combined with other advantages of the OW technique (practicality, reasonably low cost, suitability for theon-line operation and intrinsic simplicity when loading, cleaningand operatingthe cell) thedevelopments mentioned above could eventually makeOWtechnique submittedtoJ.Amer.OilChem.Soc. C0 2 Laser Infrared Optothermal Spectroscopy for Quantitative Adulteration Studies 55 a good candidate technique not only for rapidly quantifying presence of unknown adulterants in the extravirgin olive oil,but for many otherapplications aswell. As to this point, it is certainly worth noting the recent application of OW based instruments for the analysis of principal components (moisture, fat, protein, carbohydrates) of cheeses at the selected near and mid-infrared wavelengths. Its performance was shown superior to that ofFTIR-ATR in terms of speed, accuracy and the ease of data analysis.13'14 Work on similar matter is inprogress. Acknowledgments The work described in this paper was partially supported by the European Community (Copernicus ERBIC15CT961003). References 1 Recommended International Standard for Olive Oil, Virgin and Refined, and for Refined Olive-Residue Oil, FAO/WHOFood Standard Program, CAC/RS 33Rome, Italy (1970) 2 Kiritsakis A. and Marakakis P. Essential Oils and Waxes, ed. Linskens H.F. & Jackson J.F., Springer Verlag, Heidelberg, Germany, Chapter 7 (1991) 3 Wilson R.H., New Physico-Chemical Techniques for the Characterization of Complex Food Systems, 1sted. Dickinson E., Blackie Academic & Professional Chapman & Hall, Glasgow UK, Chapter 8 (1995) 4 Tsimidou M. and Macrae R., Detection and Quantitative Determination of Adulteration of Olive Oil,Int. Analyst 2, 29-34 (1987) 5 Salivaras E. and McCrudy A.R., Detection of Olive Oil Adulteration with Canola Oil from Triacylglycerol Analysis by Reverse-Phase High-Performance Liquid Chromatography, J. Amer. Oil Chem. Soc. 69, 935-938 (1992) 6 Goodacre R., Douglas B.K. and Bianchi G., Food Adulteration Exposed by Neural Networks,J. Sei. FoodAgric. 63,297-307 (1993) 7 Wesley I.J, Barnes R.J. and McGill A.E.J., Measurement of Adulteration of Olive Oils by Near-Infrared Spectroscopy,J. Amer. Oil Chem.Soc. 72, 289-292 (1995) 8 Lai Y.W., Kemsley E.K. and Wilson R.H., Quantitive Analysis of Potential Adulterants of Extra Virgin Olive OilUsing Infrared Spectroscopy, Food Chem.53,95-98 (1995) 9 NEN 6304 Plantaardig en Dierlijke Oliën en Vetten. Gaschromatografische Analyse van Methylesters van Vetzuren met Capilaire Kolommen 10 Heiander, P., A Method for the Analysis of the Optothermal and Photoacoustic Signals, Meas. Sei. & Technol. 4, 178-185 (1993) 11 Favier J.P., Bicanic D., Bovenkamp P. van de, Chirtoc M. and Helander P., Detection of Total Trans Fatty Acids Content inMargarine: An Intercomparison Study of GLC, GLC + submitted to J. Amer. Oil Chem. Soc. 56 Chapter3.2 TLC,FT-IRand Optothermal Window (Open Photoacoustic Cell),Anal.Chem. 68,729733(1996) 12 Hale G.M. and Querry M.R., Optical Constants of Water in the 200-nm to 200-p.m WavelengthRegion,Appl.Optics 12,555-563(1973) 13McQueenD.H.,WilsonR.,KinnunenA.,NearandMidinfrared PhotoaocusticAnalysisof Principal ComponentsofFoodstuffs, TrendsinAnal. Chem. 14,482-492(1995) 14McQueen D.H., Wilson R., Kinnunen A., Jensen E.P. Comparison of Two Infrared SpectriscopicMethodsfor CheeseAnalysis, Talanta42,2007-2015(1995) 15 WilsonR., InstituteofFoodResearch,Norwich,England,Privatecommunication 1997. submittedtoJ.Amer.OilChem.Soc. 3.3 Optothermal Detection of Infrared Radiation-Induced Absorption in Aqueous Solutions of Carbohydrates: Lactose and Corn Starch basedon JanPaulFavier,DaneBicanic,OttoDoka,MihaiChirtocandPerHelander JournalofAgricultural & FoodChemistry45,777-780(1997) Abstract The optothermal window method, a variant of photoacoustic spectroscopy, was used at 10 |i.m wavelength to explore its feasibility for direct investigation of aqueous solutions of lactoseandcomstarch.Present limitsofdetectionattainablebythismethodare0.19%(w/w) forlactosesolutionand0.6%(w/w)for cornstarchgel,respectively. Keywords:Photoacoustics,infrared spectroscopy,non-destructive analysis 58 Chapter 3.3 Introduction Chemical-physical analyses are becoming steadily more important when the quality of raw agricultural products and foodstuffs is evaluated. " Most of these techniques not only require some kind of sample treatment before detection is attempted but are also time consuming. The need for a rapid analysis while maintaining simultaneously sample integrity has therefore stimulated development of non-destructive inspection methods, among which are also various spectroscopic techniques. ' In particular, the infrared (IR) region with specific absorption frequencies and their band intensities uniquely characterizing chemical compounds, was utilized for qualitative and quantitative studies of foods.5"8 As many foodstuffs contain water which has a strong absorption in the IR, accurate quantitative spectroscopic measurements on such samples are not trivial. The use of traditional transmission cells for studies of pasty or viscous samples also causes problems associated with filling and cleaning procedures. These were partially alleviated by the introduction of the attenuated total reflection technique (ATR).6 This paper exploits the prospects of a new technique, i.e., that of optothermal window (OW), used in combination with a C0 2 laser (emitting between 1100 and 900 cm"1)for quantitative IR analysis of lactose and corn starch in water. A peculiar feature of this new technique, when compared to other methods, is the fact that the temperature variations induced in the sample by the absorption of radiation are sensed outside the area of the sample irradiated by the excitation source. Corn starch and lactose were selected as test samples because of their important role infoods.2 Starch isa major food constituent of foods and serves as a model for many foodstuffs. In the infrared corn starch has strong absorptions at 1681, 1053-952 and 855 cm"1.7On the otherhand, lactose isan important constituent of milk and dairy products. A sample of real milk was also investigated to validate the feasibility of the OW technique. Infrared analysis of milk is currently performed by determining the content of lactose (1042 cm"),protein (1549 cm"),and fat (1745 cm")usingdedicated filter instruments. ' All OW studies described here were carried out only at a few discrete wavelengths of the laser. However, the evaluation of results obtained in this experiment is important to estimate the prospects of a new, experimental technique based on a combined use of continuous tunable (powerful) infrared diode lasers as the radiation source and the OW concept. Finally, incorporating this new concept into that of FTIR spectrometry for quick scanning of spectra isworth considering. Journal ofAgricultural &Food Chemistry 45,777-780 (1997) OptothermalDetection ofInfrared Radiation InducedAbsorption inAqueousSolutions 59 Basictheoreticalconcepts ofthe OWmethod The OW technique is a variant of photoacoustic spectroscopy.11'12 The heat, produced in a condensed phase sampleby the absorption of modulated frequency f (s") radiation, diffuses intothedisksupportingthesample.Expansion ofthediskgeneratesanacousticwavewhich isdetectedby apiezoelectrictransducer (PZT).Adecisiveparameter isthethermal diffusion length (n)ofthesample,defined as ^= I— witha (mV) beingthethermal diffusivity of V 7tf the sample. Physically, (J. isthe distance across which the amplitude of a generated thermal waveisreduced to e"1ofits initial value.Only theheatoriginatingfrom a layerof (x deepis detectedbythePZT. When the thickness of the sample is larger than its thermal diffusion length, the sample is termed as"thermally thick".Forsuchasamplethat, inaddition, isinagoodthermal contact withthedisk,thenormalizedOWsignal S(ß)11 isgivenby S(P)= , ^ 2 ; V(nß) + (nß+ 2) 2 where ßis the wavelength dependent absorption coefficient (m1) of the sample. In general, theOWsignaldecreasesathigherfrequencies; this isalsotruefor thesensitivity ofthePZT sensor.Theabsorptioncoefficient ßofthesamplecanbeobtainedfromEq.(1)providedtheOW cellisproperlycalibrated.Inprinciple,recordingtheOWsignalatvaryingwavelengthsofincident laserradiationenablesonetoconstructanabsorptionspectrum.Theopticalpenetrationdepthmust belargerthanthethermaldiffusion length,i.e.,ß"1>n,iftheOWspectrumistocoincidewiththe trueabsorptionspectrum. ExperimentalProcedures The OW experimental arrangement (Fig. 1) comprised a homemade c.w. C0 2 waveguide laser which was modulated (here 221 Hz) by a chopper (EG&G model 179). A 2 mm diameter diaphragm (1),wasused to reducethe size of the unfocused laser beam, sothat it couldpassfrom below (atanormal incidence)through a20mmdiameterand 1.5 mmthick ZnSedisk(4)(JanosTechnology,Inc.)withtheannularPZT(3)(impedance3MQat 100Hz) bondedtothelowersurface. ToavoidFabry-Perotresonances,thediskwasprovidedwiththe antireflective coatingonthesidefacing theincident laserbeam.Inordertoprevent radiation striking thePZT directly, which would giveriseto unwanted false OW signals, a reflective goldlayer(5urnthick)wasdepositedonthesurface ofthediskbondedtothePZT,insucha wayastoleaveaclear(uncoated),central circularareaabout4mmindiameter. Adropletof the sample (200 (J.L for lactose) ora small quantity of starch gel wasdeposited directly atop the disk. The strength of the preamplified OW signal (after current to voltage amplifier JournalofAgricultural& FoodChemistry45,777-780(1997) CD 60 Chapter3.3 conversion)wasmeasuredby alock-in amplifier (Stanford Research model SR850DSP)and normalized tothe incident intensity measured with apower detector (Spectra Physics model 407A) placed above the empty OW cell. The emission wavelength of the laser was determined by inserting a movable mirror to divert the laser radiation toward the spectrum analyzer. Figure 1 Heart of the experimental OW set-up, displaying an adjustable diaphragm (1), gimbal mount (2), piezoelectric transducer (3),ZnSedisk(4),andadjusting micrometerscrews(5) Thealignment of theOW cell (the disk andPZT assembly) wasrelatively easy to perform. With the empty OW cell in a gimbal mount (2) (Microcontrole), the lock-in signal was minimized(typically0.7±0.4 (xV)byadjusting themicrometer(5)andmaximizingthelaser power throughput (typically 30-100 mW depending on the laser line) measured behind the cell. Both the alignment and the level of the OW signal were regularly checked between successive measurements as well as after completing the disk-cleaning procedure (simply wipingoutthesampleusingcottonswabsandapapertowel). JournalofAgricultural & FoodChemistry45,777-780(1997) OptothermalDetectionofInfrared Radiation InducedAbsorption inAqueousSolutions 61 Results Standard solutions varying from 2.5%to 20 % (w/w) lactose (Merck CAS-No. 10039-26-6 7660) and from 2.5%to 10% (w/w) corn starch (Merck CAS-No. 9005-25-8 11686) were prepared by dilution in distilled water. The starch solutions were shacked vigorously after heating(30secondsat80°C)inamicrowaveoven;thegelwasthenformed oncoolingdown toroomtemperature. Initially, the spectrum of 10%lactose wasrecorded with aFTTRspectrophotometer (Biorad BST-7, resolution 4 cm"1)to determine appropriate excitation wavelengths of the C0 2 laser needed for OW studies. The maximal absorption of the lactose solution was found at 1041 cm"1. The limit of detection (LOD), calculated as a ratio of 3 times the standard deviation obtainedfrom theblank (water)andthe slopeofthefitted calibration curve,is0.1%forthis FTIR experiment. The spectra of 2.5%and 5% corn starch samples were recorded using a ZnSe ATRaccessory. Difficulties were experienced whentrying to evenly spread a layerof gel alongtheflat surface ofthe ATR crystal (the gel tendsto fractionate easily). TheFTTRATRspectra ofthecorn starch sampleswereonlyqualitative indicating theabsorptionpeaks around 1053 cm'. xlO5 0 5 10 15 20 Lactosecontentinwater(%w/w) Figure2 Absorption coefficient ß for aqueous solutions containing a varying lactose content, measurementwasperformed atthe9P26(1041cm') lineoftheC0 2 laser Thecalibrationoftheexperimental set-upwasperformed13 withdistilledwaterbecauseofits well-known optical and thermal properties. The 9P26 (1041 cm"1) line of the C0 2 laser (nearly coinciding with the recommended analytical wavelength for lactose) was used for calibration; the absorption coefficient ßfor water at this wavelength is 5.94x10 m". The JournalofAgricultural & FoodChemistry45,777-780(1997) The Chapter3.3 62 calibration curvefor lactosewaslinearized (Eq. 1)andthe calculated absorption coefficients areshowninFig.2;eachdatapointisanaverageoffour measurements. The average regression coefficient of the calibration curve in Fig. 2 is 0.9992, and the calculated LOD of lactose in water is 0.19% (w/w). The 0.4% relative error for distilled water (theratio ofthestandard deviation a =0.03 tothenormalized OWsignalof 6.70) was higher than 0.2%(a =0.01 divided by 7.45) and 0.3%(a =0.02 divided by 7.89) obtained for 2.5%and 5%lactose solutions, respectively. The LOD of 0.19% (w/w) is substantially below the average lactose content found in milk. The straight line intercepts the y-axis at 5.92x10 m" which is very close to the absorption coefficient of pure water at 1041 cm'. Finally,acommercial milk samplewasused totestthefeasibility oftheOWtechnique.The 5.5%lactosecontentfound inmilkis 1%abovetheexpectedvalue,whichisnotsurprisingas no correction was made for the contribution to the OW signal due to absorption of protein andfat.ForcomparisonaFTIRapparatusyieldedaLODof0.05-0.1%forlactoseinmilk.10 xlO 4 2 4 6 8 10 Cornstarch contentinwater(w/w) Figure3 Absorption coefficient ß for a varying content of corn starch in water. The measurement was performed atthe9P12(1053cm"1)transition oftheC0 2 laser Theresults of measurements on corn starch samples are shown inFig. 3;at the 9P12 (1053 cm"1)lineoftheC0 2 laser,theLODinwateris0.6 %(w/w)andtheregression coefficient is 0.999. Theprocedure in deriving these value was the same as described above for lactose. TheinterceptinFig.3gives5.83xl04 m"1 whichcorrespondswellwiththeßofwaterat 1053 cm"1.14Thelargerrelative error(0.5-1%)for measurements of cornstarch gelsisascribed to the power instability of the laser. To check the reliability of the OW technique, the same samples were also examined at the 10P10 line of the C02 laser (953 cm"1) where their absorption is expected to be negligible. The OW signal was found independent of the JournalofAgricultural& FoodChemistry45,777-780(1997) Optothermal Detection of Infrared Radiation InducedAbsorption inAqueousSolutions 63 concentration, and the relative error was 0.6%. The difference between the OW signal strengths observed at 9P12and 10P10lasertransitions isduetooptical characteristics ofthe sample(gel)ratherthantothermalones. Conclusions TheOWmethod with the C0 2 laser here wasused for quantitative and direct determination oflactoseandcorn starch contents inwater. Presently attainable LOD's are0.19%(w/w)for lactoseand 0.6%(w/w)for cornstarch gels.FortheOWsensordescribed here, thephysical condition (liquid or gel) of the sample is irrelevant, since a good thermal contact between sampleandsensoristheonlyconditiontobemet. ApparentlytheFTIRmethodisslightly moresensitivefor determination oflactosethanOW, although the relative error (0.02%) is comparable to that of the OW method. For precise determination of lactose in milk, the OW is at present inadequate due to the availability of only a limited number of wavelengths which precludes the application of multiple linear regression. Forcorn starch gelsonly qualitative studies couldbe madewithFTIR-ATR. The effect of gel fractionation was less pronounced in the OW approach because the area that mustbecoveredismuchsmaller. Contrary to ATRandtheenzymatic methods (usedto determine incorporated starch),which both arevery expensive,15the low-cost OWdevice is easy to clean and moreover offers the possibility for on-line studies of optically opaque and thermally thick samples that are otherwisenotaccessiblebyothertechniques. Atpresent, duetoa limited tunability ofthe C0 2 laser, it isnotpossibletoobtain thewhole spectrum of the samples with the OW method. In the near future continuously tunable infrared diode lasers (gradually emerging on the market) are expected to provide power levels (strength oftheOWsignal isproportional to thepower density) sufficient to generate an acceptable signal to noiseratio. Therefore, sucha device when combined with the FTIR technique might be regarded asa new, candidate method for rapid analysis of realistic food samples throughout the entire infrared region. Additional potential practical applications of the OW method include the on-line control of processes (hydrophonic growth) and the quality control inthefood industry,for example,transfatty acids content in margarine13and phosphatesensorfor soft drinks.16 The performance of other thin IR transparent materials characterized by large values of thermal conductivity andthermal expansion coefficients might eventually surpassthatofthe ZnSe disk used in this experiment. Additional enhancement of the sensitivity is expected from adifferential OWconcept. JournalofAgricultural& FoodChemistry45,111-1%0 (1997) 64 Chapter 3.3 References 1 Alexander R.J. and Zobel H.F., Developments in Carbohydrate Chemistry, American Association of Cereal Chemists, St. Paul, MN (1992) 2 Belitz HD. and Grosch, W., Food Chemistry, Springer-Verlag, New York, Berlin, Heidelberg (1987) 3 Chang S.K.C., Holm E., Schwartz J. and Rayas-Duarte P., Food, Anal. Chem. 67, 127R153R(1995) 4 OsborneB.G., Fearn T. and HindleP.H., Practical NIR Spectroscopy with Applications in Food andBeverage Analysis, 2nded, Longman Scientific & Technical, Singapore (1993) 5 Belton P.S. and Tanner S.F. Determination of the Moisture Content of Starch Using Near Infrared Photoacoustic Spectroscopy,Analyst 108,591-596 (1983) 6 Belton P.S., Saffa A.M. and Wilson R.H., Use of Fourier Transform Infrared Spectroscopy for Quantitative Analysis: A Comparative Study of Different Detection Methods,Analyst, 112, 1117-1120 (1987) 7 Kochhar S.P. and Rossell J.B., Applications of Infrared Absorption Spectroscopy in Food Industry, Spectroscopy 4, 34-40 (1989) 8 Wilson R.H. and Belton P.S., AFourier-Transform Infrared Study of Wheat Starch Gels, Carbohydr. Res. 180, 339-344 (1988) 9 Biggs DA., Performance Specifications for Infrared Milk Analysis, J. Assoc. Off. Anal. Chem. 61, 1211-1214(1979) 10 Luinge H.J., Hop E., Lutz E.T.G., Hemert J.A. and Jong E.A.M. de,Determination of the Fat, Protein and Lactose Content of Milk using Fourier Transform Infrared Spectroscopy, Anal. Chem.Acta 284,419-433 (1993) 11 HelanderP., AMethod for the Analysis of Optothermal and Photoacoustic Signals, Meas. Sei. Technol. 4, 178-85 (1993) 12 Bicanic D., Chirtoc M., Chirtoc I., Favier J.P. and Helander P., Photothermal Determination of Absorption Coefficients in Optically Dense Fluids: Application to Oleic Acid and Water at COLaser Wavelength,Appl. Spectrosc. 49, 1485-1489 (1995) 13 Favier J.P., Bicanic D., Bovenkamp P. van de, Chirtoc M. and Helander P., Detection of Total Trans Fatty Acids Content inMargarine: an Intercomparison Study of GLC, GLC + TLC, FT-m. and Optothermal Window (Open Photoacoustic Cell) Anal. Chem. 68, 729733 (1996) 14 Hale G.M. and Querry MR., Optical Constants of Water in the 200-nm to 200-nm Wavelength Region,Appl. Optics 12, 555-563 (1973) 15 Biochemical Analysis Food Analysis, Boehringer Mannhein GmbH Biochemica, Germany (1986) 16 Vonach R., KellnerR. and Lippitsch M., APhosphate Sensorfor Soft-Drinks Based on IR Spectroscopy, Proceedings Eurofood VIII, Vienna, Austria (1995). Journal ofAgricultural &Food Chemistry 45,777-780(1997) 3.4 Compact,OpenandGeneralPurpose CellofVariableEffectivePathlength: DirectAbsorptionMeasurementof S042"inWater basedon JanPaulFavier,DaneBicanic,MihaiChirtocandPerHelander FreseniusJ. AnalyticalChemistry355,357-358(1996) Abstract The use of an optothermal window (OW) was proposed for the direct (no need for sample preparation) spectroscopic,non-destructive measurementof S042"inwater at 1078cm"1.The presently determinedlimitofdetection (LOD)of 1 mmol/Liscomparabletothatprovidedby C02 laser photoacoustic spectroscopy, but about one order of magnitude superior to that obtainablebytheATRmethod. 66 Chapter3.4 Introduction Manysamplesofenvironmental, nutritional andbiochemical interestrequirethe examination inaqueoussolution.Yet,theuseoftraditionalaswellasnewspectroscopicmethods (specific for analytes to be measured) at infrared (IR) wavelengths, is not straightforward due to the intrinsically strongabsorptionofwaterinthisregion.Theproblemsrelated tovery shortcell pathlengths when working with such samples are partly alleviated by attenuated total reflection spectroscopy (ATR). Although in the principle ATRprovides a simple means for obtaining spectra of liquids, their quality is influenced by factors such as sample surface textureandrefractive index.Furthermore,distinctdifferences areobservedbetween ATRand transmission spectra,mainly causedbythe effect ofwavelength dependent penetration depth (being smaller at shorterwavelengths) of the radiation into the sample.1 Theseproblemsare avoided by the OW method, the feasibility of which has been demonstrated previously2'3 when measuring absolute absorption coefficients of strongly absorbing and pure liquids at shorterwavelengths. ThispaperreportsonafirstuseoftheOWmethodforthedirectmeasurement (at 1078cm"1) of S042"concentrationsinaqueoussolutions.Themainobjective oftheundertaken studywas to estimatetheanalyticalpotential ofthiscandidate spectroscopic method, andto compareit tothat ofATR. Sulfate waschosen asatest analytebecauseof its generally recognized role inenvironmental andagriculturalsciences.4 Basicconcepts The new approach implies a modulated laser radiation passing through a transparent disk with a liquid sample atop. The heat generated in the sample due to the absorption of the radiation, diffuses into the window (ZnSe disk 1.5 mm thick, 20 mm diameter) causing its periodic expansion that is sensed by a piezoelectric detector bonded to the rear side of the disk. For a "thermally thick" sample making a good thermal contact with the sensorbeing itself "thermally thick",thereexistsasimplerelationship2betweenthemagnitude oftheOW signalandtheproduct oftheabsorption coefficient ßofthesampleandthethermal diffusion length (x. Theeffective pathlength ofthesampleisdeterminedby nandcanbe controlledby varyingthemodulation frequency. Experimental Mechanically chopped C0 2 laser radiation (200 mWand 223Hz)passed through a 1.5 mm diameter diaphragm before reaching the OW cell. A series of aqueous S042" solutions was prepared ((NH4)2S04 and CuS04 Merck). The OW signal was measured by a lock-in amplifier andthannormalizedtotheincidentlaserpower.TheOWcellwascalibrated at953 FreseniusJ. AnalyticalChemistry355,357-358(1996) Compact, OpenandGeneral PurposeCellofVariableEffective Pathlength cm"1 using distilled water (ß=792 cm"1)5as a reference sample; under the given experimental conditions pfj. is 1.1. Results andConclusions The calibration curve for S042" in distilled water is shown in Fig. 1. The linearity of the normalized OW signal is observed overall concentration range spanning more than two orders of magnitude. Themeasuredbackgroundsignalduetotheabsorption ofthedistilled water is high (corresponding to ß of 520 cm"1) and limits at present the sensitivity for S0 4 2 to 1 mmol/L. This value, close to that reported by another IR photoacoustic experiment,6 exceeds the LOD of ATR by nearly one order of magnitude. The non-spectroscopic methods such as nephelometry, ion chromatography and titration, ' capable ofbetter S042"LOD's all have an indirect character. 12.0 11.0 10.0 9.0 8.0 7.0 6.0 5.0 0 50 100 150 200 250 300 350 400 450 500 concentration sulfate (mmol/L) Figure 1 The normalized OWsignal at 1078cm"1plotted versusconcentration ofS042"in distilled water.The solid curve represents thebestfit(r=0.9996) The compact and open OW cell, does not only require small quantities of samplefor analysis, but it is also easy to load and clean. Possible interferences10 due to the presence of other anions (in particular P043") in solution were not investigated here; yet their overall effect on the OW signal is expected tobe comparable tothat observed with ATR and IR spectroscopy. Although the present LOD obtained with the OW is not sufficient for purposes of trace analysis, the moderate sensitivity of the OW method and its applicability for direct, on-line work, makes it suitable for monitoring higher S0 4 "contents. Examples are found in studies of S0 4 "to H2S reduction due to the reaction of anaerobic micro-organisms, combating the Fresenius J.Analytical Chemistry 355,357-358 (1996) 67 68 Chapter 3.4 corrosion in concrete, process control in pharmaceutical, dye and textile industries, monitoring of S0 4 2 asaplant nutrient inthe soil-less growth of many crops, etc. The use of both the differential concept and the coated optics in designing a new OW cell is expected to further improve the LOD. Finally, the OW approach is not restricted to the IR region scince one and the same OW can be universally used provided however the window transparency is guaranteed and the excitation of the analyte canbe achieved. References 1 Osland R.C.J., Principles and Practices of Infrared Spectroscopy, Pye Unicam Limited, Cambridge (1985) 2 Helander P., AMethod for the Analysis of Optothermal and Photoacoustic Signals, Meets. Sei. Technol. 4, 178-185 (1993) 3 Bicanic D., Chirtoc M., Chirtoc I.,Favier J.P. and Helander P.,Appl. Spec. 49, 1485-1489 (1995) 4 Sonneveld C. and Straver N., Nutrient Solutions for Vegetables and Flowers Grown in Water or Substrates, Serie: Voedingsoplossingen glastuinbouw no 8ISN 116174 (1988) 5 Hale G.M. and Querry M.R., Optical Constants of Water in the 200-nm to 200-nm WavelengthRegion,Appl. Opt. 12, 555-563 (1973) 6 Kanstad S.O. and Nordal P.E., Infrared Photoacoustic Spectroscopy of Solids and Liquids, Opt. Comms. 29, 413-422 (1981) 7 Application Manual Miran 1A,Foxboro, Bristol Park MAUSA (1985) 8 Fresenius W., Quentin K.E. and Schneider W., Water Analysis, Springer-Verlag Berlin Heidelberg New York (1988) 9 Die chemische Untersuchung von Wasser, Merck, Darmstadt (1992) 10 Cross A.D. and Jones R.A., An Introduction to Practical Infrared Spectroscopy 3 rd ed. London Butterworth (1969). Fresenius J.Analytical Chemistry 355, 357-358 (1996) 4 Thermal Characterization of Different Samples 4.1 Thermal Diffusivity of Hard Boiled Candy Obtained by Photothermal Beam Deflection and Standard Photopyroelectric Method 4.2 Photopyroelectric (PPE) Measurement ofThermal Diffusivity in Low Density Polyethylene (LDPE) and Polyvinyl Chloride (PVC) Foils 4.1 Thermal Diffusivity ofHardBoiled Candy Obtained by Photothermal BeamDeflectionandStandard Photopyroelectric Method JanPaulFavier,DorinDadarlat,JürgenGibkes,CorneliusvandenBerg,andDaneBicanic acceptedbyInstrumentation Science& Technology Abstract Two different photothermal techniques, photopyroelectric (PPE) and photothermal beam deflection (PTBD) method were used for the thermal characterization of a glassy sugar system, commercially availablehardboiled candy.Thermaldiffusivities atroomtemperature obtained by both techniques were comparable (i.e. 14.7 xlO"8and 12.0 xlO"8 m V for PPE and PTBD respectively). In addition, the PPE technique was also used to measure temperature dependence of the thermal diffusivity in the -30°Cto 70°C temperature range. Theglass-rubbertransition, underwentbythesample inthistemperature rangewasobserved andcomparedtoresultsobtainedbydifferential scanningcalorimetry (DSC)experiment. 72 Chapter4.1 Introduction Thermal and optical characterization of foodstuffs are important for their industrial manufacture, development and improvement, prediction of stability, quality assessment, and otheraspects.Withregardtomanufacturing andassessingthe stability of sugarcandies,data associated with phase transitions (melting and glass-rubber transitions) are of particular importance.1'2 Thermal properties of these products allow one to predict heat transfer rates during the manufacturing process. Assuming constant values of thermal properties during heating and cooling cycles isan oversimplifying approach. Atpresent, one computes thermal diffusivity from values for thermal conductivity, specific heat, and mass density.1'2 While the specific heat can be estimated with reasonable accuracy from the product composition, the thermal conductivity and mass density generally must be determined by experiment. Thermal properties are known to alter significantly during glass and other phase transitions, and as suchareusedtodeterminetransitiontemperatures.3 Hard boiled candies are mainly amorphous materials in the glassy state; they are solidified liquids(withanextremely highviscosity) supportingtheirownweight. Typicalpropertiesof suchsugarglassesaretheirbrittlenessandtransparency. Thesequalitiesaredecisivetextural and optical characteristics for sweets. The extremely slow changes occurring in the glassy stateareoften referred toasphysicalageing.' Water is a perfect plasticizer for glassy sugar systems, and its content strongly affects the glasstransitiontemperature(Tg)oftheproduct. Generally,Tgofamorphousfood components hasbeenrecognized asone of the major factors in controlling the shelf life of low-moisture andfrozen foods.1 Forhardboiled candies,Tgisaboveroomtemperature. Recently, two new photothermal (PT) methods, i.e. photopyroelectric (PPE)4"7 and photothermal beam deflection (PTBD) ' were proposed for calorimetric investigation of foodstuffs. In the PPE method, the amount of heat developed in a sample, induced by the absorption ofamodulatedradiation, ismeasuredwithapyroelectric sensor.10'11 VariousPPE configurations wereproposed inorderto studythermal propertiesand/orphasetransitionsin foods.4"6 Among these, the so called "standard geometry", with sensor and sample both thermally thick and an optically opaque sample, proved the most suitable, because the temperature behavior of all sample thermal parameters can be obtained from a single measurement. On the otherhand thePTBD zero crossing method isa well established laser based technique for thermal diffusivity measurements ofbiological and solid materials (with atypicalaccuracyof5%). acceptedbyInstrumentation Science&Technology Thermal Diffusivity of Hard Boiled Candy Obtained by PTBD and PPE 73_ Due to unavailability of literature data of thermal parameters for sugar systems such as candies, the zero crossingPTBD and PPE methodwere applied in this study to determine the thermal diffusivity of hardboiled candy. The influence of temperature on thermal diffusivity, and the glass transitions were investigated inthe -30°C to 70°Ctemperature range. Principles ofphotothermalbeamdeflectionmethod The illumination of an absorbing sample by a periodically modulated focused laser beam (the pumpbeam) causestheperiodicgeneration ofheat.Propagation ofthermal wavesfrom the sample into the surrounding fluid can be described by appropriate heat diffusion equation. Since the refractive indexofafluid istemperaturedependent,thegenerated spatialtemperature gradient can be sensedby a second laserbeam (theprobebeam). Asa result,theprobebeam will be deflected asittraversestheheatedregionparalleltothesurface ofthesample. The deflection is a vector having normal and transverse components that refer to deflections in a planeperpendicularandparallel tothe surface ofthe sample.For strongly damped thermal waves, deflection signalsareduetothecontributionofheatoriginatingwithinonethermal diffusion length Us (m)inthesample;thelatterisgivenby "•- ë <» wheref (s")isthemodulationfrequency ando^isthermal diffusivity (mV) ofthesample,related tootherthermalparametersby « s = — PsCs (2) where K,,, ps and cs are thermal conductivity (Wm'K 1 ), density (kgm3) and mass specific heat (Jkg'K"1)ofthesample,respectively. In order to accurately measure the thermal properties of the material sample, the probe beam shouldpassascloseaspossible(distancessmallerthanonethermal diffusion lengthinthefluid are acceptable) andparallel to its surface. If the distance between the probebeam and the surface of the samplebecomes larger, the signal will be dominated by thermal properties of the surrounding fluid (mostlyair). Astraight lineisobtained when the zero crossing distance isplotted versus Vf .12"14The obtained slopemisrelatedtothethermaldiffusivity as m =-v/W* accepted by Instrumentation Science & Technology (3) 74 Chapter 4.1 and allowsfor itsdirect calculation. InEq. (3)yis 1.44 for opaque, thermally thick samples and 1 for transparent samplesaswell asfor opaque, thermally thin samples. The PTBD technique isnot describedinmoredetailhere,becauseofitslimitedpracticabilitycomparedtothePPEmethod. Principles ofthephotopyroelectricmethod In the standard PPE configuration, the radiation impinges on the front surface of the sample, while the pyroelectric sensor, itself in a good thermal contact with the rear side of the sample, measures its temperature variation. A sensor or a sample are said to be thermally thick when their geometrical thickness' are larger than the thermal diffusion lengths in the materials. If in the standard configuration the sample and sensor are both thermally thick and the sample isoptically opaque, the amplitude Sand the phase cp ofthe PPE signal are given by the following equations15 S=S0x— —, - (4) +1 (5) In Eq. (4), S0is an instrumental constant depending on the intensity of the incident radiation, the chopping frequency, thermal, electrical and geometrical parameters of the sensor, and s is the thermal effusivity defined asJicpc . The subscripts p, s and m refer to the pyroelectric sensor, sample and medium in front of the sample (air in the case discussed here), respectively. The Eq. (5) suggests that the phase cpof the PPE signal depends solely on the thermal diffusivity of the sample, allowing for its direct and absolute measurement, providing lsandf are known. Experimentalsection The experimental set-up used for PTBD measurements of thermal diffusivity of candies is a modified version of the apparatus described previously.9,,fi The major difference is that the probe beam(SpectraPhysics 126,HeNelaser)bouncesonthesurfaceofthesample.Itfirstpassesthrough the neutral density glass filter (1% transmission; the filter is required to avoid saturation of the position sensitive quadrant detector) before reaching a 50 mm plano-convex converging lens (focusing is necessary to reduce the probe beam diameter). The probe beam is then defocused accepted by Instrumentation Science & Technology Thermal Diffiisivity of Hard Boiled Candy Obtained byPTBD and PPE 75_ behind the sample using another 50 mm plano-convex lens; the detection takes place with a sensitivequadrantdiode.Theoutputsignalfrom thediodewasfed intoanITHACO396IB lock-in amplifier coupledtoapersonalcomputer. Thesamecomputeralsoprovidedthestepwiseincreasing voltageoutputusedtodrivethemirrors(lock-insignalsweremeasuredateachmirrorsetting). The experimental set-up used for PPE investigations was discussed in extenso elsewhere;4 only somedetails are presented here. The PPE cell is a cold-finger based system, allowing the operation at temperatures below and above that of the ambient.5 The typical temperature variation rate was 1°C per minute with an acquisition at each 0.1°C. The pyroelectric sensor was a 300 um thick single crystal of LiTa0 3 . The opacity of the sample was achieved by placing a thin (10 um) blackened Al foil atop of it. Thin layers of silicon grease between Al foil, sample and sensor provided a good thermal contact. A 5 mm diameter diaphragm in front of the sample prevented direct illumination of the sensor. The signal from the detector was processed with a Stanford Research SR 850 lock-in amplifier. The radiation source was a 30mW Melles Griot diode laser chopped electronicallybythe internal oscillator ofthe lock-in amplifier, and a personal computer wasusedfor data acquisition. Initially, a frequency scan was performed at room temperature in order to find the range of appropriate modulation frequencies that satisfy the requirements imposed for this SPPE configuration (thermally thick regime for the sample and sensor) and to obtain an absolute calibrating valuefor the thermal diffusivity at room temperature. Commercially available hard boiled candies (20x20x15 mm3, type LONKA) were used as test samples. Typically, they contain about 65%sucrose, 30% glucose syrup solids and less than 4% water. For the PPE experiments samples (0.5-1.0 mm thick, area about 10x10 mm2) were cut from the middle of the candy (flatness and uniform thickness of the samples are obligatory). Duplicate samples (approximately 20 mg) of the same sugar candy in 20 ul aluminum cups were scanned at 5 and 10 Kmin"1 in Perkin Elmer DSC-2, provided with computerized data acquisition. An empty cupwas used as a reference. The results ofPPE investigations of phase transitions were compared tothese obtainedbyDSC. Resultsanddiscussions Using the experimental procedure described above,PTBD transverse signals were recorded at various chopping frequencies. An average of 50 successive measurements was taken as representative for each position of the pump laser beam. The candy was assumed transparent for thepump laser radiation, and thePTBD experiment wasrepeated three times. The average valuefor the thermal diffusivity calculated by linear regression from mvalues, 0.6198, 0.5765 and 0.9893 m m V respectively, is 12xl0"8 m V (standard deviation o =7xl0~8). accepted byInstrumentation Science & Technology Chapter 4.1 76 As to the PPE measurements, a frequency scan of the phase of the PPE signal allows one (Eq.(5)) to obtain the value for thermal diffusivity at room temperature. The results obtained for a 900 (xmthick sample are presented inFig. 1(curve '+' and 'o'); The slopes of -4.05 and -4.27 give an average a of 14.7xl0"8 m2/s (a = lxlO"8) which agrees rather well with the value obtained byPTBD. -^/Frequency (•N/HZ) Figure 1 The phase of the PPE signal plotted versus square root of frequency, (+) sample 900 urn thick not heated;(o)sample900 umthickheated upto30°C;(*)sample400 urnthick heated upto70°C -40 -30 -20 -10 0 10 20 30 Temperature (°C) 40 Figure2 The magnitude ofthephotopyroelectric signal amplitude plotted versus temperature, (+) heating and (o)cooling accepted byInstrumentation Science & Technology Thermal Diffusivity ofHardBoiledCandyObtainedbyPTBDandPPE As seen in Fig. 1, the requirement for thermally thick sample imposed that the modulation frequencies mustbelargerthan 1 Hz.Consequently, 1.5 Hzwasusedinthetemperaturescan. At this frequency the Al foil and silicon grease layers are thermally very thin and do not influence theresults. When the measured datawereprocessed the temperature dependenceof the PPE signal obtained from the unloaded sensor was used for normalization;4,5 in such a way the temperature dependence of the pyroelectric coefficient and electrical capacitance of thesensorareeliminated. -40 -30 -20 -10 0 10 20 30 Temperature(°C) Figure3 Thephaseofthephotopyroelectric signalplottedversustemperature,(+)heatingand (o)cooling -40 -30 -20 -10 0 10 20 30 40 Temperature(°C) Figure4 Thermaldiffusivity versustemperature,for asamplethicknessof900p.m; (+)heatingand(o)cooling acceptedbyInstrumentation Science& Technology 77 Chapter 4.1 78 As the temperature of the candy was varied from -30°C to 30°C, the thermal diffusivity decreased monotonically with increasing temperature. The heating-cooling process was reversible (seeFigs. 2 and 3for the amplitude and phase of the PPE signal and Fig. 4 for the thermal diffusivity). After several repeated cooling-heating cycles, the same room temperature value for a (in the error limit) was obtained, using the frequency scan method too (see Fig. 1, curve '*' with a slope of-2.05 and thickness of400 ^m; a of 12.0xl0~8 mV 1 ). 20 40 60 Temperature (°C) 80 Figure 5 The temperature dependent amplitude of the photopyroelectric signal plotted versus temperature; (+) heating and (o) cooling 150 20 40 60 Temperature (°C) 80 Figure 6 The phase ofthe photopyroelectric signal plotted versus temperature; (+) heating and (o) cooling acceptedby Instrumentation Science & Technology Thermal Diffusivity of Hard Boiled Candy Obtained by PTBD and PPE As the temperature exceeded 40°C, the candy undergoes a glass-rubber transition, i.e. a supercooled melting. The results for the amplitude and the phase of the PPE signal are presented in Figs. 5and 6. The increase in amplitude and phase observed on cooling, was due to the reduced sample thickness (400 urn) during melting (see also strong increase heating trace). The DSC measurements also confirmed the existence of a glass transition (Fig. 7) at 30°C (midpoint, corrected for scan speed). This transition was reproducible, only the first heating showed a small overshoot peak inthe DSC trace atthe end of glasstransition (Fig. 7), which is indicative for aging relaxation of the glass. The negative slope of the DSC trace is commonly observed for comparable sugar systems. Above 32°C the sample became a supercooled melt with fluid properties (crossing point of slopes in Fig. 8). However, at the time scale of measurement the sample became liquid near 40°C, where its thickness changed spontaneously. 0 20 40 60 80 100 120 Temperature (°C) Figure 7 TheDSCsignalversustemperature The width of the glass transition region is rather wide (approximately 8 centigrades from onset to endset, scan speed 5Kmin"1). At 10°C the trace started to deviate from the baseline, which is closetothefirst change in the slopefound byPPE measurements (Figs. 5-8). Despite an excessive thermal treatment, the value of thermal diffusivity at the room temperature remained practically the same (Fig. 1, curve '*')• Additional heating-cooling cycles (-30°C to 60°C) showed a well reproducible monotonous decrease of a for increasing temperature (Fig. 8), with comparable room temperature value (Fig. 1, curve '*' is reproducible). These facts lead to the following conclusions: (i) temperature changes up to 30°C did not influence the thermal properties of the candy; the thermal diffusivity displays a gradual, reproducible decrease (from 2.4xl0"7 to 1.4xl0"7 mV 1 ) with increasing temperature. accepted byInstrumentation Science & Technology 79 Chapter4.1 80 (ii)Physicalageing,asaquality parameter, couldnotbedetermined from aroomtemperature frequency scan (the room temperature value of the thermal diffusivity remained the same), however physical ageing could be detected by a PPE temperature scan. This was due to the fact (also confirmed by DSC) that the initial heating (up to 70°C), always produces a relaxation oftheglass. 0 20 40 Temperature(°C) 60 Figure8 Thermaldiffusivity versustemperaturefora400 \unthick sample;(+)heatingand(o)cooling AnimportantremarkconcernstheuseofEq.(4).TheamplitudeofthePPEsignaldependson two sample related thermal parameters (i.e.thermal diffusivity and effusivity). If a literature value of a second thermal parameter (at room temperature) would be available, the measurements of amplitude of the PPE signal could be calibrated. Consequently, the temperature dependence of all thermal parameters could be computed. Unfortunately, no quantitative information about the thermal parameters during the melting process could be obtainedbecausethethicknessofthesamplechangedinanuncontrollable manner. In conclusion, two new photothermal techniques (PTBD and PPE) were used to obtain previously unreported value for thermal diffusivity of a hard boiled candy at room temperature. Itwasdemonstrated thatthePPEmethodwascapableofmeasuring temperature dependence of the thermal diffusivity and to detect phase transitions in such sugar systems and the results were confirmed by DSC measurements. The occurrence of a glass phase transition,canbecorrelatedwiththethermalhistoryofthesample. Acknowledgments One ofthe authors (D.Dadarlat) acknowledges the receipt of avisiting scientific fellowship from theDutchOrganizationfor Scientific Research (NWO,TheHague)that madethiswork acceptedbyInstrumentation Science& Technology Thermal Diffiisivity of Hard Boiled Candy Obtained byPTBD and PPE 81_ possible. The research described here is also supported by EC program Inco Copernicus (proposal ERBIC 15CT961003) References 1 RoosY.H.,Phase Transitions inFoods, AcademicPress, SanDiego (1995) 2 Mohsenin N.M., Thermal Properties of Foods and Agricultural Materials, Gordon and Breach Scientific Publishers, New-York, London, Paris (1980) 3 SladeL. and LevineH., CRC Crit.Rev. Food Sei. Nutr. 30, 115-360 (1991) 4 Dadarlat D., Bicanic D., Visser H., Mercuri F. and Frandas A. Photopyroelectric Method for Determination of Thermophysical Parameters and Detection of Phase Transitions in Fatty Acids and Triglycerides. Part I: Principles, Theory and Instrumentational Concepts, J.Amer. Oil Chem. Soc. 72,273-279 (1995) 5 Dadarlat D, Bicanic D, Visser H, Mercuri F and Frandas A. Photopyroelectric Method for Determination of Thermophysical Parameters and Detection of Phase Transitions in Fatty Acids and Triglycerides. Part II: Temperature Dependence ofThermophysical Parameters. J.Amer. OilChem. Soc. 72,281-287 (1995) 6 Dadarlat D., Bicanic D., Gibkes J., Kloek W., Dries I. van den and Gerkema E., Study of Melting Processes in Fatty Acids and Oils Mixtures. A Comparison of Photopyroelectric (PPE) and Photopyroelectric Observations of Melting Differential Scanning Calorimetry (DSC), Chem.Phys. Lipids 82, 115-123 (1996) 7 Bicanic D, Dadarlat D, Gibkes J, Chirtoc M, Favier J P and Gerkema E, The Photopyroelectric Approach to Thermal Characterization of Liquid and Pasty Foodstuffs and an Optothermal Accessory for Obtaining Infrared Spectra of Optically Dense Fluids, Acta Chim. Slov. 42, 153-173 (1995) 8 Brown S.M., BaessoM.L., Shen J. and Snook R.D., Thermal Diffusivity of Skin Measured byTwoPhotothermal Techniques,.4MO/. Chim.Acta 282, 711-719 (1993) 9 Brown S.M., Bicanic D. and van Asselt K., Photothermal Beam Deflection Measurement on Agricultural Produce,J. Food Eng. 28,211-223 (1996) 10 Mandelis A. and Zver M.M., Theory of Photopyroelectric Spectroscopy of Solids, J Appl. Phys. 57, 4421-4430(1985) 11 Chirtoc M. and Mihailescu G., Theory of the Photopyroelectric Method for Investigation ofOptical and Thermal Materials Properties,Phys.Rev. B 40, 9606-9617 (1989) 12 Sell J.A., Photothermal Investigation of Solids and Fluids, Academic Press Inc., Boston (1988) 13 Bertolotti M., Li Voti R.L., Liakhou G. and Sibilia C , On the Photodeflection Method Applied toLow Thermal Diffusivity Measurements, Rev. Sei.Instr. 64, 1576-1583 (1993) accepted by Instrumentation Science & Technology 82 Chapter 4.1 14 Bertolotti M., Li Voti R.L., Liakhou G., Sibilia C , Reply to Comments on the Photothermal Method Applied to Low Thermal Diffusivity Measurements, Rev. Sei. Instr. 66, 277 (1995) 15 Marinelli M., Mercuri F. and Zammit U., Pizzoferrato R., Scudieri F., Dadarlat D. Photopyroelectric Study of Specific Heat, Thermal Conductivity and Thermal Diffusivity of Cr 2 0 3 at theNeelTransition, Phys.Rev. B 49, 9523-9537 (1994) 16 Gibkes J., Dadarlat D., Favier J.P., Bicanic D., Bein B. and Gerkema E., Thermal Diffusivity Measurement of Selected Metals, Technical Graphites and Magnetic Materials: Zero Crossing Points and Phase Methods Versus Photopyroelectric Technique- an Intercomparison Study,Prog.Nat. Sei. 6, S273-277 (1996). acceptedbyInstrumentation Science & Technology 4.2 Photopyroelectric Measurementof Thermal Diffusivity (-30to70°C)in LowDensity Polyethylene (LDPE)and Polyvinyl Chloride(PVC)Foils JanPaulFavier, DorinDadarlat, Klaas JanRiezebos, Cornelius VandenBerg,DaneBicanic andEdoGerkema Abstract The PPE technique in the standard configuration was used to measure the temperature dependence of thermal diffusivity of some low density polyethylene andpolyvinyl chloride foils used for food packaging purposes. The glassy phase transitions observed in these plasticswithin -30to70°Crange, causeda7-20%increaseofthermal diffusivity. Additional measurementsbydifferential scanningcalorimetry confirmed thevalidityofPPEresults. 84 chapter4.2 Introduction Due to a wide applicability of plastics, research on these materials constitutes an important topic for a number of years.1 Among these, low density polyethylene (LDPE) and Polyvinylchloride (PVC) are especially important because of their use for packaging applications (food and other products) and other purposes (hardware items, toys, etc.). The mostwidelystudiedaremechanical (tensileandyield strength, elongation, Young's modules, burst,impactandtearstrengths,stiffness, flex resistance,coefficient offriction, blocking)and other physical and chemical properties (optical features, permeability to gases and water, density,heatsealability,resistancetolight,heatandcold)andagingbehavior.2'3 Investigations concerning thermal parameters of plastics, and in particular of thermal diffusivity, conductivity and effusivity were rarely reported.3 The magnitude and the temperaturebehavior ofthesequantities haveagreat importance especially whenplasticsare produced and in applications where high and low temperatures are involved (i.e. packaging and storage offresh products). Moreover, the information provided by classical calorimetric techniques is usually "discontinuous", that is to say data are collected in increments of 1020°C.Consequently,thisimpliesapossibilityforfailing todetectananomalousbehavior.3 Inthispaperweproposeanewcalorimetricmethod,thephotopyroelectric (PPE)technique,to measure the temperature dependence of thermal diffusivity, a, for one LDPE and twoPVC foils all used as food packaging materials. The PPE method is capable of providing a continuousinformation ona withinthetemperaturerangeofinterest. Basically, thePPE method isconcerned with thedetection oftemperarure changes developed inanlightabsorbing samplewhenexposedtomodulated radiation. Thepyroelectric sensoris placed in a good thermal contact with the sample.4'5 The quantity measured in a PPE experiment is the photopyroelectric voltage (generated by the pyroelectric sensor), that has both amplitude and phaseboth depending onthethermal parameters ofthe sample. Various experimental configurations and PPE cell geometries wereproposed for calorimetric studies, and many types of materials (magnetics, ferroelectrics, superconductors, foodstuffs, raw agricultural and biological products) were investigated.6"9 Due to its ability to provide the temperature dependence of thermal parameters, the PPE method can also be used to detect phase and glass transitions as well as other anomalous behavior of materials.7"9 This paper reportstheresultsobtained in aPPE studyfrom LPDEandPVCfilmsand comparesthemto thoseobtainedbydifferential scanningcalorimetry(DSC). Theory Plasticsfilmsare optically opaque, constant thickness and flat what make suitable for PPE investigations. The most suitable PPE geometry is the standard (back) configuration with submittedtoPolymer Photopyroelectric (PPE) Measurement ofThermal Diffiisivity in LDPEand PVCfoils 85 thermally thick sample and sensor.6 In such scheme the front surface ofthe sample is exposed tothe modulated light, while the pyroelectric sensor (glued to the rear side of the sample with a coupling fluid) measures the temperature variation ofthe sample. Agiven medium is called "thermally thick" if its geometrical thickness, ls,is larger than the thermal diffusion length, u., defined as ^ (1) CO where cois the angular modulation frequency and a s is the thermal diffiisivity of the sample. The latter is related to other thermal parameters, thermal conductivity, KS, effusivity, Ss, and volume specific heat, Cs,through the relationships K =C s a s and e =yjC^ (2) Marinelli etal.6demonstrated that for optically opaque sample and thermally thick sensor and sample, (with the sensor working in a voltage or current mode), the amplitude and the phase ofthe complex PPE signal are describedby S=S 0 -1. V2a, ex — 8 (3) P + £s and I- J2a„ where S0 is a calibration factor depending on the radiation intensity, geometrical, electrical and thermal parameters of the sensor. Subscripts p and s in above equations refer to the pyroelectric sensor and sample, respectively. The phase of the PPE signal depends only on a s , allowing for its direct and absolute measurement (provided the sample thickness ls is known). The expected 9 versus Vf dependence islinear, and the slopemofthisplotis relatedtoa s through: lm ; Experiment The experimental procedure, the apparatus and PPE cell design were extensively discussed elsewhere;7'8 only the specific details for the experimental setup used in this study are given submitted to Polymer (5) 86 chapter4.2 here. The PPE cell is a cold finger and allows measurements both below and above room temperature. The radiation source was a diode laser (Melles Griot, 830 nm) electronically modulated by the internal oscillator of the lock-in amplifier (Stanford Research SR 850) which was used to process the signal from the detector (300 um thick single crystal of LiTa03).Thescanrateofthetemperaturewasabout0.5°Cmin ' withacquisitions each0.1°C. The operating temperature ranged from -30 to 70°C. The investigated samples included a LDPE(167umthick; Gent,Belgium)andtwoPVCfoils (160and90urnthick;4P,Forschein Germany) all manufactured for packaging practice. The irradiated foil area was about 20 mm2. The opacity of samples was achieved by blackening the 10 urn thick Al foil glued (with silicon grease) to the front surface of the sample. The same silicon grease was also used to obtainagoodthermalcontactbetweenthesample,sensorandthecoldfinger.Formodulation frequencies usedinthisexperimenttheAlfoil andthesilicongreaselayersarethermallyvery thin and their properties do not affect the measured data (for example, at 9Hz, the thermal diffusion lengthinAlisabout2 mm). Afrequency scanwasmadeat roomtemperature inorder(i)tocalibratethephaseofthePPE signal,and (ii)tofindtheoptimal range ofmodulation frequencies that satisfy the condition imposed by the special PPE case (thermally thick regime for the sensor and sample). The heating and cooling procedures were always performed in the same fashion, i.e. the sample was initially cooled down to -30°C and than heated to 70°C before cooled to room temperatureagain. APerkin Elmer DSC-2 equipped with computerized data acquisition and analysis was used for differential scanning calorimetry (DSC) experiments.1011 The reported curves were obtainedfromsamples consisting ofa stack ofsix layers in 60 ul sealed stainless steelcups. This was necessary to obtain sufficient heat effect. All DSC plots show thefirstand second heatingruns;thescanspeedwas10°Cmin"'. Resultsanddiscussion Thethickness ofthefoils was measured (with a micrometer) before the frequency scans and after completion ofthePPE measurements and nochangesduetotheheating were observed. Thefrequency scansbefore heating areshown inFigs. 1-3 (curve '+');the room temperature values of the thermal difiusivity are 6.7xl0"8, 3.2xl0"8, 5.5xl0"8 mV' for LDPE, PVC (90 um)andPVC(160 um),respectively.Thesevaluesagreeratherwellwithliteraturedata.312"15 submittedtoPolymer Photopyroclcctric(PPE)Measurement ofThermal Diffusivity inLDPEandPVCfoils 0.5 1 1.5 2 2.5 3 35 4 4.5 ^/Frequency (•vHz) Figure 1 The phase of the PPE signal obtained from 167 fxmthick LDPE foil plotted versus the square root of frequency. (+) before and (o) after heating •frequency IvHzj Figure 2 The phase of the PPE signal obtained from 90 ^.m thick PVC foil plotted versus the square root of frequency. (+) before and (o) after heating At the same time, Figs. 1-3, indicate that frequencies exceeding 3 Hz are suitable for temperature scans. Thefrequency of 9Hz was used for LDPE and PVC (160 \xm) samples, while 12HzwasusedforthethinnerPVC(90\im)foil. submittedtoPolymer 87 chapter 4.2 88 3.5 4 4.5 -^Frequency (•vHz) Figure3 The phase ofthe PPE signal obtained from 160 um thick PVC foil plotted versus the square rootof frequency. (+)before and(o)after heating LDPE 160 um The temperature dependence scan of the phase of the PPE signal, in -30 to 70°C range, is shown in Fig. 4. The initial heating produces a pronounced decrease of the phase for temperatures as high as 50 to 60°C. The material undergoes a relaxationjust before the onset of the melting transition, which implies melting of crystallites and molecular rearrangement followed byan increase ofthe phase. Transposing the temperature behavior of the phase into that of thermal diffusivity (Eq. (4)), one obtains results shown in Fig. 6. Apparently, heating affects the thermal diffusivity at room temperature. Thevaluefound by coolingthe sample, is in agreement with the value of 7.2x10" 8 m V found with thefrequency scan (Fig. 1,curve 'o'). This finding was also confirmed by results of DSC measurements, (Fig. 5), where a tiny exothermal event (dip at 41°C, Fig. 5) is followed by an endothermal (peak at 55°C, Fig. 5) enthalpy relaxation which preceeds the melting of crystalline parts of LDPE. The PPE measurement comprised the relaxation event but not the melting. Since the measured heat effects are relatively small, the slope of the baseline appears strongly negative due to magnification (this istruefor all DSC measurements). Depending on the amount of plasticizers, polyethylene generally has a glass-rubber transition around -110°C. However, the melting ofpolyethylene occurred overtens ofdegrees centigrade ending at 112°C. This is due to the melting of nanoscale crystallites, a process having a considerably decreased melting point, in succession of their size. Reported DSC measurements include the range of initial melting. submitted to Polymer Photopyroelectric (PPE)MeasurementofThermal Diffusivity inLDPEandPVCfoils 89 20 40 60 80 Temperature (°C) Figure4 Thephaseofphotopyroelectric signal plotted versustemperature, (+)heating and (o) cooling. Several runsweremadetoprovetheabsenceofchanges -0.3 -20 0 20 40 60 80 Temperature(°C) 100 Figure S DSC signals of LDPE 167 \im(13.48 mg) plotted versus temperature. The first scan shows theexoandendothermal peaks Followingtheinitial heating, successivecooling-heatingcycles(withPPE)exhibited thesame trend:i.e.notransitionwasobserved(Fig.4andconfirmed byDSC,Fig.5).ThesecondDSC tracewasrecorded after heating the sample upto 120°C (beyond the melting transition).No influence ofthescanspeedonDSCresultsfrom thissamplecouldbefound. Nochange inthermal propertiesasa result ofaging was observed either. APPE experiment performed with the same sample after one month yielded the same results as before. The relaxation event can mostprobablybeascribed tothefilmmanufacturing process and not to aging. submittedtoPolymer chapter4.2 90 Intercomparison ofDSCand PPE results showed (Figs.4-6) that the PPE technique is more sensitivethantheDSCmethodfor obtaining thermal properties. Only asingle layer offoil is needed here to obtain the result, as compared to six layers necessary to observe the thermal eventusingtheDSCmethod. 20 40 60 80 Temperature(°C) Figure6 Thermaldiffusivity versustemperature, for LDPE 167(xm; (+)heatingand(o)cooling PVC90um The glass-rubber transition and the melting point of PVC are near 90°C and 212°C respectively. Depending on the amount of plasticizer, the position of the glass transition appearstovaryover 10°Cormore.Thefirst observableweakening often occursalreadyabove 40°C. Reported DSC measurements cover the range up to (PVC 90 \xm) and well beyond (PVC 160ji.m) theglass-rubber transition. Thetemperaturebehavior ofthermal diffusivity for thePVCfoil (90 um)isshown in Fig.7, the shape resembles that ofLDPE.The room temperature value of3.8xl0"8 m V for a after heating ofPVCwas also confirmed bythefrequency scan made at later stage (Fig. 2, curve '°'). Duringtheinitial heatingtheDSCresults showenthalpy relaxation events linkedtotheglass transition (Fig. 8).Thedifferent peaksprobably originatefrom delayed thermal events inthe stack of different layers. The second heating run showed a small glass transition at 80°C (uncorrected for scan speed). The results obtained from the thick PVC film (160 um) were morepronounced. submittedtoPolymer Photopyroelectric (PPE) Measurement ofThermal Diffusivity in LDPE and PVC foils y 10 2.5 -40 -20 20 40 60 Temperature (°C) 80 Figure7 Thermal diffusivity versustemperature,for PVC90 um;(+)heatingand (o)cooling -0.4 20 40 60 80 Temperature (°C) 100 Figure8 The DSCsignal for PVC 90 um (18.77 mg)against temperature; results ofthefirstand second run. Thetraceassociated withthefirstrun showedthermal events PVC 160 (Am For this material, the initial heating suggests (Fig. 9)two minima concerning the temperature behavior of thermal diffusivity, the first at -15°C and the second at about 50°C. The room temperature value for a after the initial heating is in agreement with 6.7xl0"8 mV 1 found from frequency scan (Fig. 3, curve 'o'). The first anomaly was not observed in DSC experiments. The second anomaly at 50°C is associated with the onset of the relaxation prior to the glass transition as indicated by DSC measurements (Fig. 10). At subsequent coolingheating cycles, the temperatures at which anomalies took place appears shifted to higher submitted to Polymer 91 chapter4.2 92 values.Asfar asweareconcernedthemostimportant minimumfor packaging applicationsis thatcloseto0°Cexistingafter theinitial heatingcycle. xlO"8 8.5 t* 8 ""X Î.7.5 >> i .ts 7 ^^ > * ^65 ta \ • ••o 6 \ j — % - ^ ^ \^^r a*.* ^^^w v / • 4.5 -40 -20 0 20 40 60 Temperature(°C) 80 Figure 9 Thermal diffusivity versus temperature, for PVC 160 ^m; (+) heating and (o) cooling 20 40 60 80 100 120 Temperature (°C) Figure 10 The DSC signal ofPVC 160 (im (34.06 mg) plotted versus temperature (results of the first and second run). The first trace exhibit endothermal events, while the second trace shows a stable glass transition around 80°C Thethermal events observed in thefirstDSC heating run (Fig. 10)occur at the onset ofthe glass-rubber transition causedbyenthalpyrelaxation. Thethreesuccessivepeaksobservedare most probably due to a poor thermal contact within the six layers stack. After heating to temperatures beyond the glass-rubber transition, repeated DSC runs show a stable glass submittedtoPolymer Photopyroelectric(PPE)MeasurementofThermalDiffusivity inLDPEandPVCfoils 93 transition at75°C(correctedfor scanspeed4.5°Cat 10min).Theglasstransition regionwith theonset at69°C isspread over 12°C.Noinfluence ofagingonthe glasstransition couldbe found, evenafter 10weeks. Sincetheglasstransition observedat 75°Cisconsiderably lower thantheexpected90°Cfor purePVC,theinvestigated foil probably contained an appreciable amountofplasticizer. Conclusions The PPE method was used to study the temperature dependence (-30 to 70°C) of thermal diffusivity of selected plastic foils. The latter was found dependent on thermal history of sample. After the initial heating (up to 70°C) the thermal diffusivities of foils were significantly higher (7-20%)andremainedconstant during successiveheating-coolingcycles. A possible explanation for this observation are structural changes (i.e. glass transitions) taking place in the material during thermal annealing. The thermal diffusivity values found forLDPEandPVCarecomparabletodatafound inliterature. Aninterestingobservationwastheappearanceofaminimuminthethermaldiffusivity forthe 160urnthickPVCfoil at lowtemperatures. Althoughnotverypronounced,theoccurrenceof thisminimumcanbebeneficialwhenusingthematerialforpackagingapplications. ThePPE methodwasshown morepractical and sensitive than the standard DSC.Except for the low temperature thermal event in foils the results of PPE measurements were generally confirmed by DSC method. An explanation for this phenomena could be the freezing of capillarywaterinthematerial. One final remark concerns the use of Eq. (3). If one is interested to obtain values of all thermalparameters(C, Kors), thePPEsignalcanbecalibratedwithathermalparameter (at room temperature); the complete temperature behavior can then be calculated. It was demonstrated7 thattheinverse (front) PPEconfiguration withthermallythin sensor andthick sample is a useful approach to obtain thermal effusivity of a sample at room temperature. Consequently, bycombining thesetwoconfigurations, the PPE methodbecomes independent ofany other method or literature data. Unfortunately, this approach cannot be used for foils becauseoftheir smallthickness. References 1 Bershtein VA. andEgorovV.M.,Differential ScanningCalorimetry ofPolymers,Physics, Chemistry,Analysis,Technology,EllisHorwood,WestSussex(1994) 2 Bakker M. and Eckroth D., The Wiley Encyclopedia of Packaging Technology, Wiley& SonsNewYork(1986) 3 BrandrupJ.andImmergutE.H.,PolymerHandbook 3rded.,Wiley,NewYork(1989) submittedtoPolymer 9A chapter 4.2 4 Chirtoc M. and Mihailescu G., Theory of the Photopyroelectric Method for Investigation ofOptical and Thermal Materials Properties,Phys. Rev. B 40, 9606-9617 (1989) 5 Mandelis A. and Zver M.M., Theory of Photopyroelectric Spectroscopy of Solids, J Appl Phys 57, 4421-4430 (1985) 6 Marinelli M., Mercuri F., Zammit U., Pizzoferrato R., Scudieri F. and Dadarlat D. Photopyroelectric Study of Specific Heat, Thermal Conductivity and Thermal Diffusivity of Cr 2 0 3 at the Neel Transition, Phys.Rev. B 49, 9523-9537 (1994) 7 Dadarlat D., Bicanic D., Visser H., Mercuri F. and Frandas A., Photopyroelectric Method for Determination of Thermophysical Parameters and Detection of Phase Transitions in Fatty Acids and Triglycerides. Part I: Principles, Theory and Instrumentational Concepts, J.Amer. Oil Chem. Soc. 72, 273-279 (1995) 8 Dadarlat D., Bicanic D., Visser H., Mercuri F. and Frandas A., Photopyroelectric Method for Determination of Thermophysical Parameters and Detection of Phase Transitions in Fatty Acids and Triglycerides. Part II: Temperature Dependence of Thermophysical Parameters,J.Amer. Oil Chem. Soc. 72,281-287 (1995) 9 Dadarlat D., Bicanic D., Gibkes J., Kloek W., Dries I. van den and Gerkema E., Study of Melting Processes in Fatty Acids and Oils Mixtures: A Comparison of Photopyroelectric (PPE) and Photopyroelectric Observations of Melting Differential Scanning Calorimetry (DSC), Chem.Phys. Lipids 82, 115-123 (1996) 10 Höhne G.W.H., Hemminger W. and Flammersheim H.-J., Differential Scanning Calorimetry: An Introduction for Practitioners, Springer-Verlag, Berlin (1996) 11 Hatley R.H.M., Franks F. and Green M., A Novel Data Acquisition, Retention and Examination System (Dares) for Differential Scanning Calorimetry, Thermochimica Acta 156, 247-252 (1989) 12 Leite N.F., Cella N., Vargas H. and Miranda L.M.C., Photoacoustic Measurement of Thermal Diffusivity ofPolymer Foils,J.Appl. Phys. 68, 3025-3027 (1987) 13 Rantala J., Wei L., Kuo PK., Jaarinen J. and Thomas R.L., Determination of Thermal Diffusivity of Low-Diffusivity Materials Using the Mirage Method with Multiparameter Fitting,J. Appl. Phys. 73,2714-2723 (1993) 14 Merté B.,Korpiun P.,Lüscher E., and Tilgner R., Thermal Diffusivity of Polymer Foils as a -Semicrystalline PETP and PE- as a Function of Drawings, J. de Physique 6 suppl. Colloq. C6, C6 463-467 (1983) 15 Mansanares A.M., Vargas H , Galembeck F., Buijs J. and Bicanic D., Photoacoustic Characterization ofa Two-Layer System,J.Appl. Phys. 70, 7046-7050 (1991). submitted to Polymer Summary,ConclusionandOutlook Arapidlyincreasingnumberofphotothermal (PT)techniqueshashada considerable impact onagriculture and environmental sciences inthe last decade. Itwasthepurpose ofthework describedheretodevelopandapplynewPTtechniquesinthisspecificfieldofresearch. Chapter 1 isa general introduction with anoverview of PTtechniques used inthisresearch. Twodifferent photoacoustic (PA)techniques usedfor optical characterization ofavarietyof condensed phase samples are discussed in chapters 2 and 3. The possibilities for thermal characterization ofsamplesaredescribedinchapter4. Inchapter2classicalPAspectroscopy withmicrophone detection wasusedto obtain spectra inthevisibleregion(350-700nm)ofpowdered (lightscattering)food samplessuchasflours, coffee and spices (chapter 2.1). Theoutcomeofthese experiments suggest thefeasibility of PA spectroscopy for quality control in the food-processing industry. Another PA cell was designed (chapter 2.2) and used in the IR region (10 (j.m). The final PA experiment was concerned with study ofvarious carboxylic acids, alcohols and alkanes at 3.39 ^m (chapter 2.3). Thefeasibility of optothermal window (OW) method, an elegant approach to determine the optical absorption coefficient of condensed phase samples is described in chapter 3. The method was not only extended to 9-11 ^mbut also proved capable of investigating opaque samples (liquids and gel) which are otherwise not amenable to conventional IR spectroscopies. The content of trans fatty acids in several margarine samples was measured with the OW technique and itsperformance compared to that of GLC, GLC + TLC and FTIR. The data obtained withthe different methodsweregenerally inagoodagreement. The improvedOW cell resulting insubstantial reduction ofthebackground signal,wasthen used to study extra virgin olive oil (chapter 3.2) adulterated by known adulterants (sunflower oil (4.5%) and safflower oil(6%)).Theachieved limitofdetection (LOD)wascomparabletothosereported inliteratureforothertechniques(FTIR-ATR,GLC,HPLCandmassspectroscopy). Many biological samples contain water that itself exhibits a strong absorption in IR. Quantitative measurements on such specimens are all but trivial. The new OW sensorwas showncapableofdirectandquantitativemeasurements (chapter3.3 and 3.4) oflactose,corn starch and sulfate inwater. Theperformance ofthe OWmethod was slightly inferior tothat of FTIR. On the other hand, unlike the OW method, FTIR-ATR could only provide quantitative resultsfor corn starch samples. Asto the study of sulfate in water, the limit of detection (1 mmolL"1)achieved with the OW method is one order of magnitude better than thatofATR. 98 Summary In the fourth and last chapter, two different PT techniques (photopyroelectric method and photothermal beam deflection) were used for thermal characterization of a candy (a model for a glassy sugar system) and different packaging materials. The thermal diffusivity of a candy at room temperature found by PPE and PTBD was 14.7 xlO"8 and 12.0 xlO8 mV' respectively. In addition, the PPE technique in the standard configuration, was used to measuretemperaturedependenceofthethermal diffusivity inthe -30°Cto 70°Ctemperature range. The glass-rubber transition, underwent by the sample in this temperature range was observedandcomparedtotheresultsobtainedbydifferential scanningcalorimetry(DSC). ThePPEtechniquewasalso used to obtain thetemperature behaviour of thermal diffusivity of low densitypolyethylene andPolyvinylchloride foils (used for packagingpurposes). Such behavior was found dependent on the thermal history of the sample. The "untreated" foils exhibit values for thermal diffusivity that were consistently lower (7-20%) than those obtained for the same samples when heated to 70°C. The increase in thermal diffusivity is associated with structural changes (i.e. glass transitions) taking place in the material during thermal annealing. Additional measurements by differential scanning calorimetry (DSC) confirmed the validity of PPE results. It was shown that the PPE method is more sensitive thanthestandardDSCindetectingchangesinthermalparameters. Finally, the PPE method allows one in principle to obtain the temperature behavior of all static and dynamic thermal parameters provided one of the remaining thermal parameters (thermal conductivety, thermal effusivity and volume specific heat) is available at a given temperature. Outlook Results of a research described in this thesis show the feasibility of PT techniques for applications to a widerange of condensed phase samples.The methods usedhere constitute only a part of techniques developed and used in our laboratory. The low-cost and compact OW device is easy to handle and moreover offers the possibility for on-line studies of optically opaque and thermally thick samples that are normally not accessible by other techniques. It is anticipated that development of infrared diode lasers will increase the potential of the OW method because desired wavelengths characterized by the highest spectral contrast will become available. The sensitivity of the OW method was shown comparabletothatofFTIRandtherefore additionaldevelopmentsmighteventually makethe OW approach a candidate technique for quantitative analysis throughout the entire infrared region. Theanalyticalpotential ofPTschemesbecomesmoreobviouswhenthey areusedas detectors in combination with separation techniques such as GLC, HPLC or capillary electrophoresis. Summary The potential of photopyroelectric technique was demonstrated by obtaining thermal diffusivity values for different kind of samples. ThePPE method is able to provide data on agingeffect, structural stability andcrystalinepolymorphism. Thetechnique isfast, sensitive and reproducible, requires small amounts of sample for analysis while providing more information than existing, classical methods used currently in thermal research. The only drawback ofPPEisthenecessityfor goodthermal contactbetween sampleandsensor.When oneisinterested to obtainthermal diffusivity data in anon-contact mannerthephotothermal beamdeflection (zero crossing method) isavaluabletool.Accurate (5%error)determination of thermal diffusivity is achievable due to intrinsically low errors in measurement of frequency andofzerocrossingposition. In conclusion, based ontheresults ofthework described here, onecananticipate that inthe yearsto comethePTmethods, aloneorcombined with existing techniques, will most likely play a more important role for variety of applications in agricultural and environmental sciences. 99 Samenvatting Hethierbeschrevenpromotieonderzoek gaat overnieuwe ontwikkelingen op het gebiedvan fotothermische meetechnieken. Demeestbekende isdefotoakoestische spectroscopie,welke doorA.G.Bellisbeschrevenin 1880.Hijtransporteerdegeluidmetbehulpvanlichtovereen afstand van 213 meter; hij was daarmee zijn tijd ver vooruit! Deze techniek werd een wetenschappelijke curiositeit, dieeen snelledoodstierf, maareenrenaissancebeleefde inde jaren 70.Deverbeterde elektronica endeintroductievan de laser droegenbij tot denieuwe ontwikkelingen. Defeitelijk doorbraak wasdeRosencwaig-Gersho-theorie, diehet mogelijk maakteomfotothermische signalenopeeneenvoudigemanierteinterpreteren. Het basisprincipe van de fotoakoestiek berust op een omzetting van licht in geluid. Moleculen die lichtenergie absorberen komen in een aangeslagen toestand en kunnen terugkeren naar hunbegintoestand door energie, in dit gevalwarmte aan de omgeving af te staan.Erzallokaal eenkleineopwarmingzijn (<0.001°C!!!!!!!), enalsmenhetlichtmeteen bepaaldefrequentie moduleert (aan-uit-aan-uit), zaldeabsorptie enopwarming met dezelfde frequentie plaatsvinden. Als een medium opwarmt zal dit uitzetten, met als gevolg een volumevergroting,watbeschouwtkanwordeneendrukgolf,welkegemetenkanwordenmet een microfoon. Bellliethet lichtdooreenroterendeschijf metgatenvallenopeenmonster(gas,vloeistofof vaste stof). Het geluid dat Bell met een stethoscoop hoorde, had dezelfde frequentie als wanneerhetzonlichtdoordegaatjes indeschijf ophetmonster scheen. Ookvondhij datde signaalsterke afhankelijk was van de lichtsterkte en de absorptie van het licht door het monster. Eenanderfotothermisch effect, dat iedereenkent, ishet"mirage effect": dezinderingen, die je bij mooi heet zomerweerboven het asfalt ziet. Doordewarmte is debrekingsindex niet constant,waardoorlichtstralen nietrechtdoorgaan,maarwordenafgebogen. Detechniek die gebruikt maakt van het mirage effect heet fotothermische deflectie en maakt gebruik van tweelasers.Deeerstelaserzorgtvooropwarming endeverandering indebrekingsindex, de tweedelasertastditgebiedje afenwordtafgebogen. Omditnauwkeurigtemetenisgoedeen gevoelige elektronicanodig. Deze fotothermische technieken verschillen nogal van de conventionele transmissie spectroscopie.Bijdetraditionelemethodevalteenbundellichtdooreenmonster,waarnaeen detectormeetwelkdeelvanhet licht erdoorkomt. Ditwordtvergeleken met een referentie, zodat de absorptie van het monster kan worden berekend. Als nu al het licht wordt geabsorbeerd, meet de detector niets en kan er dus geen uitspraak worden gedaan over de absorptie. De oplossing is dan om de meetcel kleiner te maken, zodat het licht er nog wel 102 Samenvatting doorheen kan. Dit levert uiteraard problemen op als de cel te klein wordt. Wanneer het monsterweinigabsorbeert, geeft ditookeenprobleemvoordedetector:omdaterbijna 100% van het licht doorheen komt, is het niet te onderscheiden van de referentie. Sterk lichtverstrooiende monsters als poeders en suspensies lenen zich ook niet voor transmissie spectroscopie, omdat al het licht wordt gereflecteerd, zodat het de detector, die achter het monsterstaat,nietkanbereiken.Fotothermische spectroscopiekentaldezeproblemenniet. Debedoelingvan dit onderzoek washet ontwikkelen en toepassen van deze fotothermische techniekenopzogezegd"moeilijke" monsters,die,zoalshetwoordalzegt, moeilijk temeten zijn mettraditioneleapparatuur. Fotothermische technieken voordebepaling vanoptische eigenschappen van materialen Klassieke fotoakoestiek is gebruikt om optische verschillen aan materialen te meten. Voor veellevensmiddelen isdekleureenbelangrijkekwaliteitsfactor. Daaromzijn inhet zichtbare gebiedvan het spectrum verschillende soorten meel gemeten.Niet alleen dekleur (wit, geel of groen)van het meel kon gemeten worden, maar ook waren erverschuilen te zien tussen fijn enminderfijn gemalenmeel. Deoptischeverschillen zijn meestal klein enmoeilijk te meten inhet zichtbare deel van het spectrum.Daaromisereennieuwefotoakoestische celgebouwddiehetmogelijk maakteom inhet infrarood te meten. Waarom infrarood? In theorie is infrarood spectroscopie deideale methode om de samenstelling van monsters te meten omdat alle biologische en organische verbinding een karakteristiek infraroodabsorptie spectrum hebben, dat terug te voeren is op demoleculaire struktuur. Deze"vingerafdrukken" vanverbindingen maken het mogelijk om zevanelkaarteonderscheiden enonafhankelijk vanelkaartemeten. Eennieuwefotoakoestische techniek, het "optothermische venster", isontwikkeld om inhet infrarood temeten.Hiermeekannietalleendematevan absorptie wordenbepaald maarook opeeneenvoudige manierdeabsorptiecoëfficiënt. Hetmeetprincipe isvergelijkbaar metdie van fotoakoestiek. De optothermische cel bestaat uit een infrarood transparant venster met daarop een kristal dat een stroom afgeeft op het moment dat daar warmte of druk opkomt. Hetinfrarode lichtgaatdoorhetvensterwaarophetmonsterligtendatdestraling absorbeert. De warmte die in het monster ontstaat diffundeert naar het venster waardoor dit een beetje uitzet.Deakoestischegolfdiedanontstaatwordtdangemetenm.b.v.hetkristal. Zoals hiervoor staat geschreven heeft iedere verbinding zijn eigen vingerafdruk in het infrarood. Doordejuiste absorptiepiek(en) tekiezen,kanhetoptothermisch venster gebruikt worden omde legiotoepassingen, die erzijn, te meten:deconcentratievan transvetzuren in margarine,deechtheidvanolijfolie endeconcentratievanlactose,maiszetmeel ensulfaat in Samenvatting 103 water. Denieuwetechniek isvergeleken metandere moderne meettechnieken enbleekvaak net zo gevoelig. Het grote voordeel van deze sensor is dat hij relatief goedkoop is en makkelijk tebedienen enschoontemaken. Fotothermische technieken voorde bepalingvanthermische eigenschappen vanmaterialen Fotothermische technieken kunnen, behalvevoor debepalingvan optische kwaliteiten, ook heel goed gebruikt worden om thermische eigenschappen van materialen te bepalen. In het laatste deel van dit onderzoek zijn de fotopyroelektrische methode en de fotothermische deflectie techniek gebruikt om de warmtevereffeningscoêfficiënt te berekenen. Deze parameterbepaalt hoe makkelijk warmte zichverdeelt in een monster (bijvoorbeeld demaat voor de snelheid waarmee een ei in kokend water de temperatuur van het water heeft gekregen). De fotopyroelektrische methode kan de warmtevereffeningscoêfficiënt bij verschillende temperaturen (-40 tot 80 °C) meten. Het is niet alleen interessant om de vereffeningscoëfficiënt te metenbij verschillende temperaturen, maarop deze manier ishet ookmogelijk fase-overgangen tebestuderen,bijvoorbeeld hetsmeltgedragvanvetten.Hieris deze techniek gebruikt om naar een glasovergang te kijken in een snoepje, wat als model staatvooreensuikersysteem. Momenteelwordtveelonderzoekverricht naar glasovergangen in levensmiddelen, omdat die van groot belang zijn voor de houdbaarheid. Met de fotopyroelektische methode kon de glasovergang goed gemeten worden. De resultaten zijn vergeleken en bevestigd met een conventionele methode, namelijk differentiële scanning calorimetrie. Als laatste is met deze twee methoden gekeken naar plastics: PVC en LDPE. Deze folies worden gebruikt voor verpakkingsmateriaal en ook hier is het van belang te meten wat de thermischeparameters zijn bij verschillende temperaturen. Bij deze metingen was er echter een groot verschil tussen de twee gebruikte technieken. De fotopyroelektrische methode bleekveelgevoeliger ennauwkeurigerdandedifferentiële scanningcalorimetrie. Dankwoord KersversindienstalsAIOkonikgaaninpakken;wemoestenverhuizenvanDuivendaalnaar hetnieuweschipopdeDreijen.Hetgingallemaal nietsoepelmaardatkwamvooraldoordat het gebouw nog niet af was en het er lawaaierig was.Een ander probleem was dat het nog negen maanden zou duren voor het lab klaar zou zijn en we de 'boel' konden gaan schoonmaken enopbouwen. Debibliotheek waseentoevluchtsoord; ikhebnognooit zoveel gelezenalstoen.Uiteindelijk lagenweineenveiligehaven, medeomdatweomgevenwaren door'schepen' metgelijkelading. Indeafgelopenjaren genootikvandegrotevrijheid dieikkreegenvandecongressen dieik konbezoekeninChina, VS,Guadeloupe, SloveniëenDuitsland.Hetwaseengoedetijd maar daarentegenmisteikwelmedeAIO'sdievooreenkritischemassakondenzorgen. Tijdens mijn promotieonderzoek heb ik regelmatig het gevoel gehad: waar doe je het eigenlijk allemaalvoor.Watikwelhebervarenisdatje hetnietalleendoet.Een grootaantal mensen zijn direct of, minstens zo belangrijk, indirect betrokken geweest bij het totstandkomenvandit promotieboekje. AllereerstwilikDanebedanken.Hij stondaltijd voormeklaarenhadeindeloosveel ideeën. Daarnaast ken ik geen begeleider die een manuscript de volgende dag gecorrigeerd teruggeeft. Voor zijn snelheid en vechtlust ('nooit opgeven') heb ik dan ook grote bewondering. Als een Don Quichot vecht hij tegen 'the monkeys' voor het behoud van natuurkundigonderzoekbinnenWageningen. Professor dr. J. Reuss wil ikbedanken voor de inhoudelijke discussies die we aan het eind gevoerd hebben en welke een duidelijke bijdrage hebben geleverd in het uiteindelijke resultaat. Omonderzoek te verrichten hebje 'spullen' nodig en dan zijn werkplaatsen onontbeerlijk. Kees, niet alleen jouw kennis, maar ook jouw inzicht in de problematiek, speelden een doorslaggevende rol in het onderzoek. Ook de medewerkers van de fijn-mechanische werkplaats endeglasblazerij zijnvanonschatbarewaardegeweestvoorhetonderzoek. Indenatuurkundegang gebeurde veel, maar ik denk dat het koffie drinken het belangrijkste ritueel was.Develebuitenlandsegastmedewerkersbrachtennaastgezelligheid inhetlabook nieuwe kennis, die de productiviteit van het werk ten goede kwam. My word of thanks to Stephen,Nicu,Mladen,Mihai,Jürgen,Andres,Thomas,ÓttoandAngela. Eenbelangrijke rolwasookweggelegd voor clubjes als 'de Jeugd', 'UnitasLunch Groepje', 'Rhine Town Tigers' en 'de Moleculairs'. Zij waren vooral belangrijk voor de mentale opvang en de resetvan de geest. Hockey is een gezelligheidssport en met RTT hebben we een grote reputatie opgebouwd. De Moleculairs hebben een speciaal plekje, omdat zij als geenanderwetenwatditwerkinhoudt! Janine ik wilje hier bedanken voor alles watjij voor mij betekend hebt en zult betekenen. KoosenGusta,jullieinteresseenmeningzijnaltijd ergbelangrijk voormijgeweest. CurriculumVitae JanPaulFavierwerdop6juli 1967geboren teLeiden. In 1985behaaldehij het eindexamen Atheneum aan de Samenwerkingsschool te Waddinxveen. Aansluitend startte hij de studie moleculairewetenschappenaandetoenmaligeLandbouwhogeschool inWageningen. Tijdens de doctoraalstudie volgde hij de fysisch-chemische oriëntatie en koos als hoofdvakken: fysische- en colloïdchemie en natuurkunde. De stage periode werd doorgebracht aan het isotopen instituut van de Hongaarse academie van wetenschappen, Budapest, Hongarije. In 1991 behaaldehijhetdoctoraalexamen. Vervolgenstradhij in 1992indienst,alsassistent inopleiding,bij hetfotoakoestisch laserlab bij de vakgroep agrotechniek en -fysica van de landbouwuniversiteit te Wageningen. De resultaten van dit onderzoek hebben tot dit proefschrift geleid. Tijdens zijn promotieonderzoek heeft hij 2 maanden gewerkt bij Department of Physical Chemistry, Cambridge University inEngeland. Listof Publications FavierJ.P.,DadarlatD.,GibkesJ.,VandenBergC, BicanicD.,Thermaldiffusivity ofhard boiledcandyObtainedbyPhotothermalBeamDeflection andStandard Photopyroelectric Method,accepted inInstrum. Sei. Technol. (1997) FavierJ.P.,BicanicD.,IerselM.van,HelanderP.,Theoptothermalapproachtoarealtime monitoringofglucosecontentduringfermentation bybrewers' yeast,acceptedin/. Biochem. Biophys. Methods (1997) FavierJ.P.,BicanicD.,DokaO.,ChirtocM.,HelanderP.,Optothermaldetectionof infrared radiation-induced absorptioninaqueoussolutionsofcarbohydrates:lactoseandcorn starch,J.ofAgriculture & FoodChemistry45,777-780(1997) BicanicD.,FrancoM.,GibkesJ.,GerkemaE.,FavierJ.P.andJalinkH.,Applicationsof photoacousticandphotothermalnon-contactmethodsinselectedareasofenvironmental andagriculturalsciences,Chapter5,inLifeandEarthSciences"Progressin PhotoacousticandPhotothermalScienceIII",A.Mandelis(ed.),SPIEOptical Engineering Press,Washington DC.(1997) FavierJ.P.,IerselM.van,DokaO.,Koenderman I.,SchoutenF.,AsseltK,van,BicanicD. andHelanderP., Applicationsofauniversaloptothermalwindowdetectorforsamplesof agriculturalpurpose,Progress inNaturalSciences6,S562-564(1996) GibkesJ.,DadarlatD.,FavierJ.P.,BicanicD.,BeinB.andGerkemaE.,Thermal diffusivity measurementofselectedmetals,technicalgraphitesandmagneticmaterials:zerocrossing pointsandphasemethodsversusphotopyroelectrictechnique-an intercomparison study, ProgressNaturalSciences6,S273-277(1996) FavierJ.P.,BicanicD.,ChirtocM.,HelanderP.,Compact,openandgeneralpurposecellof variableeffective pathlength:directabsorptionmeasurementsofS042"inwater, Fresenius J. Analytical Chemistry355,357-358(1996) FavierJ.P.,BicanicD.,BovenkampP.vande,ChirtocM.,HelanderP.,Determinationof totaltransfatty acidscontentinmargarine:anintercomparison studyGLC,GLC+TLC, FTIRandoptothermalwindow(openphotoacousticcell),Analytical Chemistry68,729733(1996) BicanicD.,ChirtocM„ChirtocI.,FavierJ.P.,HelanderP.,Photothermal determinationof absorptioncoefficients inopticallydensefluids: application tooleicacidandwateratco laserwavelengths,AppliedSpectroscopy49,1485-1489(1995) BicanicD,DadarlatD.,Gibkes,J,ChirtocM.,FavierJ.P.,GerkemaE.,Thephotopyroelectric approachtothermalcharacterizationofliquidandpastyfoodstuffs andonoptothermal HO accessory for obtaining infrared spectra of optically danse fluids,Acta Chimica Slovenica, 42, 153-173 (1995) Bicanic D.,Chirtoc M., Chirtoc I., Veldhuizen B.van,Favier J.P., Helander P.,New technique for measuring absorption coefficients of strongly absorbing liquids: optothermal study of sunflower oil,oleicacid and itschloroform solutionsat3.39 microns, Spectroscopy Letters 28, 101-110 (1995) Favier J.P., Bicanic D., Asselt K. van, Miklós A.,Organic compounds measured with infrared (3.39 Mm) photoacoustics,/. de Physique IV 4,495-497 (1994) Bicanic D, ChirtocM, Chirtoc I,Vanveldhuizen B.,Favier J.P., Helander P., Harryvan J., Cozijnsen J„ LubbersM., New approach in determining absorption coefficients of strongly absorbing liquids: optothermal studies sunflower oil,oleicacid (and its chloroform solutions) at 3.39 microns,J.de Physique TV4,487-490 (1994) FavierJ.P.,Buijs J., Miklós A.,Lorincz A.,BicanicD:Photoacousticcharacterization of different food samples,Zeitschriftfur Lebensm.-l]ntersuchung und-Forschung 199, 59-63, (1994) Favier J.P., Miklós A.and BicanicD.,New and versatilephotoacoustic for studies of powdered specimens across broad spectral range,Acta Chimica Slovenica, 40, 115-122 (1993) Buijs J., Favier J.P., DokaO. and Miklós A., Photoacoustic signal generation and spectra of powders, Springer Series inOptical Sciences Vol69, Photoacoustic and Photothermal PhenomenaIII,page 168-170, editor D.Bicanic,Springer-Verlag Berlin, Heidelberg, New York (1992) Norde W. and Favier J.P., Structure of adsorbed and desorbed proteins, Colloidand Surfaces 64, 87-93 (1992) Strauss E.,Favier J.P.,Bicanic D.,Asselt K. van,Lubbers M„ Sensitive colorimetric determination of ammonium ion inwaterby laserphotothermal detection, The Analyst, 116, 77-79 (1991) BicanicD.,Favier J.P., StraussE.,Lubbers M. andFleuren G., Low-cost colorimetric measurement of phosphate tracelevels in waterand soilsolutionsby the collinear photothermal beam deflection method, Internat. J. Environ. Anal. Chem., 38, 623-628 (1990) Bicanic D., Strauss E.,Favier J.P., Miklós A., Torfs P.and Lubbers M., Photothermal colorimetry of phosphorus and ammonium in aqueous phase for use in the agricultural practice, Springer Series in Optical Sciences,Vol 62,Photoacoustic and Photothermal Phenomena II,445-447, editors:J.C. Murphy e.a., Springer-Verlag, Heidelberg (1990) View publication stats