Home & Garden

An inexpensive, portable and microcontrolled near infrared LED-photometer for screening analysis of gasoline

An inexpensive, portable and microcontrolled near infrared LED-photometer for screening analysis of gasoline
of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
   Available online at www.sciencedirect.com Talanta 75 (2008) 792–796 An inexpensive, portable and microcontrolled near infraredLED-photometer for screening analysis of gasoline Edvaldo da N´obrega Gai˜ao, S´ergio Ricardo Bezerra dos Santos, Vagner Bezerra dos Santos,Elaine Cristina Lima do Nascimento, Renato Sousa Lima, M´ario C´esar Ugulino de Ara´ujo ∗ Universidade Federal da Para´ıba, CCEN, Departamento de Qu´ımica, Caixa Postal 5093, 58051-970 Jo˜ ao Pessoa, PB, Brazil Received 4 September 2007; received in revised form 13 November 2007; accepted 8 December 2007Available online 23 December 2007 Abstract Amicrocontrolled,portableandinexpensivephotometerisproposed.Itusesanearinfraredlightemittingdiode(NIRLED)asradiationsource,aPbSe photoresistor as infrared detector and a programmable interrupt controller (PIC) microcontroller as control unit. The detector system presentsa thermoresistor and a thermoelectric cooling to control the detector temperature and keep the noise at low levels. The microcontroller incorporatedtotal autonomy on the proposed photometer. As its components are inexpensive and of easy acquisition, the proposed NIR LED-photometer is aneconomical alternative for chemical analyses in small routine, research and/or teaching laboratories. By being portable and microcontrolled, it alsoallows carrying out field chemical analyses. The instrument was successfully applied on the screening analysis to verify adulteration in gasolinesamples.© 2007 Elsevier B.V. All rights reserved. Keywords:  Near infrared; Light emitting diode; Photometer; Microcontroller; Screening analysis; Gasoline 1. Introduction Thelightemittingdiode(LED)[1]isasemiconductordevice usually adapted on portable photometers (LED-photometers) asradiation source because it confers some advantages for opticalinstruments[1,2]assimplicity,easyoperation,lightstabilityand low power and cost.LED-photometers are usually applied as dedicated instru-ments since ordinary LEDs emit only a fairly narrow bandradiation with half maximum intensity of about 20–50nm [1,2]. When spectrophotometers are constructed for general purpose,complexconfigurationstoadaptahighernumberofLEDs[3,4],interference filters [5,6] or grating monochromators [7] must be adapted to select the desired wavelengths. These are modifica-tions that naturally increase the complexity and consequentlythe cost of the instrument.Since the beginning, LEDs for analytical purposes arebasically emitters of visible radiation [8] and most of the LED- photometersareappliedonspectrometricdeterminationsinflow ∗ Corresponding author. Tel.: +55 83 216 7438; fax: +55 83 216 7437/7117.  E-mail address:  laqa@quimica.ufpb.br (M.C.U. de Ara´ujo). injection analysis, area where they are widely used [9–15]. On theotherhand,thesortofinfrared(IR)LEDemitterswaslimited[1,2] and so was its analytical applications. Advances in the IR LED construction enabled to develop good near infrared (NIR)LEDs[2]f oranalyticalpurposes.Also,chemometrictechniques [16] allowed to apply NIR technology [17] to analyze very diverse materials [18–22].Although the application of analytical procedures based onNIR methods have increased in recent years [23,24], NIR LED- photometers developed and commercialized based on screeningphilosophy were not elaborated. The screening analysis [25]is a qualitative procedure based on binary answers like pos-itive/negative or yes/no useful to take timely decisions foridentification of measurements or sample classification. In thescreening analysis, for example, only the samples that surpassan established threshold, for example a cut-off concentration,are directed for a deeper analytical procedure. As advantages,fast analytical assays and minimization of cost are achieved.In this work was constructed a portable and microcontrolledNIR LED-photometer (NLP) applied on the screening analy-sis of gasoline type C, a Brazilian blended gasoline having 25%(v/v)ethanol,aimingattoverifyitsconformityinregardtosomeofitsprincipaladulterants(solventsandkerosene)inBrazil.The 0039-9140/$ – see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.talanta.2007.12.014   E.d.N. Gai˜ ao et al. / Talanta 75 (2008) 792–796   793 NLPwasdevelopedtopresenttotalautonomyandalltheadvan-tageouscharacteristicsoftheLED-photometers.Asdetectorwasused a PbSe photoresistor equipped with a temperature sensorand refrigerated by a thermoelectric cooler. The detector systemkeptthenoiseatverylowlevelspermittingtocarryoutverypre-cise measurements. With the microcontroller unit was possibleto control external devices, perform analytical data acquisitionand display the results on a liquid crystal display (LCD). 2. Experimental 2.1. Sample, solvents and kerosenes Thirty-five samples of pure gasoline, 10 of solvent and 10 of kerosene,allofdifferentmanufacturerswerepurchased,respec-tively, from gas stations and markets at Jo˜ao Pessoa and Bayeuxcities, state of Paraiba, Brazil.Twenty-five samples of gasoline were used as blank and 10were used to prepare adulterated gasolines. Thus, solvent adul-terated gasolines having a adulterant concentration of 2.0, 3.0,3.5, 4.0, 4.5, 5.0, 5.5 and 6.0% (v/v) were prepared by spik-ing the respective volume of solvent into pure gasoline samples.Also, kerosene adulterated gasolines were prepared by addingthe respective volume of kerosene into pure gasoline in orderto obtain 2.0, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0 and 7.0% (v/v) of adulterant. For each adulteration level were prepared 10 sam-ples. 2.2. NIR spectra A FT-NIR/MIR spectrometer PerkinElmer, model GX,equipped with a quartz cell presenting an optical path lengthof 1.0cm was used to register the NIR spectra of the gasolinesamples.Aspectralresolutionof4cm − 1 and16scanswereused.TheNIRregionintherange1000–1700nmwasadoptedforthisstudy. The absorbance values for wavelengths above 1700nm,which are associated to the first overtone of OH and CH vibra-tional transitions, are very intense and lead to saturation of thedetector. 2.3. The NLP The observation of the NIR spectra (Fig. 1) allowed defining the spectral region around 1550nm as the most adequate foridentification of the gasoline adulteration. As showed in Fig. 1,clearly there exist very different response intensities betweenadulterants and pure gasolines which are not so expressive inother regions and that enable an accurate distinction betweenthese materials. At 1550nm, the absorbance intensities fromgasoline samples are lower than that from solvent but higherthan that from kerosene. Thus, it is expected an increasing of the absorbance intensities when solvent is added in the gasolinesamples and a decreasing when the adulteration is carried outby kerosene.Anotheradvantageofthisregionisthebroadabsorptionbandpresented by the analyzed samples that compensates the broadNIR LED emission band ( ± 50nm). Closer wavelengths as that Fig. 1. NIR spectra of 10 samples of solvent (dashed line), 35 of gasoline (soliddark line) and 10 samples of kerosene (solid light line). ones around 1200, 1400 and 1650nm could not produce a linearanalytical signal due to the NIR LED bandwidth limitation.Therefore, a NIR LED (L1550-35k42, Epitex incorp.) withan InAsGaP substrate and 1550nm emission wavelength wasused as radiation source in the NLP.For detection of the NIR LED radiation, a PbSe photore-sistor (PR2-27-20-320, RMT Ltd.) with 2mm sensitivity areaand detection limit at 2700nm was used. The detector pack-age involves three devices: the PbSe photoresistor chip thatgenerates the analytical signal under infrared radiation, a NTCthermistor for temperature control and a thermoelectric coolerfor refrigeration of the photoresistor.The NIR LED and detector were adapted 3cm apart on twoPerspex ® fixedbarsanda2cmquartzcellwasfittedonasupportbetween them. The support presents a central hole with 1cmdiameter which allows radiation to pass throughout the cell andto reach the detector.Fig. 2 shows details of the components of the NLP.All the electronic components were adapted into a box of 20cm × 15cm × 10cm and the optics is adapted into the cellcompartment upon this box.Fig. 3 shows the block diagram of the NLP. The instrumentusesascontrolunitaPIC16F877microcontrollerprogrammablewith C language and driven by a stabilized power supply orby 9V batteries. The microcontroller presents a 10-bit multi-channel analog to digital converter (A/D), 8k  × 14 words of flash program memory, 256 × 8 bytes of EEPROM data mem-ory, 368 × 8 bytes of RAM, three timers and three ports witheight channels (multiplexed pins) each, for general purpose,that can be configured for data acquisition. The microcontrollerallows drastic reduction of the number of electronic compo-nentsonthephotometerandmakespossibletointroduceoralterfunctionalities in the instrument without any hardware modifi-cation.The NIR LED control module pulses the NIR LED radiationat 50Hz. The module turns the NIR LED on for readings of the analytical signals and off for determination of instrumental  794  E.d.N. Gai˜ ao et al. / Talanta 75 (2008) 792–796  Fig. 2. Details of the NLP. noise. These signals are detected, filtered and amplified in thedetection module of the NLP. After being processed, both sig-nalsaretransmittedtotheA/Dconverter,selectingtherespectivechannel, to generate digital analytical signals that are convertedinto absorbance data. After digitalization the absorbance dataare sent by the microcontroller for the LCD module whichuses a WH-1602A character type LCD with an eight bits dataand a three bits control buses. In this work, the communica-tion between microcontroller and LCD was always carried outwith seven bits, four for data transmission and three for datacontrol.Thetemperaturecontrolmodulecontrolsthedetectortemper-ature. Temperature signals furnished by a NTC thermistor with Fig. 3. Block diagram of the NLP. a temperature coefficient of  − 3.4%/deg (20 ◦ C) located into thedetectorpackage,enablesthemicrocontrollertochangetheelec-tric current on the thermoelectric cooler to keep the operationtemperature of the detector on a fixed value (5 ◦ C).Aclockelaboratedwitha4MHzpiezoelectriccrystalisusedfor synchronization of all microcontroller internal functions. 2.4. Operation of the NLP The measurements with the NLP are performed in a sim-ple way. The quartz cell is filled with the sample and adjustedon the cell support. The cell compartment is closed and thetransmittance appears automatically on the LCD screen. Trans-mittance measurements of pure and adulterated samples of Cgasoline were carried out to define a reliability region that indi-cates the unconformity of the gasoline type C with the qualityrequirements. 2.5. Analytical procedure for the NLP calibration Twenty-five samples of gasolines were used to register theblank signal (b) and its standard deviation (s) on the devel-oped photometer which yielded the limits of detection, LD(blank plus three times its standard deviation, LD= b ± 3 s ), andquantification, LQ (blank plus 10 times its standard deviation,LQ= b ± 10 s ). In these measurements, 100% of transmittance(zero of absorbance) was determined when the cell was emptyand the zero of transmittance was determined when the LEDwas off.The reliability of the NLP analysis was verified by per-formance curves [25]. The parameter used to establish the cut-off adulterant concentration and the unreliability regionwas the limit of quantification considering samples of puregasoline as blank. It is an appropriated parameter due to its nar-rower unreliability region in regard to the detection limit whatmeans higher precision on the screening classification. A linearregression was carried out with the points around 50% of cor-rect response in the performance curves to elaborate straightlines used to define the concentrations at which is possibleto achieve a confidence level of 95% in the screening analy-sis. 3. Results and discussion The NLP was applied on the analysis of 25 samples of puregasolines (blank) in order to determinate the standard devia-tions of measurements. The mean and standard deviation forthis sample set were 0.498 ± 0.003.The adulterated gasolines (see Section 2.1) were prepared with solvent and kerosene whose spectral behaviors were closerto that from pure gasolines (Fig. 1) in such a way to difficult the adulteration detection. Fig. 4 shows the analytical curvesfor both adulterations. Each point on the plot is an average of 10 measurements from 10 samples. Based on the  s  value forthe blank, it was found values of 1.2 and 1.4% for kerosene andsolventLDs,respectively.Samplesrepresentingtheadulterationlevel of 1% (v/v) were not analyzed.   E.d.N. Gai˜ ao et al. / Talanta 75 (2008) 792–796   795Fig. 4. Analytical curves obtained when pure and adulterated gasolines wereanalyzed by the NLP. The analytical curves obtained for solvent and kerosene aregiven by the following equations: A solvent = 0 . 00850 C + 0 . 495 ( r = 0 . 9815) (1) A kerosene =− 0 . 00650 C + 0 . 496 ( r = 0 . 9890) (2)where  A  is the absorbance and  C   is the concentration of theadulterant.The LQs for solvent and kerosene was found to be 4.2and 4.8% (v/v), respectively. These values were used as cut-off concentrations to elaborate performance curves in order todetermine an unreliability region related to the verification of gasolineadulterationbysolventorkerosene.Intheconstructionofthesecurvesacorrectresponseisapositiveresponseforadul-teration and a correct negative response is positive response forno adulteration.Fig. 5 shows the performance curves obtained when gasolinesamples having different adulteration levels were analyzed bythe NLP using LQ as cut-off concentration. The unreliabilityregions on these graphs were calculated by regression curveselaborated with the points around 50% of correct response foradulteration (Fig. 5). Thus, at 95% of confidence level, adulter- ation by solvent and kerosene can only be confirmed (correctpositive) if their concentrations are at least 5.1% (v/v) (Fig. 5a) and 5.9% (v/v) (Fig. 5b), respectively. At this confidence level, Fig. 6. Results for gasoline verification of no adulteration by solvent. A correctnegative is a confirmation for no adulteration with a confidence level of 95%.Legend: concentration in percentage (v/v). is possible to find 5% of false negative responses. On the otherhand, a confirmation for adulteration absence (correct negative)is established when the concentration levels are below 2.7 and4.0% (v/v), for solvent and kerosene, at a confidence level of 95% (5% of false positive responses).It is interesting to analyze that if false negative responsesappear in a screening procedure elaborated using the correctpositive region of the performance curves in Fig. 5, adulterated gasolines can be classified as good ones. Therefore, it is not agood criterion to analyze adulteration. On the other hand, if theprocedure is elaborated to verify absence of adulteration usingthe correct negative region, false positive responses will directthetestedsamplestoadeeperanalysistoconfirmornottheinitialsuspicion. In spit of the confirmation for “no adulteration” canonlybecarriedoutforconcentrationsbeyond2.7and4.0%(v/v)for solvent and kerosene, respectively, these concentrations arebetter thresholds for verification of gasolines quality than thecorrect positive thresholds. Thus, the NLP was calibrated usingthe LQs of solvent and kerosene as cut-off concentrations andthe correct negative region given by the performance curves of Fig. 5 to confirm “no adulteration” on the gasolines type C.Fig. 6 shows the results obtained when the NLP was appliedon the quality control of eight gasolines samples where solventwas spiked at the levels 0.0, 2.0, 3.0, 4.0 and 5.0% (v/v). The Fig. 5. Performance curves elaborated for evaluation of the unreliability region for the screening analysis of adulterants in gasoline: (a) adulteration by solvent and(b) adulteration by kerosene. Equations are the regression curves used to define thresholds for correct positive and negative response regions at 95% of confidencelevel.  796  E.d.N. Gai˜ ao et al. / Talanta 75 (2008) 792–796  Fig.7. Resultsforgasolineverificationofnoadulterationbykerosene.Acorrectnegative is a confirmation for no adulteration with a confidence level of 95%.Legend: concentration in percentage (v/v). cut-off was 4.2% (v/v). Thus, when the concentration level wasat 3.0% (v/v) or below it, all samples were classified as notadulterated. A certain level of false positive responses at 3.0%(v/v) was expected but it did not appear on these assays. At thelevelof4.0%(v/v),halfoftheadulteratedsampleswasclassifiedas not adulterated, generating 50% of false positive results, asexpected. At 5.0% (v/v) of solvent, all analyzed samples wereclassified as adulterated.Fig. 7 shows the results obtained by eight assays carriedout using gasolines where the kerosene concentrations were atthe levels 0.0, 3.0, 4.0, 5.0 and 6.0% (v/v). The cut-off wasestablishedat4.8%(v/v).Samplescontainingkeroseneatlevelsbelow 3.0% (v/v) were classified as not adulterated. At levelsof 4.0 and 5.0% (v/v), 88 and 50% of the samples, respectively,were classified as not adulterated by the NLP. At 6.0% (v/v) allsamples are classified as adulterated.Asitcanbeseenfromtheresults,“noadulteration”responsesfor the gasoline screening analysis were obtained when the con-centration of solvent and kerosene were below 3.0 and 5.0%,respectively. If the levels of adulteration surpass these limits,samples can be classified as adulterated and must be directed toa deeper analytical procedure.Itisimportanttoconsidertheadulterationcausedbyadditionof both solvent and kerosene. In this situation a determinationof the adulteration could only be possible if the concentration of oneoftheadulterantswaslargerthentheother.Incasesatwhichthe adulteration is caused by a close amount of both adulterantsthe NLP will fail on its screening analysis. A way to overcomethis drawback is being studied. 4. Conclusion A portable and microcontrolled NIR LED photometer wasdeveloped for screening analysis of gasoline type C aiming atto verify adulteration by solvent and kerosene. The NLP usesa NIR LED emitting at 1550nm and a PbSe photoresistor asdetector whose noise is kept at low levels by a control systembased on thermistor and thermoelectric cooling. To calibrate itthe analytical signal of pure and adulterated samples of gasolinetype C were registered by the photometer and cut-off concen-trations for solvent and kerosene were calculated as a thresholdto classify the sample as adulterated or not. The correct neg-ative region yielded by the performance curves related to the“no adulteration” response was used to calibrate the NLP forthe screening analysis of the gasolines. At a confidence level of 95%,samplesofsolventandkerosenewereconsiderednotadul-terated when the levels of adulterant were below 3.0 and 5.0%(v/v), respectively. Thus, the NLP showed to be an economicaland viable alternative for screening analysis of gasoline type Cadulterated by solvent or kerosene. Acknowledgments The authors gratefully acknowledge the support by CNPq,Brazil (Proc. No.: 478961/2001–4) and FINEP (Proc. No.:478961/2001–4) and CAPES and CNPq scholarship, respec-tively. References [1] P.K. Dasgupta, H.S. Bellamy, H. Liu, J.L. Lopes, E.L. Loree, K.J. Morris,K. Petersen, K.A. Mir, Talanta 40 (1993) 53.[2] P.K. Dasgupta, I.Y. Eom, K.J. Morris, J. Li, Anal. Chim. Acta 500 (2003)337.[3] A.M. Tan, J.L. Huang, J.D. Geng, J.H. Xu, X.N. Zhao, J. Autom. Chem.16 (1994) 71.[4] M.K. Cantrell, J.D. Ingle, Anal. Chem. 75 (2003) 27.[5] P.C. Hauser, T.W.T. Rupasinghe, N.E. Cates, Talanta 42 (1995) 605.[6] J. Malinen, M. K¨ans¨akoski, R. Rikola, C.G. Eddison, Sens. Actuators B 51(1998) 220.[7] H. Flaschka, C. McKeithan, R.M. Barnes, Anal. Lett. 6 (1973) 585.[8] D. Betteridge, E.L. Dagless, B. Fields, N.F. Graves, Analyst 103 (1978)897.[9] D. Betteridge, Anal. Chem. 50 (1978) 832A.[10] C. Pasquini, I.M. Raimundo Jr., Quim. Nova 7 (1984) 24.[11] P.J. Worsfold, J.R. Clinch, Anal. Chim. Acta 197 (1987) 43.[12] J. Huang, H. Liu, A. Tan, J. Xu, X. Zhao, Talanta 39 (1992) 589.[13] H. Liu, P.K. Dasgupta, Anal. Chim. Acta 289 (1994) 347.[14] E.N. Gai˜ao, R.S. Honorato, S.R.B. Santos, M.C.U. Ara´ujo, Analyst 124(1999) 1727.[15] M.C.U. Araujo, S.R.B. Santos, E.A. Silva, G. Veras, J.L.F.C. Lima, R.A.S.Lapa, Quim. Nova 20 (1997) 137.[16] E.A. Pereira, A.A. Cardoso, P.K. Dasgupta, Quim. Nova 24 (2001) 443.[17] F.R.P. Rocha, P.B. Martelli, B.F. Reis, J. Braz, Chem. Soc. 15 (2004) 38.[18] M. Cocchi, C. Duarte, G. Foca, A. Marchetti, L. Tassi, A. Ulrici, Talanta68 (2006) 1505.[19] M. Blanco, J. Pag`es, Anal. Chim. Acta 463 (2002) 295.[20] E.L. Bergman, H. Brage, M. Josefson, O. Svensson, A. Spar´en, J. Pharm.Biomed. Anal. 41 (2006) 89.[21] F.S. Falla, C. Larini, G.A.C. Le Roux, F.H. Quina, L.F.L. Moro, C.A.O.Nascimento, J. Petrol. Sci. Eng. 51 (2006) 127.[22] C.C. Fel´ıcio, L.P. Br´as, J.A. Lopes, L.L. Cabrita, J.C. Menezes, Chem.Intell. Lab. Syst. 78 (2005) 74.[23] J.V. Stanfford, G.S. Weaving, J.C. Lowe, J. Agric. Eng. Res. 43 (1989) 45.[24] J.V. Stanfford, C.R. Bull, G.S. Weaving, J. Agric. Eng. Res. 43 (1989) 57.[25] C. Gonzalez, E. Prichard, S. Spinelli, J. Gille, E. Touraud, Trends Anal.Chem. 4 (2007) 315.
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks